Thermal INSULATOR Pad to Reduce Housing Temperature? (Desperate!)

[TimNJ]

Hi all,

I've hit a point of desperation at work. I inherited a project from my boss (who left) and I'm running out of steam and tensions are getting high. We are designing a power supply for one of our medical customers. There is a housing "touch temperature" requirement in IEC60601 that we have to meet. Basically, cannot have hot-spots on user accessible points that might cause burns/injury.

We have been maxed out on efficiency (for this topology) for months now. We have a few hot-spots on the enclosure that are just a degree or two too hot.

The drawing below shows the basic mechanical construction of the power supply. Very ordinary. PCB assembly with 90degree heatsink inside of a plastic housing.



Our issue is on the top of the housing. There is a small gap, maybe 1 or 2mm, between the top heatsink and the plastic housing. We've tried filling that with thermal RTV among other things. But now, at my wits end, I'm wondering if perhaps a thermal insulator on the hotspot could get us the reduction in temperature we need.

Does anyone have any experience with these kind of materials?

https://www.srpco.com/insulators/thermal-insulators-isolators/

Does anyone have  manufacturer, series, or part number of a product they like?

Any other suggestions on how to make this work?

Thanks a lot.
Tim

[edpalmer42]

Random thoughts:

- Go to Home Depot or Lowes and see what ideas you can come up with.  I'm not suggesting you buy something there, although you could, just look for ideas.
- Does the case have to be sealed?  Maybe ventilation would help.
- If the heatsink in that spot is too hot, what would happen if you drilled out that spot?  Would the hot spot spread out to other parts of the heatsink and so, reduce the peak temperature?
- Similar to above, could you make the heatsink thicker so it would spread the heat better.

Ed

[Gyro]

Likewise, if you have that much of a hotspot on an Aluminium heatsink inside a closed space, then it's probably not thick enough.

As for thermally insulating foam, you need something that trap sufficient air in cells without being conductive enough to offset the benefit. Not easy in a thin section. You need something as close to aerogel (obviously not practical) as possible. It probably needs to be flame retardant too.

On balance, I think to will be better using some of the space to thicken up the Aluminium and spread the heat more evenly.


P.S. Any chance of changing the heatsink from Aluminium to more thermally conductive copper instead? (I've seen it done in premium quality laptop PSUs).

[SilverSolder]

My "gut feeling" is that adding insulation will only raise the internal temperature and make matters worse...

A band of copper foil on the inside of the plastic box to help spread the heat?  Or an aluminum plate glued to the inside of the box, same idea?

[Mr. Scram]

My "gut feeling" is that adding insulation will only raise the internal temperature and make matters worse...

A band of copper foil on the inside of the plastic box to help spread the heat?  Or an aluminum plate glued to the inside of the box, same idea?

It's all about evening out what's there. A very thick aluminium casing will be very uniformly warm. Isolating parts may reduce hotspots even if average temperature rises.

[TimNJ]

Random thoughts:

- Go to Home Depot or Lowes and see what ideas you can come up with.  I'm not suggesting you buy something there, although you could, just look for ideas.
- Does the case have to be sealed?  Maybe ventilation would help.
- If the heatsink in that spot is too hot, what would happen if you drilled out that spot?  Would the hot spot spread out to other parts of the heatsink and so, reduce the peak temperature?
- Similar to above, could you make the heatsink thicker so it would spread the heat better.

Ed

Thanks Ed. I like your Hope Depot idea. I bought some fireproof fiber-paper a while back to try to block radiant heat from one component to another...I think benefit was small.

Yes the case has to be sealed, mainly from a safety perspective, but also to meet IP rating. Interesting idea about the drilling. The hot spot is caused by the PFC inductor directly underneath. My intuition says "less heatsink material means the heatsink gets hotter somewhere" but, your guess is as good as mine.

We designed this new product around an older platform with all heatsink tooling complete. So, unfortunately that's basically set in stone.

Thanks for your suggestions.

Likewise, if you have that much of a hotspot on an Aluminium heatsink inside a closed space, then it's probably not thick enough.

As for thermally insulating foam, you need something that trap sufficient air in cells without being conductive enough to offset the benefit. Not easy in a thin section. You need something as close to aerogel (obviously not practical) as possible. It probably needs to be flame retardant too.

On balance, I think to will be better using some of the space to thicken up the Aluminium and spread the heat more evenly.


P.S. Any chance of changing the heatsink from Aluminium to more thermally conductive copper instead? (I've seen it done in premium quality laptop PSUs).

Thanks for these ideas. I just dremeled out part of the plastic housing to check the actual clearance between housing and the heatsink. Less than I thought. Somewhere between a 0.5-1.0mm air gap. Therefore, as you say, any thermal "insulator" might tend to act more as a conductor in that tiny space.

Thanks for the idea about the copper...This right angle heatsink is actually a "two part" heatsink. That is, two layers of 2.0mm aluminum (but only on the top). Maybe we can change the second layer to copper instead of aluminum. I'd bet that would work.

My "gut feeling" is that adding insulation will only raise the internal temperature and make matters worse...

A band of copper foil on the inside of the plastic box to help spread the heat?  Or an aluminum plate glued to the inside of the box, same idea?

Also my gut feeling...hence, why this is my last resort!

We wrapped the whole thing in copper foil and it worked great, from an assembly standpoint, it doesn't fit well. Not enough clearance on the sides in particular. Maybe we can try some 3M 9876 foil, or similar, stuck to the top heatsink.

Thanks all.

[grifftech]

active cooling?

[KaneTW]

Thermal insulation on the hot spots is actually exactly what other people are doing. Look at the disassemblies of the recent Dell XPS 13. They added an insulator pad (from aerogel or something? don't remember) to the parts that are user-facing so they don't get hot.

[jbb]

Ah, I feel your pain.  I recently had trouble with heat spreading in a product and it took me weeks to work out what was going wrong.

Here are some thoughts:

• Can you bond a flat heat pipe to the heatsink near the trouble area? Preferably between the inductor and the heatsink if space allows?  This would spread the inductor heat out.
• I would be a bit nervous about bonding aluminium to copper vis a vis differential metal corrosion - but this assumes some moisture gets in.
• Assuming max heating corresponds with max load, can you get the software team to help you out?  If you can reduce the standing load of the appliance, and only come up to 100% power for brief (and controlled) periods, you might be able to squeak through. (Only possible if your supply is for your equipment, though.)
• Can you apply some kind of surface coating (e.g. a fancy label) to the affected area such that a higher spot temperature is allowed by the standard?

Now, at risk of being an a**hole, I'd like to ask some questions about losses. I apologise in advance if you've already tried all these.

• Can you split the inductor into 2 smaller inductors? This will cost more but could spread the heat out.
• Have you looked at a variety of core materials for the inductor, not just 'whatever we used last time'? You may be able to improve the situation with a better ferrite or some kind of specialty powdered material (powder core materials don't need a discrete gap, which can reduce fringing losses. But ferrite is generally lower loss).
• Depending on the switching frequency and operating current, how are your winding losses?  Moving to Litz wire might cost a little more but could help with skin and proximity losses.
• Have you looked at Silicon Carbide (SiC) and Gallium Nitride (GaN) parts yet?  These have (all things being equal) way lower switching losses than Silicon (Si) parts and allow operation at higher frequencies.  This reduces inductor losses.

[TimNJ]

Ah, I feel your pain.  I recently had trouble with heat spreading in a product and it took me weeks to work out what was going wrong.

Here are some thoughts:

• Can you bond a flat heat pipe to the heatsink near the trouble area? Preferably between the inductor and the heatsink if space allows?  This would spread the inductor heat out.
• I would be a bit nervous about bonding aluminium to copper vis a vis differential metal corrosion - but this assumes some moisture gets in.
• Assuming max heating corresponds with max load, can you get the software team to help you out?  If you can reduce the standing load of the appliance, and only come up to 100% power for brief (and controlled) periods, you might be able to squeak through. (Only possible if your supply is for your equipment, though.)
• Can you apply some kind of surface coating (e.g. a fancy label) to the affected area such that a higher spot temperature is allowed by the standard?

Now, at risk of being an a**hole, I'd like to ask some questions about losses. I apologise in advance if you've already tried all these.

• Can you split the inductor into 2 smaller inductors? This will cost more but could spread the heat out.
• Have you looked at a variety of core materials for the inductor, not just 'whatever we used last time'? You may be able to improve the situation with a better ferrite or some kind of specialty powdered material (powder core materials don't need a discrete gap, which can reduce fringing losses. But ferrite is generally lower loss).
• Depending on the switching frequency and operating current, how are your winding losses?  Moving to Litz wire might cost a little more but could help with skin and proximity losses.
• Have you looked at Silicon Carbide (SiC) and Gallium Nitride (GaN) parts yet?  These have (all things being equal) way lower switching losses than Silicon (Si) parts and allow operation at higher frequencies.  This reduces inductor losses.

Thanks for all of these well-formulated thoughts. I am deeply appreciative.

• With respect to the heatpipe, is your idea to improve  thermal conductivity between the inductor and the aluminum heatsink? It's an RM10 size inductor; I suppose we could use a wide, flat tube? I honestly don't know how those are even bonded...some sort of epoxy?
• Good point about bonding two different metals to each other...I'm not sure how that would pan out in reality. Though, it seems like one of my only "reasonable" options at the moment
• Actually yes, they've already done some significant "load shedding" to get the power down. They don't think they can take off much else.
• Interesting idea. Per the standard, I think anything surface which is "user-accessible" is fair game. But, if you have any examples of this, I'd love to hear/see them

• Space wise, pretty jam packed, also have internal (already tooled) support bracket that really gets in the way
• Yeah this is TDK PC95. We've also experimented with ACME P45, which is supposedly even lower loss. They are the lowest loss ferrites on the market. We've tried a powder core inductor, a few designs, (using molypermalloy), but ferrite still prevails
• Using 0.1mm strand litz. The bobbin window utilization is almost 100%. I don't think we can fit more copper
• The diode is SiC, which is hugely beneficial, but the MOSFET it Si. I remember looking at SiC MOSFET and GaN MOSFET prices a few months ago and they were pretty staggering. $5, $10+ Maybe I ought to re-visit that area. If anyone knows of any reasonably priced SiC/GaN MOSFETs, I'd love to check them out. Also, our MOSFETs are TO-220, so the whole lead-inductance situation isn't that great, for GaN in particular

Thanks for all of these ideas. It means a lot!

[KaneTW]

GaN FETs get pretty cheap, depending on what you need. https://www.digikey.com/products/en/discrete-semiconductor-products/transistors-fets-mosfets-arrays/289?FV=ffe00121%2Cfffc0395&quantity=0&ColumnSort=1000011&page=1&pageSize=25

2 N-channel GaN in one package, 100V, 1.7A continuous for $0.75/pc @ 2500pc. Lots of other offerings too.

[jbb]

Hi Tim

You’re welcome. Some quick replies:
- the idea on the heat pipe was to distribute the heat load from the inductor all the way along the aluminium part. This would mean a warm stripe rather than a hot spot. No need to try to bond the inductor to the heat pipe and heat sink too,  this would make assembly annoying. Simply glue the heat pipe to the hestsink next to the inductor and see what happens. Beware that thermal fillers tend to degrade mechanical strength, so don’t just get the thermal epoxy with lowest thermal resistance; you need to survive vibration and drops.
- what I meant with surface coating is that the standard might say “maximum X degrees if metal and Y degrees if plastic,” which might let you get away with a higher spot temperature
- looks like you’ve done he hard work on the inductor
- SiC Schottky diodes are great :-)
- SiC and GaN devices do carry a cost premium, which you have to recover by total system design. Given you’re using TO-220 devices, the package inductance is such that a TO-220 SiC device might not do much better than a Si one.

I had a further thought. It’s really hackey. Can you make a 0.5 - 1.0mm deep indent in the OUTSIDE of the enclosure (keep electrical safety and drop resistance in mind) and use a neoprene (or other foam) label to cover the hot spot? You might be able to argue that even if the spot is touched, the heat flux through neoprene is limited. Could have trouble with porosity and sanitary control, though.

[TimNJ]

Hi Tim

You’re welcome. Some quick replies:
- the idea on the heat pipe was to distribute the heat load from the inductor all the way along the aluminium part. This would mean a warm stripe rather than a hot spot. No need to try to bond the inductor to the heat pipe and heat sink too,  this would make assembly annoying. Simply glue the heat pipe to the hestsink next to the inductor and see what happens. Beware that thermal fillers tend to degrade mechanical strength, so don’t just get the thermal epoxy with lowest thermal resistance; you need to survive vibration and drops.
- what I meant with surface coating is that the standard might say “maximum X degrees if metal and Y degrees if plastic,” which might let you get away with a higher spot temperature
- looks like you’ve done he hard work on the inductor
- SiC Schottky diodes are great :-)
- SiC and GaN devices do carry a cost premium, which you have to recover by total system design. Given you’re using TO-220 devices, the package inductance is such that a TO-220 SiC device might not do much better than a Si one.

I had a further thought. It’s really hackey. Can you make a 0.5 - 1.0mm deep indent in the OUTSIDE of the enclosure (keep electrical safety and drop resistance in mind) and use a neoprene (or other foam) label to cover the hot spot? You might be able to argue that even if the spot is touched, the heat flux through neoprene is limited. Could have trouble with porosity and sanitary control, though.

Thanks for the ideas/clarification about the heatpipe. I think I'll ask our mechanical guys what they think about it.

Do you think there would be any benefit of dropping in an SiC MOSFET without changing switching frequency, inductor redesign, etc? Unfortunately, trying to think of a quick way out here.

To my knowledge, plastic materials are allowed to be the hottest per medical safety standards, and I think a foam might fall under that category, though I can't be sure. If the enclosure was metal, then yes, for sure, we could put something plastic on those areas and get an additional 10-15C. The enclosure does need to be wiped down/sanitized, so I'm not sure how the porous foam would go over. 

Thanks.

[TimNJ]

GaN FETs get pretty cheap, depending on what you need. https://www.digikey.com/products/en/discrete-semiconductor-products/transistors-fets-mosfets-arrays/289?FV=ffe00121%2Cfffc0395&quantity=0&ColumnSort=1000011&page=1&pageSize=25

2 N-channel GaN in one package, 100V, 1.7A continuous for $0.75/pc @ 2500pc. Lots of other offerings too.

Thanks. I am looking for 600-800V MOSFETs unfortunately!

[thermistor-guy]

...
Our issue is on the top of the housing. There is a small gap, maybe 1 or 2mm, between the top heatsink and the plastic housing. We've tried filling that with thermal RTV among other things. But now, at my wits end, I'm wondering if perhaps a thermal insulator on the hotspot could get us the reduction in temperature we need.
...
Any other suggestions on how to make this work?

You could try adding adhesive-backed metallic sheet (copper or aluminium foil) to the top of the housing (inside surface). The sheet reflects infra-red radiation away from the top of the housing, spreads the heat a little via conduction, plus the adhesive provides a small amount of thermal insulation. Together, these small improvements may provide just enough hotspot reduction.

Since you have 1mm to work with, you could also add a thin insulating layer between the foil and housing. This would be a cheap solution if it works. Once you get a working prototype, Parker Chomerics (http://www.chomerics.com/) may be able to help you with a professional production-ready solution.

[ChristopherN]

+1 for the metal foil on the outside.

We did that in one test for a product, using a metal label like this: https://www.schilder-buhl.de/produkte/metallschilder/untereloxaldruck# (random google link). The labels are available in different versions with different thicknesses, materials and so on.

Other ideas:
Use a label thats on a foam backing, there are ones that are a bit like thick double sided tape.
A third option could be a domed label, those are available in different versions as well and can add a few mm to the outside of the housing.
You could also get a plastic label that comes in form of a wave, something like this: ^^^^^. The raised spots would likely stay cooler that the rest, but I don't now if thats allowed since you can still reach the hot part if you press hard.

Can you specify the orientation the power supply must be used in? If thats the case you could specify that its only to be used with the hot spot on the bottom, that could change the heat distribution in the supply.

[OM222O]

you mentioned needing to seal the box for IP rating, and it seems like active cooling is the best and cheapest bet here, so here are a few suggestions that would probably help:

is there a metal plate / bar that the PSU gets mounted to? if so, can you find empty places on the aluminum heat sink that allow it to be screwed onto the plate? this doesn't give electrical isolation, but it certainly allows for IP rating.

can you leave one of the sides of the plastic box open (only pot to the top of the aluminum plate, don't go over it and seal the entire thing) and leave the aluminum plate with the hot spot exposed to some internal part of the device, then have a fan blowing directly on the hot spot?

make the copper area around the inductor larger and use thermal vias if possible, this will increase the surface area of the "hot spot" which reduces the maximum temperature.

as a basic test, buy some non conductive thermal paste and apply it around the inductor over a "large area" (about 1 inch by 1 inch should be more than good enough) to see of that reduces the temperature at all.

I don't think insulating the hotspot is a great idea, it will only create more in different parts. just try different mounting orientations so it's not where it could be touched during normal operation. if nothing else works, you should reserve to watercooling: have a pipe in the middle, or a water block that directly contacts one of the aluminum plates and soaks the heat. another alternative would be sticking a vapor chamber to the back of the aluminum plate. they are thin and very good at conducting the heat away. that way your entire aluminum plate heats up evenly, avoiding hot spots. I'm not sure if 1mm is a large enough gap, but 2 or 3mm should be more than good enough if you can fit it between the plastic and the aluminum.

[TimNJ]

...
Our issue is on the top of the housing. There is a small gap, maybe 1 or 2mm, between the top heatsink and the plastic housing. We've tried filling that with thermal RTV among other things. But now, at my wits end, I'm wondering if perhaps a thermal insulator on the hotspot could get us the reduction in temperature we need.
...
Any other suggestions on how to make this work?

You could try adding adhesive-backed metallic sheet (copper or aluminium foil) to the top of the housing (inside surface). The sheet reflects infra-red radiation away from the top of the housing, spreads the heat a little via conduction, plus the adhesive provides a small amount of thermal insulation. Together, these small improvements may provide just enough hotspot reduction.

Since you have 1mm to work with, you could also add a thin insulating layer between the foil and housing. This would be a cheap solution if it works. Once you get a working prototype, Parker Chomerics (http://www.chomerics.com/) may be able to help you with a professional production-ready solution.

Thanks. I really like this idea. I'm setting up the thermal chamber to give it a try now. I tried a few layers of copper foil. Will compare the housing temperature with and without.

[TimNJ]

+1 for the metal foil on the outside.

We did that in one test for a product, using a metal label like this: https://www.schilder-buhl.de/produkte/metallschilder/untereloxaldruck# (random google link). The labels are available in different versions with different thicknesses, materials and so on.

Other ideas:
Use a label thats on a foam backing, there are ones that are a bit like thick double sided tape.
A third option could be a domed label, those are available in different versions as well and can add a few mm to the outside of the housing.
You could also get a plastic label that comes in form of a wave, something like this: ^^^^^. The raised spots would likely stay cooler that the rest, but I don't now if thats allowed since you can still reach the hot part if you press hard.

Can you specify the orientation the power supply must be used in? If thats the case you could specify that its only to be used with the hot spot on the bottom, that could change the heat distribution in the supply.

No, unfortuantely, cannot specify orientation, so we have to pass at the worst case. Thanks.

[TimNJ]

you mentioned needing to seal the box for IP rating, and it seems like active cooling is the best and cheapest bet here, so here are a few suggestions that would probably help:

is there a metal plate / bar that the PSU gets mounted to? if so, can you find empty places on the aluminum heat sink that allow it to be screwed onto the plate? this doesn't give electrical isolation, but it certainly allows for IP rating.

can you leave one of the sides of the plastic box open (only pot to the top of the aluminum plate, don't go over it and seal the entire thing) and leave the aluminum plate with the hot spot exposed to some internal part of the device, then have a fan blowing directly on the hot spot?

make the copper area around the inductor larger and use thermal vias if possible, this will increase the surface area of the "hot spot" which reduces the maximum temperature.

as a basic test, buy some non conductive thermal paste and apply it around the inductor over a "large area" (about 1 inch by 1 inch should be more than good enough) to see of that reduces the temperature at all.

I don't think insulating the hotspot is a great idea, it will only create more in different parts. just try different mounting orientations so it's not where it could be touched during normal operation. if nothing else works, you should reserve to watercooling: have a pipe in the middle, or a water block that directly contacts one of the aluminum plates and soaks the heat. another alternative would be sticking a vapor chamber to the back of the aluminum plate. they are thin and very good at conducting the heat away. that way your entire aluminum plate heats up evenly, avoiding hot spots. I'm not sure if 1mm is a large enough gap, but 2 or 3mm should be more than good enough if you can fit it between the plastic and the aluminum.

Active cooling as in a fan? This is an external power supply. Looks like a laptop power brick essentially. It does not have a permanent mounting location. It can be placed wherever the user wants...on the ground, in a corner, on a table, etc. I like your idea about using the PCB area as additional heatsinking. Maybe some fine tuning can be done there, though I think the thermal resistance from the inductor to the board is pretty high.

[larrybl]

How about this? https://www.thermalcontrolproducts.com/our-products/thermal-barriers/
I have used this stuff to prevent engine muffler from heating the hood.

[Nauris]

If you are desperate, you can put a fan inside just blowing air around the enclosure without any holes to outside to keep it water-tight. That should equalize temps nicely. Smallest fans on digikey catalog are 10x10x3mm that is TINY. Put one blowing on that inductor...

[Mr. Scram]

If you are desperate, you can put a fan inside just blowing air around the enclosure without any holes to outside to keep it water-tight. That should equalize temps nicely. Smallest fans on digikey catalog are 10x10x3mm that is TINY. Put one blowing on that inductor...

That's a proper bodge. Having a nice solid state solution and then introducing moving parts that will inevitably fail. What happens when they do?

[NiHaoMike]

Could it be possible to glue a heatsink, maybe as simple as an aluminum bar, on the outside of the case?

[OM222O]

Active cooling as in a fan? This is an external power supply. Looks like a laptop power brick essentially. It does not have a permanent mounting location. It can be placed wherever the user wants...on the ground, in a corner, on a table, etc. I like your idea about using the PCB area as additional heatsinking. Maybe some fine tuning can be done there, though I think the thermal resistance from the inductor to the board is pretty high.

If it is an external power brick, then most of what I said will be useless 
regarding the thermal resistance, it's actually pretty low since the inductor is just a piece of copper which has really good thermal conductivity to the copper layers on the PCB. if you look at GPU designs, most of them use a thermal pad on the other side of the PCB (under the inductor) which transfers heat to the back plate of the GPU. the backplate gets quite warm and acts like a heat sink (i.e: helps cool the inductors if a fan blows on it). that is an insane amount of thermal resistance (1.6mm of FR4 material, 1 or 2mm of thermal pad and then the back plate itself not to mention the 3 different interfaces (boundaries between these layers)), but it still helps to a large degree. I think using copper foil tape as others suggested would be your best bet if you're running out of space. just make sure to use plenty of thermal vias (you can use vias in pad method if you're really out of space) and have the piece of copper foil tape directly under those vias.

[Nauris]

If you are desperate, you can put a fan inside just blowing air around the enclosure without any holes to outside to keep it water-tight. That should equalize temps nicely. Smallest fans on digikey catalog are 10x10x3mm that is TINY. Put one blowing on that inductor...

That's a proper bodge. Having a nice solid state solution and then introducing moving parts that will inevitably fail. What happens when they do?

Well then there is also the possibility to fill the whole device with fine sand. That solution avoids any moving part.

[German_EE]

Is there any possibility of using a copper heat spreader between the active device(s) and the heatsink? Radio hams who build their own power amplifiers use this technique with aluminum heatsinks on a regular basis.

[TimNJ]

Active cooling as in a fan? This is an external power supply. Looks like a laptop power brick essentially. It does not have a permanent mounting location. It can be placed wherever the user wants...on the ground, in a corner, on a table, etc. I like your idea about using the PCB area as additional heatsinking. Maybe some fine tuning can be done there, though I think the thermal resistance from the inductor to the board is pretty high.

If it is an external power brick, then most of what I said will be useless 
regarding the thermal resistance, it's actually pretty low since the inductor is just a piece of copper which has really good thermal conductivity to the copper layers on the PCB. if you look at GPU designs, most of them use a thermal pad on the other side of the PCB (under the inductor) which transfers heat to the back plate of the GPU. the backplate gets quite warm and acts like a heat sink (i.e: helps cool the inductors if a fan blows on it). that is an insane amount of thermal resistance (1.6mm of FR4 material, 1 or 2mm of thermal pad and then the back plate itself not to mention the 3 different interfaces (boundaries between these layers)), but it still helps to a large degree. I think using copper foil tape as others suggested would be your best bet if you're running out of space. just make sure to use plenty of thermal vias (you can use vias in pad method if you're really out of space) and have the piece of copper foil tape directly under those vias.

The inductor is a big through-hole type. Looks like the below. Direct conduction from the copper wire is probably okay but it all has to go through essentially two 0.8mm pins. I know our factory does not like to put adhesive underneath the magnetic parts because of the heat during wave soldering. But I could see how getting heat from the bottom of the inductor to the top-sound ground plane could be helpful.

Thanks.

 

[TimNJ]

Is there any possibility of using a copper heat spreader between the active device(s) and the heatsink? Radio hams who build their own power amplifiers use this technique with aluminum heatsinks on a regular basis.

Can you show an example of this?

Thanks.

[TimNJ]

If you are desperate, you can put a fan inside just blowing air around the enclosure without any holes to outside to keep it water-tight. That should equalize temps nicely. Smallest fans on digikey catalog are 10x10x3mm that is TINY. Put one blowing on that inductor...

That's a proper bodge. Having a nice solid state solution and then introducing moving parts that will inevitably fail. What happens when they do?

Well then there is also the possibility to fill the whole device with fine sand. That solution avoids any moving part.

Hmm, interesting. This also raises the possibility of potting the unit, which could act as a giant heat-spreader. I think potting is not really preferred from a production standpoint since it's time and labor intensive. I know (good) potting compound doesn't cause an issue but I wonder if sand has any safety issues related to it, specifically for creepage paths.

Thanks.

[Mr. Scram]

Can you perhaps elaborate how you ended up in this situation with all kinds of parts tooled but before finalising the design? Perhaps we can learn from the situation.

[German_EE]

Please see http://www.w6pql.com/images/2m/inside.jpg and http://www.w6pql.com/images/2m/leftside.jpg for examples. The active device is soldered to the copper heat spreader and then the whole assembly is bolted to a heatsink. Sure, it's a bit bigger than your power supply, but the principle of the heat spreader is the same.

[TimNJ]

Can you perhaps elaborate how you ended up in this situation with all kinds of parts tooled but before finalising the design? Perhaps we can learn from the situation.

Ah, I thought I explained this earlier, but looks like not.

We have an existing product line (in production for about 2 years) which uses this housing and metal hardware. For this customer, we estimated that re-using the existing tooled hardware would be sufficient, since the power level is similar. However, it turns out that this power supply runs a little hotter on top of our customer being much more stringent (than most) on this specification

It was a gamble. Half the reason the customer chose us is because we hardly had any tooling costs in our quote. Unfortunately, did not have a say in that process. And now we are struggling to make this work.

[SilverSolder]

It sounds like you are frustratingly close to making it work... 

Can power consumption be lowered somewhere else in the system to bring you over the goal line?  Usually you can always find a watt or two...

[Mr. Scram]

It sounds like you are frustratingly close to making it work... 

Can power consumption be lowered somewhere else in the system to bring you over the goal line?  Usually you can always find a watt or two...

The opening post states efficiency has been maxed out for a while which suggests they've been looking.