flexible/ low modulus highish thermal conductivity potting compound

[glenenglish]

I'm looking for somethign with a bit of give, IE not epoxy , that has better thermal conductivity (TC) than good old neutral cure silcone.
operating temperature of about 125C max.

there's lots of similar threads, couldnt find one with moderate - high TC

stuff like this is good, high TC is low.  ( Chemtools PCT-7000Y )

ideas ?  suggestions  ?  with thanks .
glen

[Berni]

Well silicone is unfortunately not particularly good at thermal conductivity in general.

Before you say, Yes they also use silicone for some thermal insulation pads for mounting semiconductors to heatsinks. They do conduct fairly well, but that is mostly because they are made as thin as possible (and usually with embedded fiberglass to make it not tear too easily at such a thin thickness), so that's how it gets a decent amount of conductivity, by placing as little silicone in the way as possible, not because that special silicone blend would be a very good conductor(tho it does likely contain a ceramic or something mixed in to boost conductivity). Even then the pad restricts thermal conductivity quite a bit as compared to just having thermal paste under there.

So if you need a lot of thermal dissipation you might have trouble with even the special more conductive types. A more effective method is likely embedding some thick aluminium plate inside your potted assembly that touches the heat generating components and conducts the heat to the sides of the potted block, there it only has to pass trough a thin layer of silicone to get out. If you can just reduce the distance the heat needs to travel by 1/2 you get the same gain as having 2x as conductive silicone.

[glenenglish]

yes.. except that  but surely it would need alot of fill powder before it improved much.
must hold 1kV between any volume
although, it will be silicon conformal coated before the pour.

[Stringwinder]

I have used potting compunds from Wacker Chemie (RTV-ME 607) and they offer a wide range.
RTV-ME 607 has low viscosity uncured and I have used it to make moulds to copy things
besides potting electronics.

https://www.wacker.com/h/medias/7094-EN.pdf

[jpanhalt]

If you are in the mood for a DIY solution, boron nitride (BN) is an electrical insulator, dry lubricant, and has good thermal conductivity (TC).

See: https://en.wikipedia.org/wiki/Boron_nitride

There are lots of papers on using it to enhance the TC of epoxy.  Presumably, it would also improve the thermal conductivity of any RTV or silicone.  It's not terribly expensive.  Be aware that its different crystalline forms have different TC properties.

[glenenglish]

jpanhalt
great suggestion - with thanks. Also, string- good range of options  from Wacker.

Another option might be to beef up the inner planes (add them) and have thermally conductive glue for the hotter passives onto the PCB and use the planes to spread it.

The problem comes from some designs that are loaded with large diodes, large resistors, large caps  etc all large to get the heat out, for designs that need to be partially potted . I've found a silicon conformal coatign (DCA is what I use) slightly reduces radiative cooling. 

Looks like I dont need to operate some sort of reactor in my backyard to obtain Boron Nitride.
Reading up on 'properties' in that wiki entry.- WOW this is an interesting material. Actually, I recall now, BN is used for substrates in some transistors. - glen

[tszaboo]

I really suggest not selecting a potting compound from a catalogue, but actually talking to your supplier. And then taking measurements. Even relative bad thermal conductivity will vastly improve the temperature of the components, because air is practically an isolator. The supplier will offer it's basic potting compound suggested for electronics, then you should test it before even looking at any of it's thermal parameter.

[glenenglish]

naturally
air is ~ 0.02 ?
RTV is mostly 0.2, some are 0.4

I know all this is thermo 101, some sometimes it helps to discuss with others to get your brain thinking about it.

If the enclosure  it is in is hotter than the components and PCB, the potting insulation will reduce radiative contribution from the hot enclosure to the PCB, but increase the conductive transfer
If the enclosure is cold, the RTV will reduce radiative transfer from PCB to the case , but increase conductive transfer.

Will need to run the calcs. Depends on how hot the thick metal enclosure gets . I've generally worked in environments where the environment is cooler than the PCB and parts.

[DiodeDipShit]

Thinking out of the box, would isolating the components from the enclosure, into their own temperature regulated environment be viable?   Vespel SF-0920, Vespel SF-0930 or Vespel SF-0940 Polyimide Foam may be worth thinking on.

[Stringwinder]

If you have large diodes that act as internal heat sources you also have nice
temperature sensors inside. With two (or four) wires to a diode you can measure
its forward voltage drop from the outside. It will be the "chip" temperature
which is the highest and most important one. Run at "full speed ahead" until its
hot and measure diode voltage immediately after shutting down. The diode voltages
(when running) are probably not compatible with external instruments. A four
wire measurement will/can have thinner wires that will not conduct heat to the
outside.

Do read german product data sheets and catalogues - "Ordnung muss sein!" is
part of their culture.

[glenenglish]

probably a good temp sensor indeed . Power diodes yeah they are the toughest nuts in the business .. 

That Vespel material reads very special. wow. Looks a bit like used-set  self-expanding foam....

This is a sensor situation where the outside can  be very hot .---- this is a hobby project --- , not a commercial one, but it brings me to new challenges, so.. interesting....

If insulated, eventually it will get hot inside . WHich drives a few decisions. I'm sure there are plenty of products faced with this sort of situation as the norm. If your own heat generation is low enough, and perhaps the insides have some thermal mass, it's possible you might not have to cool it within the operation time. Otherwise tolerate it / mitigate the effects  , or remove the heat....  with some sort of heat exchanger.

[tszaboo]

If you have large diodes that act as internal heat sources you also have nice
temperature sensors inside. With two (or four) wires to a diode you can measure
its forward voltage drop from the outside. It will be the "chip" temperature
which is the highest and most important one. Run at "full speed ahead" until its
hot and measure diode voltage immediately after shutting down. The diode voltages
(when running) are probably not compatible with external instruments. A four
wire measurement will/can have thinner wires that will not conduct heat to the
outside.

Do read german product data sheets and catalogues - "Ordnung muss sein!" is
part of their culture.

Meh, just glue some thermocouples to the hottest points and do the potting.
You don't really need to complicate it.

[Stringwinder]

Meh, just glue some thermocouples to the hottest points and do the potting.
You don't really need to complicate it.

Great idea if you want to know the temperature of the glue heated through the diode
plastic body and cooled/heated by potting compound.

[DiodeDipShit]

"into their own temperature regulated environment"

That Vespel material reads very special. wow. Looks a bit like used-set  self-expanding foam....

This is a sensor situation where the outside can  be very hot .---- this is a hobby project --- , not a commercial one, but it brings me to new challenges, so.. interesting....

If insulated, eventually it will get hot inside . WHich drives a few decisions. I'm sure there are plenty of products faced with this sort of situation as the norm. If your own heat generation is low enough, and perhaps the insides have some thermal mass, it's possible you might not have to cool it within the operation time. Otherwise tolerate it / mitigate the effects  , or remove the heat....  with some sort of heat exchanger.
[/quote]

I was thinking piped coolant into the isolated components or ducted air from the outside. Exterior environment considered.

[tszaboo]

Meh, just glue some thermocouples to the hottest points and do the potting.
You don't really need to complicate it.

Great idea if you want to know the temperature of the glue heated through the diode
plastic body and cooled/heated by potting compound.

For any heat generating part, the difference of temperature between a case temperature and the thermocouple will be negligible.
The diode method is only useful if you want to know the junction temperature of a diode that was switched off a while ago, cooling down...

"into their own temperature regulated environment"

That Vespel material reads very special. wow. Looks a bit like used-set  self-expanding foam....

This is a sensor situation where the outside can  be very hot .---- this is a hobby project --- , not a commercial one, but it brings me to new challenges, so.. interesting....

If insulated, eventually it will get hot inside . WHich drives a few decisions. I'm sure there are plenty of products faced with this sort of situation as the norm. If your own heat generation is low enough, and perhaps the insides have some thermal mass, it's possible you might not have to cool it within the operation time. Otherwise tolerate it / mitigate the effects  , or remove the heat....  with some sort of heat exchanger.

I was thinking piped coolant into the isolated components or ducted air from the outside. Exterior environment considered.
[/quote]
What's hot? 85 degrees on the outside?

[glenenglish]

Hi Tszaboo
Outside box still air temp is 112C-125C. Radiative black body temp the box will see is about 90C (there is  a heatshield) .
Internal heat - 0.35W (device 1 - electronics good for 105C  ambient ).  and 10W (device 2 electronics good for 100C jn, 80deg ambient)
Run time  - 6 hours

Basically I have to calculate whether sufficient internal thermal mass + insulation might do the job for device  # 1. That's easy enough

Device #2 will need the plumbed-in cooling.

Originally, I was going to have a internal cold plate/ cold box, with the thermally conductive potting material coupling the heat to the cold plate (cooled via piped air ) (hence the question about conductivity and potting)  , and insulation to the 'outer box'
If the outer surface of the outer box box is shiny, that will reflect the radiant heat, reducing that heat load input, leaving only conduction (from air) .

Piped air might be the best cooling solution, since water will need a loop and the consequences of a leak are fewer.
Air is surprisingly 1J/g/K  specific heat, and with air density call is 1kg / cu3 ---  a cubic meter of air per second will remove 1kW of heat per K difference.
In this case I want to remove 10W, and if the input air temp is 50deg C and desired internal temp is 80 deg C, I have a 30K difference and need 0.333grams/second of air.
or about 0.34 litres per second (20 lpm)  assuming a 1/4" ID pipe , 5m long , pressure drop at 10 m/s airflow will be only 20mBar.  easy.
So, plumbing in some low pressure airflow seems the easiest option.

Will do the sensor calcs next...IE thermal mass+ insulation.  6 hours at 0.35W input is total 7.56kJ . Ally is 0.9J/g/K, so if it starts at 50 deg C and terminates at 105C, I have 55K of difference, and I need 152 grams of Ally (57 cu cm) or a plate 5mm thick, 10x10cm... a bit big so I will need to thicken it up (or use copper - approx 2.5x higher specific heat). Copper it will be.

That's without any external heat input.  External heat input area perhaps 0.015m2 area. Now i need to calculate the heat transferred to the box  total area of 0.015m2due to conduction from the 125C air temperature.
 ( will assume all sides of a cube are receiving heat equally to simplify the job)

[IconicPCB]

Glen do check the toxicity of boron .
I seem to recall RF power transistors used to be ensconced in a white ceramic like material ( boron? ) and i had to be particularly careful not to chip it as the particles if inhaled would be deathly.

In the mean time do consider a double-dable approach.
a dip into thin silicone fluid cured and packaged in say aluminium loaded epoxy.

[coppercone2]

boron is boring you thinking of berylium which should be safe enough to use but the thing to know is that you don't want to powder the pads i.e. saw it off to make it fit on a old HP heat sink with a lip or something.

I noticed with the TO-3 packages its easy to get into a situation where you want to modify the ceramic thermal pads, so if you happen to have berylium oxide pad, and it does not fit, use something else

Often there is a fit problem of like 0.5mm because you got a different but similar pad handy. It is tempting to want to sand this, but obviously don't if you suspect berylium ceramic (white, should feel metalic). not to be confused with greyish aluminum nitride

[coppercone2]

as for your problem, how about heat pipes? They have heat pipe kits.

not sure how well it would work or how well it looks like but it might be possible to make your design forced air friendlier with heat pipes. I just wanna hear about a custom heat pipe solution for those heat pipe kits


and my opinion about thermal mass... not sure what your application is but specs are liable to change. Someone is gonna want 8.5 hours or ask if can be left in there indefinitely. This is just a specification I would personally be suspicious about if someone asked me for it in corporate world.. test time is not something that ends up being accurate with them.

Not to mention if you are making a series and they all work the same then your devices will have simpler documentation and the training requirement will be lower. Before angry dumb people start complaining they could not start a test because there was a defective box without cooling or whatever after they throw away the manual to cover their ass. Or they melt the big one because the little one worked without forced air and they thought they could wing it by lowering the temperature by 3 degrees to "compensate" (aka gamble irresponsibly) and plug the air holes with crumpled paper

And of course there is always forgetting about it, lying about the test time, and then you have to prove its not possible!!

I have seen corporate burn down a test chamber. I don't trust them with heaters. They will think of something. And blame you for it. Like that your box tripped the breaker controlling the safety thermostat or something.

[jpanhalt]

@Iconic:

The toxicity is to be concerned about is of boron, but boron nitride.  From Wikipedia:

Boron nitride (along with Si3N4, NbN, and BNC) is generally considered to be non-toxic and does not exhibit chemical activity in biological systems. Due to its excellent safety profile and lubricious properties, boron nitride finds widespread use in various applications, including cosmetics and food processing equipment.

  Boron fibers, which are used in some very light composites, can be quite dangerous.  They are radiolucent and had to find.  BN doesn't share that property.

BN is like a perfect atomic mating.  Boron lacks 2 electrons in its p orbitals.* Electron configuration:  [He] 2s2 2p1. Nitrogen has two more than it needs (i.e. non-bonding).  Configuration  configuration:  [He] 2s2 2p3

*Boric acid is ulike acids like water, HCl and H2SO4.  Rather than losing a proton as those other acids do, boron adds a hydroxyl from water (usually) and releases the proton from water.

[coppercone2]

I guess you gotta be careful when your doing body work on tomcat f14s

or as usual, golf clubs. I wonder when they figure out how to put plutonium in them

[jpanhalt]

I guess you gotta be careful when your doing body work on tomcat f14s

An when building superlight rubber-band powered models that stay aloft under power indoors for an hour.

[coppercone2]

https://www.sandia.gov/labnews/2018/11/21/f1d/


a better one is to see how long you can keep a plastic bag in the air near some apartment towers. I remember once there was basically a stable orbit with a plastic bag which turned into a nascar plastic bag eventually before getting stuck in some leaves

[IconicPCB]

damit.... Beryllium not boron.
Sooo close ...

Thanks Coppercone.

[coppercone2]

at least we know to avoid fishing poles and golf clubs now

I mean its probobly a 99.99% chance I would still wear a mask because it would get mistaken for fiber glass