ESC Cooling Concept- Theory and Practice

[Drag Boat Bob]

Just so everyone knows, you did sand the FET's flat before attaching the copper plate, correct?

At least it appeared that way in the picture. Flush contact with the FET's I believe is essential. One FET failure = total failure.

Unfortunately the Chinese products are not uniform in this.

Don't stop now, you're on a roll...

[questtek]

found a windscreen washer pump that might work as a pump for the closed loop system, anyone else got any ideas for a small pump?

If you really want to go with a closed loop system, here is probably the ideal pump...and under $10. Works from 6 to 12 vdc so you could use either a 2s or 3 s pac.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=250513456085&ssPageName=STRK:MEWAX:IT

Good luck and I look forward to hearing more about your closed loop system.

[questtek]

Just so everyone knows, you did sand the FET's flat before attaching the copper plate, correct?

At least it appeared that way in the picture. Flush contact with the FET's I believe is essential. One FET failure = total failure.

Unfortunately the Chinese products are not uniform in this.

Don't stop now, you're on a roll...

Yes, glad you mentioned it. You should be able to see it in one of the close up pictures I posted. I used some 220 grit paper on a flat sanding block. It was interesting to note that at least 3 or 4 FET's appeared NOT to be making contact with the heat sink on the "stock" Mystery controllers.

It is clear to me these FET's are seeing lots more heat than anyone probably anticipated. Most ESC temperature measurements are done either by touch on the outer part of the heat sink or a temperature probe/thermistor on the outside. The real trick is to measure the temperature inside, on the FET's.

As a side note, if you examine the temperature data I posted, you can see that even my copper sink with tubes (although I did not have any water running thru them) permitted the FETS to run about 20 degrees cooler. (109.5 - 89.4)

So why is that? The amount of heat transfered is proportional to the mass of the heat sink and its thermal conductivity. The mass or weight of my copper heat sink is 3.5 times that of the stock Mystery aluminum heat sink (2.1 vs 0.6 oz) and the thermal coefficient of copper is 1.6 times that of aluminum, (401 vs 250). Thus, my copper heat sink, when compared to the stock aluminum heat sink is 5.6 TIMES better (3.5 x 1.6), better in terms of overall thermal cooling performance. The downside is that the copper heat sink adds 1.5 oz or 40 grams additional weight.

Glad you are enjoying this forum but its a big thanks to all those contributing unique ideas and innovative suggestions!

[crabstick]

If you really want to go with a closed loop system, here is probably the ideal pump...and under $10. Works from 6 to 12 vdc so you could use either a 2s or 3 s pac.

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=250513456085&ssPageName=STRK:MEWAX:IT

Good luck and I look forward to hearing more about your closed loop system.

Thank you sir, I might see if I can source one locally, but if not will grab one of those.

[Jeff Wohlt]

Good stuff.

I might suggest that when you build your plate (for water) that you use 3 cooling lines so you have a water tube over each fet row and not on the edge like other builders do. Why try to cool the plate from the edges when you can solder them right over the fets and draw the heat directly away.

For the pump method this would be best. Even rudder cooling this is certainly a better system then edges.

[Drag Boat Bob]

I guess that the next thing to determine is exactly what size container of Blue Ice would be necessary (with a bit of margin) to complete a run for a given size ESC.

I can see purchasing a small can of specific size (Jeff W. link) then:
1. place liquid blue ice in a small baggie (the cans with hinges don't appear water tight).
2. place the baggie in the can and ensure that you get it full.
3. freeze both.
4. remove the frozen block from the baggie and 'seal-a-meal' it to size.
5. repeat as many times as you need for ice blocks.

If you had a bunch of frozen blocks, you could just open the lid of the can and replace the frozen part.

Still need to figure out a good method to attach the can to the copper plate/esc and keep constant pressure on it. Of course keeping the lid closed also.

[questtek]

I guess that the next thing to determine is exactly what size container of Blue Ice would be necessary (with a bit of margin) to complete a run for a given size ESC.

I can see purchasing a small can of specific size (Jeff W. link) then:
1. place liquid blue ice in a small baggie (the cans with hinges don't appear water tight).
2. place the baggie in the can and ensure that you get it full.
3. freeze both.
4. remove the frozen block from the baggie and 'seal-a-meal' it to size.
5. repeat as many times as you need for ice blocks.

If you had a bunch of frozen blocks, you could just open the lid of the can and replace the frozen part.

Still need to figure out a good method to attach the can to the copper plate/esc and keep constant pressure on it. Of course keeping the lid closed also.

Again, you certainly ar on the right track. Maybe what I have done and the pictures I included will give you some more ideas or help solve some of your questions.

In additional testing I concluded the metal box is a poor choice. Sweats too much, hard to seal, not readily/easily available, etc. I have gone back to the plastic container....remember you want an insulator for the container EXCEPT where the copper heat sink of the ESC contacts it. This minimizes sweating.

Finding the perfect size Tupperware container, (4 for 99 cents), I mount it in the Mean Machine as shown in the picture. The container has a good, solid seal.

I then add an internal finned sink to contact more of the Blue Ice and tie that into a copper sink that exactly fits the outer top lid of the container. I fill this with Blue Ice gel, (leave a little ai space for expansion), and out it in the freezer, copper side down.

In the pictures you will also see how I have modified the Mystery controllers
1. Flat lapped the Fets and put on a copper heat sink that mates to the heat sink on my tupperware Blue Ice container.
2. Changed the power wires 180 degrees so they come out of the front.
3. put a thermistor inside to measure actual FRT temperature.
4. Put on an additional capacitor bank, (three more caps at 470 uF each or 1410 uF aditional to the two 1000 uF caps already on the board. This gives me a total of over 3100 uF, not far from the 5000 uF cap board that ETTI sells.

I wll instrument the entire MM with Eagle Tree sensors including GPS and X-Y acceleration along with all the normal sensors to get rpm, current, pac voltage, etc.

The advantage of this system now for me appears to be ease of use. I can easily build a half dozen of the tupperware containers with Blue Ice and the heat sinks pretty cheap and just keep them in the freezer. The ESC with its copper heat sink on the bottom is in pressure contact with the copper cold plate on the Blue Ice flled tupperware container. I use strong, wide rubber bands to insure there is good copper to copper plate contact. This also makes it very easy to change out the Cooling pac.

Hopefully I can post some actual test results soon. I want to see the effectiveness of cooling at much higher current rates on my 540 XL. Remember now the ESC is passively cooled with no water pipes. This way I can use direct water cooling for the motor meaning it will see colder, ambient water temperature rather that a higher temperature from the water first passing through the ESC first.

Comments and Suggestions?

[FE Wannabe]

All good ideas, Brad. I have two different heat pipe setups that I have been playing with. They are made from 1/4" copper tubing and have an incredible fin arrangement. From the experiments I did I was not impressed with the amount of cooling the heat pipes gave me.......even with a big finned radiator on the heat pipes. There is a degree of complexity with this system so I put it on the shelf for the time being.

If you see my latest thread, I am tending to favor a passive system that uses the Blue Ice gel and NO water cooling. The test results were very promising and I have a few ideas to considerable improve its effectiveness and practicality. The new system will fit perfectly in my Mean Machines and provide incredible passive cooling. I just may be able to keep the inexpensive Mystery ESC's operating for a reasonable time.

Details to follow in a later post. I need to go to my local industrial metal supply store and buy some more 1/4" thick copper plate. I am so glad I have a solution that does not require soldering the copper cooling tubes to the thick copper heat sink. That was a REAL pain.

Thanks for the reply! I was wondering if the heat pipes were worth the effort to attempt, now I know.

Another question I have is, can you get the FETs too cold?

Keep up the good work!
Brad

[questtek]

Thanks for the reply! I was wondering if the heat pipes were worth the effort to attempt, now I know.

Another question I have is, can you get the FETs too cold?

Keep up the good work!
Brad

FET Temps

As the FET temp increases, the current through it decreases. Increasing the temperature of a FET tends to decrease the mobility of the charge carriers in the channel, effectively reducing the current through the channel……..This is out of the Physics lab of the university of California at Berkeley.

Please clue me in when you understand it! Seriously, I believe the vey high temps the FETS see are the major cause of ESC failures. I was really amazed at just how high the temps get even at modest power as measured by a thermistor placed next to the FETS. I cannot imagine there ever being a condition where the temps are too cold...unless possibly if you run them in liquid nitrogen. Even with a Blue Ice sink the temps were never too cold. So for now I will just try to keep them as cold as possible.

New Mystery Build
I have made an number of modifications to the 100 and 200 Amp Mysery ESC's
1. Rerouted power (battery lines) 180 degrees and replace with 10 guage flex silicon wire and large bullet connectors.
2. Integrated an "anti Cap Spark" line with 100 ohm 1 wat resistor (red silicon line, 16 guage).
3. Replaced the stock caps with Four 1000 mfd caps, 35 vdc on the top of the board. Total now 4000mfd.
4. added a thermistor to the caps to monitor their temp (see pic for close up)
5. added thermistor to measure direct FET temp.
6. Flat-lapped Fets and mounted a large heat custom copper heat sink with 2 copper cooling tubes (this can be expanded to 4 tubes if necessary.
7. Used a special 2-part Epoxy that had tremendous thermal conducton properties and great sealing ability to bond the FET's to the copper heat sink.
8 I will use this thermally conductive expxy to encapsulate the entire board to make it waterproof as well as the ability to provide optimum heat transfer
9. rewired the 3-wire RC receiver cable two thermistor cables so ALL cables exit from the same side.

Lots of work but I just took all the suggestions on the forums and incorporated all that made engineering sense.

Stay tuned for the testing!

[Drag Boat Bob]

If possible show pics of how you encapsulate in epoxy.

Too bad you cannot obtain even smaller tupperware containers given the temps so far.

[questtek]

Picture is attached but its not real pretty. The thermal conductive encapsulating epoxy the texture of honey only black in color. It takes about 24 hours to room cure or only a couple of hours at an elevated temperature.

I build dam out of black electrical tape and just drop the thermally conducting but electrically passive epoxy into it. It would be easy if I built a small silicon mold, but this is not mas production, just testing.

Not sure how small you can go on the plastic container. Remember I was running at 75 watts and that 540 xl should crank close to a thousand watts peak I belive.

My new Blue ice packs have a black heat sink inside that is thermally bonded to the outer commper plate that the ESC rests on. This way the frozen blue ice gel is always in contact. Pic is attached on this heat sink arrangement. Hope this helps.

[questtek]

Good stuff.

I might suggest that when you build your plate (for water) that you use 3 cooling lines so you have a water tube over each fet row and not on the edge like other builders do. Why try to cool the plate from the edges when you can solder them right over the fets and draw the heat directly away.

For the pump method this would be best. Even rudder cooling this is certainly a better system then edges.

I took me a while to machine a new copper heat sink out of solid stock but I made several of them; both 2- tube and 4- tube versions This is one MOTHER of a heat sink. I can not wait to see the test results. The end tubes are sealed with black thermal conductive epoxy.

Keep up the good suggestions.

PS, I also added a cooling place and thermistor to the upgraded Cap package. I doubled the caps to 4000 uf and with a termistor between them it will be interesting to see what happens. On one version I even copper heat sinked the caps with a thermistor under them to monitor the cap temps at high amp rates.

[rawsonreyes]

great r&d guys

hope to see some type of proto: type for those chinese esc's but nice attemp for helping spare some bucks for the newbe in fe as you all know it's cost a lot to import these high end german or eu stuff.

[Jeff Wohlt]

What is the chance of putting any kind of plate between those boards? Anything would help. I thought putting something between them where the bottom of the fets and then board will not short by using a piece of rubber or something like that.. File any tips off so they are not as sharp. IF we could get those other 2 boards on the 200 amp version to cool a bit. I have not taken one apart and do not have one. That was the reason I was trying to find single layer boards with more than 125 amp rating...the suppo model is one layer for HV 200 amp.

NICE job on the block....I would suspect 4 outlet version that can add a large amount of cooling and make it more even.

Great work!

[questtek]

What is the chance of putting any kind of plate between those boards? Anything would help. I thought putting something between them where the bottom of the fets and then board will not short by using a piece of rubber or something like that.. File any tips off so they are not as sharp. IF we could get those other 2 boards on the 200 amp version to cool a bit. I have not taken one apart and do not have one. That was the reason I was trying to find single layer boards with more than 125 amp rating...the suppo model is one layer for HV 200 amp.

NICE job on the block....I would suspect 4 outlet version that can add a large amount of cooling and make it more even.

Great work!

I have a number of 100 and 200 AMP Mystery controllers, (and others), that have gone south. Not uniquely, the main, top control board looks to be identical on most of these controllers listed under various brand names. This is why I have included a picture of the controller side of the ESC so those who have different controllers can match it up.

On the flip side of the top controller board are 18 FETs arranges in three rows of 6 each row. These manufactures rate this single board (controller and 16 FETs on the flip side) at around 60 AMPs. Ass a second board with 16 FETs only and you have it into a 100 AMP rated controller. Add an identical (to the second) FET board and now you have a 200 AMP rated ESC. They do change the caps on each of the models. Also, to throw buyers off, the different mfgs add different types and arrangements of caps, ie: two 100 uf caps, five 300 or 540 uf caps, etc. Aside from these changes, (also the shrink wrap and label), they all look like they came out of the same litter.

Back to your point about cooling the center FET board. Excellent idea and this is what I am working on/testing. At this time to me, it does NOT appear practical or feasible to sandwich a metal plate between them. Instead I plan to fill this void with the black conductive epoxy I am using. The product is excellent. Using it will contact ALL of the FETs plus make it waterproof. Remember, this is also considered an encapsulating epoxy. I think this might be the best, workable solution, at least at this time.

My 4-tube copper heat sink is a real jewel. It adds a bit of weight but so would a V-12 engine if you dropped it into a Pinto.

What I am finding is there at least 5 or 6 different mods that should be made to this type of ESC to make it potentially operate reasonably in a FE. My approach is to build the water test stand and instrument the pants off the buggers so that some logical, engineering-based mods can be made to improve its performance and reliability. I believe the basic design of these ESC’s is good however for operation in a demanding FE…………well, that’s a different story that is unfolding quite slowlyyyyyyyyyyyyyyyy.