V10F Mosfet heatsink pads with paste????

[Hal Farrenkopf]

Been away for a while and had to make a new Facebook profile and hence it looks like I am new here. I finally got to open my V10F to do the battery waterproofing (yes I know, I procrastinate too much and had other things to do and only rode it in good weather) and I see that the motor mosfets are using the heat sink silicone pads and there is also paste on both sides of the pad. I am quite sure that using both pad and paste is not proper. Has anyone else asked about this issue on their inmotion wheel and possibly a cause for the overheating of the first batch of wheels which this one is of?

 

[Hal Farrenkopf]

Here is a photo of my V10f mosfets 

 

https://m.facebook.com/photo.php?fbid=851418262038558&id=100015111764656&set=a.809913862855665&source=48

Edited May 1, 2020 by Hal Farrenkopf
Photo added

[Chriull]

13 hours ago, Hal Farrenkopf said:

I am quite sure that using both pad and paste is not proper

Have never heard of the application of both or could even hink.of any reason to do so. But 'it "should not be real fatal".

...If it's heat paste and not some kind of thermally insulating glue for easier assembly...

13 hours ago, Hal Farrenkopf said:

and possibly a cause for the overheating of the first batch of wheels which this one is of?

Wasn't this a temperature threshold issue they set a bit low, because they measure at the heatsink where they get higher and "faster" temperature peaks by this?

[Hal Farrenkopf]

I’m aware of the Nikola initial problem with the glue use in the assembly process that caused those mosfets to fail. This paste is white and is still paste - not rtv or glue. But paste is used between heatsink and the transistor housing and or the mica insulator under pressure which thins the thickness of the paste. There isn’t this pressure with this installation because the pad is soft and compressionable which suggests the paste might be thick. This is speculation right now without actually taking it apart. 
 

I’m really wondering if anyone else had questioned this method used in the V10f in the initial batch and whether this method is still the same in later production runs. 
 

I never experienced a high temp alarm or cutout but I never really stressed it like Marty does. I’m likely going to take it apart and use paste and mica insulators instead of the thermal pad if I can get the mica soon enough.  It will require me drilling and tapping 6 holes in the heatsink. 

[FreeRide]

I seems to remember a thread on this, sounds like you know that paste is thermally better than the pad, but the pad is quicker and if it is also insulating and insulation is needed then I'm not sure if using a thermal pad and paste has any issues.  Maybe the pad is not thermal at all and just insulating, so having the paste is good.

The photo shows that the case is not physically secured, so this seems really bad for thermal transfer, but maybe within spec.  I hope they later improved this, but maybe because isolation is needed and they didn't want to spend the time to secure each one with a good screw with insulating sleeve they took the cheap route.  Still I remember someone really digging into this on here somewhere, but maybe it was another machine or place, the memory is not as good as it once was.

Edited April 27, 2020 by FreeRide

[Seba]

17 hours ago, Hal Farrenkopf said:

I am quite sure that using both pad and paste is not proper.

As long as this paste isn't a electrically conductive paste (and it seems to be a standard, silicone paste that isn't electrically conductive), it doesn't pose any risks. It's just unnecessary, but doesn't harm.

[Hal Farrenkopf]

1 hour ago, Seba said:

As long as this paste isn't a electrically conductive paste (and it seems to be a standard, silicone paste that isn't electrically conductive), it doesn't pose any risks. It's just unnecessary, but doesn't harm.

But I think it adds additional resistance to heat conduction to the heatsink which is bad. 

 

1 hour ago, FreeRide said:

I seems to remember a thread on this, sounds like you know that paste is thermally better than the pad, but the pad is quicker and if it is also insulating and insulation is needed then I'm not sure if using a thermal pad and paste has any issues.  Maybe the pad is not thermal at all and just insulating, so having the paste is good.

The photo shows that the case is not physically secured, so this seems really bad for thermal transfer, but maybe within spec.  I hope they later improved this, but maybe because isolation is needed and they didn't want to spend the time to secure each one with a good screw with insulating sleeve they took the cheap route.  Still I remember someone really digging into this on here somewhere, but maybe it was another machine or place, the memory is not as good as it once was.

There were a bunch of videos on the Nikola wheel and different mounting materials which clearly showed paste with the mica insulator to be the best for heat transfer. The Nikola 84V version had the TO220 mosfets with screw fastenings. 
 

This V10f looks like the mosfets are held down by the PC board and 4 screws beside the TO220 cases. 

[Chriull]

24 minutes ago, Hal Farrenkopf said:

But I think it adds additional resistance to heat conduction to the heatsink which is bad. 

Depends on the choosen pad and paste. There are pastes with better thermal condictivity and there exists pads comparable to micra and great paste...

Without knowing the datasheets of the products nothing can he said...

And the amount of paste seems to not be overdone, from the little bit that can be seen on the photo.

[Hal Farrenkopf]

I was able to find mica and screw spacers In my parts bin to replace the pad so I took the board out and drilled and tapped 6 holes in the heatsink.
The paste was not put on both sides of the pad evenly  It was actually put on carelessly. 

The temperature sensor is close to the mosfets in the heatsink so hopefully improving the heat conduction to the heatsink doesn’t spike up the overtemp readings. 
There is lots of empty space that they could have made a lot bigger heatsink in this V10f wheel. 

[Rabal]

A month ago, I changed my mainboard (under warranty) because of two mosfets broken.

I used 3M 5590H thermal pad. It has two thermally conductive layers. one of them is soft and tearable by finger easily, other is durable thermally conductive layer. 

 

https://multimedia.3m.com/mws/media/920112O/3m-thermally-conductive-acrylic-interface-pad-5590h.pdf

[houseofjob]

Has anyone tried Aluminum Nitride thermal pads instead of mica / other ceramics?

An NYC user on Telegram was saying how Aluminum Nitride is more effective for cooling the MOSFETs, quoting this Russian article.

And this AliExpress listing is saying thermal conductivity for their Aluminum Nitride chips are 160-190 W/m.K, which I believe is much higher than Mica.

[Chriull]

1 hour ago, houseofjob said:

Has anyone tried Aluminum Nitride thermal pads instead of mica / other ceramics?

An NYC user on Telegram was saying how Aluminum Nitride is more effective for cooling the MOSFETs, quoting this Russian article.

And this AliExpress listing is saying thermal conductivity for their Aluminum Nitride chips are 160-190 W/m.K, which I believe is much higher than Mica.

Yes that is - wikipedia says about the same (180-220). For Mica i found 0.5 - 7 W/(mK). So there's a factor between 20-380...

But in your Aliexpress link the state a thickness of 0.35mm - mica plates are available at 0.05 mm. So here wins mica by the factor of 7...

So from the numbers AlN should win - but no idea if any other important properties are missing?

 

1 hour ago, houseofjob said:

 quoting this Russian article.

Just read this article (and some further reading).

The in the article used Al2O3 already destroys mica and thermal pads!

 

AlN should blast away Al2O3.

 

Quote

And this AliExpress listing is saying thermal conductivity for their Aluminum Nitride chips are 160-190 W/m.K, which I believe is much higher than Mica.

The thermal pad caused 70°C temperature difference for a to247 case dissipating 50W.

From.another site i got that mica should be a bit better?

But Al2O3 just dropped ~8°C! AlN should cause even less.

But as you see in the aluexpress link 50 pieces cost 40€!

But for ones beloved EUC this amount should be worth the benefit!

[houseofjob]

29 minutes ago, Chriull said:

But in your Aliexpress link the state a thickness of 0.35mm - mica plates are available at 0.05 mm. So here wins mica by the factor of 7...

Hmmm, shouldn't AlN still win since the thermal conductivity of mica is so comparatively low?

[Hal Farrenkopf]

Mica and paste is superior to that Chinese pad. Aluminum Oxide is better than Mica and Aluminum Nitrate is even better. I had the mica insulators on hand and knew it was significantly better than the pad that was supplied with the V10F.  The ceramics are a better choice if you can get them. Beryllium Oxide should be avoided because it is a health hazard if it gets powdered.