Thermal compound and insulator sheet like orignal or change to Sil-pad?

[jacamo]

Hello,

I am doing a repair on a ballast and was wondering what the "proper" or correct way I should re attach these power devices to the heat sink for the best thermal transfer.

I have attached a photo of my situation. The orange insulating sheet that came from the factory with two mica insulators for two of the devices and the rest had heatsink compound on the back of the components up against the insulating sheet. I do think they had heatsink compound on the side of the sheet that touched the actual heatsink as well but its hard to tell. This is a pretty old device and it was all dried and caked on there.

I have the full set of schematics and it shows the sheet on the assembly instructions but doesn't mention heatsink compound just the placement of the insulating sheet.

I have the regular heatsink compound and could restore it to how it came out of the factory. My only question is if I do it that way should I put heat sink compound on the side of the insulator sheet that touches the heatsink? Presumably like how they had it, but more importantly how should it be done? 

I also have the blue sil-pad type stuff that is fully insulated and I am now wondering if I should just use that stuff. I hate using that heat sink compound because it gets everywhere if your not super careful.

Any downsides to using the sil-pad instead of the heatsink compound and the insulating sheet like it came from the factory?

Wanted to get some input from other people and don't have anyone to ask in person.

Thank in advance.

[inse]

Is the orange sheet polyimide film, probably?
Not the best thermal conductor but a good electrical insulator.
How much power do you have to dissipate with which packages?
What is your insulation requirements?
Do you have a datasheet of your Sil pad?
In combination with a estimated rth of your heatsink, you try to guesstimate the allowed rth of the thermal bond.

[jonpaul]

What lamp, wattage?

Clean, plane flat HS, isopropyl alcohol

Kapton sheet, very thin coat HS compound, over both sides, apply

attach semiconductors

Use torque wrench as reccomended


Jon

[jacamo]

This is an 800W HMI.

Not sure exactly what the orange sheet is made from.

Here is a link to the Sil-pad material I have

https://www.mouser.com/datasheet/2/1/57f1350510b9a3775205d4884e036f6f6f5eab97-2951684.pdf

[jacamo]

What lamp, wattage?

Clean, plane flat HS, isopropyl alcohol

Kapton sheet, very thin coat HS compound, over both sides, apply

attach semiconductors

Use torque wrench as reccomended


Jon

Thank you for the response Jon.

[jonpaul]

Rebonjour Cher monsieur


in 1970s..1990s, we designed and built electronics HID ballasts 70 W..12 kw. So lots of experience. Troubleshooting an 800 W HID ballasts is not trivial. DANGER HIGH VOLTAGE!
At least use an isolation transformer and Variac to feed the mains input.



1. the yellow film is Dupont Kapton (Tm) https://www.dupont.com/products/kapton-mt.html
very high temp rating, fine voltage rating.
Specifically designed for high thermal conductivity.  Bergquist Sil pads, mica, etc may not be as thermally conductive or resistant to puncture.

2. you would have to research dissipation, ambient and devices rating to determine any other insulators conductivity.  Safety compliance is also affected with mains and 360V bus.

3. Beware the device and heatsink must be smooth flat and clean before application of thermal compound or insulator.

4. Hardware must be insulating eg nylon shoulder washers.

I hope this helps you out.

Bon courage

Jon

[jacamo]

Rebonjour Cher monsieur


in 1970s..1990s, we designed and built electronics HID ballasts 70 W..12 kw. So lots of experience. Troubleshooting an 800 W HID ballasts is not trivial. DANGER HIGH VOLTAGE!
At least use an isolation transformer and Variac to feed the mains input.



1. the yellow film is Dupont Kapton (Tm) https://www.dupont.com/products/kapton-mt.html
very high temp rating, fine voltage rating.
Specifically designed for high thermal conductivity.  Bergquist Sil pads, mica, etc may not be as thermally conductive or resistant to puncture.

2. you would have to research dissipation, ambient and devices rating to determine any other insulators conductivity.  Safety compliance is also affected with mains and 360V bus.

3. Beware the device and heatsink must be smooth flat and clean before application of thermal compound or insulator.

4. Hardware must be insulating eg nylon shoulder washers.

I hope this helps you out.

Bon courage

Jon

Again thank you for the feedback Jon. Sounds like with your experience you would be the a very good source of information on this subject and would love to have that type of experience.

I am a certified electronics tech so I am well aware of the dangers that these ballast pose and do absolutely use an isolation transformer and variac whenever I am testing these. 
Looks like I am just going to use the Kapton sheet and thermal compound instead of trying to find something else that will work. I was just curious if the tech for silpads had become much more advance that replacing the kapton would be a no brainer but seems like that is not the case.

Thanks for the feedback.

[jonpaul]

Rebonjour Cher Jacamo,


comparison of heatsink insulation....check thermal conductivity, rated voltage, capacités, etc.

SilPad is from Bergquist, silicone impregnated fabric
Kapton is Dupont
Otherwise alumina or berilum oxide ceramic

We generally avoid, with isolated separate HS, and semiconductor directly mounted, no insulation.

Jon

[T3sl4co1l]

Yeah, put back as original.  Maybe cut a new sheet if you can identify pinholes in it, that's about it.

1. the yellow film is Dupont Kapton (Tm) https://www.dupont.com/products/kapton-mt.html
very high temp rating, fine voltage rating.
Specifically designed for high thermal conductivity.  Bergquist Sil pads, mica, etc may not be as thermally conductive or resistant to puncture.

Hmm, I'm not aware that there's anything they can (or do) do, to affect the thermal conductivity.  It's actually a rather pitiful material in that respect, that is, it's a rather good insulator both electrically and thermally.  It's just such a good dielectric that a very thin layer will do -- which makes it usable, if still unimpressive.  To be clear, not that it's terrible, there are certainly plastics with lower conductivity; but it's also a long ways away from, say, alumina or graphite.

Mica is the same way, terrible conductor thru-plane but can be cleaved very thin so it does alright.

By "unimpressive", I mean on the order of 0.5-2 K/W added thermal resistance (RthC-HS), depending on film thickness and package size.  Typically the heatsink is the limiting factor anyway, so it's not a big deal.  (The heatsink here looks like it's going to be more like 2 K/W overall, and say we divide that by however many devices are on it, that's still not much above the 0.5-2 range.  So you are right to guess, thermal performance could be improved perceptibly here, with improved materials.)

There are better materials these days, with conductivity 6 W/(m*K) and higher -- compare with classic (firm rubber, usually gray) Sil-Pads(TM?) in the 0.5 range (I think?) -- but beware they are also much, much softer.  And usually thicker.  A different clamping strategy is needed -- and especially for components that were positioned for a ~0.5 mil film, they're going to be in all the wrong place once you get 20 mils or more in there.

There are also self-adhesive and phase-change kinds; these stick in place better, and are stiffer (or, are once melted in place), so might be better replacements for Sil-Pads as such.  But again, I wouldn't recommend replacing film with either kind, for the above reasons.

Edit: Ahh, fascinating.  Kapton(R) HN I guess is the "default" kind, and says 0.2 W/(m*K).  Or "FPC" at 0.12.  The linked MT type is a whopping 0.46 -- they do indeed do something with it!  And in a thin layer, indeed that isn't bad at all.  Take a typical TO-220 at 10x15mm, a 0.025mm layer is 0.36 K/W.  Add a little (~0.1 K/W?) for grease and you aren't much higher than the package itself (TO-220s are often in the RthJC ~ 0.3 to 2 K/W ballpark).

Tim

[jacamo]

Edit: Ahh, fascinating.  Kapton(R) HN I guess is the "default" kind, and says 0.2 W/(m*K).  Or "FPC" at 0.12.  The linked MT type is a whopping 0.46 -- they do indeed do something with it!  And in a thin layer, indeed that isn't bad at all.  Take a typical TO-220 at 10x15mm, a 0.025mm layer is 0.36 K/W.  Add a little (~0.1 K/W?) for grease and you aren't much higher than the package itself (TO-220s are often in the RthJC ~ 0.3 to 2 K/W ballpark).

Tim

Very interesting!  Thanks for that info