Page 1 of 1

RTC Quartz datasheet available?

Posted: Sat Mar 16, 2019 11:41 am
by Hanilein
I try to understand a very odd behaviour of the RTC quartz of the PyBoard v1.1, and would like to have a look at the datasheet.
It would also help to know the frequency-temperature characteristic.

Posting of attachments (pictures) is in this forum not possible? It seems not to work for me...

Thank you

Re: RTC Quartz datasheet available?

Posted: Sun Mar 17, 2019 1:56 am
by shaoziyang
Most 32K quartz crystals have similar properties. You may get datasheet in some manufacturer.

Re: RTC Quartz datasheet available?

Posted: Sun Mar 17, 2019 4:44 am
by Hanilein
Not really. I tried a search using DigiKey, but even when narrowing down to 3.2*1.2mm 32KHz SMD, there are still dozens of hits left with sometimes very different special properties. The I tried the cheapest ones of those to see if the device marking can help, but nope.

Hence my question.

Re: RTC Quartz datasheet available?

Posted: Mon Mar 18, 2019 9:09 pm
by Roberthh

Re: RTC Quartz datasheet available?

Posted: Thu Mar 21, 2019 10:06 pm
by chrismas9
What is the very odd behaviour?

Re: RTC Quartz datasheet available?

Posted: Fri Mar 22, 2019 8:43 am
by rhubarbdog
You can post images when bb-code is on.
New users it defaults to off to prevent spam.
Make some more posts and bb-code will be switched on.
I think you may be able to make a post, edit it, manually switch on bb-code (a tick box near the submit button), try and add an image or code.

Re: RTC Quartz datasheet available?

Posted: Fri Mar 22, 2019 9:17 am
by rhubarbdog
Just out of interest is it an official pyboard from the store
Or is it one of those half price ones on ebay coming out of china

Re: RTC Quartz datasheet available?

Posted: Fri Mar 22, 2019 9:57 am
by Hanilein
My plan is, to write a code that fine-tunes the adjustment of the RTC depending on the temperature.

On my particular pyboard (bought from DigiKey), the correction value for ~25°C was -293, which i can determine using either of two different algorithms that use a GPS PPS signal as a reference. The algorithms are both giving the same results, and the clock is running pretty precise when adjusted using the correction value these algorithms provide.

I mounted a temperature sensor (DS18B20 in a TO92 housing) upside-down above the 32 kHz quartz, initially using a thermal compound.
The algorithms calculated the correction value now with -257 for ~25°C, and the clock was running precise with this value.

I first assumed the thermal compound may be (slightly) conductive, so I removed it using a toothbrush and methylated spirits, and the correction value was computed being back to the original -293.

The next test was done with another thermal compound (non-conductive), this time the correction value changed to -261.

Then I just removed the sensor, but left the compound on the quartz. The correction value stayed around -260.

Again I removed the compound, but this time the value did not came back to -293, it stayed at -289, even after deliberate use of toothbrush and alcohol. BTW, after cleaning, i dried the PCB with a hairdryer, to remove any potentially conductive remains of the alcohol.

For the next test I mounted the sensor without any compound, and the correction value is -286. And that is that, if the sensor is very close to the quartz or touching it, the value is -286, and moves to -289, when the sensor is moved away for more than ~5mm.

Robert thankfully posted a link to the app note which explains a lot of the sensitivity around the quartz (or crystal resonator), I assume, the stray capacitance is changing when I apply the thermal compound or mount the sensor. Doh'.

To me it seems that the quartz/crystal resonator is very sensitive towards stray capacitance, and that is a challenge - even just a shield mounted on top of the pyboard can alter the clock frequency...

Regarding the photos - This is post #24 in this forum, my first post was July last year. BB code is active since my third or forth post. And with a 43 kB file I get an error: "Sorry, the board attachment quota has been reached."

Re: RTC Quartz datasheet available?

Posted: Sat Mar 23, 2019 2:13 am
by chrismas9
The 32 kHz crystal is sensitive to stray capacitance and also noise. The crystal requires 6pF load. Most of this (5pF) comes from stray capacitance - PCB and STM32. Only 1pF comes from the load capacitors. Any additional stray capacitance will affect frequency.

Noise is also an issue. If you place a temperature probe on the crystal and it is connected to a noise source, eg I2C or digital meter it will couple noise into the crystal. Note there is a ground guard ring around the crystal on the PCB to prevent the 12MHz clock and other signals coupling into the 32kHz.

I suggest you attach your temperature probe to an area of ground pour on the back of the PCB and this will give a reasonable approximation of the crystal temperature.