DS3231 RTC Accuracy

[Jerry]

Ive been using a DS3231 RTC module with a pic MCU, over the past month to make an accurate animated clock,ive now finished my code and all works as it should..............BUT,on checking time with my PC (after a sync to server) i had noticed my clock was reading fast by a few seconds a day,these DS3231 supposed to be accurate to +-0.432 a day,code was checked in case of errors many times,the code seems fine.So i plugged in another RTC DS3231 module,(i had bought 5) tested over another week,and my clock was slow by 30 seconds,this wasnt what i expected.Next stage i had a closer look at the RTC module,the chip on it had all the right writing apart from no  manufacturer logo!!,mm,China?
I ordered from a well known supplier,and a few days later a new DS3231 came,i removed the previous IC and soldered this one in its place,did another weeks testing,and all is great,cant really tell how accurate,but it syncs to my PC clock now,so if you want an accurate time piece dont buy from China.
When Eb@y Ebay sells RTC DS3231 modules for couple of £ dont expect accuracy,they do work,and prob ok for testing or learning,but from now on im avoiding these cheap knockoffs.
I forgot to mention, i did check the onboard battery first,it was fine.
By the way a legit Maxim DS3231 IC costs around £6 + p/p.There is a reason for this!

no point writing Eb@y

[metal]

life lesson: don't buy components from china

[zuisti]

I'm sorry but I have a better experience with these cheap Chinese DS3231 modules. I bought 5 of them on Aliexpress last year, for $1.5/piece, with free shipping. These modules called ZS-042 also include a 24c32 eeprom in addition to the DS3231M (M and not SN, unfortunately) RTC.
I tried the modules in this project:
https://www.sonsivri.to/forum/index.php?topic=52263.msg196717#msg196717
The result - after a proper setting of the RTC's aging register (address: 0x10) with a few tries - the accuracy of the clock was far better than 1 sec/week (for all five modules).
True, I'm using a DCF (radio controlled) clock (an not the PC clock) to compare the time deviation.
zuisti

[optikon]

I wouldn't say don't buy from China.. its a calculated risk.. sometimes you make out ok with great parts.. and sometimes you find the counterfeit parts.

Rule of thumb: don't buy the cheapest you can find.

[Vineyards]

Since it is a budget chip, you don't expect a 10ppm crystal in this unit. It is probably a 100ppm one. A simple calculation will show about +/- 30s per week of error in the case of a 100 ppm crystal. The manufacturers make their calculations on the assumption that the ambient temperature will remain 25 degrees Celsius and, incidentally the crystal produces the highest number of pulses at that temperature too. Thermal stability plays a great role here. There used to be RTC ovens in the past where the RTC would be placed in an enclosure and the internal temperature would be kept at a constant 25 degrees. These days there are thermally compensated RTCs with built-in or external ADC's measuring temperature from a sensor for calibration. There may still be issues like self-heating etc which may affect some types of RTCs; add to that crystal aging which though insignificant may effect systems that must be accurate for extended periods of time. In other words, RTCs are not perfect, and that's when your engineering skills kick in. I believe, any imperfect RTC can turn into a reliable time-keeping unit at the hands of Zuisti.

[Xwing]

On the DS3231 Maxim declares an accuracy equal to +/- 2 ppm in the temperature range 0 ° - 40 °, obviously we are talking about original ICs.
The ICs mounted in the low cost chinese modules are either clones or they are original second choice ICs, that is functioning but that have not passed all the quality tests, in the case of the RTC it is highly probable to find "faulty" components in terms of precision.
I have some boards where I have mounted some DS3231 purchased on Mouser, therefore certainly original, and I can guarantee that the accuracy is as stated.

[Vineyards]

Yes, I took a look at it too. It is a TCXO so it is already temperature calibrated. There is usually an attainable tolerance band with these. If it is a counterfeit, you can forget about those numbers.

[titi]

Try to change the value in "Crystal Aging" Offset (10h) register, this can speedup or speed down the internal oscillator.

May be this can help to reduce the time gap ?

Best regards.

[Jerry]

I tried the ageing offset register thank you,and was able to get acceptable results,the point i was making was if a chip sais it has +-5ppm accuracy over a given temperature range,you would expect it to be so,i was nieve to expect a £2 part to perform like a £6 part.
On a different note,ive had switching reg ics & power transistors from China and the spec varies all over the place,so i wasnt really surprised by the RTC being out of spec,just warning others its not always your fault a circuit doesnt perform as it should.
Anyway im happy with the legit DS3231 from Maxim,the China ones might have other issues down the line,i know that the common 7219 led driver from China suffered from some not resetting cleanly on power up,this was proved a while ago on a forum i visited,that the IC's, although worked, some were out of tolerance.
At the end of the day if your design is used in a critical application,buy from manufacture or known good source,its really that simple.

[comlekciler]

There is a saying that "I am not rich enough to pay for cheap goods" and it is true.

[ero]

Pls. pay attention to the following points on using DS3231 RTC module.
- Do not give 5V directly to the module try to give a voltage between 4V-4,5V even 4,6V will be much better. This will help to the modul to keep it cool.
- Use only one pullup resistor on the CLK and DTA line. ( I am telling this for the people which are replecing the module with DS1307.
- Do not place the module just near the voltage regulators or transistor which are producing heat. Pay attention to the temperature inside the box. It should be around room teperature.
These are my experiences on using DS3231 modules. I always buy them from Chine. Some of my clocks are working very accuracy since 5 years. Some of them has not so much accuracy and the reason is the temperature inside the box of the clock. I change the design some of them and provide them more cool air and they are working now better then before.

Ero

[Catcatcat]

I started using this watch quite recently, checked the operation both with 3.3 volts and 5 volts, tested the accuracy (subjectively) and was very pleased with these chips. I made backup power on an ionistor. In general, I am pleased with both the accuracy of the stroke and the capabilities (you can read not only the time, but also the temperature). The functions of alarms and the formation of interruptions for the given time intervals suit me completely.

[simma]

The reason for inaccuracy is may be due to usuage of DS3231M which uses MEMS based resonator. It is cheaper than DS3231Sx.

[Jerry]

Good point about the resonator,i was unaware of that detail,im using 5 volts which the datasheet sais is ok.

[Flash50]

I use maxim ds3232 in one of my projects and its price is high. I was excited when I saw the prices of ds3231 modules on aliexpress. I ordered and tested immediately. But I saw that these are not the same pieces. Again, I continue to use the previous original expensive parts, unfortunately.

[Catcatcat]

I bought here https://lcsc.com/product-detail/Real-time-Clocks_MAXIM_DS3231MZ_DS3231MZ_C107410.html

[Jerry]

Flash50,what was the diference between the expensive and the cheaper ones you had,did you test them?


Posted on: September 20, 2020, 10:15:11 22:15 - Automerged
Catcatcat,those are the 8 pin varients,which i would imagine would cost less than the 14 pin ones i use,also they have the MEMS Resonator oposed to the internal XTal on the MAX3231,+-5ppm still acceptable for most projects.

[Flash50]

Jerry,
Yes I tested them.
Maxim IC deviation is max 1-2 seconds in a month, but cheap module deviation is 3-4 seconds in a day. (Environment temperature 15 C degree to 30 C degree)

[microstick]

Hi,
I have implemented different approach on this and until now working very well.
I using a TCXO insted of crystal, and my stability is 1.5 ppm.
BR

[cesare]

Hi All,


The DS3232M do not have any pin connected to the internal oscillator that can modify its frequency and so, in my opinion, the only reason for the frequency error is an out of spec part.

In the past I had nasty experiences with components from brokers because sometimes they sell out of spec or partially failed parts. This can also explain the low price.

Regards

[Jerry]

Yes, sometimes even the big suppliers get burnt,bit of a minefield out there,there were several cases a while back,very difficult to police,i supose batch testing every so often would help,its all about cost at the end of the day for the suppliers.If its silly cheap avoid i guess.
Cheers all.

[Catcatcat]

After the final assembly, absolutely all the chips, if they fail the tests, are sent for recycling. And those that have not reached the disposal site appear on the market at a surprisingly low price.