RTC drift

[Vineet Nair]

Posted on June 08, 2018 at 21:35

I am Using the #RTC #M41T94 for keeping time on my project.Have noticed on multiple occasions that there is a significant #drift in the time keeping. Have reviewed the datasheet multiple times to check if my schematic or firmware is doing something wrong but no success.

Was wondering if anyone here has had similar issues ?

Any help is appreciated#.

Thanks & Regards,

Vineet Nair. Note: this post was migrated and contained many threaded conversations, some content may be missing.

[Vitaliy Chernobay]

Posted on June 11, 2018 at 11:47

What is your drift value?

[Vineet Nair]

Posted on June 11, 2018 at 15:00

over a duration of 20 days, the drift was over 6 days.

So 20 days was reflected as 26 days.

It also seems to get extreme at lower temperature 0-5 degee C.

[Tesla DeLorean]

Posted on June 11, 2018 at 15:26

Pound Me Too?

http://www.st.com/en/clocks-and-timers/m41t94.html

 

supposedly has an onboard crystal, but not a TCXO, but I wouldn't expect an excursion of this magnitude. If the part can output a 1PPS or 512 Hz signal you might want to scope that and see what sort of accuracy you observe there.

If it is your own crystal you'd want to review the exact component specs and loading for the design you have created.

Not sure ST has any support staff here for the RTC/Timing products, you should probably work with your distributor and FAE on some sort of resolution here.

[Vineet Nair]

Posted on June 14, 2018 at 01:05

I output a 512 Hz waveform and checked it on an oscilloscope.. and i could clearly see the drift.

Another thing, am using an 3 pin oscillator which has been tested and I know for sure is accurate.

Now the output from the oscillator is connected to the XO pin instead of the XI, so at the output of the internal amplifier of the oscillator.

Any thoughts ?

[Tesla DeLorean]

Posted on June 14, 2018 at 02:14

>>Another thing, am using an 3 pin oscillator which has been tested and I know for sure is accurate. 

Ok, does it have a part number and supporting specifications?

>>Any thoughts ?

Tried calibration settings?

[Vineet Nair]

Posted on June 14, 2018 at 17:02

Hi Clive,

   The oscillator I am using is CXOXHG4SE-32.768k/-/-/80/M. The calibration setting would not give me more than a few minutes and since these were working without drift before, I am not sure if this is the reason. 

Let me summarize all observations again,there is some info that i missed to point to before. This will help pinpoint the issue. 

1. When the circuit was built/tested. There is barely any drift observed on the RTC and works as expected. 

2. First time we observed a drift was when the it was subject to a low temperature like -5C to +5C. In this case we observed that the drift was very high, it was actually more like a jump of like 4 seconds instead of 1 second.

3. We saw similar drift/jump behavior even with high temperature of > 130C.

4. Later as we kept using this circuit we observed drift even at room temperature.And this seems to be happening with almost all of these boards we had. 

5. The drift seen in all cases is more than what the calibration setting can help fix, 20 days reflected as 26 days. At extreme temperature it was even worse.

At this points with all my observations I feel like there are 2 possible reasons.

1. My circuit seems to be causing damage to the RTC in someway. Cause it works wells at first ( few months) then seems to have issues.

2. Bad Batch of IC. All were boards were built with the same set of IC's.

Below is the circuit diagram.

[Tesla DeLorean]

Posted on June 14, 2018 at 17:17

So a fixed frequency source, with an ST part counting clock pulses, digitally, I'm hard pressed to see how the IC could be the cause of the failure here.

I wouldn't short XO to ground.

Couldn't find a PDF for the clock source.

No sure how accommodating the RTC will be to a rail-to-rail input. Haven't dug up an app note for that. Would probably expect AC coupling required.

[Vineet Nair]

Posted on June 14, 2018 at 17:28

XO can be disconnected from ground, luckily i just have remove R66.

Could not find a link, so attaching datasheet as images.

[Vineet Nair]

Posted on June 15, 2018 at 23:38

So I finally got this fixed. :)

Did a lot of reseach online yesterday to see what was the best way to connect clock source to an RTC chip. Went through examples of RTC chip connected to clock source felt that i should connect the clock source to XO instead of XI.

Will be testing the board at high temperature to validate that this is actually solved.

Here is what i did.

1. Disconnected the Xo from gnd.

   - This reduced the drift from the original 1.3 to ~1.15.

2. Connected the clock input to the XO pin and left the XI pin floating and there was no drift now.

So to summarize this whole ordeal,

1. Connecting to XI works fine, no issues. Problems only show after using boards for sometime ( 2 months.)

2. on a bad chip, disconnected the clock source from XI and connect it to XO and problem seems to have disappeared.

A big thank you to Clive & Vitaliy for your inputs.