change the heat sink electrical insulator to better one?

[zombiess]

Has anyone tried changing out the mosfet electrical insulator over to something better and measured any difference? I talked to Lyen and he said he didn't know exactly what the stuff was rated for that he is using in the controllers (looks like silpad). I have a whole bunch of TO220 Mica insulators which could work well and I just read and and am going to try switching over to it but won't have a before/after results unless I get another stock controller. I'm currently modding this one like Markcycle (details found here http://endless-sphere.com/forums/viewtopic.php?f=2&t=31044&p=449701#p449701) did by connecting the B+ directly to the tabs on each 3 transistor group and then connecting the output phases together by the tabs. I'll post pictures when I'm closer to finished.

I do have some questions though and that is should I parallel the original connections with the new ones or just use the new ones or the old ones. I'm worried about creating inductance (but shouldn't paralleling the bottom traces and the tabs half the inductance and resistance) if the wire is too long, but figure if I keep it short it should be OK. Need someone who is knows EE and power design to answer this one if possible. I'd appreciate it.

This is the post I read.
http://www.diyaudio.com/forums/pass-labs/37262-mica-goop.html

If I was building the board from scratch I'd probably lap the mosfet backs too, but I'm not going to bother desoldering these and go that extreme unless I don't like the look of them when I get the mounting bar off.

I have read that it's not good to reuse silpad after it's been compressed once because it's thermal resistance increases due to the permanent deformation that occurs.

 

[liveforphysics]

Thin Kapton tape, no adhesive is the best you can do (aside from a thin layer of anodizing). Lap the back of each fet before mounting them on it, and a very slight de-burr on the corners will avoid cutting through the Kapton film. Lap the sink to be flat as well, or you could end up with non-contact pockets that will ruin your conductivity that you could have had with just using the silicone thermal pad.

If you have space, you can also mount each cluster of FETs that are in parallel to a heat spreader, and then separately insulate that heat spreader (which will have greater contact surface to the sink). That's doing way more work for minimal improvement though.

 

[zombiess]

Thin Kapton tape, no adhesive is the best you can do (aside from a thin layer of anodizing). Lap the back of each fet before mounting them on it, and a very slight de-burr on the corners will avoid cutting through the Kapton film. Lap the sink to be flat as well, or you could end up with non-contact pockets that will ruin your conductivity that you could have had with just using the silicone thermal pad.

If you have space, you can also mount each cluster of FETs that are in parallel to a heat spreader, and then separately insulate that heat spreader (which will have greater contact surface to the sink). That's doing way more work for minimal improvement though.

I know about kapton, but as you mentioned, gotta be very careful. How does mica compare? I can probably split the insulators I have in half making them even thinner. It should be an improvement over the stock pad, correct? Grease applied to both sides.

***EDIT***

just read a few posts by Jeremy Harris and other that say go Mica vs the stock stuff. Sweet, finally to something to use all these extra insulators I have for I purchased many years ago.

 

[zombiess]

And here is the update, all is done but the phase outputs.

First side of battery to tabs done and heat insulated with heat shrink to reduce risk of short. Plenty of room in the case.

Botom Side

Getting the alignment right, wire is 10 AWG and I can tell you after soldering the tabs to it, it should also help as additional heat sinking, lots of extra mass directly on the tab. I was surprised how hot some of the section just 6" away were and this is only Alum wire.

Tabs fitted to bare sections and soldered up heavily. Went over them with a small torch to reflow the solder so there were no sharp jaggies to poke through the heat shrink

Completed and ready to install

Spreader bar removed, hard to see here but the board is actually warped a bit from the trace beefing that was done, won't hurt anything, just noticed it.

Disassembled and ready to start

This really shows how mounting to-220 devices provides an uneven clamping force, look at the deformation around the top vs the bottom.

close up of the clamping problem. I should probably figure out a way to clamp the body of the FETs to the spreader bar for better thermal transfer

The pain staking cutting of the mica insulators, first few took a little work but started to go pretty quick. Wouldn't want to do a 36 FET controller, that's for sure.

Spreader bar after I lapped it using 800 grit sand paper. Made it much much smoother than it was. Did both sides. Nice to see the holes are counter sunk which helps with getting more even force with the screws on the FET tabs.

Checking out the mica insulators to make sure they line up to my satisfaction. Some tiny gaps but close is good enough as long as two different tabs from different banks don't touch the spreader bar.

Close up of the Mica as I was playing around trying to line it up, a challenge in itself. Got the gaps even smaller than this after playing Tetris with the insulators.

Ceramic thermal paste applied very thinly between the spreader bar and mica insulator (just beyond the point of being able to see the raw alum) and then a little bit more on top of the mica that mounts to the FETS to account for the variances in the FETs themselves.

Spread bar attached to all 18 FET's. All 3 banks which see B+ are linked together by their tabs, but I have yet to connect the open end to the B+ side yet

Sow now that I share some questions.

1. Should I just put B+ directly to the FET tab connection or should I connect it to both the tabs and the OEM location?
2. When I do the phase wires tomorrow should I connect just the tabs to the phase wires or should I also connect them in parallel with the wires coming out of the OEM locations on the board?

I don't want to create any problems after going through all this work, hoping someone can answer the questions and also give me the answer on why to do it that way to add to my knowledge bank.

Thanks, hope you guys enjoy.

A huge thank you to Markcycle for talking to me and also for posting what he's done. He said with these mods + a few extra caps (I'm not adding more caps, not really any room) he's able to pull quite a bit more amps without having FETs go pop. I was doing 60A before on my 2806 and controller barely got warm, but that could be because of the poor heat transfer from the FET to the case which is one problem I'm trying to solve, I would like if this will allow me to bump it to 80A down the road when my new motor gets here, but that might be too much even to ride at 125V

Observations about the controller. Lots of room for improvement, FET's could use a better clamping setup. The spreader bar is drilled by hand and not machine so nothing lines up perfect vertically. These things are in there really tight! Right up against each other with very little room to breathe. I think a fan could be a big help on these controllers to keep the internal case temps down.

Now that I have this one almost totally modded, I want to buy a stock one so I can log the thermal differences between the two and find out how much if any my mods have made. At least I know it lived at 60A previously on a 6 turn low inductance motor.

I can't wait to get my darn bike back up and running, I miss this more than my race car.

 

[BatteryMooch]

Nicely done!
Yea, mounting TO-220-cased FETs by their tabs is a real pain. The worse part is that the more you tighten the mounting screw, the more the tab warps and the the more the FET lifts off the heat spreader! Add on that commonly used split-rings washers are useless for preventing mounting-screw loosening and those TO-220's drive me crazy (use Belleville washers, stacked, and compressed to 50% of height).

You'd need a new heat spreader, and you might not have room, but the TO-case FET-mounting clips from Aavid, Laird Tech. and others work great. I've used them for TO-220, TO-247 and TO-264 FETs. One great advantage is that the clip mounting screw can be threadlocked in for vibration protection and no worries about overtightening. For additional pressure on the FET, you can "stack" two clips.

http://www.futureelectronics.com/en/technologies/electromechanical/thermal-management/thermal-tapes/Pages/4749347-118300F00000G.aspx (and similar)

 

[zombiess]

Nicely done!
Yea, mounting TO-220-cased FETs by their tabs is a real pain. The worse part is that the more you tighten the mounting screw, the more the tab warps and the the more the FET lifts off the heat spreader! Add on that commonly used split-rings washers are useless for preventing mounting-screw loosening and those TO-220's drive me crazy (use Belleville washers, stacked, and compressed to 50% of height).

You'd need a new heat spreader, and you might not have room, but the TO-case FET-mounting clips from Aavid, Laird Tech. and others work great. I've used them for TO-220, TO-247 and TO-264 FETs. One great advantage is that the clip mounting screw can be threadlocked in for vibration protection and no worries about overtightening. For additional pressure on the FET, you can "stack" two clips.

http://www.futureelectronics.com/en/technologies/electromechanical/thermal-management/thermal-tapes/Pages/4749347-118300F00000G.aspx (and similar)

I've been thinking about building something to squeeze the body of the FETS against the spreader bar to make a sandwich. Since the ceramic bodies aren't conductive I could use aluminum and it would act as even more heat sinking. I'm just trying to think of ways to improve it. Just going to the mica insulators + ceramic compound should be a huge improvement, I didn't go crazy tightening them down, but I am glad to see that the holes on the spreader bar are countersunk which prevents pull up from the threads and holding the whole FET away from the spreader. Jeremy Harris mentioned that after going mica and remounting his controller like this he only had a measured 3 C/W thermal resistance to the FET junction which is pretty darn good.

One idea I had is there seems to be enough room on the top of the spreader bar to drill/tap/mount spring tabs, that could really help hold them down. Wish I could just put a binder clip over each phase and call it good, but then the spreader bar wouldn't mount to the case. Maybe I can cut some in half and drill/screw them to the top of the spreader bar and have a DIY heat sink clip? Sounds easy, but it's more work, not sure what the pay off will be, but heat is the enemy and clamping force is good. Just grabbed a few from work right now.... will do a little checking for fit tonight. A few holes + tap/die + screw + half binder clip could be pretty cheap and a very nice improvement.

The mica insulators was quite the task, but I think I could do it again pretty quick now that I have an idea of what I'm doing. Now I need to squeeze some room for new phase wires.

I was thinking of feeding the B+ in a loop from both ends to keep the voltage equal all the way down the rail, don't know if this matters or not . The stock setup has the last bank pretty far from the B+ feed, maybe just attaching that end of my new tab mount to B+ would be an improvement in helping to keep the caps charged especially at high frequency when the farthest cap is getting hit hard and the other caps might be charging before it. Right now I just have one end ready to connect in parallel with the original B+ feed. Once again, I don't know if this would create any issues I don't know about, would really like some experienced advice here.

I really hope I don't goof anything up and pop this thing, got a lot of hours into it now.

 

[zombiess]

Ran into some problems doing the phases tonight, 1st was room, I ran out, 2nd and biggest was I created shorts to the buss bar. No damage because I use a meter to check everything before it sees power.

Scrapped the tab taps temporarily until I can get some longer insulator inserts, that should solve the issue. I also found I messed up one mica insulator during install so I pulled the whole thing apart and double checked it all over again, replacing the one bad mica insulator. Seems like I have a solid insulator now with almost no gaps to the spreader bar. Couldn't find any by pressing and flexing FETs and bar while checking for continuity to the buss bar from every high/low bank so I'll call it good. Kinda bummed that I don't get to do the tab mounted feeds out yet, but I'd rather have a working controller than a shorted phase due to impatience.

Some sort of non conductive screw would be awesome.

 

[BatteryMooch]

I've been thinking about building something to squeeze the body of the FETS against the spreader bar to make a sandwich. Since the ceramic bodies aren't conductive I could use aluminum and it would act as even more heat sinking. I'm just trying to think of ways to improve it. Just going to the mica insulators + ceramic compound should be a huge improvement, I didn't go crazy tightening them down, but I am glad to see that the holes on the spreader bar are countersunk which prevents pull up from the threads and holding the whole FET away from the spreader. Jeremy Harris mentioned that after going mica and remounting his controller like this he only had a measured 3 C/W thermal resistance to the FET junction which is pretty darn good.

I've used beefy aluminum u-channel clamped against a row of FETs and it worked pretty well. Not as flexible as using individual clips though. The amount of heat it removes though is almost nil (compared to what comes out the "back" of the FETs). It certainly doesn't hurt though.

Another option is to have your heat spreader Type III hardcoat anodized. If you can get it tossed in with someone else's batch, it will only cost a few dollars. That anodizing is insulating up to hundreds of volts, is incredibly tough, and has a verrrrrrry low thermal resistance. I measured about 0.04C/W thermal resistance (for the anodizing) for some heat sinks I use in my electronic loads. Unless you have LOTs of drill bits and taps on hand though, drill and tap before having it hardcoated.

 

[bigmoose]

John, I was thinking of your idea of a U channel to clamp the FET body against the sink, but adding a little flexibility. This idea has a lot of piece parts and hand labor, but likely less than the typical assembly sequence. Shear 0.063 6061-T6 Al sheet the size of the FET plastic body. Epoxy it to the FET. Where the center of the FET heat sink would be on your U channel countersink a small dimple to index a bellville washer. For assembly add a very small dallop of RTV to hold the bellvilles onto the U channel.

What do you think?

 

[BatteryMooch]

Interesting idea!
It would certainly work (you're right though, lots of parts!) but I'm not sure if any flexibility is needed. A big advantage of clamping with just the bar is the high, even pressure across the entire case of the FET and the simplicity of the system.

But, if the surface of the spreader/heat sink wasn't flat, flexibility would be critical to compensate for the variations in the spreader/heat sink surface height! Not sure if FET clips or your channel/sheet/belleville method is better. I like clips (proven, inexpensive, easy to use), but that might be only because I haven't tried your method.

[Edit] Another method to add flexibility is to use just the u-channel but have the mounting screws compress springs that are against the channel, similar to how CPU sinks are often mounted to a motherboard. Still doesn't compensate for a wavy heat sink though.

 

[zombiess]

quick update, will post pics tomorrow.

Threads on controllers are 3mm 0.5mm pitch. Got a 6-32 tap and redid them. Purchased nylon machine head screws, brass washers and lock washers. Reason for brass is it has more copper and should transfer heat better since they are both directly connected to the tab in addition to the hoop which goes to my buss bars. This will make more sense when I edit and post up the pics tomorrow.

Had to deburr the FET edges to stop any shorts from happening through the mica gaps since the mica is really thin (way thinner than that stock silicone material, maybe 1/2 to 1/3rd the compressed thickness of the silicone). Used a dremel with a small sanding wheel on low speed and made sure the direction of rotation didn't create any burrs on the mounting surface, this solved several shorting issues and was really fast to do and only removed a tiny bit of material. Keeping some of the mica pieces straight was a little bit of a challenge but I managed. I now have 4 of the 6 buss bars installed on each bank. I'm then going to solder 10-12 AWG wire to the output for each phase and one for each B+ bank.

I'm into this for way more time than I ever thought I would be, but it's going to be done this weekend. I'm quintuple checking for shorts to the spreader bar due to paranoia. I should have just overnighted some kapton adhesive tape, slapped it on the spreader bar and called it good, but was already pretty far along on the mica. Won't be doing this mod again in this way any time soon, that's for damn sure!

I really hope this mod is worth it by letting the controller run cooler and letting me push 80ish amps continuous from it at 125V when needed. Now that I've done this mod once I know many ways to make it go faster. Still don't have a good idea for clamping yet. Good news is since I'm using a nylon screw my brass washer is almost the exact same diameter top to bottom as the tab which means it will have a much more even clamping force than the normal 3mm hardware would ever come close to so a clip probably isn't necessary. In addition, the nylon screws prevent over tightening, but are plenty strong to squeeze the lock washer shut. I have a strong feeling these mods will be making a large difference in thermal performance. I'll need or order another identical 18FET 4115 controller and measure it's temps at the same current/riding conditions to see how much of a difference. I'll most likely do that just because I'm curious if my many hours of labor are actually worth it. I'm also thinking of tossing 2-4 fans in the case creating a cross flow to help with cooling. At 60A the controller only managed to get to 80F with a 60F ambient temp, but I'm not sure how hot those FETs were getting due to the lack of even clamping force. I was going to just stick thermocouples in and measure it, but then decided to mod first and could always buy another stock controller later. Another thought I had with the fans is to use longer nylon screws and create bigger internal heat sink off the inside of the tabs with brass washers by alternating their sizes to act as fins by building up mass and then going wider once past the FET body. Could probably build a cardboard duct so one fan blew directly over them and an exhaust fan to remove the internal heat. Wish I had more money to test stuff out and see what works well and what doesn't, I always enjoy that.

I will post the pics tomorrow morning, I'm just about to wrap up this totally time consuming mod, guess it's better than being constantly bored. I wanna get some speed seriousfrom this greyborg motor with a 20" bicycle wheel and 21.5" diameter scooter tire all while still looking like a bicycle. Should be riding the 9C motor this weekend.

 

[liveforphysics]

I wish you the best of luck with the Nylon screws. I personally see creep making it impossible to hold the desired clamping loads, but I hope I'm mistaken.

 

[markcycle]

I wish you the best of luck with the Nylon screws. I personally see creep making it impossible to hold the desired clamping loads, but I hope I'm mistaken.

I agree nylon screws are a mistake. If you have to take shrink tubing and put it over the threaded part of the screw that might/can touch the tabs, use two layers.

Also split lock washers are useless bevel washers are a must for top performance. Get them from Mcmaster Carr

Edit:
I just wanted to add you can do everything just right except one thing and lose a FET it's very unforgiving you have to do everything correct or don't bother doing anything.

One thing you are not doing is adding caps and this is one of the most important things you can do

I'm only showing electrolytic Caps added but I would also add polystyrene and or silver mica caps also for transient spike reduction.

 

[liveforphysics]

http://www.precision-ceramics.co.uk/nutsandbolts/index.htm


Make sure the ceramic you choose actually has high electrical resistance. A couple types are unfortunately a bit conductive.

 

[zombiess]

The reason I went with nylon screws is because the insulated bushing shoulder isn't long enough with the extra eyelet in the mix. A TO-220D has a longer shoulder but the only source I found was in Hong Kong and I'm not going to wait that long to get them.

I'm really surprised at the strength of the nylon and torqued one to fatigue which was beyond the force you would want to use even with a metal screw, don't forget that I went form the 3MM x 0.5mm pitch to 6-32 which is larger in diameter. Don't know if they will hold up or not but at this point I'm going to try.

I'd like to find some lowis ESR electrolytic caps to add, not sure if I'll have any at Fry's, but I'll take a look tonight and see if I can add some onto the rails and still fit it in the case. I have a large selection of mylar caps at home, just need to check the voltage ratings and I'll add some of those as well.

I'm against the mica insulator due to the complexity of getting them just perfect, but don't know anyone who sells kapton tape locally after several phone calls. Kinda wish I would have just left it alone now so I could go ride, but it's too late to turn back now. Once you pop the FETs off that silicone pad it's done and I don't know where to get replacement locally.

In the end, if I blow this up, I'll just buy another one, nothing ventured, nothing gained. Should probably just spend the money now to get another one so I have it ready to swap out. I'm tired of dealing with theory now, time for real world testing.

 

[zombiess]

Finally just about done. All I have to do is insulate a few solder connections and hopefully add a little more capacitance and then it's finally done Been over it time and time again for shorts, but will check multiple times again after it's in the case for operation. I think I have some higher voltage caps from an older switching supply I ripped apart, gotta dig through my junk box and see what I have.

This job was much much harder than I had originally anticipated. Many hours of work so I hope I'm not let down.

I spoke to two different people who used nylon hardware in mounting hot running to-220 case transistors and other hot applications and they both said that for what I'm using them for there should be no issue at all. For the guy who mentioned the split washer, I use them to help judge how tight I'm torquing a bolt. Flatten then add 1/2 a turn and call it good. Helps keep me from getting over zealous with the tighter is better thinking.

I already ordered a spare board in case I pop this one. If this work I'm very curious to see how high I can go amp wise. I'll most likely call it quits at 80A which means a very brief 100A burst. For more current I"ll go 24 or 36 FET or just get a Kelley that can do 200V

 

[zombiess]

here are the last of the pictures of the mods I've done now.

Connection from buss bar to phase wire, not the prettiest, but it's solid, took a few tries because the heat sink works so well causing cold solder joints. Had to get the iron really hot using a torch to get it to stick well. I normally like nice smooth solder connection but was frustrated and settled at this point since they were solid when I pried on them with a small screw driver which is how I found the first solder job was crap.

Overall shot of the board, starting to apply some gorilla tape to the connections to protect them from shorting and any possible jaggies from poking through. Think I got them all, but can't be too safe.

Everything all taped up and zip tied down to fit back into the case. Whats not pictures is that I put a piece of gorilla tape inside the case right above the buss bars just in case something makes it through the two layers of heat shrink and the tape on the buss bars. If it does, it then needs to go through another layer of gorilla tape + the clear anodizing on the case. At least I think it's anodized because it doesn't conduct unless you scratch it.

I took the time to install two K-type therm couples, one to monitor FET body temp and one to measure inside ambient case temp.

This is the ambient case temp thermocouple it's sorta near the shunts which are near the middle of the case. Figured this spot would have the hottest ambient temps.

Sandwiched a thermocouple between the bodies of two FETs in the same bank in the middle of the case where I thought it would have one of the highest thermal loads due to be surrounded by the other FETs. The mica is full size on the middle FET so it can't touch the buss bar, I have the mica trimmed behind the other two FETs and butted up against each other as good as possible. Ceramic thermal paste on either side of the mica and then again between the spreader bar and the enclosure.

I noticed that the spreader bar doesn't mount very flush against the case. A simple mod to improve this would be to drill and tap some small screw holes on the top of the spreader bar so it makes more solid contact with the enclosure to aid in heat transfer.

Unfortunately I did not find the space to add addition capacitance to this setup so I'm going to run it as is. There is a total of over 1800uF on the board rails as it is now, just really hard to fit more in this tiny case.

Fired it up for the first time tonight. 100% throttle was 74 mph unloaded with a 2.0A draw. Ambient room temp was 26C, FET body was 28C, ambient enclosure temp got up to 34C. Should be able to have the bike ride able tomorrow to see how it performs. One difference I did notice was that the motor sounded smoother than it ever has. Could just be my imagination though or a harmonic from the new swing arm, but it sounded quite a bit different and smoother at really low RPM.

I tied the B+ FETs to the original B+ location because I'm not sure if feeding the FET's from the other end and creating a possible differential would cause issues, especially since I didn't get to add any extra capacitance.

I don't think any of the mods I've done will make the controller any worse, but I'll keep an eye on things because my goal was to improve the design. They should promote more equal current sharing between each bank of 3 FET's if nothing else which means more reliable and higher current handling. I'm quite surprised at how those little buss bars soaked up the heat too, really noticed it when soldering. Being so insulated probably isn't that great, but it does prevent shorts and I might install a blower/puller fan setup if I see ambient temps inside getting pretty warm. A cool controller is a happy longer living controller. Making it through a Las Vegas summer under hard riding conditions will be the ultimate test when it's 45C+ outside.

If I have a good Xmas I'll buy another identical 18 FET 4115 from Lyen and install thermocouples in the same place to see what the difference is in temps at the same settings on each controller. I'd buy one now just to have as a spare, but I've been spending way to much this last month on this bike project and it's still not over but it's close Another $400 should end it, then I hope to put 100 miles a month on this thing.

I need to get my hands on a 36 FET controller which has lots of room, convert to 18 T0-247's and change the way the buss bars mount. Anyone know of a decent low RDSon 200V TO247 FET?

 

[BatteryMooch]

I wish you the best of luck with the Nylon screws. I personally see creep making it impossible to hold the desired clamping loads, but I hope I'm mistaken.

I'm with you on this one.
Even when cold, nylon can creep (stretch). And as it gets hot, it will stretch even more over time. And with TO-220 FETs mounted directly, i.e., not via spring clamps, even a little bit of loosening can cause a huge heat increase in the FET.

zombiess, I'd check the tightness every once in a while. just to make sure all is OK.

 

[full-throttle]

Nice work on getting the most out of the controller!

It may sound like splitting hairs, but have you considered Loctite315 as an alternative to mica?
Its thermal conductivity is marginally better, but the self-shimming thickness is 4x smaller and there's no need for thermal paste.

Just a thought..

 

[zombiess]

Nice work on getting the most out of the controller!

It may sound like splitting hairs, but have you considered Loctite315 as an alternative to mica?
Its thermal conductivity is marginally better, but the self-shimming thickness is 4x smaller and there's no need for thermal paste.

Just a thought..

never heard of it, would have to read up on it and how to use it. Cutting the mica was a PITA but worth it. Controller still lives even after putting 100A of battery current through it yesterday and today.

 

[The Mighty Volt]

@Zombie: thanks for the information and the PM. This looks very sophisticated, to be sure. I am running 4110's so maybe it won't be necessary.

 

[zombiess]

@Zombie: thanks for the information and the PM. This looks very sophisticated, to be sure. I am running 4110's so maybe it won't be necessary.

If you take the time and want the maximum end result, the mod is worth it, but if what you have already works fine, you don't need to go to these extremes. This was a project born out of curiosity and boredom. I'm no longer curious, it works unbelievably well to get very high current out of a 4115 controller and I'm no longer bored because I'm lacking free time lately because I have so many projects that this mod got me thinking about

One of my biggest concerns is how hot I see the internal temps on the controller rise now. When I'm pushing it I get internal ambient temps in the 60C range which is pretty warm. It's not even summer here yet when it will be 40C+ when I ride. I need to think about adding ventilation to the controller to remove the trapped heat. The FETs cool off quite quickly, but the internal ambient temp takes a lot longer to come down. Ventilation will probably improve both.

 

[gensem]

A single small fan ll drop the controller temperature alot...

 

[crusoe]

I had problems with shorts on the fets too. Those stupid white washers are awful! We ended up going the shrinkwrap route as mentioned in one of the previous posts instead of using the white washers. Held up fine on a 10km ride tonight!

Hey Gensem, did you actually mount a fan to your controller? Or is that just a picture for sizing purposes? My controller is encased in frame and gets pretty hot. I don't want to move the controller, so I'm thinking the fan route like you mentioned. I'm just not sure it will fill in an 18fet 4115 controller. Here's a pic of my setup http://imgur.com/a/D7Bcs#0

 

[gensem]

Not yet... but I think I ll in my new bike.
Things are going much slower but I should be able to come with something fast and reliable in a couple months.