heat dissipation in potting compound

[Simon]

How much heat can i dissipate from a potted circuit ? How much should I de-rate parts that will be potted ?

[IanB]

Given that electrical insulators are commonly also thermal insulators, I would think not very much heat at all. If you want a good service life, I would not pot any circuits that dissipate more than a few mW.

[Simon]

well no component will dissipate more than 100mW, the whole thing should only dissipate no more than 400mW. So say I have a resistor that dissipates 0.1W, what size should I use to compensate ? unfortunately it has to be potted to make it water proof.

[IanB]

Unfortunately, 400 mW is quite a lot of heat relative to small confined spaces. If you trap that heat in a small space and don't give it a way to get out the temperature will go way up.

I would start by checking the data sheets and application notes for the potting compound you plan to use and see what advice they give about thermal conductivity and heat dissipation.

If you simply use a larger resistor in an insulated environment it won't necessarily help. The larger resistor has a higher power rating only because it has a larger surface area to conduct and radiate heat away. If you wrap it in insulation the larger surface area will be to no avail.

But another problem is that the trapped heat will increase the temperature of other adjacent components in the circuit like capacitors and transistors and will reduce the service life of those.

Have you considered a small aluminium box enclosure with a rubber seal? The free air inside will transfer heat to the walls and the aluminium walls will conduct the heat to the outside.

Also, rather than potting your circuit you could apply a conformal coating. Conformal coating and putting inside a weatherproof enclosure should handle most automotive environments.

[Simon]

hm, I'll have to see it's for work so they just tell me what they want and expect me to make it work, I'll make some enquiries. we are not potting it so it's also dependent on what the subcontractor assembling it decides

[NiHaoMike]

It depends on the potting compound. But in general, solid materials conduct better than air. (Thermal insulation is actually mostly air or some other gas.)

[Simon]

individual resistors do not dissipate more than 100mW

[IanB]

It depends on the potting compound. But in general, solid materials conduct better than air. (Thermal insulation is actually mostly air or some other gas.)

That's true, but to make air an insulator you have to trap it in foam or wadding so it can't move. If the air is free to circulate it can convect heat away quite effectively.

[HackedFridgeMagnet]

If the circuit is going to be enclosed in a box the the air flow within the box may not be that great anyway.
At work we have a circuit that generates over 2watts of heat that is potted and works quite happily, although we did run a copper strap from the warmest part of the board to the Al case.
But you can feel the warmth of the potting mix so I think there is a reasonable amount of heat conduction.
If you want to be sure pot an NTC into one or your circuits and measure it.
Over temperature is one of the most common causes of failure, so I would test this aspect carefully before going into production.

[peter.mitchell]

can you just put all the high heat devices on one side of the device, very close to the edge of the potting compound, then heatsink that side?

[SgtRock]

Dear Simon:

--It seems to me the best and simplest solution for you may be to pot one up and measure. Or pot something up that is close in terms of shape and power consumption, and measure at the highest likely temperature and duration you expect. Then you should get some idea, of how to proceed. You seem to get real engineering jobs with real problems. Excellent experience.

"Measure twice. Cut once"
Norm Abram 1950 -

Best Regards
Clear Ether

[robrenz]

Use low viscosity potting material and heavily load it with tetrahedral boron nitride powder. It is the highest thermal conductivity next to diamond that is non conductive. Then pot a properly sized heatsink base right into the potting material.  Try to minimize the thickness of potting material between the heat generating components and the heat sink base.

[Simon]

You seem to get real engineering jobs with real problems. Excellent experience.

The problem I got is an employer that does not like testing or doing R&D and finding stuff out and then using the information. they like someone else to take the bollock 

[chickenHeadKnob]

 There is a standard approach to this problem, just check the firewall in the engine compartment of an automobile. The automotive guys generally  mount all heat sinked components (to-220 and such) to an aluminum angle bracket that forms one side of a potted enclosure with the other sides plastic with an open top. Potting compound is poured in giving hermetic seal then unit is mounted by the exposed heatsink/wall/bracket to some other surface that dissipates the heat. Cheap and cheerfull and functional.

[jeroen74]

You know that potting brings a whole lot of issues with it? Mainly stresses on solder joints while the potting compound cures and differences between thermal expansion of the compound, the components and the PCB. You might find that components are simply ripped off the PCB with temperature extremes. Some components don't like potting. Silicone based stuff and relays that are not hermetically sealed do not go along.

Not to mention extra cost, unservicability etc.

When I worked at Philips Lighting the rule was to avoid potting at absolutely all costs. Unfortunately you sometimes just can't escape it. And you would not believe it, but all of their electronic HID ballasts are potted, here it comes, in asphalt That stuff they pave roads with.

Asphalt does have a few advantages: when components get hot, the asphalt melts slightly, forming a good thermal interface. It's cheap. Does not cure, just solidifies. And you can remove it by soaking the product in the appropriate solvent.

You can't add potting into the equation as an afterthought and expect no problems.

[SeanB]

Most often a flexible conformal coating is a better bet, along with a heatsink area, and placing all power dissipation there. If you need full encapsulation use a conformal coat and then a flexible potting compound afterwards, as this will allow thermal expansion to occur without risk of fracturing the components.

Make the board, test and then dip in a varnish and allow to cure fully ( possibly at an elevated temperature) then use a flexible encapsulant to fit it in the housing.

[jeroen74]

Another important point: potting can influence the EMC behaviour of the product. It can easily decrease the margin by 2dB. Not a problem if have 10, but it is if you only have 2 or even less.

[Simon]

well potting is not an afterthought, it's a stated requirement to make it water tight, I've kept dissipation in each resistor to under 100mW, the problem here is i'm running a circuit on 5V from 24/28V so I have to to something with 28-5*I.

[G7PSK]

Many of the AVR boards on the alternators that I have used are potted in a black compound that does not melt with a hot air gun making them irreparable but they seem to last ten years on average. They usually seem to control a field current of between 30 to 90 volts at 1 or 2 amps. On some of the units there is no heat sink just a FR4 board in an inch or so of potting compound both sides of the board.

[lewis]

Simon, just go for it. We pot almost all our products one way or another in PX700K epoxy from Robnor. It's very thermally conductive and we regularly pot components dissipating a few watts (we do ensure there's lot of copper fill on the PCB to spread the heat, it's important to avoid hotspots). We heat the unmixed epoxy to 50 degrees to reduce its viscosity prior to mixing, and owing to its low exothermicity and 24h cure time it does not seem to cause any component stress whatsoever.

Coefficient of thermal expansion has never been an issue for us, we regularly pot 0603 passives, QFPs and QFNs on standard FR4 with no problems of components being ripped off whatsoever. It's far more thermally conductive than air, even free-circulating air, unbeatable for ingress protection, and great for protecting against tin whisker growth, vibration damage and even arc-flashover. We've never had EMC problems using it either.

We ensure component temperature is limited to less than 70 degrees in the resin under worst case conditions (if we're not sure we'll pot an LM35 or a thermocouple right at the component in question to test it). The 3-yr failure rate on our potted products is less than 0.1%.

We use the 1kg twinpacks of the PX700K resin, it's cheaper for me to employ somebody to fill the products by hand than it is to buy and maintain the dispensing machine from Robnor. We cure under an IR lamp at about 30 degrees for 24 hours. I suspect problems occur if you try to cure faster.

We make a lot of IP rated stuff and stuff for marine environments, and potting is indispensible. In fact I try to use it as much as possible. It's costly and labour intensive, but the benefits are numerous in all respects. I don't see what all the fuss is about....

[jeroen74]

The fuss is valid if you manufacture millions of your product per year that contains high power electronics that's are difficult to get EMC compliant in the first place, have a ten year outdoor lifetime requirement, where every penny counts, competition is fierce, and where failure rates of 0.1% just don't cut it

[Simon]

oh well they just decided to put the whole unit being controlled by the circuit in a box that can house the circuit too so no need to pot anymore grrrr, time wasters !

Thank you for all your help, will no doubt need it again soon

[Neilm]

There are some potting compounds out there that are electrically insulating, but thermally conductive. I have used one that was epoxy based - mix the two components then pour. Leave for 48 hours then it is done. I used to pot a small transformer and some circuitry. It probably dissipates about 3 Watts with no problem.

Neil

[lewis]

The fuss is valid if you manufacture millions of your product per year that contains high power electronics that's are difficult to get EMC compliant in the first place, have a ten year outdoor lifetime requirement, where every penny counts, competition is fierce, and where failure rates of 0.1% just don't cut it

I think automotive electronics, in which potting is extensively used, fulfils all of those criteria. 

[jeroen74]

I think so too, and I'm sure those manufacturers put in a lot of effort to make sure the potting does not cause any problems One Ford ignition module I once took a look at had its hybrid module covered in an extremely soft and clear compound, still very flexible.