[time-nuts] Re: Power and heat re: Heathkit WWV clock / where are the good oscillators?

[ASSI]

On Sonntag, 7. August 2022 23:21:46 CEST Brooke Clarke wrote:
> The DS3231 is specified at 2ppm/year out of the box.  But it can be tweaked
> to much better performance.

That's roughly a minute per year off or about 1 second per week (both rounded 
down).  I have one DS3231M (the one with an actual quartz) that I tuned up and 
then unplugged about a year ago (or maybe a bit longer already), I should 
probably have a look how far it went off.  Anyway the new (pretty much the 
only ones available right now) DS3231SN come with a MEMS oscillator instead 
and are very different beasts as they need much more aggressive intervention 
to compensate the larger temperature dependency.  I implemented a software PLL 
that keeps it within ~50µs of system time (set from GPS / PPS) while connected 
to the computer and it takes about ±6 LSB of the aging register to keep it 
there since the PLL also fights against the temperature compensation algo on 
the chip that obviously runs on different time scales (note: the DS3231M looks 
at the aging register once every 64 seconds, the DS3231SN every second).  The 
upshot is that while the RTC easily meets spec, it can run past the ±2ppm 
boundary in the short term when presented with large and fast temperature 
changes and only produces nominal performance if nearly correctly compensated 
for the mean temperature over the longer time span.  I would think that this 
is not easy to ascertain for many applications, so beware.


Regards,
Achim.
-- 
+<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+

Factory and User Sound Singles for Waldorf rackAttack:
http://Synth.Stromeko.net/Downloads.html#WaldorfSounds