[time-nuts] Re: 32.768Khz Crystal Trimming

[Bob kb8tq]

Hi

The typical 32 KHz crystal has a parabolic tempco. It peaks at kinda 
sorta room temperature. This makes it work pretty well on a normal 
wrist. 

As you get away from ~ 25C, the frequency drops. Since it’s a 
parabola, the further away from room you get the faster it drops. 
Something in the 1 to 2 ppm / C is not a bad guess for typical “on a 
board” sort of temperatures. 

It’s pretty hard to predict what temperature the board you just made 
will operate at out in the field. Thus there are limits on what you can 
expect from a watch crystal. 

10 seconds over 100,000 seconds ( so a bit over a day) is hundred
ppm. That’s a lot of shift and yes, it’s worth taking care of that
sort of offset. 

At 1 second per 100,000 you are at 10 ppm. That is in the range of 
“questionable temperature”. You can get it set that well ( or better) 
on your bench. As soon as it leaves your bench …. all that work may 
not have been worth the effort. 

No, temperature isn’t the only variable. The crystals have a calibration
tolerance at room ( +/- 20 ppm on this crystal). They most certainly do 
age ( 3 ppm  / year on this one). They also shift when you solder them 
into a board. That shift may or may not relax over the next few days. 

If your application *needs* a second a day sort of performance, 
something sort of improvement over a basic watch crystal will be 
needed. At the very least, the 20 ppm cal tolerance will need to be 
adressed. 

Bob

> On Apr 1, 2022, at 12:34 PM, Richard (Rick) Karlquist <richard at karlquist.com> wrote:
> 
> No one mentioned tempco, so I will.  Ideally you should do your
> calibration at a temperature corresponding to the long term
> average in your workshop.  If the crystal is in a piece of
> equipment with a temperate rise, it should be accounted for,
> and then going forward you have to leave the equipment powered
> up 24/7.  The crystal is probably a tuning fork, meaning it
> won't be AT cut.  It may have a substantial tempco around
> room temp.  In which case that old time-nuts insult may apply:
> 
> "congratulations, nice thermometer."
> 
> Rick N6RK
> 
> BTW, I go back 48 years with crystals.
> 
> On 4/1/2022 8:30 AM, Bernd Neubig wrote:
>> Hi,
>> If you do not want to make it a time-nuts style research project, but just
>> look for a quick fix - here is a rule of thumb:
>> This kind of crystal usually has a trimming sensitivity of around -10
>> ppm/pF. This means, if you increase the value of both capacitors on either
>> side by 2 pF will increase the load capacitance by 1 pF and thus lower the
>> frequency by  about 10 ppm.
>> If you need to vary one cap slightly more than the other, (to get a finer
>> resolution), please use the one at the output side of the on-chip oscillator
>> stage.
>> Take care of the start-up margin (safety to get a reliable start-up after
>> power on). The larger the capacitors, the lower becomes the margin,
>> sometimes the margin drops rather quickly. Therefore, after having made the
>> changes, test the start-up behavior by switching on and off several times,
>> preferably with a slow voltage ramp.
>> Enjoy the crystal world (as I did for the last 45 years and still doing).
>> Bernd
>> -----Ursprüngliche Nachricht-----
>> Von: Dan Kemppainen [mailto:dan at irtelemetrics.com]
>> Gesendet: Freitag, 1. April 2022 14:47
>> An: time-nuts at lists.febo.com
>> Betreff: [time-nuts] 32.768Khz Crystal Trimming
>> Hi,
>> I've got a 32.768Khz (USA number format) crystal on a RTCC oscillator of a
>> small micro, and it's running fast. Around 10 seconds per day or so.
>> This is a bit more than an order of magnitude more than the datasheet
>> states.
>> The 9 seconds per day error should be a good measurement. The RTCC is
>> running a 1 second counter, and that's being compared to a 1 second counter
>> derived by clocking the micro from a 10Mhz EXT clock reference.
>> This is consistent between multiple copies of the board.
>> I'm assuming, the C1/C2 load capacitors to ground needs to be higher in
>> value to trim that oscillator closer to the correct frequency. Is this
>> correct? Any quick back of the napkin calculations how much additional load
>> capacitance would be needed?
>> For Ref, this is the crystal:
>> https://abracon.com/Resonators/ABS06.pdf
>> ABS06-32.768KHZ-1-T
>> Thanks,
>> Dan
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send an
>> email to time-nuts-leave at lists.febo.com To unsubscribe, go to and follow the
>> instructions there.
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send an email to time-nuts-leave at lists.febo.com
>> To unsubscribe, go to and follow the instructions there.
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.