Cooling and insulating fluid

[David Lesher]

I'm working on cooling the heat sinks on an electric car project.

We are considering DOT5 brake fluid, but I also talked to
3M. They have a line called "Novec" and we may use their 7500.
It's not cheap but is available in small quantities..

Call them up and see what they recommend for your project.

3M Engineered Fluids is the name you want.

 

[Robert Baer]

De-ionised water would work. I've seen that used to cool the floating
anode of an X-ray tube running at that kind of voltage. It's got a
higher heat capacity and a lower viscosity than most oils.

An alternative would be a heat-pipe, if you could guarantee that the
fluid being evaporated could be spread evenly over all your zeners.

Plugging them all into a block of a alumia - which does have a
respectable thermal conductivity - 29 W/m/K - could work. Aluminium is
better - at 201 - as is copper at 385 but both conduct. Diamond is
brilliant (pun intended) at 900, and it is a good insulator - a vapour-
deposited layer might be worth the trouble and expense if you were
really pushing the state of the art.

What i did (before test of PCB) is use 0.020 Getek, did a "pour"
around each part for a heatsink, and a rectangle on the back as added
heatsink.
Each PCB has 168 zeners, a test at 1.4mA (about 25W) yields about
150mW/zener which has a max rating of 225mw "On FR - 5 board using
recommended solder pad layout"; the PDF showing _only_ pads for the SOT-23.
At a max 1mA use, i have a lot of elbowroom if i had used their
"layout"; so i am guessing at least a factor of two for "still" air.

I thought of water, but it is too easily contaminated with ions to
make it practical except for use in a controlled lab-type environment.

This project is for hobby use; re-build Victoreen Corotron(TM) HV
regulators used in some TVs,so cost is a definite factor.

 

[Robert Baer]

Will they sell you a small quantity? That is always the problem with
"cool stuff". Often it is easier to sweet talk a rep than to buy the stuff.

That said, I've always seen mineral oil used, but that doesn't mean FR3
isn't better. It is just a matter of availability.

I know zip about regulating 25kV, but meditating on such a string of
zeners, the first thing that comes to mind is a string of devices just
needs one device to fail for the whole string to fail. Is failure
tolerable for your application. I means, if nothing catches fire or
blows up, that might be OK. If a zener failing means you have a IED, I'd
consider a back up safety scheme.

Failure is an option as long as lawyers don't get involved.

I saw the post about using water, but you might want to consider
expansion. I don't know what it takes to boil water in a small tube.
Well actually I could compute it, but well you know...that would be work.

It is a bitch to get electrical specs for any oil, and in some cases
hard to get the boiling point as well (flash point seems easy).
The FR3 seems to be better than the "plain" mineral oils
electrically, boiling point and thermal.
But there is an indication of many mineral oils,and then the
possibility of using silicone oil.

Sigh, choices....with so little data to help.

 

[Robert Baer]

3.2 beer is an insulting liquid

How about the burp?

 

[Robert Baer]

ATSM1816 is a 0.100" gap.
This is similar to Shell Diala AX which is about 32kv for ATSM 1816.

It will be fine, as long as you have the right spacings and creepage
distances.
You need a expansion thingy there as it will probably expand 8% or so
from 25c to 40c.
and make sure your seals are Buna N.

I take it that Shell Diala AX is an insulating oil?
This is typical; zero electrical info; those ASTMs not in the MSDS
because they do not care about anything not related to "safety".
Yes, it does seem that it is an oil for use like i need, but zero
info makes it useless.
Fortunately, there is a _different_ data sheet pointing to
unavailable D877 and D1816.
At least,the numbers are larger than what the stress is going to be..

 

[Robert Baer]

Robert is proposing 25 KV over about 6 inches. That's 350 times the
voltage gradient of the RCA transmitter, so assuming linear scaling
he's have to flush it about every day. And I don't think he'd include
a circulating system or purity testing. It wouldn't last long.

Oh, it is WORSE than that..the six inches is the vertical spacing;
there will only be about 0.11 (max?) spacing from the edge of a centered
PCB and the inside of the cylinder.
Crude representation (not to scale):
+----------+
| | cylinder wall
| |
| +------+ |
| | | |
| | | |
| | | | PCBs inside
| | | |
| | | |
| | | |
| +------+ |
| |
| |
+----------+

 

[Robert Baer]

Maybe a DOT 5 brake fluid?

Well, i _is_ silicone oil; no electrical specs tho..
And i see that it has an additive - the coloring, which could mess it
up..

 

[Robert Baer]

Yeah, with ten feet of looped hose running back from the anode (maybe
less, but always some length, even if the DI is fresh).

At mA, inside a case, without pump, filter or replacement, you're nuts.

DI goes bad after a while, because there are always ions to dissolve, even
from nonobvious sources. Analytical chemistry texts go into grotesque
detail when ultrapure water is desired: even quartz glass apparatus
contains enough traces of sodium to foul up a sensitive ICAA test. The
solubility of silica alone becomes noticable.

As oil goes, how well it cools depends on how well convection works. In a
tight case, even water will end up significantly worse than potting
compound. A good epoxy or silicone potting compound, advertised for high
thermal conductivity ("high" meaning ~1 W/(m.K) or thereabouts), will do
better than stagnant water.

Tim

I have always considered the so-called high thermal conductivity
epoxies to be excellent thermal insulators; just slightly less than the
"standard" ones.
And i have yet to see electrical and thermal specs (together) for any
of them..

 

[Robert Baer]

A good - heavily loaded - potting compound should do better than
stagnant water, but only if it actually does fill the volume involved.
You've got to evacuate the volume you want to fill and de-air the
potting compound before you let it into the - evacuated - volume.
Water does have a conveniently low viscosity.

Farnell does silicone potting compounds with a thermal conductivity of
0.9 W/m/K which isn't all that much better than water

http://www.farnell.com/datasheets/1633577.pdf

http://www.farnell.com/datasheets/515246.pdf

There's an epoxy which does appreciablyt better at 1.25 W/m/k, but
it's not cheap.

http://www.farnell.com/datasheets/1484800.pdf

Thanks.
What pisses me off is that i had an ad from one of the trade mags
(EDN or ??) of a very pourable "thermally conductive" HV epoxy..and i
tossed it because i consider the conductive part to be a convenient lie.

 

[Robert Baer]

Have you seen what that does to paint? God only knows what it does to FR4
and component encapsulants longterm.

OOPS! Thanks!!!

 

[Robert Baer]

dot 5 is paint safe, it's a different formulation, mostly silicone.

It is the "mostly" part i am worried about..that damn purple coloring
additive..

 

[Robert Baer]

I'm working on cooling the heat sinks on an electric car project.

We are considering DOT5 brake fluid, but I also talked to
3M. They have a line called "Novec" and we may use their 7500.
It's not cheap but is available in small quantities..

Call them up and see what they recommend for your project.

3M Engineered Fluids is the name you want.

Now THAT data sheet has it all; no MSDS bush beating.
I see they have other 7xxx; is the 7500 top of the thermal line?

 

[josephkk]

Less corrosive, too. When ultrapure water becomes less pure, by
dissolving metals out of the things it contacts, those metals are...
well, no longer doing their original jobs. The better a job your
deionizing cartridge is doing, the greater of the tendency of the
purified water to eat away at your components.

Wright arm. Water really is the universal solvent, ultrapure water
dissolves PTFE Teflon, polyethylene, gold, glass and everything else
(though several of these very slowly). I remember stories of
semiconductor fabs being surprised that their ultrapure water was
dissolving enough Teflon tubing to contaminate some processes.

 

[Martin Riddle]

I take it that Shell Diala AX is an insulating oil?
This is typical; zero electrical info; those ASTMs not in the MSDS
because they do not care about anything not related to "safety".
Yes, it does seem that it is an oil for use like i need, but zero
info makes it useless.
Fortunately, there is a _different_ data sheet pointing to
unavailable D877 and D1816.
At least,the numbers are larger than what the stress is going to be..

Here is a data sheet,
<http://www.nttworldwide.com/docs/diala-axbrochure.pdf>

Now I don't know if you can get it in small properties, but it looks
the same as the FR3.

D877 and D1816 are just test methods.


Cheers

 

[Robert Baer]

At your power levels, any epoxy should work. We use a Thermalloy
thermally conductive epoxy, but it's expensive and not radically
better a thermal conductor than plain epoxy.

Use cheap epoxy and degass it after you fill the tube. It's messy, but
oil is messy too.

How about some wax? It's cheap, removable, and not messy like oil.

Problem with most of the waxes that have _some_ info available, is
the low MP and flash point - making them a definite hazard.
So hi MP waxes may be better, but there seems to be no info on them
except that they presumably exist.

 

[Robert Baer]

Put a tiny fan and a filter at one end, and air-cool the stack. Extra
credit if you can get the fan to run from the 1.4 mA stack current.

Not too bad idea for some users; the picky ones want their TV sets to
look original.
Know of any fans that have 100K resistance? That way,the higher the
current, the faster it would run: 100V at 1mA for 25W cooling.

 

[Robert Baer]

Motor oil is an excellent insulator. I used to use plastic tubes
filled with motor oil to extend the end of a car ignition coil
bushing. Got 6" sparks, with a big oil cap and a thyratron blasting
the primary. The old car coils were oil-filled inside, too.

Yes, many are good - but zero electrical specs and hard to find
thermal specs.

 

[Robert Baer]

Here is a data sheet,
<http://www.nttworldwide.com/docs/diala-axbrochure.pdf>

Now I don't know if you can get it in small properties, but it looks
the same as the FR3.

D877 and D1816 are just test methods.


Cheers

Yes, that is the one i found. Note the unavailable "D" specs and most
especially note zero thermal info.

 

[Bret Cannon]

"Robert Baer" wrote in message

At your power levels, any epoxy should work. We use a Thermalloy
thermally conductive epoxy, but it's expensive and not radically
better a thermal conductor than plain epoxy.

Use cheap epoxy and degass it after you fill the tube. It's messy, but
oil is messy too.

How about some wax? It's cheap, removable, and not messy like oil.

Problem with most of the waxes that have _some_ info available, is
the low MP and flash point - making them a definite hazard.
So hi MP waxes may be better, but there seems to be no info on them
except that they presumably exist.

Apiezon waxes are high melting point and low vapor pressures and some
reported electrical properties.
See <http://www.sisweb.com/vacuum/sis/apiezon3.htm>

 

[Robert Baer]

some sort of boiling liquid is going to do better than an unforced circulating
liquid. build a heat-pipe.

* How? Method, materials, etc needed.

25W is a lava-lamp scale energy flow. I don't think there's any need to
chase the maximum

* "chase"?