ESC direct cooling - Failed


#1

Hi Everyone !

I’m working on a cheap way to make sure everything is cooled down and went for a direct cooling of the Esc (a VESC 4.12 in my case). I’ve bought a supposedly “non conductive” thermal fluid (conductivity < 2µs/cm) https://www.amazon.fr/gp/product/B00AXW07TE/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1.

I’ve put the Vesc in a casing filled with this fluid. Unfortunately, eventhough everything worked fine without fluid poored in, as soon as I put the liquid, it stopped working.

What I don’t understand: this fluid is meant for PC watercooling and some use it to fill an aquarium with it and drop ALL the computer’s components on it, so it should work in my case too right ? Maybe it only works with low power ?

After removing the liquid and drying the Vesc, everything works again, meaning there is no short and the PCB isi not damaged.

Does anybody have any idea on what could have happened ?

Thanks =)


#2

I have a modified vesc 4.12, too.
What does not working mean? No Leds illuminated or no communication over uart/usb or no reaction to the remote?
It might only be that the connectors dont‘t work in the fluid. I would try soldering an uart cable to it and monitoring it with the vesc tool while filling the fluid in then.


#3

It means that right before I filled the box with the fluid, the motor was spinning, and the LED on the 2,4ghz receiver was ON. Then I poored in the liquid and the LED started blinking until it stopped.
I’m now considering using thermal paste to cool the ESC down.


#4

Everything in fluid even receiver ?


#5

I tried the exact same thing with thermal paste in my vesc 4.10’s, they both fried when load was put on the motor. I have no idea why as this stuff was non conductive.


#6

The fluid is for thermal exhanges but even with low conductivity may cause this kind of interference.
Gill


#7

It is meant for liquid cooling to prevent corrosion when used in a water cooling circuit, not for direct cooling.


#8

@Alexandre has a good point:
If your receiver in inside the fluid, it’s likely that signal won’t get to it.
at least for sure it can’t go through water…
may be try to get the fluid level above the ESC, but keep your receiver out of it (it doesn’t need cooling anyway…)


#9

The receiver was outside of the fluid, otherwise the 2,4Ghz signal wouldn’t be able to connect to the receiver.
After thinnking a lot about it, I coated the ESC with Epoxy to waterproof it.


#10

If you search the forum you will find some cooling efforts for the vesc 4.12 by @PowerGlider and myself.
I would suggest, if you make the effort to modify your vesc, to go the whole way.
I have flipped / remounted the Mosfets on the vesc and soldered 3mm copper plates to it, on this a thin (!) silcone-fireglass insulator pad and then there is a watercooler mechenically pressed against it. To go for higher currents than about 90A on the motor, you need to switch to 0,5mOhm shunts (half the normal size). Then you need to recompile the software and allow higher peak currents (set mine to 290A) and higher average currents ( set mine to 160A). I have already pushed 140A for a while, but as a good experimenter I also drastically reduced the water flow and now it is too low to sustain this power … In the long run a new PCB and a efoil specific vesc will be the most time effective solution.
Edit: here is a quick CAD of my old version to give you an idea:
image


#11

Thanks Flo ! That’s very interesting.
Unfortunately, my Vesc died in the Epoxy resin and I’m now planning on buying some experimental 100A Vesc.
I don’t really understand the need for higher current ESC, 80A is the maximum my motor can handle anyway, and studies have proven that we should be able to be foiling at 1000W, so 20A.
Why do you need to go up to 160 A ? Is your motor a 10kW one ?


#12

How did this happen?


#13

One of the cables going to the receiver de-soldered from the VESC while on the Epoxy, I could remove the epoxy and try to solder it again to the board but it’s a lot of work and I’ve found an alternative solution