The only way I’d use water cooling for electronics

Yes you can, exposed heatsink or forced air intake to heatsink funnel :wink:

Either way you can still ensure water resistant enclosure and allow good airflow for the critical components.

1 Like

Touché. :wink:
Good point. I was assuming everything was inside the enclosure.

1 Like

I did a test, from room temperature, used a blowdrier on different components to see how long it takes to make them bump in temperature.
Motors spiked up in about 10 seconds, but vescs in the deck took more than 30 seconds and the temperature was still slowly rising even 5 minutes after. Wood may not have a big thermal capacity but is good enough with enough surface exposed.

Problem is, on hot days, that black griptape on top soaking heat, with copper under giving even more, things get toasty…

Also, any short on the battery to the copper would be pretty damn bad

2 Likes

Channeling Aliens…“Yea, but it’s a dry heat” :grin::grin:

1 Like

Been digging out old 2005 copper GPU blocks and old pc cooling bit to try cobble something together.

I got a 3d printer recently and planning on making something to mount vescs to the copperblock… This stuff sounds ideal :slight_smile:

Might be insane to try it but I got laughed at for trying it on my PC back in the day lol

Temp gains were worth it then for messing with overclocking… I can’t help but wonder how much fun any gains might be on esk8 lol

1 Like

Do you have a good heat sink you can use, putting it and the ESC in front of a fan, to compare against the water-cooling? That would be great to see.

Plan on testing the differences properly, and not just on the bench on a board :slight_smile:

1 Like

We don’t shake apart our PCBs (usually), so I don’t believe that 3D printed heatsinks (which don’t experience the stress of trucks) are the first things that will fail due to vibration.

Even then: mount the entire ESC in a simple, urethane shock absorber. IMO all eSkate electronics need better vibration damping than we have now.

3 Likes

Isn’t that what I said? Unless you were addressing the other guy.

1 Like

Mooch you should check out this beast from battery hookup

Mercedes OEM L09-1 EV Battery is pretty cool.

For $99 it includes an awesome water-cooling block that you could also hook up to a motor or esc (esc should be just mounted directly to it)

1 Like

I was able to dissipate 250W on a single to220 mosfet with standard heatsink+fan from old single core AMD CPU.

1 Like

What was the MOSFET junction temp at that power level though?

At 250W, using a fairly standard MOSFET junction-to-ambient thermal resistance value of 0.5°C/W that results in a 250W * 0.5°C/W = 125°C temp rise. Using an okay performing standard heat sink setup with a sink-to-ambient thermal resistance of 0.5°C/W that adds another 250W * 0.5C/W = 125°C temp rise.

If we assume a 25°C ambient temp then we have a junction temp of 125°C + 125°C + 25°C = 275°C. The highest junction temp rating I have seen is 175°C for a standard MOSFET. :scream:

That MOSFET can be run at levels that high but IMO it won’t last very long.

Even if we use an amazingly efficient heat sink. 0.1°C/W thermal resistance, that results in a 175°C junction temp for the MOSFET. And that is only assuming a perfect distribution of heat across the heat sink (which can’t happen) and a room temp heat sink under the MOSFET (which is impossible). You’ll still be running way above the MOSFET’s rating.

It’s fun to set up stress tests like this though! I’ve done that with some TO-264 FETs I was testing for my electronic loads and it sounded like a 22LR was fired when the MOSFET’s blew. :grin:

LOL…beast is right…omg.
I’ve seen that pack mentioned here but never read the posts. A verrrry interesting option for a power hungry setup that needs a lot of cooling.

Once you add on the cooling system it’s even more of a monster. With all the “free cooling” we can get when moving I’m wondering if it all ends up being worthwhile? I’m the esk8 noob though so I’m not familiar enough with the setups where using this pack would be a great fit.

Wow…I’d love to play with one or two of those things though. Just to see what can be done. Time and my wallet make that impossible though. :grin:

Thanks for the link!

1 Like

I used SUP57N20 soldered with it’s base to IHS of some CPU, then using standard thermal paste to the heatsink, temperature on the package was 100C I think.
If I would be to use modern CPU cooler with thermal pipes I could reach up to 350W.

1 Like

100°C on the epoxy case?

yes

Then you were way, wayyyyyy over that FET”s max junction temp. :slightly_smiling_face:
The case temp is measured at the junction of the drain lead to the epoxy case, the base of the drain tab, or (the best way) on the middle of the FETs rear drain plate, using a small thermocouple. The thermal resistance of the epoxy case itself is incredibly high resulting in a large drop in temperature from the drain to the top of the epoxy case.

We can’t confuse a capability of a component with a rating. Sure, we can have TO-220 MOSFETs dissipating 350W, and they might even do that for a while, but it is far beyond the MOSFET’s ratings and the life of that MOSFET will be very short. I was taking those TO-264’s up to 700W, and some lasted a decent amount of time, but they all eventually blew when doing reliability testing at much above 150W using a 100mm sq. 0.17°C/W pin fin heat sink and high speed/pressure fan.

In my experience a good fan-cooled heat sink with a single TO-220 can get up to about 100W before you start approaching the max junction temp of the MOSFET. High performance cooling might get you a bit higher but you are bottlenecked by the TO-220’s high junction-case resistance and tiny contact patch.

Yeah resistance of the epoxy is high but there’s nothing cooling it from the top so measurement of 100C would result in actual temperature of the die being 110-120 the mosfet is rated for 175C
I’ve never heard of mosfets aging.

Thermal resistance (and the temp difference it causes) exists whether there is cooling or not. :grin:
It’s a function of the material itself. Cooling can help but it’s still there.

How are you determining that 20°C temp rise from the epoxy to the junction?
From the drain pin or rear of the FET (the datasheet doesn’t say) to the junction it is at least a 125°C difference at 250W and that is a MUCH lower thermal resistance path than from the epoxy top to the junction.

Check out thermal fatigue of semiconductors. There is failure from just running at high temperatures (but still below the junction rating) and there are additional failure modes from thermal cycling. Once mounted to a circuit board there can be other failures due to the different coefficients of thermal expansion between the FET and the circuit board. This is a big problem for leadless MOSFET packages where there is no “flex”.

Check out hot carrier injection, time dependent dielectric breakdown, bias temperature instability, and hotspotting/thermal runaway of switching FETs when biased in their linear region.

That last item is a particular problem when switching FETs are used as an electronic load. Even a lot of power engineers aren’t aware of this issue. Your FET has DC plot line in its SOA graph though, which is good. Just stay as far away from the DC line as you can, especially at higher voltages.

1 Like