[time-nuts] Temperature stability for Thunderbolt: results
avollhar at physik.uzh.ch
Wed Jan 26 08:11:36 UTC 2011
After some tweaking of the temperature controlled fan setup (not always
to the better), I finally settled on a setup with great results.
- Peak-peak variations in stable environments are down to 0.02K.
- opening window for few minutes disturbs only 0.03K down.
- TBolt sits now nice and stable at 36 degC.
Summarizing the important points (and please if somebody wants to
add/correct, feel free!):
- Using Lady Heather's temperature controller helps A LOT.
- Using a too small fan results in too small control range. Large
temperature disturbances cannot be compensated.
- Direction of flow DOES make a difference. First attempt blew the air
from the (warmer) power supply to the TBolt.. which make you kind of
wondering why the temperature shoots up while the fan is running 100%.
- I minimized the thermal mass *between* the fan and the TBolt. It is
still in his own case, but no passive shielding around. Slows down the
control loop and seems to make things worse. Some thermal mass around
the fan control loop might be a different story.
- Once the temperature has settled, 'ka4' optimizes the controller
parameters. Another improvement in response speed and temperature residuals.
A big thank you to Warren and Mark for implementing this feature (sorry
if I forgot somebody) and all others who helped in this list. I think
this feature is a must-have for all TBolt setups out there..
73s Achim, DH2VA
More information about the time-nuts