PC fan on 30 volts

[Gustav]

Hmm,liquid Nitrogen,sounds like a plan.

What i really want is little mini watercoolers on the FETs,with a little pump to send coolant to a little radiator in the front grill.You're a technical guy Brian,can't you sort that out on my MGM? Oh and it needs to weigh less than 100 grams too:018:

[BrianG]

Originally Posted by Gustav

Hmm,liquid Nitrogen,sounds like a plan.

What i really want is little mini watercoolers on the FETs,with a little pump to send coolant to a little radiator in the front grill.You're a technical guy Brian,can't you sort that out on my MGM? Oh and it needs to weigh less than 100 grams too:018:

Watercooled, eh? Let's see.

- Radiator
- Attach a 80mm, 12v fan to the radiator.
- Waterblock to attach to your FETs using thermal epoxy.
- Of course you need a pump.
- And lastly, you'll need some tubing.

However, I don't see this weighing 100g - more like 900g+.

IMO, just get something like this.

[Gustav]

Yeah! Now that's what i'm talkin' about.

I was wondering how the cooling tubes on the shulzes transfer heat from the FETs,they just look attached to the sides of the PCB,Just a thermal compound or mat or something? Those little waterblocks would be the buisness though.

[BrianG]

The FETs used in most ESCs I've seen are just a huge array of surface mounted devices (no mounting "tab"). Therefore, they mostly use the PCB traces as a heatsink. Since there are so many FETs, the PCB is nowhere near big enough.

Schulz apparently attaches heatsinks to the sides of the PCB where the traces end.

Other designs seem to have multiple FET PCBs stacked on top of each other with the heatsink on the top layer. Each layer has the PCB above it laying on top of the FETs so the heat transfers to the layer above it (that sounded confusing even to me). BKs and Quark seem to do this. They have to make sure that each PCB firmly touches the FET below it or the heat won't be conducted as effectively.

[Edit]: Personally, each design leaves room for error and some FETs aren't cooled enough. I would like to see a slab of Aluminum between each layer of FETs, maybe machined so there are little raised squares where the actual FET is to prevent shorts. Each slab is then attached to a heatsink on the sides. Each layer gets cooled more or less equally that way. Although, it would be substantially larger, weigh considerably more, and be more expensive...

[Gustav]

I always think the stacked designs look like they must suffer with low surface area:volume ratio,i wonder how much hotter they are in the middle.

[BrianG]

I think the designers rely on the fact that heat tends to travel from a higher concentration to a lesser concentration. The heatsink on top cools the top layer, which then "attracts" the heat from lower levels. I'm sure those lower levels must be hotter, but FETs conduct less the hotter they get, so they are self-regulating in a way. Of course there are limits, which is why some burn up.

IMO, they should add more heatsink area or more FETs. Adding FETs would do two things:
1: Increase current handling (duh :))
2: Decrease total output resistance, which in turn causes less voltage drop on the devices, which then reduces generated heat, and increases efficiency.

[Gustav]

Air goes in the scoop,then it's forced through a sort of letter box,then it narrows again at the end before it hits the esc.Self explainatory really,just look at the pics.I'll test it out,should work pretty good me thinks.:005:

That looks like it will work a treat, so long as it stays on its wheels. :026:
Did you work out the size for the in/out.

[Gustav]

Nah didn't work that out,just did it,had these scraps of lexan left over from an undertray i was making for a g-maxx.Didn't take long so i won't cry too much when it gets destroyed:030: ,it's just 1mm lexan superglued together.

[BrianG]

Ah, a highly scientific "yeah, that looks about right" procedure? :dft012:

I did a little reading up on the venturi effect and it seems it needs a little more than a funnel. From the materials I read, you need an input funnel with a slope of ~30 degrees, and an output funnel with a slope of ~5 degrees (for least amount of drag). I think you also need a good amount of pressure forcing air into the input because otherwise the air will simply go around the restriction where it is easier to flow (around the hole instead of in it). It might work better at higher speeds since air doesn't have as much of chance to change directions and so is forced into the opening. IMHO, I think the only gain in cooling you will see is from simply having some air coming in directly onto the ESC and not from any venturi effect. Once you test this, try simply making a hole in the windshield with a panel that directs the air directly to the ESC (instead of a tube). I think the effect will be the same or better.

No matter the outcome, good or bad, at least you are trying something different! As for me, I'll just stick to using larger heatsinks. :)

Something else you might want to try; Simply blowing air onto a heatsink is not really air flow; you get eddy currents. A lot of OEM computer makers (Dell, Gateway, HP, etc) sometimes use a ducted system with a fan on the output to pull air through the hot device. For this to work, an air shroud must be fashioned around the heatsink so the air is targeted directly across the fins instead of simply going around them. And since the air is being pulled through (instead of pushed), there is a cleaner flow of air (read: less turbulent).

Good luck! It will be interesting to see how this all works for you!

[Gustav]

Yeah very scientific:005:

I agree,just channeling the air across the esc is what i'm thinking of doing.I was just messin' about really,it's kind of the way i think out loud.I think the air will probably manage to 'dodge' the scoop at these low speeds.

There's no direct route for air onto the esc on this truck unless i raise the esc to the level of the windscreen,that might be the best way to go,with a lexan 'box' between the windscreen and the esc.And maybe an exhaust channel straight out the back too.

I've got some heatsinks coming too and some thermal adhesive,i'd like to try to get away without a fan if i can and just use more heatsink and find the best way to channel as much airflow over it,like you say.

[squeeforever]

Is the heatsink on the MGM's removable? It looks to have screws that hold it on. Maybe you can remove the heatsink and put a larger one directly on the board.

[Gustav]

I think it's stuck on there aswell,i was just going to add to the heatsinks.

[BrianG]

Once you remove the screws, you can try gently twisting the heatsink as the thermal paste may have dried a little causing it to stick somewhat. Also, be careful with adding more heatsinks to the MGM. It looks like the current HS may be directly attached to the FETs which might put physical stress on the FETs if you add more weight.

That's kinda why I like the Quark a little more; the case is the heatsink and adding more doesn't physically stress the FETs.

[Gustav]

Maybe the screws support the weight of the heatsinks aswell though,and they're screwed from the side aswell.i was just going to add some small ramsinks,not alot of extra mass really,thanks for the heads up on that though.