[time-nuts] Tbolt temperature sensor
SAIDJACK at aol.com
SAIDJACK at aol.com
Thu Feb 5 17:31:12 UTC 2009
Hello Tom,
>Ah, yes, I see that one also shows the same correlation between
>oven current (temp) and EFC voltage, but the TI wanders a lot
>more. What are the large rapid swings; almost like ringing?
This is a brand new crystal, and so the unit is learning about it, and the
aging is rather high in the first couple of days. The "ringing" is caused by
the heater going on in the lab. The phase errors happen because the loop needs
them to properly change the EFC voltage (proportional and integral gain).
After some more time this unit will perform as well as the three I had sent
earlier (without the large phase swings).
By the way, I disconnected the GPS antenna last night, and you can now see
that compensation continues and the EFC curve still follows the thermal curve.
It does look to me like the unit is over-compensating a little bit though;
it has not finished fully learning about the OCXO yet.
>I remember a few years ago there was a long thread about the
>effect of changes in oven current on EFC potential due to
>common ground lines.
We checked for this in some extensive testing, and it is not an issue, even
with oven currents going to 0.5A (this oven only needs about 0.05A). There is
no measurable effect due to ground currents causing voltage drops. You can
see this by turning temperature compensation completely off, and running a
double oven in a thermal chamber. It performs the same even with ground loops
removed (by soldering the OCXO can directly to the PCB ground through massive
straps etc). There is no change in performance. This does require some tricky
PCB layout to work properly though.
>In your plots the EFC correction seems to match so well with
>oven current (without clear evidence of the usual thermal lag)
>could it be that what you're seeing, and what you're correcting,
>is not resonator temperature at all but simply the voltage offset
>due to ground currents?
This is hard to see due to the long time scale, there is a small lag, but
way to small to see in 50+ hour plots.
BTW: the ground current effect is not an issue in our designs, since it is
the total current that we measure and correct for. Ground loops could change
the temperature compensation gain up or down, but will be fully measured and
compensated for.
bye,
Said
More information about the time-nuts
mailing list