Discussion:
A look inside the DS3231
(too old to reply)
Didier Juges
2017-07-30 13:53:51 UTC
Permalink
Raw Message
That device also has analog circuitry for the oscillator itself and the
temperature sensor and the temperature compensation.
I believe I have read an app note some time ago, it may have been from
Maxim describing a kind of ring oscillator being used as a temperature
sensor which drew much less power than a bandgap or a PN junction and
directly produced a digital output.
The DAC itself, or whatever circuit they use for temp compensation also has
analog components and must use pico power.
Quite amazing.
On Sun, 30 Jul 2017 12:23:17 +0200
- I find it remarkable that this circuit can operate on less than a
microamp during normal usage, including temperature conversion.
That's not so remarkable. If you make the transistors long, then
you get very low leakage. Couple that with small clock frequency
and you use very little current. Modern ICs only use so much current
because they have so many transistors, which are also optimized
for being fast, rather then low leakage.
Good point! I admit the details of optimizing transistors for different
purposes is beyond my ken, and I appreciate the insight.
There are multiple optimization points. One is to select a prodcution
process that is optimized for low leakage. I.e. thick gate oxide
and high threshold voltage. Both of these parameters imply higher
suplly voltage.
Then, in the design, you make your transistors long and large.
The problem here is, that power consumption scales proportional
to the square of supply voltage, the gate capacitance and the
switching frequency. This means, if you choose a low leakage
process, and thus high supply voltage, your power consumtion
will go up. The same goes for choosing large transistors.
Hence it becomes a trade-off between static (leakage) and
dynamic (through gate capacitance) power consumption.
The DS3231 has on-board temperature monitoring to correct the crystal
frequency: is this something where they would have bothered putting a
separate sensor next to the crystal itself, or are the die and the
crystal are close enough and in the same package that they could use
an
on-die sensor like a diode and call that "good enough"?
My guess would be that it's a PN-junction or a bandgap temperature
sensor somewhere on the chip. Adding another part increases the cost
of production quite considerably.
Indeed. At first glance, I was surprised not to see tiny discrete
capacitors within the chip package itself, as I assumed (incorrectly)
that getting sufficient capacitance to steer a crystal a little would
require larger capacitors than could be easily put on a die, but then I
remembered that each LSB in the aging register only changes the
frequency by 0.1ppm at 25C, so that wouldn't need a large amount of
capacitance.
As a rule of thumb, you can assume that in an "old" (aka large node size)
process the gate capacitance is approximately 1nF per mm^2. So, you can
build quite easily 10-100pF of capacitors on-chip.
Attila Kinali
--
You know, the very powerful and the very stupid have one thing in common.
They don't alters their views to fit the facts, they alter the facts to
fit the views, which can be uncomfortable if you happen to be one of the
facts that needs altering. -- The Doctor
_______________________________________________
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.
Tim Shoppa
2017-07-30 12:15:57 UTC
Permalink
Raw Message
On the subject of low-current 32kHz oscillators:

DS3231 spec says typical 1uA for timekeeping and circa 600uA for
temperature conversion. I understand they periodicailly kick the
temperature conversion on but only for extremely short duty cycles and this
is included in the 1uA.

Standard DS12887 spec was 500nA with the oscillator and counter logic
running. This did not have any temperature conversion/compensation.

RCA published a 4007-based 32kHz oscillator that was circa 1uA but I think
that spec was at 1.5V. RCA got a patent on putting a resistor in the drain
of the first stage to slow it down and reduce power consumption to get down
to 1uA. So in the DS12887, Dallas figured out how to go at least a factor
of two lower in power. I would imagine there's a series of patents by watch
companies on this subject as well probably all back in the 1970's and
1980's.

Tim N3QE
Looks like it still says "DALLAS SEMICONDUCTOR" to the left of Maxim.
Maybe Maxim only wanted to change the mask enough to find some empty
space to sign it?
It does indeed say "DALLAS SEMICONDUCTOR".
I managed to get some high-quality photos using the microscope's
on-board camera and have updated the photo album at
https://imgur.com/a/0zudj with the newest ones (they're the
all-rectangular photos below the two circular photos). There's some
high-resolution composite images.
- There's a section just above the "Maxim" part that has several
snippets of text ("17A3", "16A3", etc.). In normal light, each of these
bits of text is a different color, where the colors correspond to
different layers of the chip. Each bit of text has a different depth of
focus, indicating they're physically closer or further from the lens.
Does anyone know what material the colors might correspond to?
- There's several square grids of circles-in-squares circuit elements. I
have no idea what these are.
- I find it remarkable that this circuit can operate on less than a
microamp during normal usage, including temperature conversion.
The DS3231 has on-board temperature monitoring to correct the crystal
frequency: is this something where they would have bothered putting a
separate sensor next to the crystal itself, or are the die and the
crystal are close enough and in the same package that they could use an
on-die sensor like a diode and call that "good enough"?
Cheers!
-Pete
--
Pete Stephenson
_______________________________________________
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.
Attila Kinali
2017-07-30 12:29:24 UTC
Permalink
Raw Message
On Sun, 30 Jul 2017 08:15:57 -0400
Post by Tim Shoppa
I would imagine there's a series of patents by watch
companies on this subject as well probably all back in the 1970's and
There are also a lot of papers and books. I can recommend those
written by Eric Vittoz, who was the mastermind behind quite a
few of the oscillator circuits of the Swiss watch industry.
In particular his book on low power oscillators[1]. He also
wrote a review of the history of low power electronics in
the watch industry about 10 years ago[2], which is also very
much worth a read.

Attila Kinali


[1] "Low-Power Crystal and MEMS Oscillators", by Eric Vittoz, 2010
[2] "The Electronic Watch and Low-Power Circuits", by Eric Vittoz, 2008
https://doi.org/10.1109/N-SSC.2008.4785777
--
You know, the very powerful and the very stupid have one thing in common.
They don't alters their views to fit the facts, they alter the facts to
fit the views, which can be uncomfortable if you happen to be one of the
facts that needs altering. -- The Doctor
_______________________________________________
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
2017-07-28 23:05:19 UTC
Permalink
Raw Message
On Thu, Jul 27, 2017, at 09:31 PM, Pete Stephenson wrote:
[snip]
Anyway, the photos are available at http://imgur.com/a/0zudj -- I will
add more photos from the petrographic microscope tomorrow. I focused
mainly on the markings on the die that indicated it was, in fact, a
Maxim chip but if there's any other region of the chip that you'd like
images of, please let me know and I'd be happy to take some more
pictures.
Hi all,

Just a quick update: I was able to look at the DS3231 at work at the
quality of the (very expensive) Zeiss microscope is dramatically better
than my $20 USB microscope at home. No surprise.

Unfortunately, due to the ancient Canon camera attached to the
microscope not being compatible with Windows 7 or Linux, I was unable to
get any high-quality photos at this time. The camera is normally used in
tethered mode with no CF card, with the camera connected to the user's
laptop. Most of my colleagues use Macs, which evidently do work with it
but I wasn't able to ask any of them today before they all left. I've
ordered a CF-to-SD adapter that should allow me to take photos without
any issues, but it will be a few weeks until it arrives. Once it's
arrived, I'll take some more photos of the chip and let people know.

I've taken a few photos with my smartphone through the microscope's
eyepiece, but they turned out quite poorly as you can see below. When
viewed directly via the eyepiece, the appearance of the chip is quite
stunning.

On a related note, the reflected differential interference contrast
(DIC) filters on the microscope make looking at multi-layer chips
dramatically more clear and interesting. Compare
http://imgur.com/7nuTooL , which was taken with with no optical
filtering using standard reflected light illumination and
http://imgur.com/P6HL9MB which was taken of a different area of the chip
using reflected DIC. The colors are different, of course, but the
contrast between elements of the chip is much improved.

If anyone has any chips they'd like me to examine under the microscope,
let me know and I'd be happy to do so.

Cheers!
-Pete
--
Pete Stephenson
_______________________________________________
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
2017-07-30 22:47:36 UTC
Permalink
Raw Message
Lower drain means a smaller battery or backup source... and in today's world of electronics smaller is better. But, past a certain point, it all boils down to a "spec waving" contest ;-)

----------------
What's the motivation for this, other than "because we can"? Aren't
existing RTC chips capable of running 10+ years from a lithium coin cell
already, to the point where the cell's self-discharge is the limiting
factor?
_______________________________________________
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
2017-07-30 16:47:46 UTC
Permalink
Raw Message
On Sun, 30 Jul 2017 08:53:51 -0500
Post by Didier Juges
I believe I have read an app note some time ago, it may have been from
Maxim describing a kind of ring oscillator being used as a temperature
sensor which drew much less power than a bandgap or a PN junction and
directly produced a digital output
Right. I always forget that there are these "digital" sensors
that are much less power hungry than the "analog" ones.


Attila Kinali
--
You know, the very powerful and the very stupid have one thing in common.
They don't alters their views to fit the facts, they alter the facts to
fit the views, which can be uncomfortable if you happen to be one of the
facts that needs altering. -- The Doctor
_______________________________________________
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
2017-07-30 13:37:22 UTC
Permalink
Raw Message
A friend of mine is an engineer for one of the biggest manufacturers of clock chips and has worked quite a bit on their clock chips and is quite familiar with the issues of building consistent ultra low power oscillators in a production product. Getting nanowatt (and now sub-nanowatt) level oscillators to do their thing consistently is not easy. Getting them to do it with customer supplied crystals is a big thing. Variations by the crystal maker regularly cause previously working products to stop working. Also they are notoriously sensitive to PCB layout issues. Older, higher power clock chips don't have nearly as many problems as the newer ultra low power designs. Competition to see who can make the lowest power clock chips seems to be one of the biggest drivers for new clock chip designs.

Oh, and although the clock chip oscillators have good long term accuracy they tend to have lots of jitter and poor ADEVs.
_______________________________________________
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
2017-07-30 19:21:42 UTC
Permalink
Raw Message
Post by Mark Sims
A friend of mine is an engineer for one of the biggest manufacturers of
clock chips and has worked quite a bit on their clock chips and is quite
familiar with the issues of building consistent ultra low power
oscillators in a production product. Getting nanowatt (and now
sub-nanowatt) level oscillators to do their thing consistently is not
easy. Getting them to do it with customer supplied crystals is a big
thing. Variations by the crystal maker regularly cause previously
working products to stop working. Also they are notoriously sensitive to
PCB layout issues. Older, higher power clock chips don't have nearly as
many problems as the newer ultra low power designs. Competition to see
who can make the lowest power clock chips seems to be one of the biggest
drivers for new clock chip designs.
What's the motivation for this, other than "because we can"? Aren't
existing RTC chips capable of running 10+ years from a lithium coin cell
already, to the point where the cell's self-discharge is the limiting
factor?

Is there some application where exceptionally low power use for a clock
chip would be of interest?

I ask as an interested amateur not familiar with the subtleties of such
designs.

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.
Graham / KE9H
2017-07-30 19:41:16 UTC
Permalink
Raw Message
The typical method of frequency correction is not to add or subtract
capacitance across the crystal, (like an old analog engineer would do) but
rather to add or subtract pulses to the stream of cycles/pulses coming out
of the crystal oscillator. More the kind of correction a digital engineer
would do. The long term end result is correct, but the addition and
subtraction of correction pulses shows up as jitter and short term errors.

As far as the need for low battery drain, everything is going for smaller,
lighter, cheaper, portable, and runs off a battery. You would probably
turn your nose up at a watch, whose battery did not last at least a year.
Most simple watches go several years. Now put an electronic display on it,
and a GPS in it and BlueTooth LE. Everything inside is under pressure to
make sure the battery lasts as long as possible.

I personally don't wear a watch any more. Get GPS time from my cellphone,
that fits in my watch pocket of my jeans. (I finally found a use for that
pocket, after wearing pants with them for 50+ years.)

But, I have to charge the thing every day, every other day, at the most.

What I want, is a cellphone that I only have to charge once a week, or once
a month.

I don't want to have to be in the battery management business.

Before we exited from the pager business, we had a customer that had a
published goal of a pager that would run an entire year on one "AA" primary
(alkaline throw-away) battery. Imagine changing the battery once per year,
on New Years day. We were up to running for nine months from a single AA
battery. Now, that was a radio receiver and a 'beeper' (and an internal
clock for management purposes.). And we had a road map to get to the full
year, but the cellphone systems killed the pager business first. It was all
about timing, and putting as much of the IC to sleep at any given time, as
you could.
Post by Mark Sims
A friend of mine is an engineer for one of the biggest manufacturers of
clock chips and has worked quite a bit on their clock chips and is quite
familiar with the issues of building consistent ultra low power oscillators
in a production product. Getting nanowatt (and now sub-nanowatt) level
oscillators to do their thing consistently is not easy. Getting them to
do it with customer supplied crystals is a big thing. Variations by the
crystal maker regularly cause previously working products to stop working.
Also they are notoriously sensitive to PCB layout issues. Older, higher
power clock chips don't have nearly as many problems as the newer ultra low
power designs. Competition to see who can make the lowest power clock
chips seems to be one of the biggest drivers for new clock chip designs.
Oh, and although the clock chip oscillators have good long term accuracy
they tend to have lots of jitter and poor ADEVs.
_______________________________________________
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.

Loading...