R
Rick C
Guest
On Friday, February 7, 2020 at 1:49:07 AM UTC-5, DecadentLinux...@decadence..org wrote:
Yes, that's the point. It does a much better job of cooling if it does evaporate. In the system I was talking about, they tried to make sure there was no or very little surface layer of liquid. If there was, the cooling effect would be dramatically reduced due to the higher thermal resistance.
"very good" is a judgemental term with no quantitative value. The point is it is not as good as it needs to be to equal what can be achieved with a spray and vaporization rather than immersion.
Don't care. I'm talking about systems that require much better cooling than either of those. One of the things they talked about was using a synthetic diamond substrate to mount components to provide a thermal path for cooling in a very high density package.
There are more technologies in mil-spec gear Horatio than are dreamt of in your philosophy.
--
Rick C.
+- Get 1,000 miles of free Supercharging
+- Tesla referral code - https://ts.la/richard11209
Rick C <gnuarm.deletethisbit@gmail.com> wrote in
news:45c6c4d5-81e3-4756-bd71-cff8fd38bded@googlegroups.com:
On Sunday, February 2, 2020 at 4:03:16 PM UTC-5,
DecadentLinux...@decadence.org wrote:
Klaus Kragelund <klauskvik@hotmail.com> wrote in
news:484e78ef-6290-493c-bbd0-7c6bfe76e4a0@googlegroups.com:
On Sunday, February 2, 2020 at 3:37:03 PM UTC+1,
upsid...@downunder.com wrote:
On Sat, 1 Feb 2020 09:03:47 -0800 (PST), Klaus Kragelund
klauskvik@hotmail.com> wrote:
Hi
I am working on something for my own business
So need to do testing at below 0 degrees C
What would be a cheapskate way to do it?
Just a household freezer?
Product is just 20x20x20
There are some issues with electronics in cold temperatures.
One problem with any processor based (or sequentially logic)
system is starting the clock oscillator at extreme low
temperatures. This can be quite unreliable. Once the clock is
running, the electronics will generate some internal heat
warming the components in the box to higher temperatures.
Thus, even i some components are initially out of specs it
will be within specs after a while.
When doing cold testing, let the device cool to the lowest
temperature obtainable with equipment power turned off. Once
the lowest temperature has been achieved, turn equipment power
on and verify that the clock starts. Once running, make sure
the box outside temperature remains in your case at -20 C
despite the 5 W heating,
We have big, super expensive chambers at work, where we always
do that LOL and UOL (Lower Operating Limit and Upper Operating
Limit), which has the purpose you mention to find when it
really fails. We would normally have the LOL at -50 and UOL at
150. But for my own business if I can do a minus 20 test and
150 then I feel pretty safe
When doing indoor cool chamber tests, the initial air contains
some absolute humidity. While the chamber cools down, the
relative humidity RH increases, finally reaching 100 %, after
which condensation or frost will accumulate. Do not put the
equipment directly on the bottom of the chamber, since there
might be some condensation water on the bottom. The
condensation or frost may cause problems with uncoated PCBs.
Yes. Also, don't open the chamber when at -40, tried that,
won't do it again (heavy condensation)
Cheers
Klaus
Place the UUT in a Fluorinert bath.
I wanted to do an entire motherbaord that way and go down below
and
clock up above what the OC boys do. Some of the videos are
amusing watching some dude pouring LN2 over a CPU.
My solution <sic> would work, and for long term too, not "I
OC'd my
PC to 10GHz" (for 1 minute long enough to run the benchmark). It
is funny watching them make claims. My PC in a fish tank method
would work down to -70°C with FC-72 Fluorinert. And run... all
day.
Maybe, maybe not. I recall an outfit that was working on cooling
military electronics. They used something like Fluorinert but
instead of immersion which actually provides thermal barriers from
limited conductivity of the fluid, they sprayed it on using
patented nozzles to obtain optimum droplet size and used the heat
of vaporization to cool. While that would not cool to -70°C, it
removes more heat than immersion would until your device heated up
to at least 10°C but actually higher due to the thermal
conduction issues of the liquid.
Fluorinert does not vaporize and the immersion unit would have the
fluorinert always flowing over the parts.
Yes, that's the point. It does a much better job of cooling if it does evaporate. In the system I was talking about, they tried to make sure there was no or very little surface layer of liquid. If there was, the cooling effect would be dramatically reduced due to the higher thermal resistance.
And oh fluorinert has very
good thermal conductivity.
"very good" is a judgemental term with no quantitative value. The point is it is not as good as it needs to be to equal what can be achieved with a spray and vaporization rather than immersion.
It would work well in seal, conduction
cooled designs as well, where no movement happens or is needed. They
work far better than a friggin thermal pad or even a directly
attached air flow heat sink.
Don't care. I'm talking about systems that require much better cooling than either of those. One of the things they talked about was using a synthetic diamond substrate to mount components to provide a thermal path for cooling in a very high density package.
There are more technologies in mil-spec gear Horatio than are dreamt of in your philosophy.
--
Rick C.
+- Get 1,000 miles of free Supercharging
+- Tesla referral code - https://ts.la/richard11209