Can an oil cooled controller take to-220s to 75a per fet?

[Olly3012]

Have you tried this ment to be the ducks nuts, COOL-GELFILM™ SZ is a new class of thermal interface material that dispenses like a film but delivers performance characteristics similar to a thermal grease or gel. COOL-GELFILM™ SZ is a modified zinc oxide admixture, crystallite filled, electrically insulating at normal voltage, and engineered for large area and gap filling interface applications. It is designed to have high compressibility and instant flow to enhance thermal transfer from power device to heat sinks. COOL-GELFILM SZ is slightly tacky on both sides for optimum instant thermal transfer performance. http://www.aitechnology.com/store/index.php?main_page=product_info&cPath=6&products_id=17

 

[Arlo1]

I used this stuff in one of my last power stage builds where the parts needed to be electricaly insultaed .13-.3 degC/W http://www.digikey.ca/scripts/DkSearch/dksus.dll?wt.z_header=search_go&lang=en&keywords=926-1157-ND&x=7&y=10

 

[Jeremy Harris]

Jeremy - I understand what you're saying. But you're not thinking outside of the box here. You aren't taking into account the cooling provided on the non-junction side of the fet nor the better cooling of the heatsink itself. But you are right. It is a bit like polishing a turd as the case itself get's HOT. Likely, this will only allow me only a bit more power.


I'd like to try that sikaflex stuff. If I can't find it locally I'll just use blue gasket builder or the stuff chalo found.

In another thread here we did the sums on this. Even with an infinitely good heatsink (i.e. one where the case remains at ambient temperature no matter what heat the FETs chuck out) you still can't get anywhere close to using the current capacity of the FETs without the junction overheating.

The silicon die is welded to the copper tab, so that has about as good a thermal contact as you're going to get. The extra heat you try to dissipate through the rest of the FET case by surrounding it with oil will be miniscule, maybe 1 or 2% better than it was with just a good heatsink.

Cooling the case better will make a small difference, but won't massively increase the current handling capability of the controller, my guess is you might see a 10 to 20% improvement at best. The single most significant thing you can do to increase the current handling capability of a controller is to replace the FETs with ones with a lower Rdson and reduce the heat being generated at source. For example, replacing the nasty IRFB 4410 FETs that a fair few Chinese controllers have with IRFB4110 FETs allows more than double the current for the same controller heat dissipation, a pretty worthwhile improvement. If you only run at around 60 V to 65V or so, then using IRFB3077 FETs would allow about three times the current for the same heat dissipation.

 

[auraslip]

Arlo - I understand what you are saying, but I don't really know what to do with it. The software settings are maxed out. I'm using the throttle to limit speed. I need to limit speed some how!


Jeremy - a 10% boost in power would be pretty awesome. Especially since this mod will only cost me $5. I don't see the difference between the two fets you listed. Did you mean replacing the chinese knock offs with real ones?

 

[auraslip]

Just picked up this. Will use today.

Controller is getting out of hand in this 100f weather. Getting up to 80c running at only 30mph and 1500w. WTF. My 48v ping + 9x7 could do that all day an the controller was barely warm to the touch. Now with my 22s pack and hs350 the damn thing is burning hot.

 

[auraslip]

I'm gonna take a guess and say that my fast wound motor is the reason the controller is heating up so fast. Kinda wish I had realized that before I got it. 50mph is nice, but really unnecessary anyways.

In any case. Progress.

I filled it in from the inside first going around the edges since I can't spread the wires open to get inside the bunches. I did the inside and then let it dry a bit so that when I did the other side it wouldn't drip down inside the controller.

Looks good. The stuff works as promised.

Don't want it to run or get get on the controller!

Here is where the temp sensor is wedged FYI

Here is the beefed up traces and r12 mod.

 

[auraslip]

Waiting for vent plug to dry now. I drilled and tapped a fill bolt as well.

Can anyone think of a household tool I could use to fill a 1/4" hole?

 

[Arlo1]

Waiting for vent plug to dry now. I drilled and tapped a fill bolt as well.

Can anyone think of a household tool I could use to fill a 1/4" hole?

Turkey baseter?

 

[amberwolf]

Those usually have pretty large openings, more like 1cm and up. How about a syringe--with or without needle. Can be had from Walgreens and the like, I expect, or anywhere with a pharmacy. Sometimes from auto parts stores, or even electronics stores.

Feed and pet stores also carry them, in larger sizes, for helping feed incapacitated animals.

 

[auraslip]

Well this has been sort of a frustrating night. First I was about half way through filling the cotronller when I noticed it was leaking. So I guess the gasket wasn't good enough. So I got out the blue rtv, and went to town.

After closing it all up and letting it dry for an hour I went for a spin.

Second frustrating thing. My temp sensor is now on the fritz. Just like the one in my motor that died right away. So mad. Cheap chinese hobbyking vt bullshit. Must be a bad batch as the one in my old motor lasted a year. Well, I have to replace the one in my motor......andddddddd controller seems to be still leaking so I'll need to open it up and redo it. Don't know if it's from the vent tube that's leaking or the gaskget. I tried to clean off the mineral oil with paper towls, a tooth brush, and rubbing alcohol but oil is everywhere. It might of caused the RTV to not take. My whole house smells like a clean babies butt.

Fill hole and vent tube. I used an ecig juice bottle to fill it up. Took ~300ml.

Now the good news. This mod does work. Did the 3 mile convenience store run at 50mph (60a) and the controller was warmish-almost-hot. I'd say it was at 60c. Finally. The motor is hot and not the controller!
Don't know if this is just a temporary gain from doubling or trippling the weight of the controller. Would I be able to push that much power continuously - I don't know. In any case, if I can get this sealed up it might be the best $15 I ever spent on my ebike.

 

[Jeremy Harris]

Jeremy - a 10% boost in power would be pretty awesome. Especially since this mod will only cost me $5. I don't see the difference between the two fets you listed. Did you mean replacing the chinese knock offs with real ones?

4410's have an Rdson of 10 mohm max, 4110's have an Rdson of 4.5mohm max. That means that an IRFB4110 can be run at more than twice the current for the same dissipation and heat. Rdson is probably the most critical parameter when it comes to selecting a TO220 FET, next to getting the right voltage rating. That's why using the IRFB3077 makes so much sense if you don't want to run over about 60 to 65V, as it has an Rdson of only 3.3 mohm max. The current rating is a meaningless number in practice, as you can never realistically run the FETs at their maximum current without cooking the junctions, no matter what heat sink you put on them.

 

[auraslip]

I may be mildly dyslexic for not noticing the difference between 4410 and 4110

 

[nieles]

I may be mildly dyslexic for not noticing the difference between 4410 and 4110

had the same problem, but i didn't noticed it until the Digikey order arrived (but i am dyslexic, so i have a excuse :wink: )
luckily it was only for 12 spare fets, not something i was waiting for.

 

[auraslip]

Found the problem with the temp sensor:

Looks like the stripped too much wire off!

Related question: What's with the bit of reflective metal on the board? It would make sense if the IC was actually thermally coupled to it. As is though?


Now I've had another problem with the controller. This happened once before.

The shunt got hot enough to melt the solder on the CA yellow wire. As a result CA saw max current and shut down the bike. Next time I'm going to make a better mechanical connection. And also limit the current a bit.


Now to tackle the leaking oil problem. It seems to be collecting at the bottom of the wiring harness If have three theories.

1) oil is spilling out the vent hole. I taped a paper towel over it, and it was dry.

2) oil is seeping through the wiring jackets themselves. possible. Can't really work around this in the current form. I'd have to reverse the controller so gravity won't let the oil siphon down.

3) the mineral oil is making all rubber and plastic brittle. the wiring is brittle. the rubber gasket is brittle. Falling apart really. It could be that the oil seeped through the gasket and out the edges.

I'm betting on number three for now since it's the easiest thing to fix. Scrape off the gasket so that the RTV has a metal to metal seal.

Just gotta clean the oil off everything enough to work on it. Mineral oil is EVERYWHERE!!

 

[auraslip]

Alright. Got some early testing.

Cruising at 30 mph 5 miles - ~1200w -

Old temp: 65c
New temp: 51c

The case is actually hotter than ever though. Which actually is probably a good thing since it means heat is transferring better.

Oil is still leaking. But only through the vent hole. The new sealant job as done well! Do I need a vent hole? Will the thing explode with out one, or will oil just seep out of the wires?

 

[sn0wchyld]

nah mate that little tongue is only to enable you to get it into tight spaces, if needed. I clipped mine off, so now it sits snugly under my stator, with the IC sitting right under my windings (held in with some high temp silicone and some thermal cement).

if the case is warmer, then yea thats a good thing ie better heat transfer, assuming that its not cooking the other components on the board. though at ~50deg it should be fine i think.
Now you just need to stick on a couple more large heat-sinks on the outside of the case! It'd be interesting to see how much more power you can push before reaching your old 'un oiled' temps too. as to the leaks, try a longer breather hole, or maybe a small baloon attached to the end, to stop the leaks but allow pressure equalisation? I wouldn't seal it... even a small amount of air in there will expand and potentially cause your controller to drop its oily guts all over the road!

Found the problem with the temp sensor:

Looks like the stripped too much wire off!

Related question: What's with the bit of reflective metal on the board? It would make sense if the IC was actually thermally coupled to it. As is though?


Now I've had another problem with the controller. This happened once before.

The shunt got hot enough to melt the solder on the CA yellow wire. As a result CA saw max current and shut down the bike. Next time I'm going to make a better mechanical connection. And also limit the current a bit.


Now to tackle the leaking oil problem. It seems to be collecting at the bottom of the wiring harness If have three theories.

1) oil is spilling out the vent hole. I taped a paper towel over it, and it was dry.

2) oil is seeping through the wiring jackets themselves. possible. Can't really work around this in the current form. I'd have to reverse the controller so gravity won't let the oil siphon down.

3) the mineral oil is making all rubber and plastic brittle. the wiring is brittle. the rubber gasket is brittle. Falling apart really. It could be that the oil seeped through the gasket and out the edges.

I'm betting on number three for now since it's the easiest thing to fix. Scrape off the gasket so that the RTV has a metal to metal seal.

Just gotta clean the oil off everything enough to work on it. Mineral oil is EVERYWHERE!!

 

[Arlo1]

Look at how a differential is vented on an ATV. They need to allow expansion and contraction of whats inside and not let watter and mud etc in. So Honda usually has a long hose on a sewer trap type setup. But Bombardier (Can Am) uses a funny accordion little house on that just attaches to the vent outlet with a zip tie. I bet if you wanted you could find the squeezable end from a turkey baster and it would work great!

 

[spinningmagnets]

http://www.corvalent.com is a company from 3M that is making a non-conductive fluid that boils at 93F / 34C, The proof-of-concept display is a desk-top computer. The hot parts are placed at the bottom, there is an air-space at the top, and there is a fan that is aircooling the outside of the top-mounted aluminum heat-sink fins which help the coolant to condense back into a liquid. The chemical coolant is "Novec 7000". Fast forward to 0:30, the rising bubbles provide all the circulation that is necessary.

http://www.youtube.com/watch?v=Z_X_hgtlJpA

[youtube]Z_X_hgtlJpA[/youtube]

 

[Chalo]

That's a cool idea (so to speak).

Phase change can remove way more heat at the source than simple convection.

Chalo

 

[auraslip]

That is awesome! I want to do that for my next computer! Throw in the right lighting, and the case itself would be visually stimulating!

 

[sn0wchyld]

shame about the cost... $250 a gallon.