Discussion:
Yet another GPSDO - locking to 10MHz
Add Reply
Murray Greenman
2010-06-27 19:14:22 UTC
Reply
Permalink
Raw Message
I have a design which locks a high performance 10MHz OCXO to a 10MHz
source which should work equally well with the LEA5, or any source of 5
or 10MHz.

I designed it for use with a distributed factory GPS reference which has
picked up noise, hum and phase modulation, in order to deliver a high
quality but GPS locked reference direct to equipment. The design adds
nearly three orders of improvement.

Essentially it divides the incoming 10MHz by 16384 and compared the
phase with a similar division from the OCXO, within an ATTiny2313 micro.
The phase detector is a D-flip-flop type implemented in software (in
interrupts), and it delivers a locked reference with ADev around 10e-12
for Tau between 1s and 20s. The micro also keeps a real time clock and
does various background monitoring and telemetry tasks. There is PC
monitoring software as well. There are only four chips in the design.
While I can't share the code (belongs to my employer), the idea is
simple enough and I could share the schematic.

Eight of these units have been built. I used the excellent Rakon
STP2402E OCXO.


Regards,
Murray Greenman ZL1BPU
Robert Benward
2010-06-28 00:59:51 UTC
Reply
Permalink
Raw Message
All this talk about interpolation reminds me of a little neat chip by Analog
Devices, AD9500. It's programmable digital delay, bit, with lops resolution
with a loons full-scale range. I believe (from app notes) you can push it
to 100us FS, giving 390ns resolution. I think the minimum jitter at the
lowest FS was about lops. The AD9500 is cell and the 9501 is TTS.

http://pdf1.alldatasheet.com/datasheet-pdf/view/48545/AD/AD9500.html

app note

http://www.analog.com/static/imported-files/application_notes/105895411AN-260.pdf

Bob

----- Original Message -----
From: Murray Greenman
To: time-nuts-***@public.gmane.org
Sent: Sunday, June 27, 2010 3:14 PM
Subject: [time-nuts] Yet another GPSDO - locking to 10MHz


I have a design which locks a high performance 10MHz OCXO to a 10MHz
source which should work equally well with the LEA5, or any source of 5
or 10MHz.

I designed it for use with a distributed factory GPS reference which has
picked up noise, hum and phase modulation, in order to deliver a high
quality but GPS locked reference direct to equipment. The design adds
nearly three orders of improvement.

Essentially it divides the incoming 10MHz by 16384 and compared the
phase with a similar division from the OCXO, within an ATTiny2313 micro.
The phase detector is a D-flip-flop type implemented in software (in
interrupts), and it delivers a locked reference with ADev around 10e-12
for Tau between 1s and 20s. The micro also keeps a real time clock and
does various background monitoring and telemetry tasks. There is PC
monitoring software as well. There are only four chips in the design.
While I can't share the code (belongs to my employer), the idea is
simple enough and I could share the schematic.

Eight of these units have been built. I used the excellent Rakon
STP2402E OCXO.


Regards,
Murray Greenman ZL1BPU


_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Attila Kinali
2010-06-29 09:26:05 UTC
Reply
Permalink
Raw Message
On Sun, 27 Jun 2010 20:59:51 -0400
Post by Robert Benward
All this talk about interpolation reminds me of a little neat chip by Analog
Devices, AD9500. It's programmable digital delay, bit, with lops resolution
with a loons full-scale range. I believe (from app notes) you can push it
to 100us FS, giving 390ns resolution. I think the minimum jitter at the
lowest FS was about lops. The AD9500 is cell and the 9501 is TTS.
Unfortunately, the AD9500 line is obsolete with no replacement.
Which means it will be quite soon not available anymore.

Attila Kinali
--
If you want to walk fast, walk alone.
If you want to walk far, walk together.
-- African proverb
Hal Murray
2010-06-29 10:15:30 UTC
Reply
Permalink
Raw Message
Unfortunately, the AD9500 line is obsolete with no replacement. Which means
it will be quite soon not available anymore.
On-Semi makes a programmable delay: MC100EP195, 2.2 to 12.2 ns in 10 ps steps

There are a couple of other similar chips.
http://www.onsemi.com/PowerSolutions/parametrics.do?id=586
--
These are my opinions, not necessarily my employer's. I hate spam.
Peter Vince
2010-07-04 12:41:05 UTC
Reply
Permalink
Raw Message
Hi Murray,

Why did you use such a large division factor - or any for that
matter? Could you not just have used a PLL with very long time
constant running at 10 MHz?

Peter
Post by Murray Greenman
I have a design which locks a high performance 10MHz OCXO to a 10MHz
source which should work equally well with the LEA5, or any source of 5
or 10MHz.
I designed it for use with a distributed factory GPS reference which has
picked up noise, hum and phase modulation, in order to deliver a high
quality but GPS locked reference direct to equipment. The design adds
nearly three orders of improvement.
Essentially it divides the incoming 10MHz by 16384 and compared the
phase with a similar division from the OCXO, within an ATTiny2313 micro.
The phase detector is a D-flip-flop type implemented in software (in
interrupts), and it delivers a locked reference with ADev around 10e-12
for Tau between 1s and 20s. The micro also keeps a real time clock and
does various background monitoring and telemetry tasks. There is PC
monitoring software as well. There are only four chips in the design.
While I can't share the code (belongs to my employer), the idea is
simple enough and I could share the schematic.
Eight of these units have been built. I used the excellent Rakon
STP2402E OCXO.
Regards,
Murray Greenman ZL1BPU
Hal Murray
2012-01-23 19:39:07 UTC
Reply
Permalink
Raw Message
Not to mention the solar noon varies by +/- 15min over the year (don't ask
me how this is called...old knowledge from my high school days)
Analemma: It's the figure 8 you see on globes in the middle of the Pacific
Ocean.

It's caused by the Earth's orbit not being circular and the axis of rotation
not being perpendicular to the orbit plane.

http://en.wikipedia.org/wiki/Analemma
--
These are my opinions, not necessarily my employer's. I hate spam.
Chris Albertson
2012-01-23 19:57:49 UTC
Reply
Permalink
Raw Message
The Earth's rotation is always refferenced to a larger reference frame
of distant stars, not the sun. If you try and use the Sun, I think
there are smaller effects like the Earth and Moon revolve around a
common center and then there is Jupiter. So they use stars.
The next question is "which stars" and I think measurements are good
enough that the question of "Which Stars?" matters. Makes sense when
you remember that a nano second is about a foot and all stars are
moving quite fast relative to each other.
Post by Hal Murray
Not to mention the solar noon varies by +/- 15min over the year (don't ask
me how this is called...old knowledge from my high school days)
Analemma: It's the figure 8 you see on globes in the middle of the Pacific
Ocean.
It's caused by the Earth's orbit not being circular and the axis of rotation
not being perpendicular to the orbit plane.
http://en.wikipedia.org/wiki/Analemma
--
These are my opinions, not necessarily my employer's.  I hate spam.
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
--
Chris Albertson
Redondo Beach, California
Poul-Henning Kamp
2012-01-23 20:01:27 UTC
Reply
Permalink
Raw Message
Post by Chris Albertson
The Earth's rotation is always refferenced to a larger reference frame
of distant stars, not the sun.
It's actually distant quarsars, and the point being that they are so
far away that any cross-field motion they might or might not have
would not represent a parallax error in our measurements of their
posisiton.

The measurements are done with VLBI, which is a good google search
to learn more about this.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk-***@public.gmane.org | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
J. L. Trantham
2012-01-24 01:50:50 UTC
Reply
Permalink
Raw Message
Is this, in any way, related to the fact the Earth has a Moon?

Ideally, the Earth rotates around the Sun and the Moon rotates around the
Earth. However, is it better described as the 'center of mass' of the
Earth/Moon rotates around the Sun? Not to mention all those other
confounding variables such as other planets and moons?

Also, getting back to the placement of the 'first pps', is there a way to
'sync' the 1 PPS output of an FE-5680A to an external signal, such as a GPS
receiver or TBolt? I would think that might be possible given their
original purpose.

Thanks in advance.

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Hal Murray
Sent: Monday, January 23, 2012 1:39 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Determination of the placement of the first pps
Not to mention the solar noon varies by +/- 15min over the year (don't ask
me how this is called...old knowledge from my high school days)
Analemma: It's the figure 8 you see on globes in the middle of the Pacific
Ocean.

It's caused by the Earth's orbit not being circular and the axis of rotation

not being perpendicular to the orbit plane.

http://en.wikipedia.org/wiki/Analemma
--
These are my opinions, not necessarily my employer's. I hate spam.




_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Jim Lux
2012-01-24 06:06:20 UTC
Reply
Permalink
Raw Message
Post by J. L. Trantham
Is this, in any way, related to the fact the Earth has a Moon?
Ideally, the Earth rotates around the Sun and the Moon rotates around the
Earth. However, is it better described as the 'center of mass' of the
Earth/Moon
barycenter is the term

rotates around the Sun? Not to mention all those other
Post by J. L. Trantham
confounding variables such as other planets and moons?
Well, yes.. there are fairly sophisticated numerical integration
programs that figure all this stuff out, if you're interested.
J. L. Trantham
2012-01-24 11:49:07 UTC
Reply
Permalink
Raw Message
Thanks.

As I was re-reading the postings, I noted additional information in an
earlier post that pretty much answered the question though the term
'barycenter' is new to me.

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Jim Lux
Sent: Tuesday, January 24, 2012 12:06 AM
To: time-nuts-***@public.gmane.org
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Is this, in any way, related to the fact the Earth has a Moon?
Ideally, the Earth rotates around the Sun and the Moon rotates around
the Earth. However, is it better described as the 'center of mass' of
the Earth/Moon
barycenter is the term

rotates around the Sun? Not to mention all those other
Post by J. L. Trantham
confounding variables such as other planets and moons?
Well, yes.. there are fairly sophisticated numerical integration
programs that figure all this stuff out, if you're interested.
_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Chris Albertson
2012-01-24 06:34:27 UTC
Reply
Permalink
Raw Message
Post by J. L. Trantham
Is this, in any way, related to the fact the Earth has a Moon?
is there a way to
'sync' the 1 PPS output of an FE-5680A to an external signal, such as a GPS
receiver or TBolt?  I would think that might be possible given their
original purpose.
The 1PPS is not in the units specs. It is just by luck that it works.
However we could adjust the phase of the 1PPS by running the unit
fast or slow for some period of time and then going back to exact
10MHz. But that method could take a LONG time, like tens of thousands
of seconds. Better I think to test the phase and if it is "off" by
more than say, 0.01 second to just power the unit off and restart and
see what luck gives you. I bet 100 power cycles is faster than
moving the phase by 0.5 seconds.

Maybe the answer is to wire up a few decade divers and divide the
10MHz to 1pps yourself. Thenyou can let a known good PPS reset your
counters and get the phase correct instantly
--
Chris Albertson
Redondo Beach, California
J. L. Trantham
2012-01-24 11:42:36 UTC
Reply
Permalink
Raw Message
Thanks Chris.

It seems such a logical feature to have, I would think it would have been
included perhaps by a serial command. My old CS clocks have this feature
though I have never taken the time to sync them.

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Chris Albertson
Sent: Tuesday, January 24, 2012 12:34 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Is this, in any way, related to the fact the Earth has a Moon?
is there a way to
'sync' the 1 PPS output of an FE-5680A to an external signal, such as
a GPS receiver or TBolt?  I would think that might be possible given
their original purpose.
The 1PPS is not in the units specs. It is just by luck that it works.
However we could adjust the phase of the 1PPS by running the unit fast or
slow for some period of time and then going back to exact 10MHz. But that
method could take a LONG time, like tens of thousands
of seconds. Better I think to test the phase and if it is "off" by
more than say, 0.01 second to just power the unit off and restart and
see what luck gives you. I bet 100 power cycles is faster than
moving the phase by 0.5 seconds.

Maybe the answer is to wire up a few decade divers and divide the 10MHz to
1pps yourself. Thenyou can let a known good PPS reset your counters and get
the phase correct instantly
--
Chris Albertson
Redondo Beach, California

_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus Danielson
2012-01-24 18:50:41 UTC
Reply
Permalink
Raw Message
Post by J. L. Trantham
Thanks Chris.
It seems such a logical feature to have, I would think it would have been
included perhaps by a serial command. My old CS clocks have this feature
though I have never taken the time to sync them.
"Jumping" the PPS into about the right phase is done within a second and
is well worth the effort. I use this myself and it works well.

Forcing "sync" on atomic clocks is badly needed. The frequency steering
range is small so frequency limit sideways would take ages.

Cheers,
Magnus
Post by J. L. Trantham
Joe
-----Original Message-----
Behalf Of Chris Albertson
Sent: Tuesday, January 24, 2012 12:34 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Is this, in any way, related to the fact the Earth has a Moon?
is there a way to
'sync' the 1 PPS output of an FE-5680A to an external signal, such as
a GPS receiver or TBolt? I would think that might be possible given
their original purpose.
The 1PPS is not in the units specs. It is just by luck that it works.
However we could adjust the phase of the 1PPS by running the unit fast or
slow for some period of time and then going back to exact 10MHz. But that
method could take a LONG time, like tens of thousands
of seconds. Better I think to test the phase and if it is "off" by
more than say, 0.01 second to just power the unit off and restart and
see what luck gives you. I bet 100 power cycles is faster than
moving the phase by 0.5 seconds.
Maybe the answer is to wire up a few decade divers and divide the 10MHz to
1pps yourself. Thenyou can let a known good PPS reset your counters and get
the phase correct instantly
J. L. Trantham
2012-01-24 19:07:01 UTC
Reply
Permalink
Raw Message
Magnus,

How did you 'jump' the PPS on the FE-5680A? Is it a serial command? How do
you 'sync' it to the external PPS from say a TBolt?

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Magnus Danielson
Sent: Tuesday, January 24, 2012 12:51 PM
To: time-nuts-***@public.gmane.org
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Thanks Chris.
It seems such a logical feature to have, I would think it would have
been included perhaps by a serial command. My old CS clocks have this
feature though I have never taken the time to sync them.
"Jumping" the PPS into about the right phase is done within a second and is
well worth the effort. I use this myself and it works well.

Forcing "sync" on atomic clocks is badly needed. The frequency steering
range is small so frequency limit sideways would take ages.

Cheers,
Magnus
Post by J. L. Trantham
Joe
-----Original Message-----
On Behalf Of Chris Albertson
Sent: Tuesday, January 24, 2012 12:34 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Is this, in any way, related to the fact the Earth has a Moon?
is there a way to
'sync' the 1 PPS output of an FE-5680A to an external signal, such as
a GPS receiver or TBolt? I would think that might be possible given
their original purpose.
The 1PPS is not in the units specs. It is just by luck that it works.
However we could adjust the phase of the 1PPS by running the unit
fast or slow for some period of time and then going back to exact
10MHz. But that method could take a LONG time, like tens of thousands
of seconds. Better I think to test the phase and if it is "off" by
more than say, 0.01 second to just power the unit off and restart and
see what luck gives you. I bet 100 power cycles is faster than
moving the phase by 0.5 seconds.
Maybe the answer is to wire up a few decade divers and divide the
10MHz to 1pps yourself. Thenyou can let a known good PPS reset your
counters and get the phase correct instantly
_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus Danielson
2012-01-24 19:18:44 UTC
Reply
Permalink
Raw Message
Post by J. L. Trantham
Magnus,
How did you 'jump' the PPS on the FE-5680A? Is it a serial command? How do
you 'sync' it to the external PPS from say a TBolt?
I was not talking about the 5680 specific. Sorry for the unclarity.

Cheers,
Magnus
J. L. Trantham
2012-01-25 01:41:51 UTC
Reply
Permalink
Raw Message
Darn.

I was hoping for that feature. I still think it should be there.

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Magnus Danielson
Sent: Tuesday, January 24, 2012 1:19 PM
To: time-nuts-***@public.gmane.org
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Magnus,
How did you 'jump' the PPS on the FE-5680A? Is it a serial command? How do
you 'sync' it to the external PPS from say a TBolt?
I was not talking about the 5680 specific. Sorry for the unclarity.

Cheers,
Magnus

_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus Danielson
2012-01-25 03:14:04 UTC
Reply
Permalink
Raw Message
Post by J. L. Trantham
Darn.
I was hoping for that feature. I still think it should be there.
Indeed. Should be in there somewhere...

Didn't see these link hit the list:
http://pastebin.com/S8UcnCMZ

http://www.dd1us.de/Downloads/precise%20reference%20frequency%20rev%200_7.pdf

http://vk2xv.djirra.com/tech_rubidium.htm

Looking at the last one it says:

"NOTE: Although this unit is marked with both 10MHz and 1pps, research
on the 'net seems to indicate that the '1pps' output has only a period
of exactly 1 second when the frequency is set to 223 Hz (8.388608Mhz).
According to those sources the '1pps' will have a period of 0.8388608
seconds when the output frequency is set to 10MHz. This should be easy
to verify. In any case I have no need for a 1pps output - I use a GPS
module to get a 1pps signal which also has the advantage of being
in-phase with real time seconds. "

Now... to speed-adjust the PPS phase, use the DDS and steer it
intentionally of frequence with sufficient delta frequence for suitable
time and you should home in pretty quickly.

Hunting some more:
http://www.qsl.net/zl1bpu/PROJ/Ruby.htm

"Without modification the units have just one output - 1pps (1 Hz).
There is a simple modification to extract 8388.608 kHz, which is of
course 223 Hz, and this frequency is used, through binary division, to
generate the 1pps output. The 8388.608 kHz output is generated by a
32-bit Direct Digital Synthesizer chip (AD9830). This output can be
steered to any other frequency within the operating range, by
interacting with the controlling microcontroller, with three provisos:

1. The unit has a peaked filter at the synthesizer output, and so the
level at other frequencies varies wildly. This can be corrected with
minor modifications.
2. When operating at any other frequency than 8388.608kHz, the 1pps
output is of course incorrect.
3. The synthesizer operating frequency can be set to within ±5 mHz
(milliHertz) of the requested frequency,
- but ONLY if the calculations, on which the command sent to it
is based, correctly use 32-bit maths. "

Which is a confirmation...

http://www.leapsecond.com/museum/fei5650a/

http://www.ko4bb.com/dokuwiki/doku.php?id=precision_timing:rubidium_oscillators

So, this would work for the 50,255+ MHz based FE 5680A. For 60 MHz
variants it works a little different, but it has two MCUs sitting there,
so some use for theme should there be.

Cheers,
Magnus
beale
2012-01-25 03:49:40 UTC
Reply
Permalink
Raw Message
Yikes, please delete that pastebin link. That was just my initial try at a FAQ. A significantly updated and more current version of the FE-5680A FAQ is located at http://www.ko4bb.com/dokuwiki/doku.php?id=precision_timing:fe5860a_faq

Scott's utility seems to be able to dump the FEI unit serial number, and perhaps the included options. If my guess is correct, all three of the units I have are equipped only with "Option 2" which is the RS-232 port. But there are many possible varieties of 5680 units out there, and some of them may have the 1 PPS behavior you mention while others do not- I have no information on that.
-------Original Message-------
http://pastebin.com/S8UcnCMZ
http://www.dd1us.de/Downloads/precise%20reference%20frequency%20rev%200_7.pdf
http://vk2xv.djirra.com/tech_rubidium.htm
Bob Camp
2012-01-25 17:42:31 UTC
Reply
Permalink
Raw Message
Hi

Based on the number of units that come in with the "1 pps output" missing,
I'd bet the $40 FE5680's were used only for the 10 MHz output.

Bob

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Magnus Danielson
Sent: Tuesday, January 24, 2012 10:14 PM
To: time-nuts-***@public.gmane.org
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Darn.
I was hoping for that feature. I still think it should be there.
Indeed. Should be in there somewhere...

Didn't see these link hit the list:
http://pastebin.com/S8UcnCMZ

http://www.dd1us.de/Downloads/precise%20reference%20frequency%20rev%200_7.pd
f

http://vk2xv.djirra.com/tech_rubidium.htm

Looking at the last one it says:

"NOTE: Although this unit is marked with both 10MHz and 1pps, research
on the 'net seems to indicate that the '1pps' output has only a period
of exactly 1 second when the frequency is set to 223 Hz (8.388608Mhz).
According to those sources the '1pps' will have a period of 0.8388608
seconds when the output frequency is set to 10MHz. This should be easy
to verify. In any case I have no need for a 1pps output - I use a GPS
module to get a 1pps signal which also has the advantage of being
in-phase with real time seconds. "

Now... to speed-adjust the PPS phase, use the DDS and steer it
intentionally of frequence with sufficient delta frequence for suitable
time and you should home in pretty quickly.

Hunting some more:
http://www.qsl.net/zl1bpu/PROJ/Ruby.htm

"Without modification the units have just one output - 1pps (1 Hz).
There is a simple modification to extract 8388.608 kHz, which is of
course 223 Hz, and this frequency is used, through binary division, to
generate the 1pps output. The 8388.608 kHz output is generated by a
32-bit Direct Digital Synthesizer chip (AD9830). This output can be
steered to any other frequency within the operating range, by
interacting with the controlling microcontroller, with three provisos:

1. The unit has a peaked filter at the synthesizer output, and so the
level at other frequencies varies wildly. This can be corrected with
minor modifications.
2. When operating at any other frequency than 8388.608kHz, the 1pps
output is of course incorrect.
3. The synthesizer operating frequency can be set to within ±5 mHz
(milliHertz) of the requested frequency,
- but ONLY if the calculations, on which the command sent to it
is based, correctly use 32-bit maths. "

Which is a confirmation...

http://www.leapsecond.com/museum/fei5650a/

http://www.ko4bb.com/dokuwiki/doku.php?id=precision_timing:rubidium_oscillat
ors

So, this would work for the 50,255+ MHz based FE 5680A. For 60 MHz
variants it works a little different, but it has two MCUs sitting there,
so some use for theme should there be.

Cheers,
Magnus

_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
J. L. Trantham
2012-01-25 19:04:50 UTC
Reply
Permalink
Raw Message
Magnus,

Thanks for the research. Now that I have read your information, I am going
to have to go back and 'measure' the PPS output on my unit. I just
'eyeballed' the '1 PPS' and it seemed close to 1 second.

It would be nice to find a serial command that would allow placing the 1 PPS
at any arbitrary point in phase to match GPS.

My impression of reading the posts about the FE-5680A is that the 1 PPS is
likely there, just not easily documented.

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Bob Camp
Sent: Wednesday, January 25, 2012 11:43 AM
To: 'Discussion of precise time and frequency measurement'
Subject: Re: [time-nuts] Determination of the placement of the first pps

Hi

Based on the number of units that come in with the "1 pps output" missing,
I'd bet the $40 FE5680's were used only for the 10 MHz output.

Bob

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Magnus Danielson
Sent: Tuesday, January 24, 2012 10:14 PM
To: time-nuts-***@public.gmane.org
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Darn.
I was hoping for that feature. I still think it should be there.
Indeed. Should be in there somewhere...

Didn't see these link hit the list:
http://pastebin.com/S8UcnCMZ

http://www.dd1us.de/Downloads/precise%20reference%20frequency%20rev%200_7.pd
f

http://vk2xv.djirra.com/tech_rubidium.htm

Looking at the last one it says:

"NOTE: Although this unit is marked with both 10MHz and 1pps, research on
the 'net seems to indicate that the '1pps' output has only a period of
exactly 1 second when the frequency is set to 223 Hz (8.388608Mhz).
According to those sources the '1pps' will have a period of 0.8388608
seconds when the output frequency is set to 10MHz. This should be easy to
verify. In any case I have no need for a 1pps output - I use a GPS module to
get a 1pps signal which also has the advantage of being in-phase with real
time seconds. "

Now... to speed-adjust the PPS phase, use the DDS and steer it intentionally
of frequence with sufficient delta frequence for suitable time and you
should home in pretty quickly.

Hunting some more:
http://www.qsl.net/zl1bpu/PROJ/Ruby.htm

"Without modification the units have just one output - 1pps (1 Hz).
There is a simple modification to extract 8388.608 kHz, which is of course
223 Hz, and this frequency is used, through binary division, to generate the
1pps output. The 8388.608 kHz output is generated by a 32-bit Direct Digital
Synthesizer chip (AD9830). This output can be steered to any other frequency
within the operating range, by interacting with the controlling
microcontroller, with three provisos:

1. The unit has a peaked filter at the synthesizer output, and so the level
at other frequencies varies wildly. This can be corrected with minor
modifications.
2. When operating at any other frequency than 8388.608kHz, the 1pps output
is of course incorrect.
3. The synthesizer operating frequency can be set to within ±5 mHz
(milliHertz) of the requested frequency,
- but ONLY if the calculations, on which the command sent to it is
based, correctly use 32-bit maths. "

Which is a confirmation...

http://www.leapsecond.com/museum/fei5650a/

http://www.ko4bb.com/dokuwiki/doku.php?id=precision_timing:rubidium_oscillat
ors

So, this would work for the 50,255+ MHz based FE 5680A. For 60 MHz variants
it works a little different, but it has two MCUs sitting there, so some use
for theme should there be.

Cheers,
Magnus

_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.


_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus Danielson
2012-01-25 19:19:16 UTC
Reply
Permalink
Raw Message
Hi Joe,
Post by J. L. Trantham
Magnus,
Thanks for the research. Now that I have read your information, I am going
to have to go back and 'measure' the PPS output on my unit. I just
'eyeballed' the '1 PPS' and it seemed close to 1 second.
Good luck. Hopefully I put you onto a working solution.
Post by J. L. Trantham
It would be nice to find a serial command that would allow placing the 1 PPS
at any arbitrary point in phase to match GPS.
My impression of reading the posts about the FE-5680A is that the 1 PPS is
likely there, just not easily documented.
Well, being lazy enough not to go into the lab and open up my 5680 I
went hunting the web and it seems like there was more out there than we
seems to have been piecing together.

Seems like a bit of more systematic research needs to be done. There are
obviously at least two basic models, one older with 50,255+ MHz
oscillators and various add-on boards and secondly the 60 MHz + DDS
variant. Reverse-engineering the schematics and functions has only
partly been done, but I haven't seen any systematic schematic popping
out of the work.

There is already plenty of information available, but it is not
coordinated and systematically done to the level that most aspects can
be written down in a service type of document.

Cheers,
Magnus
Chris Albertson
2012-01-25 19:54:40 UTC
Reply
Permalink
Raw Message
I'm sure people will figure out all these undocumented features. But
there is a danger is using any of them because some day your Rb
oscillator will fail and you will need to replace it. You can't count
on the replacement unit to have the same set of undocumented features.

I do intend to lock mine to a T-Bolt. but I'll phase lock the 10MHz
signals not the PPS.
I think I can send both 10MHz to the 74HCT9046 a uPwill read the
74HCT9046 and then send rs-232 commands to adjust the fe5680 to reduce
the phase or frequency error

The uP might also check the temperature of the FE5680 and keep a table
of DDS words vs. temperature and use this table for hold over when the
GPS is disconnected. I don't know. First steps first need to play
with 74HCT9046 and need to write fe5680 library for Arduino.

On Wed, Jan 25, 2012 at 11:19 AM, Magnus Danielson
Post by Magnus Danielson
Hi Joe,
Post by J. L. Trantham
Magnus,
Thanks for the research.  Now that I have read your information, I am
going
to have to go back and 'measure' the PPS output on my unit.  I just
'eyeballed' the '1 PPS' and it seemed close to 1 second.
Good luck. Hopefully I put you onto a working solution.
Post by J. L. Trantham
It would be nice to find a serial command that would allow placing the 1 PPS
at any arbitrary point in phase to match GPS.
My impression of reading the posts about the FE-5680A is that the 1 PPS is
likely there, just not easily documented.
Well, being lazy enough not to go into the lab and open up my 5680 I went
hunting the web and it seems like there was more out there than we seems to
have been piecing together.
Seems like a bit of more systematic research needs to be done. There are
obviously at least two basic models, one older with 50,255+ MHz oscillators
and various add-on boards and secondly the 60 MHz + DDS variant.
Reverse-engineering the schematics and functions has only partly been done,
but I haven't seen any systematic schematic popping out of the work.
There is already plenty of information available, but it is not coordinated
and systematically done to the level that most aspects can be written down
in a service type of document.
Cheers,
Magnus
_______________________________________________
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
--
Chris Albertson
Redondo Beach, California
Attila Kinali
2012-01-25 23:21:46 UTC
Reply
Permalink
Raw Message
On Wed, 25 Jan 2012 20:19:16 +0100
Post by Magnus Danielson
Reverse-engineering the schematics and functions has only
partly been done, but I haven't seen any systematic schematic popping
out of the work.
Now that i have received my units, i thought could help with the
schematics. But having had a look at mine, i dont think it's worth
the effort. The PCB has at least 4 layers, more likely 6. I don't
think it's 8 as the PCB looks cost optimized.

A few distinct subsystems can be identified by their position, but
their connections and the connections within are hard to get. It would
take countless hours with good measurement equipment to get anything
halfway usable.

I'd rather say, save the time, buy another spare unit if you think
one of those you have is going to fail.



For those who wants to open their unit here a few hints:

* Unscrew the unit from it's base plate
They are either Torx or Hex. But this doesn't matter at this size,
the difference is hard to tell and is very small anyways.
I used a T6 Torx screw driver, as that was what i had at
hand that did fit. Be carefull, the screws are cheap and very
soft. It's easy to rip the torx/hex bit apart. Apply ample pressure
to your screw driver.
* Unscrew all screws that you can see from outside
* two in the center of the bottom plate
* two underneath the DB-9 connector
* two on the side of the DB-9
* two on the sides
These are Philips-1 and Philips-2 screws
* There is a screw under each lable at the top
* After this, you can take the unit apart. The heat spreader in the middle
comes out together with the PCB
* Do NOT force the PCB out. If you need to, wiggle gently. If you have
to use force, you've forgotten a screw.
* The two screws that you see on the PCB stay there (they hold the PCB
on the head spreader). I do not recommend to unscrew them.

* Take lots of pictures while you are at it :-)

If possible, i'd be interested in two Revision markings on the PCB.
Mine has a sticker at the bottom, that reads "RevF".
Then there is a PCB number on the top side, at the edge near the
reference crystal (that with the PTC soldered to it). My unit reads there
"217421-30352" "Rev G"

Attila Kinali
--
Why does it take years to find the answers to
the questions one should have asked long ago?
Attila Kinali
2012-01-25 23:29:08 UTC
Reply
Permalink
Raw Message
On Thu, 26 Jan 2012 00:21:46 +0100
Post by Attila Kinali
* Do NOT force the PCB out. If you need to, wiggle gently. If you have
to use force, you've forgotten a screw.
Additional note: wiggle holding the heat spreader, not the PCB.
Post by Attila Kinali
If possible, i'd be interested in two Revision markings on the PCB.
Mine has a sticker at the bottom, that reads "RevF".
Then there is a PCB number on the top side, at the edge near the
reference crystal (that with the PTC soldered to it). My unit reads there
"217421-30352" "Rev G"
And a correction: the sticker reads "406" "RevF".

I think the sticker denotes the what population option was choosen,
while the PCB marking denotes the type and revision of the PCB.

Additonal note to self: do not take any device apart, when already
being half asleep. Also do not try to write emails and trying to make
any sense.

^^;

Attila Kinali
--
Why does it take years to find the answers to
the questions one should have asked long ago?
J. L. Trantham
2012-01-26 03:13:52 UTC
Reply
Permalink
Raw Message
Magnus,

I went back to the shop tonight and checked both my FE-5680A's.

Both pin 3 and pin 6 are 'high' when power is first applied. When the unit
'locks', both pin 3 and 6 go 'low' but pin 6 then puts out a 5 uSec wide 1
PPS pulse, as judged by my 'calibrated eyeball' which means that I used a
stop watch and counted the time for 10 sweeps of my scope, triggered by pin
6 and set for 5 mSec/cm sweep. These are, indeed '1 pulse per second'
pulses.

Now, I need to move the power supplies, scope and FE-5680A to where my 5370B
and TBolt are located. Don't hold your breath on that though, too many
irons in the fire.

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Magnus Danielson
Sent: Wednesday, January 25, 2012 1:19 PM
To: time-nuts-***@public.gmane.org
Subject: Re: [time-nuts] Determination of the placement of the first pps

Hi Joe,
Post by J. L. Trantham
Magnus,
Thanks for the research. Now that I have read your information, I am going
to have to go back and 'measure' the PPS output on my unit. I just
'eyeballed' the '1 PPS' and it seemed close to 1 second.
Good luck. Hopefully I put you onto a working solution.
Post by J. L. Trantham
It would be nice to find a serial command that would allow placing the 1 PPS
at any arbitrary point in phase to match GPS.
My impression of reading the posts about the FE-5680A is that the 1 PPS is
likely there, just not easily documented.
Well, being lazy enough not to go into the lab and open up my 5680 I
went hunting the web and it seems like there was more out there than we
seems to have been piecing together.

Seems like a bit of more systematic research needs to be done. There are
obviously at least two basic models, one older with 50,255+ MHz
oscillators and various add-on boards and secondly the 60 MHz + DDS
variant. Reverse-engineering the schematics and functions has only
partly been done, but I haven't seen any systematic schematic popping
out of the work.

There is already plenty of information available, but it is not
coordinated and systematically done to the level that most aspects can
be written down in a service type of document.

Cheers,
Magnus

_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Scott Newell
2012-01-26 03:57:15 UTC
Reply
Permalink
Raw Message
Post by J. L. Trantham
Both pin 3 and pin 6 are 'high' when power is first applied. When the unit
'locks', both pin 3 and 6 go 'low' but pin 6 then puts out a 5 uSec wide 1
PPS pulse, as judged by my 'calibrated eyeball' which means that I used a
stop watch and counted the time for 10 sweeps of my scope, triggered by pin
6 and set for 5 mSec/cm sweep. These are, indeed '1 pulse per second'
pulses.
I attempted to test a FE-5680A pps against the tbolt. I had the 10
MHz tbolt out to my (somewhat flaky) 5370A's ref in. Here's what I
got for 100 events, which took about 3:20 to complete:

mean: 999.999 900 993 ms
sdev: 253 ps
min : 999.999 900 49 ms
max : 999.999 901 68 ms
evnt: 100


Now this was a few days ago and I was confused so don't put too much
trust in these numbers. FYI, the tbolt pps output measured as (not
freq, not period):

mean: 1.000 000 099 83 Hz
sdev: 289 pHz (?)
min : 1.000 000 099 04 Hz
max : 1.000 000 100 62 Hz
evnt: 100
--
newell N5TNL
Hal Murray
2012-01-24 23:11:47 UTC
Reply
Permalink
Raw Message
But how do you untangle longitude and time? How do you know that you are
looking exactly south (or north)?
If I understand what you are asking, it's the same problem as navigating a
ship without a clock.

Classic navigation with a sextant needs a clock and sightings on 3 objects in
the sky. Each sighting gives you a circle on the globe, or a line if you
know roughly where you are. The lines form a triangle. The size of the
triangle is an indication of the accuracy. You pick the objects so the
triangle is roughly equilateral.
http://en.wikipedia.org/wiki/Celestial_navigation
You can get time from the moon, so in theory at least, that's an answer to
your question.
http://en.wikipedia.org/wiki/Lunar_distance_%28navigation%29

Years ago, when a friend was learning navigation, he was reading one of the
old classic texts. There was a good story about the guy off the coast of
England/Ireland who didn't trust his clock, so he did the calculations again
assuming his clock was a bit fast and again with it slow. That gave him 3
parallel lines for each sighting. Anybody recognize that story?


Longitude by Dava Sobel is a good read, especially for time-nuts. There is
also a version with lots of good photographs.

One of the techniques they actually considered before Harrison built good
enough clocks was to derive time from Jupiter's moons. They knew enough to
correct for the time shift due to speed of light delays as the Earth-Jupiter
distance changed. (I don't know if they knew if was due to speed of light.)

You can synchronize two clocks if both sites can see the same event in the
sky, Occultations are often used for this.

With modern technology, radio telescopes are very very good at this. In
order to do VLBI, you need to know where the telescopes are located. With a
big collection of data you can do least-squared fit type calculations to
refine the location and clock calibration.
--
These are my opinions, not necessarily my employer's. I hate spam.
J. Forster
2012-01-24 23:19:10 UTC
Reply
Permalink
Raw Message
Is the USNO almana/ephemeris still published in hard copy every year? That
had moon timing, etc.

-John

===============

-John

================
Post by Hal Murray
But how do you untangle longitude and time? How do you know that you are
looking exactly south (or north)?
If I understand what you are asking, it's the same problem as navigating a
ship without a clock.
Classic navigation with a sextant needs a clock and sightings on 3 objects in
the sky. Each sighting gives you a circle on the globe, or a line if you
know roughly where you are. The lines form a triangle. The size of the
triangle is an indication of the accuracy. You pick the objects so the
triangle is roughly equilateral.
http://en.wikipedia.org/wiki/Celestial_navigation
You can get time from the moon, so in theory at least, that's an answer to
your question.
http://en.wikipedia.org/wiki/Lunar_distance_%28navigation%29
Years ago, when a friend was learning navigation, he was reading one of the
old classic texts. There was a good story about the guy off the coast of
England/Ireland who didn't trust his clock, so he did the calculations again
assuming his clock was a bit fast and again with it slow. That gave him 3
parallel lines for each sighting. Anybody recognize that story?
Longitude by Dava Sobel is a good read, especially for time-nuts. There is
also a version with lots of good photographs.
One of the techniques they actually considered before Harrison built good
enough clocks was to derive time from Jupiter's moons. They knew enough to
correct for the time shift due to speed of light delays as the
Earth-Jupiter
distance changed. (I don't know if they knew if was due to speed of light.)
You can synchronize two clocks if both sites can see the same event in the
sky, Occultations are often used for this.
With modern technology, radio telescopes are very very good at this. In
order to do VLBI, you need to know where the telescopes are located. With a
big collection of data you can do least-squared fit type calculations to
refine the location and clock calibration.
--
These are my opinions, not necessarily my employer's. I hate spam.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
jmfranke
2012-01-24 23:47:10 UTC
Reply
Permalink
Raw Message
For a rough determination; you are facing due south, or due north when the
elevation of a celestial body stops increasing with time. The elevation is
highest when the body is on the observer's local meridian. There are
exceptions, for instance when observing a body below Polaris, then the body
reaches the lowest elevation when crossing the observer's local meridian,
but reaches its highest elevation twelve sidereal hours later.

John WA4WDL

--------------------------------------------------
Post by Robert Benward
Post by Hal Murray
But how do you untangle longitude and time? How do you know that you are
looking exactly south (or north)?
If I understand what you are asking, it's the same problem as navigating a
ship without a clock.
Classic navigation with a sextant needs a clock and sightings on 3
objects
in
the sky. Each sighting gives you a circle on the globe, or a line if you
know roughly where you are. The lines form a triangle. The size of the
triangle is an indication of the accuracy. You pick the objects so the
triangle is roughly equilateral.
http://en.wikipedia.org/wiki/Celestial_navigation
You can get time from the moon, so in theory at least, that's an answer to
your question.
http://en.wikipedia.org/wiki/Lunar_distance_%28navigation%29
Years ago, when a friend was learning navigation, he was reading one of the
old classic texts. There was a good story about the guy off the coast of
England/Ireland who didn't trust his clock, so he did the calculations again
assuming his clock was a bit fast and again with it slow. That gave him 3
parallel lines for each sighting. Anybody recognize that story?
Longitude by Dava Sobel is a good read, especially for time-nuts. There is
also a version with lots of good photographs.
One of the techniques they actually considered before Harrison built good
enough clocks was to derive time from Jupiter's moons. They knew enough to
correct for the time shift due to speed of light delays as the Earth-Jupiter
distance changed. (I don't know if they knew if was due to speed of light.)
You can synchronize two clocks if both sites can see the same event in the
sky, Occultations are often used for this.
With modern technology, radio telescopes are very very good at this. In
order to do VLBI, you need to know where the telescopes are located.
With
a
big collection of data you can do least-squared fit type calculations to
refine the location and clock calibration.
--
These are my opinions, not necessarily my employer's. I hate spam.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Jim Lux
2012-01-25 01:52:35 UTC
Reply
Permalink
Raw Message
Post by J. Forster
Is the USNO almana/ephemeris still published in hard copy every year? That
had moon timing, etc.
You can download pieces from the Astronomical Applications website at USNO.

Or you can buy a copy of the Nautical Almanac for about $20 from a
variety of sources. You could also download the pdf (but printing it
would cost you more than the $20)..

Amazon has it, for instance.

http://www.usno.navy.mil/USNO/astronomical-applications/publications/naut-almanac

will find it, but the GPO version is more expensive than the commercial
versions..
Neville Michie
2012-01-25 13:34:02 UTC
Reply
Permalink
Raw Message
Finding your location without GPS is not all that difficult.
You need a quality theodolite, but even an ordinary one will read to
1 second of arc.
You observe circumpolar stars at night to obtain a true azimuth.
(North and South)
and also the latitude by the inclination of the pole.
On a time photograph these stars draw circles around the pole, the
centre of the circle
is the celestial pole and its elevation above the horizon gives the
latitude.
You can also use an almanac and a calendar to determine your latitude
by observing stars
with the theodolite.
You observe the sun at noon to find the local time and set your local
clock. You then
wait for an event like an eclipse of a planets moons to establish the
relationship
between your local time and the time at a known site.
A theodolite has a telescope that can be "plunged" i.e. used upside
down and this
technique is used to get a very accurate level from a striding level.
No pool of mercury
is needed.
The setting up of a theodolite uses sitings and reversed sitings to
set the vertical level.
The main error is the atmospheric refraction which scatters
individual observations,
so many repeated observations are needed. The local time observations
need to be
repeated for good accuracy.
A sextant is a less accurate instrument that has the main redeeming
feature that when
reading it you superimpose the image of a star or the sun with the
image of the horizon.
Although the image seen may be rolling around, the position of the
sun on the horizon
is rock steady and is adjusted by the thimble for coincidence. The
elevation is then
read off the vernier. A theodolite needs a solid base to work from
and would be useless
on a ship.
cheers,
Neville Michie
Post by Jim Lux
Post by J. Forster
Is the USNO almana/ephemeris still published in hard copy every year? That
had moon timing, etc.
You can download pieces from the Astronomical Applications website at USNO.
Or you can buy a copy of the Nautical Almanac for about $20 from a
variety of sources. You could also download the pdf (but printing
it would cost you more than the $20)..
Amazon has it, for instance.
http://www.usno.navy.mil/USNO/astronomical-applications/
publications/naut-almanac
will find it, but the GPO version is more expensive than the
commercial versions..
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/
time-nuts
and follow the instructions there.
J. Forster
2012-01-25 15:49:40 UTC
Reply
Permalink
Raw Message
Post by Neville Michie
Finding your location without GPS is not all that difficult.
You need a quality theodolite, but even an ordinary one will read to
1 second of arc.
Ordinary? You mean something like a Wild T-2 or Kern DKM-2. Even then
getting close to 1 arc-second requires a lot of care.
Post by Neville Michie
You observe circumpolar stars at night to obtain a true azimuth.
(North and South)
and also the latitude by the inclination of the pole.
Not quite so straight forward. You have to have accurate siderial time and
an almanac. Polaris is only near the pole, not at it.
Post by Neville Michie
On a time photograph these stars draw circles around the pole, the
centre of the circle
is the celestial pole and its elevation above the horizon gives the
latitude.
You can also use an almanac and a calendar to determine your latitude
by observing stars
with the theodolite.
Not so easy. At the celestial equator the stars are moving in Hour Angle
at 15 arc-seconds per second.

-John

==============
Post by Neville Michie
You observe the sun at noon to find the local time and set your local
clock. You then
wait for an event like an eclipse of a planets moons to establish the
relationship
between your local time and the time at a known site.
A theodolite has a telescope that can be "plunged" i.e. used upside
down and this
technique is used to get a very accurate level from a striding level.
No pool of mercury
is needed.
The setting up of a theodolite uses sitings and reversed sitings to
set the vertical level.
The main error is the atmospheric refraction which scatters
individual observations,
so many repeated observations are needed. The local time observations
need to be
repeated for good accuracy.
A sextant is a less accurate instrument that has the main redeeming
feature that when
reading it you superimpose the image of a star or the sun with the
image of the horizon.
Although the image seen may be rolling around, the position of the
sun on the horizon
is rock steady and is adjusted by the thimble for coincidence. The
elevation is then
read off the vernier. A theodolite needs a solid base to work from
and would be useless
on a ship.
cheers,
Neville Michie
Post by Jim Lux
Post by J. Forster
Is the USNO almana/ephemeris still published in hard copy every year? That
had moon timing, etc.
You can download pieces from the Astronomical Applications website at USNO.
Or you can buy a copy of the Nautical Almanac for about $20 from a
variety of sources. You could also download the pdf (but printing
it would cost you more than the $20)..
Amazon has it, for instance.
http://www.usno.navy.mil/USNO/astronomical-applications/
publications/naut-almanac
will find it, but the GPO version is more expensive than the
commercial versions..
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/
time-nuts
and follow the instructions there.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Neville Michie
2012-01-25 22:29:06 UTC
Reply
Permalink
Raw Message
Post by J. Forster
Post by Neville Michie
Finding your location without GPS is not all that difficult.
You need a quality theodolite, but even an ordinary one will read to
1 second of arc.
Ordinary? You mean something like a Wild T-2 or Kern DKM-2. Even then
getting close to 1 arc-second requires a lot of care.
A wild T1 reads directly to 6 seconds, but with repetition will get 1
second.
Unlike digital instruments you need a little bit of skill and
persistence to get the best measurement from an analogue instrument.
Post by J. Forster
Post by Neville Michie
You observe circumpolar stars at night to obtain a true azimuth.
(North and South)
and also the latitude by the inclination of the pole.
Not quite so straight forward. You have to have accurate siderial time and
an almanac. Polaris is only near the pole, not at it.
No need for time, you follow the azimuth of the star until it turns
around and then again until it turns back. Half the difference gives you
the azimuth of the pole very accurately. Fit your observations to a
parabola to get a good result.
Works best in Winter when the sun is down for more than 12 hours. A
good technique as refraction errors cancel.
Post by J. Forster
Post by Neville Michie
On a time photograph these stars draw circles around the pole, the
centre of the circle
is the celestial pole and its elevation above the horizon gives the
latitude.
You can also use an almanac and a calendar to determine your latitude
by observing stars
with the theodolite.
Not so easy. At the celestial equator the stars are moving in Hour Angle
at 15 arc-seconds per second.
As I said, analogue measurements need some skill and perseverance.
If you added more modern technology you could track your theodolite/
telescope with a clock so you would get a longer period to adjust/
observe
the observations and set your clock.

Neville
Post by J. Forster
-John
==============
Post by Neville Michie
You observe the sun at noon to find the local time and set your local
clock. You then
wait for an event like an eclipse of a planets moons to establish the
relationship
between your local time and the time at a known site.
A theodolite has a telescope that can be "plunged" i.e. used upside
down and this
technique is used to get a very accurate level from a striding level.
No pool of mercury
is needed.
The setting up of a theodolite uses sitings and reversed sitings to
set the vertical level.
The main error is the atmospheric refraction which scatters
individual observations,
so many repeated observations are needed. The local time observations
need to be
repeated for good accuracy.
A sextant is a less accurate instrument that has the main redeeming
feature that when
reading it you superimpose the image of a star or the sun with the
image of the horizon.
Although the image seen may be rolling around, the position of the
sun on the horizon
is rock steady and is adjusted by the thimble for coincidence. The
elevation is then
read off the vernier. A theodolite needs a solid base to work from
and would be useless
on a ship.
cheers,
Neville Michie
Post by Jim Lux
Post by J. Forster
Is the USNO almana/ephemeris still published in hard copy every year? That
had moon timing, etc.
You can download pieces from the Astronomical Applications website at USNO.
Or you can buy a copy of the Nautical Almanac for about $20 from a
variety of sources. You could also download the pdf (but printing
it would cost you more than the $20)..
Amazon has it, for instance.
http://www.usno.navy.mil/USNO/astronomical-applications/
publications/naut-almanac
will find it, but the GPO version is more expensive than the
commercial versions..
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/
time-nuts
and follow the instructions there.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/
time-nuts
and follow the instructions there.
J. Forster
2012-01-25 23:07:42 UTC
Reply
Permalink
Raw Message
Post by Neville Michie
Post by J. Forster
Ordinary? You mean something like a Wild T-2 or Kern DKM-2. Even then
getting close to 1 arc-second requires a lot of care.
A wild T1 reads directly to 6 seconds, but with repetition will get 1
second.
Unlike digital instruments you need a little bit of skill and
persistence to get the best measurement from an analogue instrument.
Assuming youcan do that w/o bias.

A T-2 ius a 1 ard second instrument, a T-3 is 0.1 arc-second. I've never
seen a T-4 in the flesh.
Post by Neville Michie
Post by J. Forster
Post by Neville Michie
You observe circumpolar stars at night to obtain a true azimuth.
(North and South) and also the latitude by the inclination of the pole.
That means observations over more than 18 hours. It'll take you most of a
year, unless you are above the artic circle.
Post by Neville Michie
Post by J. Forster
Not quite so straight forward. You have to have accurate siderial
time and an almanac. Polaris is only near the pole, not at it.
No need for time, you follow the azimuth of the star until it turns
around and then again until it turns back. Half the difference gives you
the azimuth of the pole very accurately.
See above.
Post by Neville Michie
Fit your observations to a
parabola to get a good result.
Works best in Winter when the sun is down for more than 12 hours. A
good technique as refraction errors cancel.
In practice, the "seeing" is nowhere near 1 arc-second for 2-3" aperture
'scopes.
Post by Neville Michie
Post by J. Forster
Post by Neville Michie
On a time photograph these stars draw circles around the pole, the
centre of the circle
is the celestial pole and its elevation above the horizon gives the
latitude.
And to do that you need a sub-arc second telescope mount. You just can't
mount a camera on a tripod.
Post by Neville Michie
Post by J. Forster
Post by Neville Michie
You can also use an almanac and a calendar to determine your latitude
by observing stars
with the theodolite.
Not so easy. At the celestial equator the stars are moving in Hour
Angle at 15 arc-seconds per second.
As I said, analogue measurements need some skill and perseverance.
That's an understatement. I've done it, both for North lines and to adjust
a 24" telescope.
Post by Neville Michie
If you added more modern technology you could track your theodolite/
telescope with a clock so you would get a longer period to adjust/
observe the observations and set your clock.
Neville
The "modern technology" just makes the angle readout direct.

-John

==============
Brooke Clarke
2012-01-24 23:45:57 UTC
Reply
Permalink
Raw Message
Hi:

The basic way to find your location anywhere in the world is to use a photo sensor.
This is the method used on tagged fish. The light level is logged and time stamped probably using a watch crystal.
When the fish is caught the logger data is read out.

Have Fun,

Brooke Clarke
http://www.PRC68.com
http://www.end2partygovernment.com/Brooke4Congress.html
J. L. Trantham
2012-01-26 04:35:23 UTC
Reply
Permalink
Raw Message
I'll have to get one of those 'round TUIT's' one of these days and check
mine.

The interval clearly is not something near .8 sec and, likely, truly 1 PPS
along with the 10 MHz output. However, by time nut standards, I have more
work to do.

Joe

-----Original Message-----
From: time-nuts-bounces-***@public.gmane.org [mailto:time-nuts-bounces-***@public.gmane.org] On
Behalf Of Scott Newell
Sent: Wednesday, January 25, 2012 9:57 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Determination of the placement of the first pps
Post by J. L. Trantham
Both pin 3 and pin 6 are 'high' when power is first applied. When the
unit 'locks', both pin 3 and 6 go 'low' but pin 6 then puts out a 5
uSec wide 1 PPS pulse, as judged by my 'calibrated eyeball' which means
that I used a stop watch and counted the time for 10 sweeps of my
scope, triggered by pin 6 and set for 5 mSec/cm sweep. These are,
indeed '1 pulse per second' pulses.
I attempted to test a FE-5680A pps against the tbolt. I had the 10
MHz tbolt out to my (somewhat flaky) 5370A's ref in. Here's what I
got for 100 events, which took about 3:20 to complete:

mean: 999.999 900 993 ms
sdev: 253 ps
min : 999.999 900 49 ms
max : 999.999 901 68 ms
evnt: 100


Now this was a few days ago and I was confused so don't put too much
trust in these numbers. FYI, the tbolt pps output measured as (not
freq, not period):

mean: 1.000 000 099 83 Hz
sdev: 289 pHz (?)
min : 1.000 000 099 04 Hz
max : 1.000 000 100 62 Hz
evnt: 100
--
newell N5TNL


_______________________________________________
time-nuts mailing list -- time-nuts-***@public.gmane.org
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2012-02-02 20:15:41 UTC
Reply
Permalink
Raw Message
Yes, if you need lots o'bits, but a single bit sampler with wide
bandwidth is easy (which is why they do it). It's basically a D-latch
at the end of the amplifier/limiter chain.
Yes, but you lose IIRC about 3dB of performance compared to a 2bit ADC.
Only if you get the gain right.

A 1-bit ADC does the right thing with any gain.
--
These are my opinions, not necessarily my employer's. I hate spam.
Hal Murray
2012-02-02 21:07:36 UTC
Reply
Permalink
Raw Message
I thought the 4th satellite was needed to determine the time. Wouldn't
it take a 5th satellite to also determine the frequency of the local clock?
Not really. There are two ways to get the postion and time derivatives. One
is to either use two fixes which give you each a (x,y,z,t) tuple, while you
know what your expected delta-t is, you can calculate the "real" delta-t and
get from that your frequency offset.
That's the sort of thing I'm looking for, but I don't quite get it yet.

I have 4 satellites. If I know f, I can solve for x, y, z, and t. If I don't
know f, I'm short an equation.

If I get two samples, I have 8 equations and I need to solve for:
x0, y0, z0, t0, and f0
x1, y1, z1, t1, and f1
That's 10 unknowns with 8 equations. I get a 9th equation by setting t1 = t0
+ 1. I'm still short one equation.

Can I do something like assume f0 = f1? That would make sense if the change
in frequency is small relative to the noise/error in all the other
calculations.
The other way is to use the doppler shifts of the satelites. You know what
position and speed relative to you the satelites have and can from this
calculate what your speed, respektive frequency is.
There is a chicken-and-egg problem in there. If I need the local clock
frequency to solve for position, I can't use position to solve for frequency.

Consider the time-nut case of 1 satellite, known position, and trying to find
time. If I know the rough time I can calculate the Doppler. That tells me
which FFT bucket to look in. Is the local clock close enough for that even
if it's off by a few/10s of PPM?
--
These are my opinions, not necessarily my employer's. I hate spam.
Dennis Ferguson
2012-02-03 07:39:10 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
I thought the 4th satellite was needed to determine the time. Wouldn't
it take a 5th satellite to also determine the frequency of the local clock?
Not really. There are two ways to get the postion and time derivatives. One
is to either use two fixes which give you each a (x,y,z,t) tuple, while you
know what your expected delta-t is, you can calculate the "real" delta-t and
get from that your frequency offset.
That's the sort of thing I'm looking for, but I don't quite get it yet.
I have 4 satellites. If I know f, I can solve for x, y, z, and t. If I don't
know f, I'm short an equation.
If you are using an undisciplined free-running oscillator, as most cheap
receivers do, you never know f. What you know is the frequency written on
the oscillator's package (call it fn, the nominal frequency), but the actual
f is a mystery. Whatever f is, however, you assume f=fn and use that
oscillator to generate a local timescale to measure signal phases against.

When you solve for x, y, z and t from data generated by measuring the phase
of the incoming signals against your oscillator, the `t' you compute is
actually a delta_t with respect to the local time scale generated from that
oscillator. The value of delta_t tells you the phase error of your local
timescale, so the rate of change of delta_t from sample to sample tells you
the error in the fn you assumed, that is (f/fn) integrated over the sample
interval.
Post by Hal Murray
x0, y0, z0, t0, and f0
x1, y1, z1, t1, and f1
That's 10 unknowns with 8 equations. I get a 9th equation by setting t1 = t0
+ 1. I'm still short one equation.
Can I do something like assume f0 = f1? That would make sense if the change
in frequency is small relative to the noise/error in all the other
calculations.
I suspect that if the local oscillator does not exhibit fairly good short
term stability there is no hope of any of this working. That doesn't matter,
though, since the GPS `t' you compute is actually a delta_t from whatever your
local time scale is, so (delta_t1 - delta_t0) directly tells you how the rate
of your local time scale differs from the rate of the GPS timescale. The GPS
receiver in fact has no knowledge of the GPS `t' other than as a function of
the local time scale. The GPS time scale is purely a paper time scale from the
receiver's point of view unless the receiver does the additional work of somehow
using that information to generate a real timescale out of the paper.

Dennis Ferguson
Hal Murray
2012-02-07 11:35:29 UTC
Reply
Permalink
Raw Message
BTW: does anyone know how these days low cost FPGAs perform in terms of
jitter? (the data sheets are kind of scarce in that regard). And how do they
compare to state of the art ECL logic?
Generally, not good.

The general problem is that they have a lot of logic and a lot of I/O drivers
and shared power/ground pins. Things are messy if you have multiple clocks.
Things are better if you only have one clock and better if you don't have any
nearby drivers switching at the same time.

For a few critical signals, you could reclock in an external FF.
--
These are my opinions, not necessarily my employer's. I hate spam.
Azelio Boriani
2012-02-07 13:06:55 UTC
Reply
Permalink
Raw Message
To enhance the PICTIC II performance can step recovery diodes be used?
Maybe the fast turn off can boost the switching capabilities of the
interpolator for best resolution...
Post by Hal Murray
BTW: does anyone know how these days low cost FPGAs perform in terms of
jitter? (the data sheets are kind of scarce in that regard). And how do
they
compare to state of the art ECL logic?
Generally, not good.
The general problem is that they have a lot of logic and a lot of I/O drivers
and shared power/ground pins. Things are messy if you have multiple clocks.
Things are better if you only have one clock and better if you don't have any
nearby drivers switching at the same time.
For a few critical signals, you could reclock in an external FF.
--
These are my opinions, not necessarily my employer's. I hate spam.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
David
2012-02-07 14:18:33 UTC
Reply
Permalink
Raw Message
Step recovery diodes turn off fast but have a relatively long storage
time. The fastest switched current integrators use schottky diodes.

On Tue, 7 Feb 2012 14:06:55 +0100, Azelio Boriani
Post by Azelio Boriani
To enhance the PICTIC II performance can step recovery diodes be used?
Maybe the fast turn off can boost the switching capabilities of the
interpolator for best resolution...
Post by Hal Murray
BTW: does anyone know how these days low cost FPGAs perform in terms of
jitter? (the data sheets are kind of scarce in that regard). And how do
they
compare to state of the art ECL logic?
Generally, not good.
The general problem is that they have a lot of logic and a lot of I/O drivers
and shared power/ground pins. Things are messy if you have multiple clocks.
Things are better if you only have one clock and better if you don't have any
nearby drivers switching at the same time.
For a few critical signals, you could reclock in an external FF.
Hal Murray
2012-04-16 08:19:25 UTC
Reply
Permalink
Raw Message
What's the quality of those chinese scopes?
http://www.febo.com/pipermail/time-nuts/2012-January/061925.html
--
These are my opinions, not necessarily my employer's. I hate spam.
Hal Murray
2012-05-23 08:54:20 UTC
Reply
Permalink
Raw Message
[snip lots of low jitter samples]
Network is a destkop - switch1 - switch2 - ntp box.
The switches are two Level1 Gbit smart switches. The desktop is a ~4y old
Xeon 2GHz system with a Gbit interface The ntp box is a AMD Geode LX 500MHz
system with a 100MBit interface Both running linux.
There aren't noticable more jitter for moderate (1-2 MByte/s) traffic.
(Probably visible if i would do a statistical analysis...but..)
1-2 megabytes/sec is 8-16 megabits/sec. You won't get into serious troubles
until you saturate a link. With modern CPUs, it's trivial to saturate 100
megabit links and not very hard to saturate 1 gigabit links.

With older/slower CPUs, you might run into problems a lot sooner.

I'm not trying to discourage using these boxes. Just don't be surprised when
you run into quirks if you are trying to use them for timing. (and don't
depend upon loud mouths like me to point out all possible ways they can screw
up)
--
These are my opinions. I hate spam.
Hal Murray
2012-12-08 09:27:21 UTC
Reply
Permalink
Raw Message
Only then he
made the comment that there might be something in that "Black Magic"
book that I had bought and distributed.
Which book would that be?
High Speed Digital Design: A Handbook of Black Magic
Howard Johnson (Author), Martin Graham

Amazon charges $81.64 (down to $63 through their affiliates ???)

I really like it. Mostly, it's covering PCB design for 1993 technology. If
you are interested in things like the recent reflection discussion this is
what you want.

It's probably not good enough for modern superfast logic.
--
These are my opinions. I hate spam.
Magnus Danielson
2012-12-08 15:00:54 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
Only then he
made the comment that there might be something in that "Black Magic"
book that I had bought and distributed.
Which book would that be?
High Speed Digital Design: A Handbook of Black Magic
Howard Johnson (Author), Martin Graham
Amazon charges $81.64 (down to $63 through their affiliates ???)
I really like it. Mostly, it's covering PCB design for 1993 technology. If
you are interested in things like the recent reflection discussion this is
what you want.
It's probably not good enough for modern superfast logic.
It is, but it does not cover modern PCB layout technologies, which is
why there is a much thicker follow-up book.

Cheers,
Magnus
Hal Murray
2012-12-27 20:27:24 UTC
Reply
Permalink
Raw Message
Here lies the big problem. While with GPS we pretty much know what the time
is that the signal takes to reach earth, we have no clue with network
packets in a loaded network.
I agree that if you are running on a busy network you are out of luck.

On the other hand, if the network is lightly loaded and the topology is
stable, you can measure the round trip time and learn the unloaded time.
Then you can ignore samples with longer round trip times.
You can guestimate that getting below 200us is not easy in a normal network,
but sub-1ms should be possible unless the network is very loaded.
I can easily get ping times of under 150 us. One switch. I've got lots of
ntp log files. I should make a histogram.
--
These are my opinions. I hate spam.
Hal Murray
2012-12-29 03:18:06 UTC
Reply
Permalink
Raw Message
From the data ntp gives me in the networks i manage. I hardly get any
jitter number below 1ms, even with unloaded network and unloaded hosts. The
200us comes from the "usual" rtt time measurements on PCs.
What sort of networks are you talking about? Are you synchronizing over LAN
or WAN? Do you have a local refclock?

On LANs with non-ancient PCs, it's easy to get round trip times under 200
usec. That makes it hard to get jitter over 1 ms. Or, if the jitter is that
bad, how can you call the network unloaded?

Here are a couple of ntpq printouts:

$ ntpq -p tim
remote refid st t when poll reach delay offset jitter
==============================================================================
*glypnod .PPS. 1 u 194 256 377 0.145 0.186 0.018
-shuksan .PPS. 1 u 95 256 377 0.115 0.126 0.030
-mini-mon .PPS. 1 u 68 256 377 0.177 0.080 0.030
-tom 192.168.1.3 2 u 57 256 377 0.200 0.161 0.086
+bob .PPS. 1 u 117 256 377 0.146 0.199 0.045
+ted .PPS. 1 u 11 256 377 0.211 0.208 0.036

$ ntpq -p tom
remote refid st t when poll reach delay offset jitter
==============================================================================
*glypnod .PPS. 1 u 37 256 377 0.130 0.043 0.014
-shuksan .PPS. 1 u 91 256 377 0.139 -0.005 0.015
+mini-mon .PPS. 1 u 191 256 377 0.172 -0.006 0.023
-bob .PPS. 1 u 198 256 377 0.392 0.186 0.050
+ted .PPS. 1 u 206 256 377 0.319 0.127 0.038
-tim 192.168.1.3 2 u 34 256 377 0.243 -0.074 0.036
-jim 199.102.46.72 2 u 31 256 377 26.243 8.357 29.291
-xo-c2 155.101.3.113 3 u 126 256 377 28.335 10.815 23.143

The last two lines are from systems synchronizing out over the
big-bad-internet. The others systems are all local to my LAN. (Calling it a
LAN is stretching things. It's just one switch.)
--
These are my opinions. I hate spam.
Hal Murray
2014-01-05 21:35:43 UTC
Reply
Permalink
Raw Message
Also keep in mind that RS-232 relies on the voltage going negative to encode
a "1". I.e. getting 0V is not enough and might only work by chance with some
RS-232 receivers.
I think there are 2 parts to this discussion. What do the specs say, and
what actually happens in the real world?

I think the specs say that -3 to +3 is no mans land. A valid signal must be
over +3 or under -3.

In practice, the receiver chip only has one power supply. It would take
extra work to make the switching threshold below ground.

There is an additional quirk in here. The original Motorola MC1489 had a
switching threshold of a diode drop (and some hysteresis). That chip was
very popular and turned into a defacto standard. If you built a RS-232
receiver chip that required a negative input voltage, all sorts of obscure
things would break and anybody who used it would have support nightmares. [1]

The typical RS-232 receiver chips actually have good specs. In particular
they spec the transition voltages in each direction.

TI Data sheet for MC1489(A) and SN75189(A)
http://www.ti.com/lit/ds/symlink/mc1489a.pdf

TI Data sheet for MAX232
http://www.ti.com/lit/ds/symlink/max232.pdf

-----------

Many years ago (early 1980s?), there was a popular brand of modems that sent out a TTL level rather than real RS-232 levels. Yes, we found that the hard way when we cut a corner.
--
These are my opinions. I hate spam.
Hal Murray
2015-02-25 09:44:40 UTC
Reply
Permalink
Raw Message
Yes. You should not use a logic gates with analog input signals. Using a
74LVC14 helps due to its Schmitt-Trigger input. I think the proper solution
here would be to use a high speed comparator instead (with hysteresis).
The Schmitt trigger mostly avoids glitches on the output. Does it do
anything to reduce timing noise if the input signal is clean enough that it
doesn't make any glitches?
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus Danielson
2015-02-25 21:02:12 UTC
Reply
Permalink
Raw Message
Hi Hal,
Post by Hal Murray
Yes. You should not use a logic gates with analog input signals. Using a
74LVC14 helps due to its Schmitt-Trigger input. I think the proper solution
here would be to use a high speed comparator instead (with hysteresis).
The Schmitt trigger mostly avoids glitches on the output. Does it do
anything to reduce timing noise if the input signal is clean enough that it
doesn't make any glitches?
No, it just avoids flipping state at the transition point(s).

The trigger jitter problem remains the same, regardless if it is at one
voltage (comparator with no hysteresis) or two voltages (comparator with
hysteresis aka Schmitt trigger), the slew-rate at the comparator voltage
and the noise will interact to create trigger jitter. If you want to
improve on that the main solution is to improve slew rate, but naturally
careful filtering can help.

It is all to often that I have encountered people to confuse the
Schmitt-trigger for improving the timing jitter. It's movement in two
different domains, voltage (schmitt trigger) and time (trigger jitter).

Cheers,
Magnus
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Charles Steinmetz
2015-02-26 01:46:59 UTC
Reply
Permalink
Raw Message
Post by Magnus Danielson
Post by Hal Murray
The Schmitt trigger mostly avoids glitches on the output. Does it do
anything to reduce timing noise if the input signal is clean enough that it
doesn't make any glitches?
No, it just avoids flipping state at the transition point(s).
Note also that the hysteresis of logic gates with Schmitt inputs is
WAY too much to be optimal for squaring sine waves (300mV minimum,
typically 400 to 450mV, for the 74LVC14). Fast comparators with
internal hysteresis are optimized for that sort of thing (the LT1719
and LT1720 have a few mV of hysteresis).

Best regards,

Charles


_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Magnus Danielson
2015-02-26 06:46:19 UTC
Reply
Permalink
Raw Message
Hi Charles,
Post by Magnus Danielson
Post by Hal Murray
The Schmitt trigger mostly avoids glitches on the output. Does it do
anything to reduce timing noise if the input signal is clean enough that it
doesn't make any glitches?
No, it just avoids flipping state at the transition point(s).
Note also that the hysteresis of logic gates with Schmitt inputs is WAY
too much to be optimal for squaring sine waves (300mV minimum, typically
400 to 450mV, for the 74LVC14). Fast comparators with internal
hysteresis are optimized for that sort of thing (the LT1719 and LT1720
have a few mV of hysteresis).
Indeed.

If you think about what large hysteresis does on a sine, it moves the
trigger points further up and down on the sine from the mid-point, which
moves them into lower slew-rate areas.

If you are picky, amplitude variations will then also move the phase
more than mid-point triggers.

A bit of hysteresis can help to avoid flipping back, but considering the
type of signal, it passes the mid-point (0 V) at highest slew-rate, so
there is very little risk of flipping back and fourth in the first
place, so hysteresis may not even be needed.

Cheers,
Magnus
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2015-03-15 10:27:06 UTC
Reply
Permalink
Raw Message
A CR2032 is a quite huge coin cell. An NVRAM module does not use much power
once Vcc goes to zero. In todays low power modules it's in the order of
100nA max specified. You can assume it to be somewhere in the range of 10nA
(probably package leakage limited) and 1uA (something has gone wrong or very
old module). ...
GPS also needs the time, so add on a 32KHz clock.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
paul swed
2015-03-15 15:05:53 UTC
Reply
Permalink
Raw Message
I agree with Atilla from what I have seen. Its actually somewhat difficult
to measure this level of current. But all is not lost. Even if the unit is
drawing 1-10ua because something is going wrong. Simply add a battery
holder and 2 X AAA or AA or ...
Whatever it takes to keep the unit going.
If you mount the batteries externally you can easily replace them and check
the discharge rate.
Its a way around the problem if you simply can not get a replacement or its
totally embedded on the board.
Best of luck
Paul
WB8TSL
Post by Hal Murray
A CR2032 is a quite huge coin cell. An NVRAM module does not use much
power
once Vcc goes to zero. In todays low power modules it's in the order of
100nA max specified. You can assume it to be somewhere in the range of
10nA
(probably package leakage limited) and 1uA (something has gone wrong or
very
old module). ...
GPS also needs the time, so add on a 32KHz clock.
--
These are my opinions. I hate spam.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Chris Albertson
2015-03-15 19:46:37 UTC
Reply
Permalink
Raw Message
I have one of these UT+ receivers. Backup is not a big deal. How long
will the power be off? Certainly not for days and weeks. The backup
battery only has to last a few seconds or maybe an hours or two. The real
problem with batteries is not how much energy they store but shelf life.
You have to change the 2032 over five to eight years or so just because of
shelf life. So some people are using "super capacitors" because these can
handle the few hours or few days of backup power and have a much longer
working lifetime r maybe 20 years or more. F

That said, for a hobby user you can do fine with zero backup. So what the
unit looses it's memory and takes a few hours to do the site survey all
over again. What happens to the OCXO during the power outage? It cools
down. This is just as bad as killing the memory in the UT+. So... if you
are worried about outages you have to backup the 12 volt power the entire
GPSDO is running on and if you back this up you don't technically even ned
the CR2032 on the GPS because it will never loose power. But then
again coin batteries are go easy to use why not use one?

If you are worried about hold over performance during a power glitch, you
need a big 12V gel cell battery that can supply the biggest load which has
to the heater on the OCXO. The coin cell is really for the convenience of
YOU the developer who has to power cycle the controller 1,000 times to make
software changes or whatever and you don't want to wait for the GPS for
each software test. Once it is running the big 12V battery means the
CR2032 is never used.
Post by paul swed
I agree with Atilla from what I have seen. Its actually somewhat difficult
to measure this level of current. But all is not lost. Even if the unit is
drawing 1-10ua because something is going wrong. Simply add a battery
holder and 2 X AAA or AA or ...
Whatever it takes to keep the unit going.
If you mount the batteries externally you can easily replace them and check
the discharge rate.
Its a way around the problem if you simply can not get a replacement or its
totally embedded on the board.
Best of luck
Paul
WB8TSL
Post by Hal Murray
A CR2032 is a quite huge coin cell. An NVRAM module does not use much
power
once Vcc goes to zero. In todays low power modules it's in the order of
100nA max specified. You can assume it to be somewhere in the range of
10nA
(probably package leakage limited) and 1uA (something has gone wrong or
very
old module). ...
GPS also needs the time, so add on a 32KHz clock.
--
These are my opinions. I hate spam.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
--
Chris Albertson
Redondo Beach, California
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Bob Camp
2015-03-15 21:00:26 UTC
Reply
Permalink
Raw Message
Hi

For a one off / home use application - enter the local “position hold” info into your
code. Let the micro send it up to the Oncore ….You will be hitting the “compile”
button enough times already that one more isn’t going to slow you down much.

Bob
Post by Chris Albertson
I have one of these UT+ receivers. Backup is not a big deal. How long
will the power be off? Certainly not for days and weeks. The backup
battery only has to last a few seconds or maybe an hours or two. The real
problem with batteries is not how much energy they store but shelf life.
You have to change the 2032 over five to eight years or so just because of
shelf life. So some people are using "super capacitors" because these can
handle the few hours or few days of backup power and have a much longer
working lifetime r maybe 20 years or more. F
That said, for a hobby user you can do fine with zero backup. So what the
unit looses it's memory and takes a few hours to do the site survey all
over again. What happens to the OCXO during the power outage? It cools
down. This is just as bad as killing the memory in the UT+. So... if you
are worried about outages you have to backup the 12 volt power the entire
GPSDO is running on and if you back this up you don't technically even ned
the CR2032 on the GPS because it will never loose power. But then
again coin batteries are go easy to use why not use one?
If you are worried about hold over performance during a power glitch, you
need a big 12V gel cell battery that can supply the biggest load which has
to the heater on the OCXO. The coin cell is really for the convenience of
YOU the developer who has to power cycle the controller 1,000 times to make
software changes or whatever and you don't want to wait for the GPS for
each software test. Once it is running the big 12V battery means the
CR2032 is never used.
Post by paul swed
I agree with Atilla from what I have seen. Its actually somewhat difficult
to measure this level of current. But all is not lost. Even if the unit is
drawing 1-10ua because something is going wrong. Simply add a battery
holder and 2 X AAA or AA or ...
Whatever it takes to keep the unit going.
If you mount the batteries externally you can easily replace them and check
the discharge rate.
Its a way around the problem if you simply can not get a replacement or its
totally embedded on the board.
Best of luck
Paul
WB8TSL
Post by Hal Murray
A CR2032 is a quite huge coin cell. An NVRAM module does not use much
power
once Vcc goes to zero. In todays low power modules it's in the order of
100nA max specified. You can assume it to be somewhere in the range of
10nA
(probably package leakage limited) and 1uA (something has gone wrong or
very
old module). ...
GPS also needs the time, so add on a 32KHz clock.
--
These are my opinions. I hate spam.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
--
Chris Albertson
Redondo Beach, California
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Pete Stephenson
2015-03-15 22:18:22 UTC
Reply
Permalink
Raw Message
Post by Chris Albertson
I have one of these UT+ receivers. Backup is not a big deal. How long
will the power be off? Certainly not for days and weeks. The backup
battery only has to last a few seconds or maybe an hours or two. The real
problem with batteries is not how much energy they store but shelf life.
You have to change the 2032 over five to eight years or so just because of
shelf life. So some people are using "super capacitors" because these can
handle the few hours or few days of backup power and have a much longer
working lifetime r maybe 20 years or more. F
True, it's not a huge deal. The receiver would be in position-hold mode
anyway, so I would have NTPd configured to send the receiver its known
position and a few other configuration options. Everything else is
easily retrievable from the GPS signal given a few minutes.

Backups of several years are overkill for me, particularly because the
ephemeris and almanac are only good for so long. Backup power in the
minutes-to-hours range would be perfectly fine for me.

My question was prompted mainly because I was seeking clarity as to the
proper use of pin #1: I didn't want to connect a non-rechargeable
battery if pin #1 was only intended for rechargeable batteries, as that
might cause damage. If that were the case I could use a supercapacitor,
but I wanted to be sure the pin (a) could supply power, which the manual
didn't mention, and (b) was current-limiting, otherwise the
supercapacitor would draw enormous currents and possibly cause damage.

Fortunately, it seems that a coin-cell battery will work perfectly. Once
the boards arrive I'll do some tests with the supercapacitor.

Many thanks to all for your help.

Cheers!
-Pete
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Attila Kinali
2015-03-15 22:40:23 UTC
Reply
Permalink
Raw Message
On Sun, 15 Mar 2015 11:05:53 -0400
Post by paul swed
I agree with Atilla from what I have seen. Its actually somewhat difficult
to measure this level of current.
It's not that difficult. You just need a good DMM. Standard ones
will not work as they have resolution limits in the range of 10-100uA.
A semi-decent Fluke handheld gets you already to 0.5uA. If you get
a reall DMM, then you will start worrying about the surface currents
on your device (FR4 has a quite low surface resistance of 10-100Mohm,
depending on humidity and whether you have any finger prints on it)
and anything you have in your setup (finger prints, small residues
of oily stuff, dirt,...)

The bigger problem is, that you have to make sure you have no floating
inputs anywhere (inputs are not clearly 0 or 1, will have more input leakage
current, additionally to the "shot trough" they cause inside the chip).
If you use diodes for decoupling parts of circuitry, you need those with
extremely low reverse current (in the low nA) and those will have high
forward voltages (>1V, the ones we used had iirc 2V). etc pp
Ultra low power electronics is probably as messy a field as ultra low
noise electronics.
Post by paul swed
But all is not lost. Even if the unit is
drawing 1-10ua because something is going wrong. Simply add a battery
holder and 2 X AAA or AA or ...
Please be aware that AA and AAA cells have a much higher self discharge
than coin cells. A CR2032 is usually of LiMnO2 chemistry, while AA/AAA's
are usually ZnMnO2. Also coin cells are optimized for long life times,
with very little current drawn, while most AA/AAA are not, or not as much.
Ie, i wouldn't expect an AAA cell, and much less an AA cell to last 10 years.

Attila Kinali
--
< _av500_> phd is easy
< _av500_> getting dsl is hard
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Charles Steinmetz
2015-03-16 02:49:23 UTC
Reply
Permalink
Raw Message
Post by Attila Kinali
A CR2032 is usually of LiMnO2 chemistry, while AA/AAA's
are usually ZnMnO2. Also coin cells are optimized for long life times,
with very little current drawn, while most AA/AAA are not, or not as much.
Ie, i wouldn't expect an AAA cell, and much less an AA cell to last 10 years.
The Energizer AA and AAA lithium primary batteries (Li/FeS2) have a
shelf life exceeding 10 years, as do their 9v batteries (Li/MnO2). I
have been mightily impressed with their performance on all counts.

Best regards,

Charles



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
paul swed
2015-03-16 02:23:41 UTC
Reply
Permalink
Raw Message
Atilla
Nor do I expect them to last 10 years, more like 1-2 years if the units
drawing the currents mentioned. Being external they are easy to change and
measure. Also cheap.
Only do this on boards that can't be replaced etc.
Regards
Paul
Post by Attila Kinali
On Sun, 15 Mar 2015 11:05:53 -0400
Post by paul swed
I agree with Atilla from what I have seen. Its actually somewhat
difficult
Post by paul swed
to measure this level of current.
It's not that difficult. You just need a good DMM. Standard ones
will not work as they have resolution limits in the range of 10-100uA.
A semi-decent Fluke handheld gets you already to 0.5uA. If you get
a reall DMM, then you will start worrying about the surface currents
on your device (FR4 has a quite low surface resistance of 10-100Mohm,
depending on humidity and whether you have any finger prints on it)
and anything you have in your setup (finger prints, small residues
of oily stuff, dirt,...)
The bigger problem is, that you have to make sure you have no floating
inputs anywhere (inputs are not clearly 0 or 1, will have more input leakage
current, additionally to the "shot trough" they cause inside the chip).
If you use diodes for decoupling parts of circuitry, you need those with
extremely low reverse current (in the low nA) and those will have high
forward voltages (>1V, the ones we used had iirc 2V). etc pp
Ultra low power electronics is probably as messy a field as ultra low
noise electronics.
Post by paul swed
But all is not lost. Even if the unit is
drawing 1-10ua because something is going wrong. Simply add a battery
holder and 2 X AAA or AA or ...
Please be aware that AA and AAA cells have a much higher self discharge
than coin cells. A CR2032 is usually of LiMnO2 chemistry, while AA/AAA's
are usually ZnMnO2. Also coin cells are optimized for long life times,
with very little current drawn, while most AA/AAA are not, or not as much.
Ie, i wouldn't expect an AAA cell, and much less an AA cell to last 10 years.
Attila Kinali
--
< _av500_> phd is easy
< _av500_> getting dsl is hard
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2015-03-18 08:28:11 UTC
Reply
Permalink
Raw Message
While for (optical/electrical) delay line oscillators, the way to go is to
add a frequency selective element, this is not done for ring oscillators.
So, how do people keep ring oscillators from oscillating at higher modes?
I think the answer is that you don't have to do anything. It takes care of
it by itself.

Suppose you have a long string of buffers and 1 inverter in a ring. Suppose
you start out with 3 transitions. That's the normal 1 transition with an
extra pulse. The key idea is that the edges don't propagate at exactly the
same speed. So one edge will catch up with another and they will self
destruct.

It would be fun to set that up and watch it on a scope. You could do that
with 3 NAND gates. Feed a reset signal into the other side of all 3 gates.
(watch the wire lengths) Maybe in a FPGA.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Tim Shoppa
2015-03-18 14:08:35 UTC
Reply
Permalink
Raw Message
The "modern digital" model of gates having inputs and outputs is in fact a
simplified case. It's unlikely that a digital logic student today would
ever have been exposed to gate elements that can work bidirectionally (I'm
not talking about tri-state, I'm talking about logic elements that have no
preferred input vs output). In general, even symmetrical ring oscillator
circuits using elements that do not have a preferred direction, will settle
down into rotating one way or the other depending on infinitesimal details
of initial conditions. See in particular the neon-light ring oscillator
here: http://donklipstein.com/sillyne2.html

This is an example of "symmetry breaking", a term I learned in quantum
electrodynamics class!, and the reason particles have mass! I don't
actually understand the Higgs Boson but I do understand that neon light
ring oscillator because I built it long before I took QED :-).

Tim N3QE
Post by Hal Murray
While for (optical/electrical) delay line oscillators, the way to go is
to
add a frequency selective element, this is not done for ring oscillators.
So, how do people keep ring oscillators from oscillating at higher modes?
I think the answer is that you don't have to do anything. It takes care of
it by itself.
Suppose you have a long string of buffers and 1 inverter in a ring.
Suppose
you start out with 3 transitions. That's the normal 1 transition with an
extra pulse. The key idea is that the edges don't propagate at exactly the
same speed. So one edge will catch up with another and they will self
destruct.
It would be fun to set that up and watch it on a scope. You could do that
with 3 NAND gates. Feed a reset signal into the other side of all 3 gates.
(watch the wire lengths) Maybe in a FPGA.
--
These are my opinions. I hate spam.
_______________________________________________
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2015-03-20 07:01:54 UTC
Reply
Permalink
Raw Message
Hmm... maybe the assumption that all edges walk around at the same speed is
wrong?
It's really really hard to make things like edges travel at exactly the same
speed. If it isn't exact, then one will eventually catch up with another and
self destruct.

The signal integrity wizards discuss eye patterns for multi gigabit serial
links. They now divide jitter into two parts: random and data-dependent. If
you have a long string of 0s as compared to a single 0 between 1s, the data
line will have a chance to get closer to a solid low. Starting from closer
to 0 takes slightly longer to make a transition. You can see it in the eye
diagram.


Does anybody have a scope on a ring oscillator? Is the signal symmetric? If
not, that says that the H-L transition travels at a different speed than the
L-H transition.

Actually, just looking at the prop times on a gate mignt be good enough. The
ring is just a handy signal generator.

It would be fun to make a ring with no inverters, inject a pulse, and watch
to see how long it lasts. I'll bet there is matastability type math that
depends on the width of the pulse. If you get the width exactly right it
will last a long time. Too long and it settles to all 1s. Too short and it
settles to all 0s.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Daniel Mendes
2015-03-21 02:08:28 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
It would be fun to make a ring with no inverters, inject a pulse, and
watch to see how long it lasts. I'll bet there is matastability type
math that depends on the width of the pulse. If you get the width
exactly right it will last a long time. Too long and it settles to all
1s. Too short and it settles to all 0s.
You mean in a coaxial cable in a loop? It would be very fun... more
points if you use a directional coupler to put the pulse in the loop.
Anyhow I doubt it would settle to 1 :)

Daniel

_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Attila Kinali
2015-03-22 14:16:18 UTC
Reply
Permalink
Raw Message
On Fri, 20 Mar 2015 21:26:44 -0700
You mean in a coaxial cable in a loop? It would be very fun... more points
if you use a directional coupler to put the pulse in the loop. Anyhow I
doubt it would settle to 1 :)
I was thinking of an amplifier in there someplace so the pulse wouldn't decay
simply due to the cable loss.
This is the description of how a delay line oscillator works.
While it is similar to a ring oscillator, there are certain things
that do not work exactly the same way. Rubiolas book[1, chapter 5]
contains a nice description of delay line oscillators and their
performance.

Attila Kinali

[1] "Phase Noise and Frequency Stability in Oscillators",
by Enrico Rubiola, 2008
--
< _av500_> phd is easy
< _av500_> getting dsl is hard
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2015-04-08 08:01:49 UTC
Reply
Permalink
Raw Message
The biggest problem would be to get the data into ntp in the right way, as I
am not sure whether ntp supports that kind of input.
If anybody ever needs help with ntp, feel free to poke me off list.

There are 2 ways to get data into ntp.
You can implement the kernel PPS API which basically reads a pair of time
stamps, one for when the signal changes from low to high and another for high
to low.

You can feed data in via the shared memory driver. GPSD uses this. Each
sample needs 2 time stamps, one for what the time should be and the second
for the OS/system time when the first time stamp arrived.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2016-05-04 16:30:50 UTC
Reply
Permalink
Raw Message
The limit for TDCs in FPGAs seems to be around 5-20ps RMS (which makes it
more like 15-50ps in "real" precision) depending on type and technology.
Going down to below 20ps usually means to take the latest tech FPGA with
lots of redundant structures, which makes the whole thing quite expensive.
FPGA prices generally scale with the amount of logic. How much do you need?
The smaller FPGAs are usually not expensive.

(Yes, the large ones can be very expensive.)
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2016-06-08 21:21:08 UTC
Reply
Permalink
Raw Message
Temperature, in an office or lab, does not change that much to cause large
differences.
Maybe in your lab.

I'd expect that will change as people get more sensitive to energy costs.
Things like turning down/off the heat/cooling at night can lead to large
swings.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Richard (Rick) Karlquist
2016-06-09 02:39:54 UTC
Reply
Permalink
Raw Message
I remember when we first got a prototype 10816
Mini-Rubidium standard working. We put in on
of those old paper strip chart recorders (this
was circa 1981). We were pretty cocky about
how it went straight down the page. You
couldn't do that with quartz. When we
came back the next day, you could clearly
see frequency steps when the air conditioning
went off at night and came on in the daytime.
People would say "nice thermometer, guys :-(

Rick
Post by Hal Murray
Temperature, in an office or lab, does not change that much to cause large
differences.
Maybe in your lab.
I'd expect that will change as people get more sensitive to energy costs.
Things like turning down/off the heat/cooling at night can lead to large
swings.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Mark Sims
2016-06-09 04:33:17 UTC
Reply
Permalink
Raw Message
When I was working on the temperature control feature of Lady Heather, I noticed that I could detect when ever I opened the refrigerator door (in the next room) or when I was in the same room as the Tbolt by looking at the EFC or temperature sensor plots... the Thunderbolt oscillator makes a nice thermometer. People are basically a 100 watt space heater.

Another time I was working on a precision temperature recorder (based upon an Analog Devices V/F chip). It could easily detect when a person walked into or left a (rather large) lab. You could even quantify the number of people in the lab (one particularly large guy counted as two people).

-----------------
Post by Richard (Rick) Karlquist
People would say "nice thermometer, guys :-(
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2016-06-26 01:17:58 UTC
Reply
Permalink
Raw Message
What determines the Q of a crystal? Is it atomic level impurities? Crystal
defects? ...

How has that changed over time? Is there a Moore's law for crystals?

How does the quality of crystals used for timing compare to the crystals used
for semiconductors?

Are there any other economically significant uses of high quality crystals?
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2016-07-22 19:28:08 UTC
Reply
Permalink
Raw Message
You can already get 24bit DAC's off the shelf from TI (DAC1282). I do not
know how stable they are in reality. ...
There are 2 markets for DACs and ADCs. I'll call them DC and RF, but the RF
goes down to audio.

In the DC market, the data sheet talks about linearity and usually covers
temperature stability.

In the RF market, the data sheet has Fourier transform plots when the input
is a clean sine wave or pair of sine waves. Think software radios, radar or
spread spectrum. The usual one term summary is ENOB: Effective Number of
Bits.

In the RF market, nobody cares about temperature drift.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Charles Steinmetz
2016-07-22 20:42:21 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
There are 2 markets for DACs and ADCs. I'll call them DC and RF, but the RF
goes down to audio.
In the DC market, the data sheet talks about linearity and usually covers
temperature stability.
* * *
In the RF market, nobody cares about temperature drift.
By far the largest market for 16+ bit converters (both ADC and DAC) is
digital audio. Virtually all current audio converters use sigma-delta
("one-bit") conversion with huge oversampling rates. For a variety of
reasons, these are quite well suited to the requirements of digital
audio, but they all have serious shortcomings WRT DC performance,
tempco, absolute accuracy, and the dreaded idle tones (Google for more
than you could read in a lifetime).

Fortunately, the sheer volume of this market means that you can get
really excellent audio (and audio-like) performance for very reasonable
cost. Unfortunately, it tempts designers to misuse the inexpensive
audio chips for general DAQ tasks (for example, the many glorified,
DC-coupled, "DAQ" PC sound cards and "instrument on a board" products
available today).

Best regards,

Charles


_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2016-07-26 22:38:08 UTC
Reply
Permalink
Raw Message
I am not sure you can apply this definition of Q onto earth. Q is defined
for harmonic oscillators (or oscillators that can be approximated by an
harmonic oscillator) but the earth isn't oscillating, it's rotating. While,
for time keeping purposes, similar in nature, the physical description of
both are different.
What do gyroscope-nuts use to describe the quality of their toys?
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Brooke Clarke
2016-07-26 23:00:03 UTC
Reply
Permalink
Raw Message
Hi Hal:

I resemble that remark.

Momentum and drift. It's interesting that the drift rate depends on the physical volume. See table at:
http://www.prc68.com/I/Sensors.shtml#Gyroscopic
http://www.prc68.com/I/Gyroscopes.html
--
Have Fun,

Brooke Clarke
http://www.PRC68.com
http://www.end2partygovernment.com/2012Issues.html
The lesser of evils is still evil.

-------- Original Message --------
Post by Hal Murray
I am not sure you can apply this definition of Q onto earth. Q is defined
for harmonic oscillators (or oscillators that can be approximated by an
harmonic oscillator) but the earth isn't oscillating, it's rotating. While,
for time keeping purposes, similar in nature, the physical description of
both are different.
What do gyroscope-nuts use to describe the quality of their toys?
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2016-08-23 06:56:12 UTC
Reply
Permalink
Raw Message
Is there any advantage of using groups.io compared to a traditional
mailinglist? If not, I would prefer a traditional mailinglist. But maybe I
am just oldfashioned :-)
The obvious advantage is that there is a professional staff keeping things
running so you don't need your own admin/wizard.

The part I'm missing is where do they get their cash? Are they adding ads?
(maybe only on the web version) Are they collecting data and selling it?
(This list is public so a lot of data is already available.)
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2016-12-30 11:06:15 UTC
Reply
Permalink
Raw Message
After that, it's just some simple math of calculating the difference between
the position of the stars and where you would have expecteded them at the
time when the picture has been taken.
You will probably have to correct for the distortion of the lens. That is a
line in the sky will have a slight curve on the pixels.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2017-03-22 18:24:19 UTC
Reply
Permalink
Raw Message
There have been a couple of discussions about doing GPSDOs using only analog
components in the past. People fare more knowledgable than me have commented
there on what the challenges would be and how to solve them. So I recommend
to go through the archives and look for those discussions. They might be a
little bit hidden, though.
I think the main problem is how to build a filter with a time constant of
many seconds. The better your OCXO the longer the time span you can
integrate over. 100s of seconds isn't an unreasonable target.

There may be other problems.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Bob kb8tq
2017-03-22 21:57:02 UTC
Reply
Permalink
Raw Message
Hi

As others have pointed out, a control loop at 100 seconds is more a gain spec than
an R/C time constant spec. The real issue is that you should have an integrator on
the loop and that *is* an R/C sort of thing. It’s also likely to have a much longer time
constant than the magic number for the loop.

Bob
Post by Hal Murray
There have been a couple of discussions about doing GPSDOs using only analog
components in the past. People fare more knowledgable than me have commented
there on what the challenges would be and how to solve them. So I recommend
to go through the archives and look for those discussions. They might be a
little bit hidden, though.
I think the main problem is how to build a filter with a time constant of
many seconds. The better your OCXO the longer the time span you can
integrate over. 100s of seconds isn't an unreasonable target.
There may be other problems.
--
These are my opinions. I hate spam.
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2017-06-08 07:41:29 UTC
Reply
Permalink
Raw Message
Keep in mind that you will need a good voltage reference as well to reach
anything close to 12bit. Your LDO is _not_ a good voltage reference
(depending on type and load/source conditions they vary eaily by 1-2%... not
to talk about their noise)
That applies when using a thermocouple. If using a bridge (or just a
divider), you can use the same voltage that powers the bridge as the
reference for the ADC and any drift cancels out.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Bob kb8tq
2017-06-08 10:57:41 UTC
Reply
Permalink
Raw Message
Hi
Post by Hal Murray
Keep in mind that you will need a good voltage reference as well to reach
anything close to 12bit. Your LDO is _not_ a good voltage reference
(depending on type and load/source conditions they vary eaily by 1-2%... not
to talk about their noise)
That applies when using a thermocouple. If using a bridge (or just a
divider), you can use the same voltage that powers the bridge as the
reference for the ADC and any drift cancels out.
If you go utterly crazy and get into the $1,000 and up “traceable standard”
thermistors or RTD’s just about everything matters :)

Bob
Post by Hal Murray
--
These are my opinions. I hate spam.
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2017-11-18 20:16:02 UTC
Reply
Permalink
Raw Message
I have a need for a GBIP adapter that I can use with Linux. It shouldn't be
too expensive, but I rather spend a few bucks more for ease of use. Where
"ease of use" means I don't have problems with weird drivers on Linux
I've been happy with the Prologix. It may not be as low cost as you would
like.

It uses one of the common USB-Serial chips, so there is no problem with
drivers on Linux.

It needs hardware flow control or long-enough pauses in the right places.
That was the only problem I had getting started.
--
These are my opinions. I hate spam.



_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Adrian Godwin
2017-11-18 20:21:13 UTC
Reply
Permalink
Raw Message
The Galvant adapter appears to use a very similar protocol to the Prologix,
but I'm unsure if it's exactly compatible.

There have been large numbers of HP adapters on ebay - they're generally
thought to be clones of varying quality.

http://www.galvant.ca/#!/store
http://www.eevblog.com/forum/testgear/flood-of-new-agilent-82357b-gpib-usb-adaptors-on-ebay-the-real-deal/150/
Post by Hal Murray
I have a need for a GBIP adapter that I can use with Linux. It shouldn't
be
too expensive, but I rather spend a few bucks more for ease of use. Where
"ease of use" means I don't have problems with weird drivers on Linux
I've been happy with the Prologix. It may not be as low cost as you would
like.
It uses one of the common USB-Serial chips, so there is no problem with
drivers on Linux.
It needs hardware flow control or long-enough pauses in the right places.
That was the only problem I had getting started.
--
These are my opinions. I hate spam.
_______________________________________________
To unsubscribe, go to https://www.febo.com/cgi-bin/
mailman/listinfo/time-nuts
and follow the instructions there.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Angus
2017-11-20 08:17:07 UTC
Reply
Permalink
Raw Message
Post by Adrian Godwin
The Galvant adapter appears to use a very similar protocol to the Prologix,
but I'm unsure if it's exactly compatible.
There have been large numbers of HP adapters on ebay - they're generally
thought to be clones of varying quality.
http://www.galvant.ca/#!/store
http://www.eevblog.com/forum/testgear/flood-of-new-agilent-82357b-gpib-usb-adaptors-on-ebay-the-real-deal/150/
Hi,

The commands in the Galvant one do seem to have been done to be quite
like the Prologix. I tried one of them 2 or 3 years ago because they
were temptingly cheap at the time, but could not get it working
consistently.

As far as I can remember, single commands and short strings would
often work for me, but longer strings and regular data like readings
every second were very unreliable. They seem to work for some folk but
not for others, which is often the story for cheap low volume ones.

In the end I gave up and bought a Prologix - which appears to be a
common end result of buying and wasting time on cheaper ones.

Angus.
_______________________________________________
time-nuts mailing list -- time-***@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hal Murray
2018-08-31 18:17:51 UTC
Reply
Permalink
Raw Message
I have somewhere a paper (which i cannot find currently, sorry) that used a
dish trained at one of the EGNOS satellites and used it as the only source
for timing. IIRC the results were promising, but not spectacular. The problem
being that all the ionospheric and tropospheric ...
There is another problem in that area. How accurately is the location of the
satellite known? published?

Geo-sync satellites actually wander around their nominal positions. How much
does that effect timing? I've seen figure-8 pictures of the pattern, but I
don't remember any data on elevation changes.
--
These are my opinions. I hate spam.




_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
jimlux
2018-08-31 18:46:56 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
I have somewhere a paper (which i cannot find currently, sorry) that used a
dish trained at one of the EGNOS satellites and used it as the only source
for timing. IIRC the results were promising, but not spectacular. The problem
being that all the ionospheric and tropospheric ...
There is another problem in that area. How accurately is the location of the
satellite known? published?
Geo-sync satellites actually wander around their nominal positions. How much
does that effect timing? I've seen figure-8 pictures of the pattern, but I
don't remember any data on elevation changes.
Their orbital elements are published and updated periodically - download
them off spacetrack or other sources.

Here's an example

The orbit data is extracted from the following two-line orbital elements,

1 43228U 18023A 18243.14094620 -.00000224 00000-0 00000+0 0 9995
2 43228 0.0377 58.1839 0003405 72.7364 229.1178 1.00272082 1950
Epoch (UTC): 31 August 2018 03:22:57
Eccentricity: 0.0003405
inclination: 0.0377°
perigee height: 35772 km
apogee height: 35800 km
right ascension of ascending node: 58.1839°
argument of perigee: 72.7364°
revolutions per day: 1.00272082
mean anomaly at epoch: 229.1178°
orbit number at epoch: 195
_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
Peter Laws
2018-08-31 18:42:09 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
There is another problem in that area. How accurately is the location of the
satellite known? published?
Which satellites? Most are here: https://celestrak.com Tracking
programs are abundant.

Your GPS receiver knows where the whole Navstar (or other GNSS)
constellation is because it has to else it can't work.

--
Peter Laws | N5UWY | plaws plaws net | Travel by Train!

_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Magnus Danielson
2018-09-01 07:12:23 UTC
Reply
Permalink
Raw Message
Hi Hal,
Post by Hal Murray
I have somewhere a paper (which i cannot find currently, sorry) that used a
dish trained at one of the EGNOS satellites and used it as the only source
for timing. IIRC the results were promising, but not spectacular. The problem
being that all the ionospheric and tropospheric ...
There is another problem in that area. How accurately is the location of the
satellite known? published?
Geo-sync satellites actually wander around their nominal positions. How much
does that effect timing? I've seen figure-8 pictures of the pattern, but I
don't remember any data on elevation changes.
The figure of 8 is just one of the artifacts of the elliptic orbit, and
it does cause distance changes and hence affect timing. Correct
ephimeris data is needed for good compensation. We do that for GPS
satellites too, which is part of the systems capability that the
receiver can do these compensations unaided with anything but the
signals from space segment.

Cheers,
Magnus

_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Gerhard Hoffmann
2018-09-01 09:31:29 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
I have somewhere a paper (which i cannot find currently, sorry) that used a
dish trained at one of the EGNOS satellites and used it as the only source
for timing. IIRC the results were promising, but not spectacular. The problem
being that all the ionospheric and tropospheric ...
There is another problem in that area. How accurately is the location of the
satellite known? published?
Geo-sync satellites actually wander around their nominal positions. How much
does that effect timing? I've seen figure-8 pictures of the pattern, but I
don't remember any data on elevation changes.
Geo-stat sats usually do their navigation via their linear transponders,
the rest is in some ground stations, so the added cost for flight hardware
is essentially nil.

The operators know the position of their sats quite precisely since,
for a phone sat as an example, the ground station transmission timing
must be aligned quite carefully to avoid both overlapping and idle time
of the channel. They switch both between cities and to give the phone
user the illusion of a continuous 2 way connection without too much delay.

The absolute position is less important as long as it is known.

Small countries like Luxembourg have just one geostationary parking lot
but operate several sats. They may have a more pronounced need to
keep the positions precise.

Overly precise position shortens the lifetime of a sat since it eats up
fuel.

The navigation is simply made by PN streams say > 20 dB below the
MPEG data.

I have made the PN generators,  bit / frame generators /synchronizers,
correlators, de/modulators for some of them.

I don't think that the exact position data is published. It is a closed
system, after all.

But a bunch of hams with enough criminal energy could probably
measure it for themselves. The down link is already there in every
household with a SAT TV.  Oh, no, I do not promote that!

best regards,
Gerhard




_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the
Mark Sims
2018-08-31 19:03:39 UTC
Reply
Permalink
Raw Message
A GPS receiver that supports SBAS, etc will tell you where the sats are. Some only report to 1 degree, others 0.1 to 0.01 degrees resolution. The beam with of a small dish at GPS freqs is not all that narrow.

Using orbital elements or processing the GNSS ephemeris message will give you a result with quite a bit better resolution.

-----------------
Post by Hal Murray
There is another problem in that area. How accurately is the location of the
satellite known? published?
_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Hal Murray
2018-09-01 05:23:14 UTC
Reply
Permalink
Raw Message
Which satellites? Most are here: https://celestrak.com Tracking programs
are abundant.
I was thinking of geo-sync, but if position info is widely available then I
guess I should be interested in any satellites that provide good time. (maybe
an are likely to continue providing good time after some nasty event)
--
These are my opinions. I hate spam.




_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
David I. Emery
2018-09-01 18:36:55 UTC
Reply
Permalink
Raw Message
Post by Hal Murray
Which satellites? Most are here: https://celestrak.com Tracking programs
are abundant.
I was thinking of geo-sync, but if position info is widely available then I
guess I should be interested in any satellites that provide good time. (maybe
an are likely to continue providing good time after some nasty event)
I hate to intrude into such an august group of experts, but as a
practical matter what sort of real world timing accuracy could one
really expect from using just ONE of the WAAS birds and a dedicated L
band dish ?

I do understand the WAAS satellites do transmit a L1 carrier
with a full and correct GPS signal and ephemerides for the WAAS
satellite in addition to their other correction information, so I
presume one could use single satellite timing mode with a variety of GPS
timing receivers that support this... given of course a suitably
accurately surveyed ground position of the dish...

There are obviously SOME degree of ionospheric corrections
possible from the WAAS data... how much good do they do in practice
for this special case of single satellite timing ?
--
Dave Emery N1PRE/AE, ***@dieconsulting.com DIE Consulting, Weston, Mass 02493
"An empty zombie mind with a forlorn barely readable weatherbeaten
'For Rent' sign still vainly flapping outside on the weed encrusted pole - in
celebration of what could have been, but wasn't and is not to be now either."


_______________________________________________
time-nuts mailing list -- time-***@lists.febo.com
To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.
Loading...