Another one has let the smoke out

sorry to here the bad news leroy

[BrianG]

Originally Posted by coolhandcountry

The one in the pics is a 9920. I have a 36120 as well. I think this is #3 I have let the smoke out of. One is in germany now getting fixed.

Sounds like a pattern there. Are they getting hot when in use or something? And I assume there isn't something wrong with the motor (like partially melted winding insulation), or mounting screws too far in? The day mine burns up, I'll cry. Those things are too expensive to blow left and right.

[squeeforever]

Originally Posted by BrianG

The day mine burns up, I'll cry.

Been there, done that...

[MetalMan]

Originally Posted by squeeforever

Been there, done that...

4 1/2 times for me... :026: It's a tough hobby sometimes, there just isn't a perfect solution (yet).

[Nick]

They said to me it could take 6 weeks to fix. So I just bought a new one.

[Serum]

Brian, i noticed the exact same thing on the BK's.

mine is even worse, the 2nd 'powerboard' is closer on one side, than on the other. So in fact, it is loose, the thick layer of silicone paste doesn't touch the underside of the heatsink equiped upper powerboard. I am not quite happy with this. I thought about taking it apart, and mounting it right..but i don't know how BK feels for it..

[BrianG]

Yeah, mine had that gap as well filled with thermal grease. That stuff is only supposed to fill microscopic imperfections between mating surfaces, not fill gaps. It's thermal conductivity goes WAY down when it is too thick (by "too thick", I mean like 0.1mm). In my case, it looks like the thickness of the motor wire solder joints were keeping the PCB from sitting where it should. I'd try to resolder it as well, but have you ever soldered/desoldered a multilayered board? It sucks. You need quite a bit of heat to keep the traces from sinking too much, but then nearby sensitive components can get too hot.

IMO, I think BK should offer a discount on an exchange program to replace ones like ours with a properly assembled one. It would benefit them in the long run for better customer service and a more reliable product. Already, I've heard a number of people with blown ESCs and for some, it's not the first they've blown...

[Serum]

Yeah, I've got experience with multilayer prints. They are a pita .1 mm of paste already is too much IMO. The less the better..

i don't like that way of constructing either. if they used an aluminum plate to transfer the heat. In my case there is this fat 1-1.8 mm gap, with some silicone stuff just floating, only touching the fetts....

If i see something like that, with that amount of paste, i highly doubt if the person who assembled or did the quality control knew what he/she was doing. It's plain ridiculous.

[Serum]

here, this is what i am talking about..

[BrianG]

If they used an aluminum plate between layers, they might as well extend the plates fully out to the sides and attach a heatsink to them.

I still think those surface mount FETs are not the way to go; the plastic casing does not transfer heat that well compared to a metal tab type of device, like a TO-220 case. They seem to try to get around that by paralleling a butt load of FETs together. Then, servicing is a PITA; not too many people have soldering irons that can work with surface mount. I like small, easy to mount ESCs as much as the next guy, but I'd be willing to trade a larger ESC to get something more reliable and solidly built. Leave the light and tiny ESCs for the buggies, planes, helis, and any other small scale vehicle.

[BrianG]

Originally Posted by Serum

here, this is what i am talking about..

Yup, that's about what mine looks like, except they dumped more grease in there...

[GriffinRU]

It is bad, but not as bad as you all think...
Yes, performance goes down, but the higher the temperature of the FET - less current goes through this FET. So, FET's with better heatsinking will do the most job, but overall number of FET's will take care of high current spikes, when they cool.
TO-220 and simular are old design, there are better new gen FET's which offer 10+ times better internal resistance. Sooner or later they will be here as well.

If you have 2 layer PCB with power FET's then assume half of the cold current ratings when temperature of ESC goes up. If you have 3 then 1/3...
Otherwise look for smoke.

Artur

P.S. It is better then no grease at all. And if you compare 0.1 mm of termogrease with plastic case of SOIC package, it won't look so bad.

[Serum]

there are to220's available with low internal resistance as well.

Brian was mentioning the TO220's because of the for him more service friendly point of view. He says it clearly.

It's not a fact that TO220 is an old design, there are still new powerfetts coming out, that i would not use in a smd package.

the better the quality of the blockwave, the less hot the controller will be. the faster the fetts, the better, and the lower the internal resistance the better too.

[GriffinRU]

Originally Posted by Serum

there are to220's available with low internal resistance as well.

Brian was mentioning the TO220's because of the for him more service friendly point of view. He says it clearly.

It's not a fact that TO220 is an old design, there are still new powerfetts coming out, that i would not use in a smd package.

the better the quality of the blockwave, the less hot the controller will be. the faster the fetts, the better, and the lower the internal resistance the better too.

I was talking about this...

There is always fine line, more amps lower R but higher C

Artur

P.S. It is cheaper to populate board with SOIC then with TO-220-like parts.

[BrianG]

Serum is right: I was looking at it from a service perspective. I also like the reduced thermal resistance to heat by heatsinking the metal tab (which conducts heat far better than plastic I'm sure you'll agree). Heatsinking the plastic case does work, it's just not as efficient.

And you are right about FETs: as they heat up, they conduct less (negative temperature coefficient). However, this is more of a circuit design parameter to allow them to be paralleled without one or more "current hogging" as compared to a traditional bipolar transistor. As far as high currents are concerned, FETs can still burn up. Any package type that does not shed it's heat will eventually melt. Some of the problem here is the use of the PCB to act as a heatsink and to thermally couple the layers of FETs together so they run at the same temperature. It's just not enough apparently.