[time-nuts] Car Clock drift - the lowly 32kHz tuning fork
albertson.chris at gmail.com
Wed Apr 12 13:55:12 EDT 2017
I worked on a project once where a part was temperature sensitive. Another
engineer, not me stuck a Peltier device on the part along with a
temperature sensor and a crude servo using an op amp. This was not
complex, a simple design and it held to about 1/2 degree C. Basically
made an oversized part with JB Weld epoxy and $15 in parts. Something
like that wold have put your clock spot-on.
Yes you can design very sophisticated controls and insolation and get
marginally better results at 100x cost. Did that one too. Used water
cooling and a vacuum chamber for insulation. That was more of a plumbing
job than an electrical job, getting the signals and water lines and vacuum
lines not to leak.
I learned to be a fan of designs that to 90% of perfect with 10% of the
cost. On my Rb oscillator I took the cheap/easy route this time. There is
a temperature sensor on the heat sink and a fan has its speed controlled to
keep the heatsink at constant temperature. Used just one chip to build
the fan controller an 8-pin AVR running an Arduino sketch. The Rb is now
Those cheap 32K crystals might be really good if you find a low cost why to
temperature control them. Strapping it so some one's wrist seems the best
and lowest cost way, just don't remove it.
On Tue, Apr 11, 2017 at 12:21 AM, Morris Odell <vilgotch at bigpond.net.au>
> > Most wristwatches do not have any temperature compensation. If worn, the
> wristwatch is pretty close at the 25°C (the human body is a quite good and
> temperature stable oven). The difference only starts to > show when the
> watch isn't worn for long periods of time.
> That explains my experience with the first microcontroller based clock I
> built years ago. I used a commercial module with a micro and some
> accessories including a watch crystal for timing. It's on a window ledge
> facing west in Australia where the temp varies during the year by 40°C. It
> was always a bit fast and I spent a lot of time checking my code to make
> sure I was dividing it by the right amount. I eventually tamed it by
> programming a short pause at 3:00 am. I'm sure the temp of the watch
> crystal is very rarely 25°C!!
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/
> and follow the instructions there.
Redondo Beach, California
More information about the time-nuts