Empathy List Archives

PS

Pete Stephenson

Fri, Jul 14, 2017 9:55 AM

Hi all,

I use some DS3231 temperature-compensated real-time clocks with some
Raspberry Pis, particularly those that might not always have internet
access. About a year ago I wrote some code [1] to characterize the
behavior of these particular chips using my Thunderbolt as a reference,
using the Arduino/atmega328 as a glorified counter.

As a somewhat longer-term test, I set the time on one of the DS3231s to
the correct time using GPS-synced NTP on one of the Pis, then set the
whole thing on the shelf for a bit over a year and forgot about it. If
relevant, the only power source was the CR2032 battery.

I checked it today, and the clock had drifted 16 seconds since June 6th
of 2016 to now. That works out to around 0.5 ppm drift over that time.
The chip is specced to +/- 2ppm. Not bad for a cheap module of
potentially dubious provenance from eBay. For those who are curious, I'd
be happy to provide a link to the specific item I purchased; contact me
off-list for details.

Cheers!
-Pete

[1] https://github.com/heypete/Frequency_Counter_32kHz

--
Pete Stephenson

Hi all, I use some DS3231 temperature-compensated real-time clocks with some Raspberry Pis, particularly those that might not always have internet access. About a year ago I wrote some code [1] to characterize the behavior of these particular chips using my Thunderbolt as a reference, using the Arduino/atmega328 as a glorified counter. As a somewhat longer-term test, I set the time on one of the DS3231s to the correct time using GPS-synced NTP on one of the Pis, then set the whole thing on the shelf for a bit over a year and forgot about it. If relevant, the only power source was the CR2032 battery. I checked it today, and the clock had drifted 16 seconds since June 6th of 2016 to now. That works out to around 0.5 ppm drift over that time. The chip is specced to +/- 2ppm. Not bad for a cheap module of potentially dubious provenance from eBay. For those who are curious, I'd be happy to provide a link to the specific item I purchased; contact me off-list for details. Cheers! -Pete [1] https://github.com/heypete/Frequency_Counter_32kHz -- Pete Stephenson

CK

Craig Kirkpatrick

Fri, Jul 14, 2017 4:36 PM

I set a DS3231 to correct time of WWVB in January 2017 and it has not gained or lost one second in over six months!

The clock chip is part of a project board that I have running on power during the day and powered off overnight meaning that for about half the day the DS3231 is keeping time using just the watch cell backup battery. Temperatures have varied in the range of 60F to about 85F.

This performance is impressive for an Integrated circuit chip costing under ten dollars.

I bought a second DS3231 and I'm thinking of a simple experiment project to compare two of the chips and how they drift relative to each other just for fun.

Best Wishes,
Craig K.

On Jul 14, 2017, at 2:55 AM, Pete Stephenson pete@heypete.com wrote:

Hi all,

I use some DS3231 temperature-compensated real-time clocks with some
Raspberry Pis, particularly those that might not always have internet
access. About a year ago I wrote some code [1] to characterize the
behavior of these particular chips using my Thunderbolt as a reference,
using the Arduino/atmega328 as a glorified counter.

As a somewhat longer-term test, I set the time on one of the DS3231s to
the correct time using GPS-synced NTP on one of the Pis, then set the
whole thing on the shelf for a bit over a year and forgot about it. If
relevant, the only power source was the CR2032 battery.

I checked it today, and the clock had drifted 16 seconds since June 6th
of 2016 to now. That works out to around 0.5 ppm drift over that time.
The chip is specced to +/- 2ppm. Not bad for a cheap module of
potentially dubious provenance from eBay. For those who are curious, I'd
be happy to provide a link to the specific item I purchased; contact me
off-list for details.

Cheers!
-Pete

[1] https://github.com/heypete/Frequency_Counter_32kHz

--
Pete Stephenson

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

I set a DS3231 to correct time of WWVB in January 2017 and it has not gained or lost one second in over six months! The clock chip is part of a project board that I have running on power during the day and powered off overnight meaning that for about half the day the DS3231 is keeping time using just the watch cell backup battery. Temperatures have varied in the range of 60F to about 85F. This performance is impressive for an Integrated circuit chip costing under ten dollars. I bought a second DS3231 and I'm thinking of a simple experiment project to compare two of the chips and how they drift relative to each other just for fun. Best Wishes, Craig K. On Jul 14, 2017, at 2:55 AM, Pete Stephenson <pete@heypete.com> wrote: Hi all, I use some DS3231 temperature-compensated real-time clocks with some Raspberry Pis, particularly those that might not always have internet access. About a year ago I wrote some code [1] to characterize the behavior of these particular chips using my Thunderbolt as a reference, using the Arduino/atmega328 as a glorified counter. As a somewhat longer-term test, I set the time on one of the DS3231s to the correct time using GPS-synced NTP on one of the Pis, then set the whole thing on the shelf for a bit over a year and forgot about it. If relevant, the only power source was the CR2032 battery. I checked it today, and the clock had drifted 16 seconds since June 6th of 2016 to now. That works out to around 0.5 ppm drift over that time. The chip is specced to +/- 2ppm. Not bad for a cheap module of potentially dubious provenance from eBay. For those who are curious, I'd be happy to provide a link to the specific item I purchased; contact me off-list for details. Cheers! -Pete [1] https://github.com/heypete/Frequency_Counter_32kHz -- Pete Stephenson _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.

BC

Brooke Clarke

Fri, Jul 14, 2017 4:56 PM

Hi Pete:

AFAICR the DS3231 has a software fine tune. Did you set that before the test?

--
Have Fun,

Brooke Clarke
http://www.PRC68.com
http://www.end2partygovernment.com/2012Issues.html

-------- Original Message --------

Hi all,

I use some DS3231 temperature-compensated real-time clocks with some
Raspberry Pis, particularly those that might not always have internet
access. About a year ago I wrote some code [1] to characterize the
behavior of these particular chips using my Thunderbolt as a reference,
using the Arduino/atmega328 as a glorified counter.

As a somewhat longer-term test, I set the time on one of the DS3231s to
the correct time using GPS-synced NTP on one of the Pis, then set the
whole thing on the shelf for a bit over a year and forgot about it. If
relevant, the only power source was the CR2032 battery.

I checked it today, and the clock had drifted 16 seconds since June 6th
of 2016 to now. That works out to around 0.5 ppm drift over that time.
The chip is specced to +/- 2ppm. Not bad for a cheap module of
potentially dubious provenance from eBay. For those who are curious, I'd
be happy to provide a link to the specific item I purchased; contact me
off-list for details.

Cheers!
-Pete

[1] https://github.com/heypete/Frequency_Counter_32kHz

Hi Pete: AFAICR the DS3231 has a software fine tune. Did you set that before the test? -- Have Fun, Brooke Clarke http://www.PRC68.com http://www.end2partygovernment.com/2012Issues.html -------- Original Message -------- > Hi all, > > I use some DS3231 temperature-compensated real-time clocks with some > Raspberry Pis, particularly those that might not always have internet > access. About a year ago I wrote some code [1] to characterize the > behavior of these particular chips using my Thunderbolt as a reference, > using the Arduino/atmega328 as a glorified counter. > > As a somewhat longer-term test, I set the time on one of the DS3231s to > the correct time using GPS-synced NTP on one of the Pis, then set the > whole thing on the shelf for a bit over a year and forgot about it. If > relevant, the only power source was the CR2032 battery. > > I checked it today, and the clock had drifted 16 seconds since June 6th > of 2016 to now. That works out to around 0.5 ppm drift over that time. > The chip is specced to +/- 2ppm. Not bad for a cheap module of > potentially dubious provenance from eBay. For those who are curious, I'd > be happy to provide a link to the specific item I purchased; contact me > off-list for details. > > Cheers! > -Pete > > [1] https://github.com/heypete/Frequency_Counter_32kHz >

PS

Pete Stephenson

Sat, Jul 15, 2017 12:27 AM

Hi Brooke,

No. I did not perform any software fine-tuning. My earlier tests using
my Arduino/Thunderbolt-based testing setup indicated its short-term
drift was <1ppm, so it didn't think it'd be worthwhile. After over a
year of testing, I'm even more impressed.

In addition, I forgot to mention that the DS3231 in question was in an
indoor environment but not one with particularly stable temperatures: in
winter, indoor temperatures dropped to around 15C, while in summer it'd
get up to 30C, with direct sunlight from the window shining directly on
the Raspberry Pi enclosure holding the device for several hours each
morning.

Cheers!
-Pete

--
Pete Stephenson

On Fri, Jul 14, 2017, at 06:56 PM, Brooke Clarke wrote:

Hi Pete:

AFAICR the DS3231 has a software fine tune. Did you set that before the
test?

--
Have Fun,

Brooke Clarke
http://www.PRC68.com
http://www.end2partygovernment.com/2012Issues.html

-------- Original Message --------

Hi all,

I use some DS3231 temperature-compensated real-time clocks with some
Raspberry Pis, particularly those that might not always have internet
access. About a year ago I wrote some code [1] to characterize the
behavior of these particular chips using my Thunderbolt as a reference,
using the Arduino/atmega328 as a glorified counter.

As a somewhat longer-term test, I set the time on one of the DS3231s to
the correct time using GPS-synced NTP on one of the Pis, then set the
whole thing on the shelf for a bit over a year and forgot about it. If
relevant, the only power source was the CR2032 battery.

I checked it today, and the clock had drifted 16 seconds since June 6th
of 2016 to now. That works out to around 0.5 ppm drift over that time.
The chip is specced to +/- 2ppm. Not bad for a cheap module of
potentially dubious provenance from eBay. For those who are curious, I'd
be happy to provide a link to the specific item I purchased; contact me
off-list for details.

Cheers!
-Pete

[1] https://github.com/heypete/Frequency_Counter_32kHz

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

Hi Brooke, No. I did not perform any software fine-tuning. My earlier tests using my Arduino/Thunderbolt-based testing setup indicated its short-term drift was <1ppm, so it didn't think it'd be worthwhile. After over a year of testing, I'm even more impressed. In addition, I forgot to mention that the DS3231 in question was in an indoor environment but not one with particularly stable temperatures: in winter, indoor temperatures dropped to around 15C, while in summer it'd get up to 30C, with direct sunlight from the window shining directly on the Raspberry Pi enclosure holding the device for several hours each morning. Cheers! -Pete -- Pete Stephenson On Fri, Jul 14, 2017, at 06:56 PM, Brooke Clarke wrote: > Hi Pete: > > AFAICR the DS3231 has a software fine tune. Did you set that before the > test? > > -- > Have Fun, > > Brooke Clarke > http://www.PRC68.com > http://www.end2partygovernment.com/2012Issues.html > > -------- Original Message -------- > > Hi all, > > > > I use some DS3231 temperature-compensated real-time clocks with some > > Raspberry Pis, particularly those that might not always have internet > > access. About a year ago I wrote some code [1] to characterize the > > behavior of these particular chips using my Thunderbolt as a reference, > > using the Arduino/atmega328 as a glorified counter. > > > > As a somewhat longer-term test, I set the time on one of the DS3231s to > > the correct time using GPS-synced NTP on one of the Pis, then set the > > whole thing on the shelf for a bit over a year and forgot about it. If > > relevant, the only power source was the CR2032 battery. > > > > I checked it today, and the clock had drifted 16 seconds since June 6th > > of 2016 to now. That works out to around 0.5 ppm drift over that time. > > The chip is specced to +/- 2ppm. Not bad for a cheap module of > > potentially dubious provenance from eBay. For those who are curious, I'd > > be happy to provide a link to the specific item I purchased; contact me > > off-list for details. > > > > Cheers! > > -Pete > > > > [1] https://github.com/heypete/Frequency_Counter_32kHz > > > > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there.

time-nuts@lists.febo.com

DS3231 drift over a year