[time-nuts] Re: 32.768Khz Crystal Trimming

[Bob kb8tq]

Hi

The cap ratio very much depends on how the internals of the chip
got done. Since you have zero info on that, it’s best to stick with the
ratios and approximate values shown in the app notes. 

The most common way to get a fine grain tune is to put something 
large in series with the crystal (maybe 120 or 240 pf)  and then pad 
it with small values (maybe 0.5 to 10 pf). Yes, we just took up more 
board space and upped the BOM … sorry about that :)

Bob

> On Apr 4, 2022, at 2:03 PM, Dan Kemppainen <dan at irtelemetrics.com> wrote:
> 
> After playing with cap values the crystal is running about 6.5ppm fast. Plenty good for what it is. Of course this was with a sample size of 1.
> 
> Being a time-nut, I did try a few different cap values just to see if it could be made any better. The Next closest cap combo gives 8.3ppm slow.
> 
> Putting a little heat on the crystal showed it tended to drop in frequency, so 6.5ppm fast is probably better than 8.3ppm slow.
> 
> Looking at the slope/intercept of the values tried, the needed value isn't easily achievable. Again, n=1, so sample to sample variation is untested.
> 
> It looks like the 'ideal' cap value might be achievable, but that would require a pretty big mismatch between C1/C2. More than 2:1, probably closer to 3:1. I'm guessing that would be a 'no-no'?
> 
> Anyway, Thanks to all who replied!
> 
> Dan
> 
> 
> 
> 
> 
> 
> On 4/2/2022 3:27 AM, time-nuts-request at lists.febo.com wrote:
>> All,
>> Ahhh, Yes. Checking notes, during the initial build I slapped a pair of 10pF caps down (all I had on hand) just to get the RTCC oscillators somewhat working. I can see now that is WAY under needed capacitance.
>> Measurements are taken right in the micro, so no external leads to influence the board. A serial debug line spits out the difference in elapsed time for the RTCC and time derived from the 10Mhz EXT REF (GPSDO). (No extra wires/leads etc.)
>> Honestly this RTCC doesn't need to be stellar. It's more of a holdover oscillator when the EXT CLK is lost. But, this is time-nuts and all, so might as well squeeze everything out of it! :)
>> Bernd,
>> Your post, and rule of thumb are what I was looking for. I will certainly archive this in my notes. This should be good enough to save destroying the board by soldering/de-soldering many 0603 SMT caps.
>> Thanks to all who replied. This should be enough to keep me busy over the weekend!
>> 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.