water cooling

Didn't Cray submerge the circuit boards in liquid freon? IBM did some heroic cooling too.

My pulser will mount deep inside the customer's gadget, and he does have a cold water system. I just need to figure out the best way to insulate the fet drain but suck the heat out. We have tools to simulate circuits, but not thermal systems. Electronic circuits have nice equipotential nodes on a planar schematic; thermal systems are

3D, distributed, diffusive, and in this case involve liquid flow. The component thermal specs aren't to be trusted either; you can thank IR for that.

Water is fabulous thermally, but it's awful electrically. The dielectric constant is huge.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin
Loading thread data ...

You pick your heat transfer fluid to be about the temp you want to stabilis e the part case at. A heat pipe is liquid cooling at the heat input end. Le ss grams are shifted but you get the advantage of the phase change - it tak es some effort to exceed the boiling point of liquid your device is submerg ed in. Heat throughput is largely controlled by the condensing end.

NT

Reply to
tabbypurr

and the condenser was a waterfall :) or freonfall.

I'm sure you've a lot more experience on that than I. I try to stay conservative - assembly does their best to undo that.

NT

Reply to
tabbypurr

I'm still learning, and the customer keeps moving the goalposts. But if I can help him sell more gadgets, it's worth some hassle. I designed a circuit that I barely understand electrically and don't understand thermally.

The SiC fets are great, but Rds-on increases radically with temperature. So there is big opportunity for thermal runaway. My goal is to get the thermal positive feedback loop gain down.

As you note, we don't want to do something that is complex or messy to assemble, or that an assembly error will mess up, as in cause a water leak or a blown fet.

I was thinking of a machined block with cavities and o-rings, so the fets bolt down with their metal bottoms in pools of flowing water. That's probably too risky, for a number of reasons.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

there's an IRF appnote on testing fets, they submerge the whole assembly in some magic Dupont refrigerant that is electrically inert and boils at 25 'C

Reply to
Lasse Langwadt Christensen

On Saturday, 20 October 2018 18:06:07 UTC+1, Lasse Langwadt Christensen wr ote:

e.

25'C

there are plenty of volatile liquids to choose from. Water is problematic, not everything is.

25C bp won't work though, it won't recondense in a lot of cases, resulting in electrical & mechanical failure.

NT

Reply to
tabbypurr

And then they spec DPAK fets for insane power dissipations, and currents that would melt the source lead in free air.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Swimming pools get a lot of sun, and use chlorine bleach to kill the critters. I put a splash of bleach in my bucket.

It's hard to measure the water temp in the bucket. When I start pulsing, the Omega thermocouple meter instantly jumps up about 5 deg C. Or sometimes 20C.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

They use sodium isocyanurate to absorb the sunlight to make the chlorine last longer, too.

Reply to
Clifford Heath

ps.

bleach is corrosive & fugitive. Fine for experiments, not for shipped produ ct.

I thought copper would do it, but not an insulator in any form that I can t hink of.

NT

Reply to
tabbypurr

Why don't you use alcohol? No bugs will grow in it and you have the added heat transfer of a state change.

Reply to
Tom Del Rosso

I have both some copper and aluminum in my coolant loop, there might be a risk of galvanic corrosion if I use plain water as a loop fluid long term.

I'm thinking about giving say a 40/60 mix of glycerol/de-ionized water a try as a long-term stable coolant

Reply to
bitrex

hips.

rm.

deionised water always dedeionises. That's why it's not popular.

NT

Reply to
tabbypurr

In a properly evacuated heat pipe - no non-condensable gasses - the boiling point of the transport fluid is the temperature of the coolest point in th e heat pipe set-up.

There's a theoretical limitation on the rate of heat transport set by the a ctual pressure of the vapour whizzing through the heat pipe to the point wh ere it condenses - gas viscosity is pretty much independent of pressure, so that mass flow rate and the heat transfer would drop at lower temperatures , but the liquid has got to wick back to the hot bits before it can volatil es, and this might equally be rate determining.

The boiling point of the working fluid at atmospheric pressure probably isn 't much of guide to the effectiveness of the heat pipe wrapped around it.

--
Bill Sloman, Sydney
Reply to
bill.sloman

e.

.

ise the part case at. A heat pipe is liquid cooling at the heat input end. Less grams are shifted but you get the advantage of the phase change - it t akes some effort to exceed the boiling point of liquid your device is subme rged in. Heat throughput is largely controlled by the condensing end.

A heat pipe cools the heat input end by having the working fluid volatilise - boil - at that end.

The heat is transferred as latent heat of evaporation in the vapour that wh istles through the heat pie to the cooling end, where it condenses.

The boiling point of the working fluid inside the heat pipe is the temperat ure at the cooling end.

If it isn't, the people who filled the heat pipe didn't get rid of all the non-condensable gases before they loaded in the working fluid. It happens, but it shouldn't, and it's easy to set up a test rig to make sure that they did the job right.

--
Bill Sloman, Sydney
Reply to
bill.sloman

Heat pipe assemblies are normally brazed together. What you get is neat and sealed, though it starts off pretty messy while it is being manufactured.

--
Bill Sloman, Sydney
Reply to
bill.sloman

hips.

rm.

That described my central heating system back in Nijmegen - we had copper p ipes, some mild steel radiators, and some cast aluminium radiators.

There's a specific additive you can buy from your local plumbing shop for p recisely such systems

formatting link

1_us.pdf

This isn't the one I used, and the one I used didn't completely stop hydrog en emission - I had the vent the system every few months - but it worked fi ne for some fifteen years

Probably not the right answer. The corrosion inhibitor I used was added as

1% of the volume of the circulating water (and working out what that was a tedious exercise, involving draining the entire 600 litre system into a buc ket, one bucket at a time, weighing the bucket, and adding up the total we ight).
--
Bill Sloman, Sydney
Reply to
bill.sloman

:

is

ic

.

ge

al

;

chips.

term.

a

Not in a sealed system. Carbon dioxide from the air forms carbonic acid in systems that are open to the air, and that makes the water a much poorer in sulator.

--
Bill Sloman, Sydney
Reply to
bill.sloman

I've had that where the heat pipe conductivity got poor over time -- likely a microscopic air leak.

That's the, well, the only failure mode of the things, really.

You can exceed the dissipation and get nonzero thermal resistance, but that's normal behavior, and that limit shouldn't change over time.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams

The total volume of my (significantly more humble-sized) loop is probably about 1/2 L, still an applicable product?

Reply to
bitrex

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.