32 kHz output of Maxim DS3231* RTC

[Paolo Cravero]

Hello.

One more thread on the subject of knock-off components...

I use cheap RTC modules that use a DS3231 RTC chip (ZS-04). They are satisfactorily accurate and I haven't had surprises in their internal calendar.

Today I measured the 32K output, which is supposed to be 32768 Hz, even though the datasheet often mentions 32K ... that I would read as 32000 Hz.

Three out of four modules measured 32699 +/- 2 Hz, and just one was at 32768 Hz (500 MHz counter with OCXO, 10 seconds gate time and Nixies ;) ). This would mean loosing 179 seconds a day, which is not the case.

One more info: the three slow guys are DS3231M, while the 32768 one is DS3231SN. I've never noticed this difference before because I use the 1 Hz output and both have it.

Has someone measured the 32k output of their modules and/or can help me understand why the slow guys still keep up with the time?

Paolo

[David Pye]

So.

The DS3231M datasheet specifically says it's internal clock output is 32.768kHz (https://datasheets.maximintegrated.com/en/ds/DS3231M.pdf)

whereas the other datasheet for SN (https://docs-emea.rs-online.com/webdocs/0f6e/0900766b80f6ec88.pdf) says its' clock should be 32KHz.

So the two chips are supposed to have different internal oscillator frequencies, but the datasheets suggest the other way round to what your measurements suggest.

David

--
You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/CABj2VaZ13z8DQsoBTp0o5jei945aTdtjJ7aJ3hYq%3DkUiPJAJ_w%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Paolo Cravero]

Actually Maxim uses both 32K and 32.768 kHz in the same document, even paragraph:

Bit 3: Enable 32kHz Output (EN32kHz).This bit con-trols the status of the 32kHz pin. When set to logic 1, the32kHz pin is enabled and outputs a 32.768kHz square-wave signal. When set to logic 0, the 32kHz pin goes toa high-impedance state. The initial power-up state ofthis bit is logic 1, and a 32.768kHz square-wave signalappears at the 32kHz pin after a power source isapplied to the DS3231 (if the oscillator is running).

Taken from the DS3231(SM) datasheet. I think it must be 32768 Hz in all cases because it is a power of 2 and can be divided down to 1 Hz. Search for "768" in both datasheets!

Paolo

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/CAOQ6x0Hm38dGSanfnvDKDKbeX6MGDZ%3DdrThb3mC2fXHvBEuhgw%40mail.gmail.com.

[Dekatron42]

I’ve used both versions to divide the 32768Hz signal down to 1Hz and I’ve never seen one which didn’t output 32768Hz on that pin. I usually use a pullup of som1-10k and then use that signal with anything from CMOS to TTL or microprocessors like Arduinos and I’ve always gotten a very stable signal with just a few soconds drift per year without compensating the factory adjustment. Ibought mine at Digikey and Mouser and other reliable sources.

/Martin

[David Pye]

Interesting...

Fair point!   Perhaps some of your ICs are non-genuine, and have different speed oscillators?  It seems very hard to explain!

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/CABj2VaYxJD6TJwuz3xV16WnJPB5icNNQk-vLLsEJttVMc%2ByOaQ%40mail.gmail.com.

[Nick]

Yep. Genuine ones are very precise

Where did you source yours from?

Nick

[Paolo Cravero]

On Mon, Feb 11, 2019 at 10:15 PM Nick <gerbilp...@gmail.com> wrote:

Yep. Genuine ones are very precise

Where did you source yours from?

I got them from Aliexpress (2 pcs in 8/2016 then 3 pcs in 7/2017) and Banggood (3 pcs in 11/2018).

I think the batch of 3 from Aliexpress that came without battery had DS3231SN, because I have three of them, but I did not keep so detailed records.

OTOH I noticed that the ZS-042 board with DS3231SN is a deeper blue color than those with DS3231M. Coincidence? Interestingly the only DS3231M in service, outputting 32701 Hz, exhibits good precision, gaining maybe 5 seconds/month, and it is not running at 25°C in there. I use the 1 Hz output to do Nixie effects and re-read date/time from RTC every 60 seconds. So, if it is a Maxim chip, it must have some sort of internal factory calibration. Otherwise it is a microcontroller pretending to be an RTC, with some sort of software calibration to hide its nature.

BTW, all boards have the charging circuitry for LIR2032 but came with CR2032, when installed.

Paolo

[gregebert]

Better check the accuracy of your frequency measurement; an error of 67 Hz will throw you off about 7 seconds per hour, or about 1.5 hours per month.

The DS32xx devices do have a trimming register to handle timing-offset, but I've never used it.

What amazes me is how many of these fake RTCs seem to be out there, and who the customers are that are putting these into their products and selling them.

[GastonP]

Well, a possible use might be as backup clock generator in dirt cheap clocks that use the line frequency as reference.

You can make millions of those, and no one is very fussy with it. I know because my wife used to have an alarm clock that was horrid when line operated (not the clock's guilt) but way worse whenever we turned off the power for electrical repairs in the house.

[David Pye]

I have to say, I have a huge pile of supposed DS3231 RTCs, which keep very good time.

I bought them dirt cheap from China, and they work well enough for my projects.

I'm not sure if the chips are genuine (doubtful), reclaimed (possible), or counterfeit (quite likely!) but they do actually seem to keep very good time.

I shall check, when I get a moment, what the "32kHz" frequency output actually is, for interest.

The chips are labelled DS3231SN, and the board has 6 pins on one edge (32K, SQ, SCL, SDA, VCC, GND) and four on the other edge (SCL, SDA, VCC, GND) and a CR2032 button cell holder on the other side.

David

--

You received this message because you are subscribed to the Google Groups "neonixie-l" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neonixie-l+...@googlegroups.com.
To post to this group, send email to neoni...@googlegroups.com.

To view this discussion on the web, visit https://groups.google.com/d/msgid/neonixie-l/744e477e-9a20-42e1-815e-1a96b20ca840%40googlegroups.com.