water cooling

If you want corrosion protection, why not go with the proven product that is designed to provide corrosion protection in mixed systems composed of rubber, plastic, aluminum, cast iron, steel, and stainless steel, and has lubricants for the pump? Of course I?m talking about automotive antifreeze. Use non-toxic Sierra brand based on propylene glycol. Given the small surface area of your system compared to a car engine, and the relative volumes, a fill should last you several years.

--
Regards, 
Carl Ijames

Adding thermal resistance, even a very low thermal resistance, can only make my fets get hotter. A heat pipe only moves heat; it can't dump it. My ultimate sink is flowing 20C water.

The fets need to be insulated, and the best insulator (aside from pure water itself) is aluminum nitride (diamond and BeO excluded). Thinner is better, until the capacitance gets too high.

AlN with water flowing on the back side is interesting. Water has a dielectric constant around 80, but that would still be better than metal against the back of the insulator; the water could be pretty thick. Some weird topologies are possible.

I recall seeing a sort of puck shaped heat pipe, like a silver dollar, that functions as a lateral heat spreader. Something like that could be bonded to the back of an AlN insulator to keep the backside maximally cold. It's probably expensive. A copper slug might help spread the heat, or make the entire baseplate from copper. Pure copper is a bear to machine, and alloys don't conduct heat as well.

Heat spreading will keep the backside of an insulator cool, but doesn't help the theta:capacitance tradeoff much.

Fast electronics is fun, because it's so hard. Keeps the amateurs away.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

It's also flourescent under UV, which makes leak spotting easy (if you're equipped).

--
  Notsodium is mined on the banks of denial.

Pure water has a dielectric constant of 80. It requires extreme attention to cleanliness. The best you will probably see has a conductivity of around

1 microsiemen per cubic centimeter. That is not pure:

You can measure the conductivity with a pure water tester (PWT), such as the DM Digital COM-100:

You may find deionized water from a local water store. The best will give around 11 microsiemens, others may give 30 microsiemens or more.

You are not likely to see pure water in any application that involves metals, tubing, and other contaminants. You will not get it out of a tap.

If you have a voltage gradient in your setup due to dissimilar metals, the purity of the water will rapidly degrade. Galvanic corrosion has been known for centuries:

You can check the compatibility of metals by referring to a galvanic table. One example is here. There are many others:

The US Navy is famous for building ships that corrode:

Engine-Corrosion.aspx

Aircraft designers use steel-aluminum joints to hold the wings on:

Piper SB 1244B Inspection of the Aft Wing Attach Fittings Steel on Aluminum Galvanic corrosion

They fall off

Embry-Riddle Piper Arrow Loses Wing and Crashes at KDAB

Am 21.10.18 um 19:42 schrieb John Larkin:

< > < >

The naked chips were attached to the plate via some ceramic filled foil. You need a little bit of of elasticity; some of the boards were HUGE. The board & the cooler do not belong together.

Don't ask me how they got the tube into the block. The block is 9.3mm thick. Probably soldered in a salt bath. That stuff is so old that the plastic interconnect hose broke to pieces. What a wet mess on my floor... :-(

Carl: > If you want corrosion protection, why not go with the proven product > that is designed to provide corrosion protection in mixed systems >composed of rubber, plastic, aluminum, cast iron, steel, and stainless > steel, and has lubricants for the pump? Of course I?m talking about > automotive antifreeze. Use non-toxic Sierra brand based on propylene > glycol. Given the small surface area of your system compared to a car > engine, and the relative volumes, a fill should last you several years.

The Agilent people were also sure in the beginning that this would be an easy piece of cake, esp. when they were in Stuttgart, the home town of Mercedes and Porsche. It turned out not to be the case. Even for them.

regards, Gerhard

I forgot to mention, do not immerse a PWT in ordinary tap water. It may take a year to flush out the contaminants so you can read down to 1 us again.

OT: what a blast. Stuttgart is where I got my first Mercedes 500 SEL. You get on the freeway and pass a Porshe at 130 mph. They wave as you go by.

Coming down from these speeds is like walking in slow motion. Even at 100 km (60 mph) everyone is moving so slowly you feel you could get out and run faster.

This is an experience everyone should have.

Fortunately, my customer will provide the water. My bench setup is just a temporary hack.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Do you specify the water purity and dissimilar metal constraints?

You need to make them aware of the galvanic corrosion problems.

Ordinary tap water is probably not going to work.

These people are seriously expert about thermals and water cooling. They will make me aware of what works. I just do the electronics.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

.

In principle your heat pipe could be big enough to act as the heatsink too. Small heat input area, large output end. Or large enough to interface with a large heatsink. Perhaps you could make much of the enclosure a heatpipe.

Vageu ideas I've never tried... If I wanted a flat slab, I might try sandwi ching a piece between 2 flat plates to sand it, hoping to avoid distortion by machining. Or just press a larger piece than needed thinner. I've no ide a what sort of temp drop a pressurised air jet direct on the FET would get you.

NT

The freeways around Stuttgart are not that free. Limits, lots of automatic & mobile radars and normally traffic jam, even at night. Absolutely no fun.

Porsches are normally not very fast. Methinks they are happy with the insight that they could if they wanted to. No point in demonstrating it. The outcome is clear to see.

Large Audis are usually the fastest ones. People who had a Rabbit GTI (?) as youngsters, and then had a career. They have stayed with their supplier.

But I disagree that everyone should have these experiences. If they happen, that means that your gut feelings/driving experience and the physics are not yet in tune. The difference is not only after driving fast but also when you start driving fast.

I see that often enough on the A81 freeway between Stuttgart and Switzerland. Switzerland is about as speed-limited as possible outside of a parking lot. Now & then some people from Switzerland awake from their locked-in-syndrome, get into their high-end car, cross the border to .de and try speeding. But you see that they have no experience. They have it built-in that all others drive as they are used from Switzerland. No situation awareness. The result is that soon there will be severe limits on most of the A81.

I must admit, that I drive my BMW quite often until the artificial limit on nightly commutes between Stuttgart & Saar county, but it took 20 years to learn it. It teaches a lot how short 500 m can be when you are fast enough. Small steps!

Gerhard

It's a lesson in perception. But I found the 500sel boring, if comfortable.

NT

My experience was some time ago. The freeways were wide open then. About the only problem was to keep a sharp lookout behind you. Some idiot in a souped- up racer would blast past you at 200 mph. One slight mistake and both of you would end up in a crumpled ball of steel.

I asked my German friend why the car insurance rates were so much lower than in the States.

He said, "There are no survivors."

You were concerned about the dielectric constant of water. Is the anode connected to the heat sink? And are you running 600 V or so on the anode? Does your customer know anything about water electrolysis?

Maybe the only thing that is saving your bench setup is the long lines to the bucket. They will present a high resistance and limit the current flow. This is great for a 5 minute demo, but not so good for weeks or months of service.

If you are putting 600 V on the aluminum heat sink, I'd have a long discussion with your customer about the problem.

Old water-cooled vacuum tubes specified very pure water - 1 uS or better. 1 uS = 1 megohm per centimeter. Here's an example:

"Tube life can be seriously compromised by water con- dition. With contaminated water, deposits will form on the inside of the water jacket, causing localized an- ode heating and eventual tube failure. To minimize

should always be one megohm per cubic centimeter or higher. The relative water resistance should be periodically checked using readily available instruments."

:

is

ic

.

ge

al

;

chips.

e

term.

er pipes, some mild steel radiators, and some cast aluminium radiators.

or precisely such systems

0_2.1_us.pdf

drogen emission - I had the vent the system every few months - but it worke d fine for some fifteen years

a

as 1% of the volume of the circulating water (and working out what that wa s a tedious exercise, involving draining the entire 600 litre system into a bucket, one bucket at a time, weighing the bucket, and adding up the tota l weight).

There's nothing - except the price - to stop you buying a 0.5 litre contain er and dosing your system with the 20 millitres it would need.

Don't pipette it by mouth.

I've now remembered the products I used, sold by Fernox. The Fernox ALU pro duct I used in the UK was superseded by the Fernox Copal I used in the Neth erlands, and seems to have been superseded again.

F1 seems to need to be added as at 4% rather than 1%.

Amazon stock an even smaller bottle

4P7PY

It isn't cheap.

--
Bill Sloman, Sydney

the

ns,

they

ely

John Larkin doesn't seem to understand how a sealed system works.

You do have make sure that brazing alloy you use to put it toghetrer isn't wildly incompatible with the metal (generally copper) used for the pipes an d the wick, but the water you put in should be de-ionised and CO2 free, and stay that way, so it doesn't encourage electrochemical corrosion.

People have been using heat pipes for a whiole now - I got my hands on my f irst one thirty years ago - and seem to have worked out what works.

It's set by the laws of physics, which do seem to be stable. It's not clear which ones might be doing the limiting.

--
Bill Sloman, Sydney 
> >

Amazon stocks some of these type of "PC coolant loop protectant" additive pre-mixes, not super-cheap, a little non-specific about what it actually contains other than "a blend of refined vegetable extracts with non-toxic corrosion inhibitors and non-toxic dyes."

.

Idiot. Any heat sinking system is an additional thermal resistance to ambie nt, but a lower one than relying on direct heat transfer from the FETs to f ree air.

The point about a heat pipe is that it lets you put boiling liquid within a fraction of a millimetre of metal of your FET, and lets you condense the l iquid being boiled off over a rather more extensive area (which can be wate r cooled or air-cooled) which can be some way away.

It's hard to set up a lower thermal resistance arrangement.

Any heat sink can only move heat - the interesting question is the thermal resistance it presents between the heat source - your FET - and ambient (wh ich can be room air or flowing cold water).

Heat transfer into a film of boiling water is remarkably fast. Steam raisin g in power stations relies on it.

A thin film of water may have a high dielectric content, but the steam bein g generated (not necessarily at 100C - in a properly sealed and evacuated h eat pipe the boiling point of the working fluid is set by coolest point alo ng the heat pipe, where the working fluid is condensing.

Specialised devices usually are. That's what keeps your business profitable .

A heat pipe will do better than than a lump of copper of the same shape.

It does if the heat is being spread by the flow of steam - it shifts a lot of latent heat of evaporation as it moves, and dumps it rapidly when it con denses.

So far your thoughts on heat sinking haven't been all that professional.

That fact that that you charge for them anyway reflects the fact that you a re a trifle over-confident.

--
Bill Sloman, Sydney