Inside the Quark 125, pics.

Here is what the insides look like.The plastic end of the ESC keeps popping open (battery wire side). So I opened it all the way. I was thermaling today even with heatsinks and fans. I think the plastic end holds it so the fets are pushing on the thermal pad. I'm thinking just bond a heatsink directly to the fets. Then shrink wrap it, but I'm a novice when it comes to electronics. Anyone have any ideas for keeping it even cooler.

[BrianG]

That is interesting. I took mine apart but the FETs were stuck pretty good to the case and I didn't want to break any, so I never got it apart that much. Looks to be well constructed. Is that an Aluminum heat spreader between the PCBs? From the looks of it, there are 12 FET IC's per phase (top and bottom of the PCB), not too bad. Can you read the part number on one of the FETs? I'd like to look up their datasheet.

Of course, anything you do to mount it differently will probably void the warranty, but I'm sure you knew that. You could mount the assembly in a better heatsink like this (from here) inside the channel, and then fabricate some kind of enclosure to cover the other parts of the ESC. NOTE: Do NOT use thermal epoxy like normal Arctic Silver. The paste will squeeze out when you press the FETs on the heatsink and the silver particles may contact the FET leads and/or PCB traces and cause problems. Even if you don't get a direct short circuit, the particles can introduce capacitance. Make sure whatever thermal bonding agent you use is electrically insulative, yet thermally conductive. You can use this instead - same company, different blend.

Also, when I removed the top plate (where the LEDs and button are), I accidentally ripped off a surface mount capacitor. I was able to resolder it, but it was a pain due to the size. You might want to check to make sure you didn't do the same. I can tell if you did or not if you could post a picture of the top board if you want.

Thanks for all the good information. Yes, there is a Aluminum heat spreader between the PCBs.

As far as the FET part numbers go, unfortunately, I just put it back together, but was able to blow up my picture and Here they are: 22N03S GAC529.
For some reason the FETs were no longer stuck to the thermal heat pad or tape or whatever it is. There is some movement to the boards inside the case, maybe due to being mounted on it's side. Can't be very good for heat transfer. I may have to do something about it.

I don't think I ripped the capacitor you were talking about. I wish I would have checked more carefully. I was planning on running it again tomorrow so, I put it back together. I will probably open it up again. When I do I will get it a picture of the other side.

It's a little scary that the Arctic Alumina Adhesive is a permanent bond.

[BrianG]

Thanks for the part number. For those interested, here is the datasheet for them. Some interesting points:

• Rdson is pretty low at ~0.002 ohms. Paralleling 12 of these brings the total value to ~0.00016 ohms.
  
• 50A rating
  
• 30v rating. Maybe this is why some of these don't work well at 6s lipos? 6s under a full charge would be 25.2v, which is quite close to the max rating of 30v. However, a look at the "3 Safe Operation Area" graph, you can see that it is good for 50A up to 20v. Past 20v, the safe current goes down. Hmm, interesting.
  
• Dissipation up to ~105W. Of course this assumes an adequate heatsink. The derating graph ("1 Power Dissipation") shows that it is 105W up to 25C. Above 25C, the power is derated about 0.9W/degreeC.

Was the thermal pad sticky at all? There has to be something to hold the FETs against the heatsink. If you don't want to use a thermal epoxy of some kind, you could get some thermal tape instead. It won't work quite as well, but I understand your point about the epoxy being permanent.

Edited for stupid spelling error

[MetalMan]

To my recollection many other brushless controllers that can handle 18 cells have FETs with a 50v rating. You did some good work, Brian, in figuring out a very possible cause of the Quark's disliking of 6s.

[BrianG]

Thanks. Although without CartWheels taking it apart, we might have never known.

At first glance, 30v looks fine, if a bit close to a 6s charge. We might also have to think about the inductive kickback (back EMF) since the signals the ESC sends to the motor are square wave AC. Back EMF under a collasping magnetic field can be extremely high depending on the inductance of the motor coils. Of course, the reverse diode in each FET is supposed to help curtail this, but if it has similar limitations, it might no be doing its job. Maybe this also explains why some people can run 6s and some can't - the type and wind of the motor. Or it could be some FET batches have a higher tolerance than others.

The periodic problems with the quark simply can't be related to the current capability. Even grossly underrating the current of each FET for temperature, it still has plenty of current capacility. If each FET only supplied 10A, that would be three phases of 120A each (assuming 12 are paralleled per phase)!

Even with this "problem" the Quark is still a VERY nice ESC and wouldn't hesitate to buy it again. The construction as seen here is very high quality. They even added a heat spreader between the PCB instead of relying on pure stacking. So, we know we should only run 5s tops and add cooling for best results.

My thermal pad was not sticky at all, but that may be due to me running in dirt while the boards were loose and a plastic end was opened up. The thermal pad just feels like a soft pad. I kind of wonder if it is part of the overheating problem in general. I know for me it is part of the problem because the FETs aren't making good contact with the thermal pad and the heatsink.

I ran it today. It ran good, but it did thermal on me after about 10 minutes. I tried to tape the boards so they wouldn't move around, but it just wasn't enough to hold the FETs against the heatsink. I have to take it apart again and I think I will remove the thermal pad and use some of the recommended epoxy to fix it once and for all.

[squeeforever]

How difficult is it to change the FET's? Not sure if its possible, but maybe a higher voltage FET?

[BrianG]

Ugg! While I'm sure possible, it would take a very long time to remove all 36 FETs and solder in new ones. You'd have to separate the PCBs for one thing, and you'd have to have a quite small tipped iron for this job. Plus, by the time you buy all those devices, the price and labor would make it preferable to simply buy a new one. Not to mention that there might be other devices with similar working voltages. You'd end up building a whole new ESC!

[squeeforever]

Well if you did, that would make it one damn good controller.

[Mike.L]

i know selvester was thinking of doing just that, he was going to take apart his 9920 and look at it and order far better components and and on top of it mill his very big alunimum heat sink. But because hes busy and lack of time, re-considered it.

he did tell me that when he comes back that hes going to add avery large heat sink to his 9920.

[BrianG]

I'll be curious to see how it all turns out. Is he looking for more current capability, or just better cooling? Any idea which FET he plans on using?

The only perfectly built ESC I've seen is the CC Phoenix HV-110. Each layer has an Al slab connecting to the heatsink and the main heatsink looks quite beefy. It's probably a PITA to assemble those since you can't just slap the PCBs together and then slap a heatsink on the top layer of FETs. The Quark is probably the next best as far as construction goes.

[thesteve]

I don't think the voltage limit of the mosfets is the problem with supporting 6S in the ESC. Mosfets will keep operating just fine at higher voltages until they fail - and when they do fail you'll know cause you'll smell that lovely electronics smell. Without having one of the controllers to look at I believe the problem with higher voltages is simply a function of how fets work, in general you need/want a higher voltage on the gate(the on/off pin if you will). To get this higher voltage pretty much all ESCs have a voltage charge pump(booster) So even though you may have a 6 cell nimh pack connected the ESC is turning the fets on and off with 12 or more volts. I am guessing the charge pump in the Quarks isn't high enough to keep the gate voltage higher then the input voltage, and as a result the fets will not turn on properly.
The ESC above appears to have 36 fets (6 per phase) which is a very solid design. Even the Schulze U-Force which is a very good controller only has 18 fets in it(3 per phase)
Also the fets they are using are quite good - rated at 28 amps in real world conditions.

[GriffinRU]

Originally Posted by thesteve

I don't think the voltage limit of the mosfets is the problem with supporting 6S in the ESC. Mosfets will keep operating just fine at higher voltages until they fail - and when they do fail you'll know cause you'll smell that lovely electronics smell. Without having one of the controllers to look at I believe the problem with higher voltages is simply a function of how fets work, in general you need/want a higher voltage on the gate(the on/off pin if you will). To get this higher voltage pretty much all ESCs have a voltage charge pump(booster) So even though you may have a 6 cell nimh pack connected the ESC is turning the fets on and off with 12 or more volts. I am guessing the charge pump in the Quarks isn't high enough to keep the gate voltage higher then the input voltage, and as a result the fets will not turn on properly.
The ESC above appears to have 36 fets (6 per phase) which is a very solid design. Even the Schulze U-Force which is a very good controller only has 18 fets in it(3 per phase)
Also the fets they are using are quite good - rated at 28 amps in real world conditions.

And this story about FET's and high voltage based on?
And do not blame fet's control if you not sure how it works...

Artur

[coolhandcountry]

What about the caps? What voltage are they rated at?