-60 deg C TEC cooled coldbath and idea for chiller

[uranium]

The first part of this has nothing to do with PCs but bear with me
Anyway, we needed a small cooling bath for the lab capable of reaching -60 deg C, and able to be set at any temperature between -60 and +10 C. So here is what I came up with:

[IMG][/IMG]

This shows how its made-on the right is the insulated (with polyurethane
foam sheet) aluminium chiller block with watercooled copper blocks (on left
and right sides) to cool hot sides of peltiers. In the middle is a section
of the internals, cut by a (£80,000!) CNC wire eroder. On the left is the
bit that came out of the middle! You can see the hole where the wire was
threaded through on bottom left. Would make a nice heatsink!

The peltiers (cascade units) are these

http://uk.farnell.com/multicomp/mcpk...-6w/dp/1639734

Next up is a commercial unit which chills the water cooling the hotsides of the Peltiers to around 10 deg C (and will dissipate 600 watts- would be nice for cooling a CPU but a bit big)- its the thing on the floor under the bench



Peltiers (2 in parallel) are powered by a 15V 20A switched mode PSU feeding the circuit on the right (commercial- called a Supercool PR59) which has fancy PID control of PWM circuit and will reverse Peltiers so they heat as well as cool! Hooked up to PC via RS232 (Serial) Interface



Just to show it works! Its cooling about 50 ml methanol, which has the best
thermal conductivity of any liquid which is still liquid at -60, in the
peltier chiller.

[IMG][/IMG]

Anyway, this got me thinking whether one could use a similar construction to make what is essentailly a TEC cooled reservoir sitting on top of a Laing DDC pump i.e. this

[IMG][/IMG]

This is a pretty rough diagram but hopefully you get the idea. It would be made out of copper not aluminium, obviously. The wire eroder will cut any electrically conducting material, and can be programmed to cut any fin cross section, e.g. sawtooth for maximum surface area. The TECS could be 4 50x50 mm units, not cascades (not enough cooling power for i920 load) but maybe the 245W QMax from Frozen CPU or similar. Whole thing would probably be around 3.5 in square by about 5 in tall (or a bit less)

What do you folks think?

[Ultrasonic2]

cool ..
IT should work

[Ultrasonic2]

i had a bit more of a look into your controller . it only handles 15 amps what isn't very much. WHY doesn't anyone make a descent controller ! arrrr

[uranium]

Actually, according to the manufacturer if you force cool the mosfets it will handle up to 30 Amps (900 watts). There is indeed a temperaure probe built in to the unit to monitor the mosfet temperatures. The unit in the lab has an additional fan blowing on the mosfets (pic 3), and at 18 amps load they never get above 35 deg C.

[Ultrasonic2]

Actually, according to the manufacturer if you force cool the mosfets it will handle up to 30 Amps (900 watts). There is indeed a temperaure probe built in to the unit to monitor the mosfet temperatures. The unit in the lab has an additional fan blowing on the mosfets (pic 3), and at 18 amps load they never get above 35 deg C.

oh i wasn't worried about your current setup i was thinking when you add more TEC's that are more powerful .Now tell me did it come with a sensor ? and was it a thin wire one ?

interesting this says 15

http://docs-europe.origin.electrocom...6b8065c181.pdf


Thanks Great work

[uranium]

The unit comes with an NTC sensor for monitoring the temperature of whatever the unit is controlling. But it can be programmed for a whole range of sensors.
Since thermistor based sensors are not great for below -30, the cold bath is using a platinum resistance thermometer.

I'm not sure this unit is ideal for controlling a PC chiller- mainly cos its pretty expensive (£250). The T-balancer + your booster works really well- but the T-balancer can't measure temperatures belwo 0 deg C as far as I can see

Do you have any thoughts on an improved fin design for the internals of a potential res chiller? Only caveat is that there has to be a hole in the centre for the wire of the wire eroder- that hole would also be useful since it would be good to have the flow return via a central dip tube to aid bleeding

[Ultrasonic2]

I think the controller you used is to expensive too. i think i've finally decided on the parts required to make my controller.. it requires money so is pretty much on hold like everything else.

AS for improvements are you running this inline with a rad ?
i presume you have a separate loop to cool the hot side of the TEC's ?

[uranium]

AS for improvements are you running this inline with a rad ?

Note I haven't built this yet, the PC chiller using the finned copper reservoir as the chiller is just an idea- I have the resources but not the time, at least for the moment. No it wouldn't be inline with a rad- just in a loop with the CPU block, if I understand your question correctly

i presume you have a separate loop to cool the hot side of the TEC's ?

Yes the hotsides would be cooled by a seperate loop with a Feser 480 rad (as in my existing chiller- see my earlier thread)

[Ultrasonic2]

well in that case you could increase the surface area by making the cooling fins more like a tree if that makes any sense

whats the internals of the hotside water blocks like ? can you change them or are they off the shelf

[uranium]

well in that case you could increase the surface area by making the cooling fins more like a tree if that makes any sense

Yes I see what you mean- I will try and draw something when I have a moment and post it

whats the internals of the hotside water blocks like ? can you change them or are they off the shelf

The ones for the cold bath are off the shelf- not very sophisticated. They are a crossed drilled design out of a solid piece of copper. If I were to make the PC chiller, I would probably make something a bit more efficient

[Delfistyaosani]

uranium

These mosfets are N-CHANNEL right?

If replace with this one that should handle more than 2KW (MOSFET max power is 2.5KW) 500V, 100A but u have to check gate-drain voltage compatibility not to burn PSU or MOSFET ))

[zipdogso]

Using a rad to cool the coolant for the hotsides will seriously put a spanner in the works unless you have chilled air through the rad.

The temp differential is taken from the hotside temp so warm hotsides will harm your proposed coldside temp. I don't think a 480 rad cooling 4 large TECs will be spectacular of course it will all depend on the heat coming off the hotsides. Generally this can be assumed to be somewhere around 2.5 times the coldside loading - as a rough ballpark figure - so it will depend on what you have in the loop.

I also think your proposed 4 sided TEC mounting will cause problems when you come to applying the necessary pressure. It's doable but I think will require rather more thought and ingenuity than you have currently applied.

[Serpentarius]

i wonder if it works ... i recall the horror of having cool/heat 4200watts of energy in 1kg water for 1°C in 1 second ..

we all know how fast the pump pushes water velocity to the waterblock ... D5 pushes 0.333kg per second