Fan cooling controller

Get all your technical information about electric bikes here.
Post Reply
deswong   10 mW

10 mW
Posts: 29
Joined: Jul 09 2010 4:42am

Fan cooling controller

Post by deswong » Nov 07 2010 9:30pm

Hi All,

I've had no luck in fan cooling my controller because the fan itself doesn't take the roughs and bumps that I travel across. I have gone through two 40mm fans in the last few months and the blades on them keep breaking!

What other ideas do people have for cooling the controller? I have an Infineon 9-FET which at the moment I am waiting replacement FETS, and will be using proper heatsinking to the bar rather than how it currently is through the Kapton tape - I think this would result in some better heat transfer efficiency than how it currently is working. I have some Adhesive heat transfer tape (the stuff they use for CPU's) but insulated stuff, and will be putting that on.

Regards,
Des.

User avatar
neptronix   100 GW

100 GW
Posts: 14196
Joined: Jun 15 2010 5:56pm
Location: California refugee living in Utah, USA
Contact:

Re: Fan cooling controller

Post by neptronix » Nov 08 2010 2:09am

ever thought of upgrading to a bigger controller? I think infineon makes some 12 fet controllers.
"Love and compassion are necessities, not luxuries. Without them, humanity cannot survive." - Dalai Lama

My first major build: 1.6kW 8T MAC motor on a Trek 4500.
The new all-arounder: Leafmotor 1500w @ 4kW on a Turner O2 full suspension.
The monster scooter: 20" eZee on a Cannondale Semi Recumbent.
Whipper-snapper: ? on a lightweight BikeE Semi Recumbent

User avatar
MrKang   100 W

100 W
Posts: 297
Joined: May 01 2009 12:52am
Location: The Netherlands
Contact:

Re: Fan cooling controller

Post by MrKang » Nov 08 2010 5:12am

deswong wrote:Hi All,

I've had no luck in fan cooling my controller because the fan itself doesn't take the roughs and bumps that I travel across. I have gone through two 40mm fans in the last few months and the blades on them keep breaking!
i wonder how you installed your controller on your bike..must be a very rough/hard ride to brake these little 40mm fans...
do you have a picture of yor setup?
MrKang E-Scooter : 24V 750W Bosch Motor modified with YK-43 48V 1kW on 12S6P Lipo - top speed 44kmph
Xinling XTD800A : LYEN 4110 72V (modified for safe 100.8V with extra cooling) - 24S4P/100.8V 25Ah Turnigy + CA with Datalogger + SOUNDSYSTEM + USB power outlet + 78km top speed(120%)
MrKang Retro II : in progress.. ** 12FET MKII LYEN + 28S 20AH LifePO4 cells **
Project link [/color]: http://endless-sphere.com/forums/viewto ... 10&t=13442

User avatar
spinningmagnets   100 GW

100 GW
Posts: 11611
Joined: Dec 21 2007 10:27pm
Location: Ft Riley, NE Kansas

Re: Fan cooling controller

Post by spinningmagnets » Nov 08 2010 10:39am

Perhaps suspend the fan? then direct the air-flow to the controller through a flexible duct of some type?

User avatar
gwhy!   100 kW

100 kW
Posts: 1099
Joined: Jan 13 2009 8:32am
Location: UK, Bristol

Re: Fan cooling controller

Post by gwhy! » Nov 08 2010 11:01am

What sort of currents are you pulling, fets are you using and what the ambit temps, I have a 6 fet controller pulling 70A and it do not get warm let alone hot.

User avatar
Jeremy Harris   10 GW

10 GW
Posts: 4208
Joined: Oct 23 2007 4:15pm
Location: Salisbury, UK

Re: Fan cooling controller

Post by Jeremy Harris » Nov 08 2010 11:10am

gwhy! wrote:What sort of currents are you pulling, fets are you using and what the ambit temps, I have a 6 fet controller pulling 70A and it do not get warm let alone hot.
My 6 FET controllers are similar, they barely get above ambient temperature, even after a fairly long hard climb at high current. Controllers usually only get hot because they are running too much current for the FETs. One of my controllers, that was fitted with the original P60NF06 FETs used to get hot at around 25A, now it's got IRFP3077 FETs it doesn't even get warm at double that current,

Jeremy
Please ask questions on the forum, rather than by PM, as it helps others and you'll get a better range of answers.

User avatar
Ypedal   100 GW

100 GW
Posts: 12524
Joined: Dec 27 2006 12:55pm
Location: Moncton NB, Canada
Contact:

Re: Fan cooling controller

Post by Ypedal » Nov 08 2010 11:40am

fans won't survive sand and muck...

Big aluminum heat sinks are better.

agreed on above comments, what motor and voltage/setup are you using this on ?
ES site status page:
http://www.ypedal.com/ES/ES.htm
----------------
http://www.ypedal.com

User avatar
lifepo4ever   10 kW

10 kW
Posts: 662
Joined: Mar 30 2009 10:02pm
Location: Québec , canada

Re: Fan cooling controller

Post by lifepo4ever » Nov 15 2010 2:30pm

Ypedal wrote:fans won't survive sand and muck...

Big aluminum heat sinks are better.

agreed on above comments, what motor and voltage/setup are you using this on ?


next week i will increase the current of my 18 fets lyen and the first idea was to put fan put because the road is very rougth here the bearing of the fan won't survive even inside my plastic box but i came wtih another idea and use cpu heat sinks but again how can I hold them there because glue is not conductive so i want
to use termal paste ? maybe tie rap or rubber band

sjacome   10 mW

10 mW
Posts: 30
Joined: Oct 23 2009 6:34pm

Re: Fan cooling controller

Post by sjacome » Nov 17 2010 9:00am

Why not squirrel cage pc fan's (used in blade pc servers and oem pc in the pci slot to cool cases)???
Attachments
350_squirrels.png
350_squirrels.png (15.78 KiB) Viewed 1652 times
Nidec_Gamma30.jpg
Nidec_Gamma30.jpg (17.22 KiB) Viewed 1670 times

User avatar
lifepo4ever   10 kW

10 kW
Posts: 662
Joined: Mar 30 2009 10:02pm
Location: Québec , canada

Re: Fan cooling controller

Post by lifepo4ever » Nov 17 2010 11:58am

like y pedal say because of the dirt of the road , also the bearing of the fan will die fast and i really thing if he put the controller face to the wind on is bike this will be enougth , i am lucky my 18 fets is waterproof so i can put everywhere i want on my ebike but i want add some heatsink but how to hold it there?

User avatar
Ricky_nz   10 kW

10 kW
Posts: 846
Joined: Aug 15 2010 6:46am
Location: New Zealand

Re: Fan cooling controller

Post by Ricky_nz » Nov 17 2010 12:10pm

You need something like this heat sink
Image
Exposed so some in moving air, even if it is moving slowly will work very well.
Fans fail, heat sinks can't (they can become clogged so use wider fin spacing like this one.)
Also most after market PC fans are cheep rubbish.
There are better fans available if you are prepared to pay for them
Latest Projects: High power vector controller for Turnigy 80-100.
Youtube channel: http://www.youtube.com/user/rwhitenz?feature=mhum
1) Specialized Hard Rock + surviving 1/2 of Elation 300W + Sick Bike Parts freewheel and sprockets + echo cranks.
2) Keewee Stealth dirt jump with MAC 10T rear gear hub

User avatar
amberwolf   100 GW

100 GW
Posts: 29844
Joined: Aug 17 2009 6:43am
Location: Phoenix, AZ, USA, Earth, Sol, Local Bubble, Orion Arm, Milky Way, Local Group
Contact:

Re: Fan cooling controller

Post by amberwolf » Nov 18 2010 1:01am

deswong wrote:proper heatsinking to the bar rather than how it currently is through the Kapton tape - I think this would result in some better heat transfer efficiency than how it currently is working. I have some Adhesive heat transfer tape (the stuff they use for CPU's) but insulated stuff, and will be putting that on.
The Kapton may well be better at conducting the heat than the adhesive heat transfer tape. You might want to check it's specifications beforehand, or at least do a before/after thermal test to see how much heat is really being moved from the FETs to the outside world.

The most effective simple way I've seen of improving heat transfer in these things is simply to remove most of the highly-overused thermal paste, and leave only enough to fill the scratches and tiny airgaps. If there are big airgaps, then either lap the heatsinks until they don't have them, and/or clamp them together more strongly by using a rigid steel bar on either side of it (not touching anything electrical) with strong bolts thru it, between the FETs, to clamp them very evenly and thoroughly.

deswong   10 mW

10 mW
Posts: 29
Joined: Jul 09 2010 4:42am

Re: Fan cooling controller

Post by deswong » Nov 24 2010 6:27pm

Bleh - Technology and I sometimes have a fight and I've been offline a while.

So this is how I have got it configured at the moment.

Image

I guess I have been a *bit* rough with my bike - just travelling as if I were pedalling but going that bit faster, so I still get caught out in a few spots and can't brake in time to slow down enough :)

(now, going back through the questions and comments...)

The Kapton may actually be better - although I have changed the FETS. The reason why I was asking originally about this (the fans) was because the FETS that I had in there (original) seemed to get quite hot, to the point where the solder has melted a bit and run back up the legs of the FETS - and a few weeks ago the bike got to the point I couldn't even pedal it - and on examination the FETS looked ok - but when tested, the 4 of them were a dead short between all the legs (tested out of circiut). I figure it is because it has overheated in use.

I've changed all the FETS now, and have had a really short test up and down the street as it has been pouring rain, so I haven't given it a real world test yet!

Usually I would use a very thin smear of thermal paste - but these FETS need to be insulated - they let out the smoke if you just clamp them to the bar. (experience has told me so...)

I have actually considered re-casing the controller and putting on a nice, proper finned heatsink - it's probably overkill for this controller though.

I don't do much off-road - I do a lot of bike paths and near-freeway riding (60k/hr) to commute to work and back in less time than what I would do drive. But yes, the bearings would get clogged up thinking about it now.

This is the setup currently:

Image

Oh, again make sure you insulate the FETS - don't just thermal paste it!!! It will damage your controller. I ended up using silicone impregnated rubber, which has a similar thermal conductivity to the Fujik white thermal paste.

The ambient temp does vary, most days I ride it is around 26*c - 30*c so it is fairly warm here.

I did consider suspending the fan - but I think now having a more finned heatsink would be better - and not having the FETS attached to an aluminium bar and then having the bar attached to the case - would be a better modification than to muck around with the fan further.

User avatar
Jeremy Harris   10 GW

10 GW
Posts: 4208
Joined: Oct 23 2007 4:15pm
Location: Salisbury, UK

Re: Fan cooling controller

Post by Jeremy Harris » Nov 25 2010 2:43am

deswong wrote:The Kapton may actually be better - although I have changed the FETS.
The Kapton tape, with a thin smear of heat sink compound, is about as good as you can get for transferring heat from the FET tab to the spreader bar. The FET is the weak point. No matter what you do outside the FET you can't overcome the rather poor thermal conductivity inside the FET, which is the primary cause of failure. Using low loss (i.e. low Rdson) FETs makes a very significant difference to FET internal heating.
deswong wrote:Usually I would use a very thin smear of thermal paste - but these FETS need to be insulated - they let out the smoke if you just clamp them to the bar. (experience has told me so...)
As you've discovered, it's a recipe for certain failure to fit FETs without some form of insulation between both the FET tab and the spreader bar and between the mounting screw and the FET tab. Heat sink compound is not an acceptable means of insulation. Either use Kapton or mica for FET tab to spreader bar insulation and always use the small 'top hat' plastic insulators under the FET screws.
deswong wrote:I have actually considered re-casing the controller and putting on a nice, proper finned heatsink - it's probably overkill for this controller though.
You'd be better off just using low loss FETs in a controller sized to do the job. If the heat isn't being generated you don't need to get rid of it, plus you don't waste so much power, either. Fans will only have a very modest effect and don't overcome the real problem, which is right inside the FET at the junction to case joint. No matter how much you cool the case you can't fix this fundamental restriction that TO220 package FETs have.

Jeremy
Please ask questions on the forum, rather than by PM, as it helps others and you'll get a better range of answers.

User avatar
CamLight   10 kW

10 kW
Posts: 729
Joined: Dec 13 2008 8:51pm
Location: New York City, USA
Contact:

Re: Fan cooling controller

Post by CamLight » Nov 27 2010 1:48pm

deswong wrote:Oh, again make sure you insulate the FETS - don't just thermal paste it!!! It will damage your controller. I ended up using silicone impregnated rubber, which has a similar thermal conductivity to the Fujik white thermal paste.
Very good advice! No insulation = very bad day for your controller. :)
Which silicone insulators are you using? I've never seen one with a thermal resistance below about 0.6-degrees C/W (and typically around 1.0) whereas the typical silicone paste can easily achieve 0.03-degree C/W resistance...about 20 times less resistance. Which makes sense though because the compound is typically only a fraction of a mil thick once the FET is mounted where the rubber insulator is several mils thick and is a thermal insulator as well as a electrical one.

texaspyro   100 kW

100 kW
Posts: 1408
Joined: May 12 2010 6:53pm

Re: Fan cooling controller

Post by texaspyro » Nov 27 2010 2:17pm

There are thermal pads out there with very low thermal resistance. Unfortunately, they cost a zillion dollars per square inch (well, OK around $40). You can actually buy diamond film insulators for less (but they have their own gotchas)...

deswong   10 mW

10 mW
Posts: 29
Joined: Jul 09 2010 4:42am

Re: Fan cooling controller

Post by deswong » Nov 27 2010 7:02pm

These insulators were a bit expensive - $8 for 10 of them. Well, expensive compared to other methods anyway. They are quite thin.

What causes FETS to die? I went on a real-world test yesterday, and I swear one of the FETS have gone again - the controller was warm (not overly hot - you could hold the controller in your hand comfortably) and the wheel feels like it has some insane resistance against it. Also there is no 5v output anymore to the CycleAnalyst so I think this time something bad really happened. :(

User avatar
CamLight   10 kW

10 kW
Posts: 729
Joined: Dec 13 2008 8:51pm
Location: New York City, USA
Contact:

Re: Fan cooling controller

Post by CamLight » Nov 28 2010 11:40pm

deswong wrote:These insulators were a bit expensive - $8 for 10 of them. Well, expensive compared to other methods anyway. They are quite thin.

What causes FETS to die? I went on a real-world test yesterday, and I swear one of the FETS have gone again - the controller was warm (not overly hot - you could hold the controller in your hand comfortably) and the wheel feels like it has some insane resistance against it. Also there is no 5v output anymore to the CycleAnalyst so I think this time something bad really happened. :(
Most of the silicone insulators are only a few mils thick, which is quite thin. :)
But, their performance is miserable compared to just grease. Which you might be seeing if another FET blew.

The outside of the controller, or even the heat sink, can be much, MUCH lower in temperature than the FET junction temperature. It can easily be cooler by more than 100-degrees C.! Especially if using those silicone insulators. In fact, the worse the cooling system works (however simple or complex it is), the cooler the heat sink or controller will feel since the heat isn't being moved from the FETs to the heat sink or controller case.

There are others but typical failure mechanisms for controller FETs are over-temperature and over-voltage.
Last edited by CamLight on Nov 28 2010 11:45pm, edited 2 times in total.

User avatar
CamLight   10 kW

10 kW
Posts: 729
Joined: Dec 13 2008 8:51pm
Location: New York City, USA
Contact:

Re: Fan cooling controller

Post by CamLight » Nov 28 2010 11:43pm

texaspyro wrote:There are thermal pads out there with very low thermal resistance.
Agreed...but the silicone ones are awful. :D
Having said that though, there is a place for them and I have used them often. But, I run the numbers first to make sure it's not going to be a problem. If so, hard-anodizing the heat sink to insulate it is my next choice.

deswong   10 mW

10 mW
Posts: 29
Joined: Jul 09 2010 4:42am

Re: Fan cooling controller

Post by deswong » Nov 29 2010 3:44am

Thanks for the replies, I'm learning a lot quickly. :)

The first time around the FET that blew was the "A group" fets - it was only one on the left if looking at the EB809 with the fets along the top row.

This time around though something weird went on, all THREE fets on the "B Group" are now dead :( I've placed an order for a new 12 FET controller, I think my days of tinkering will stay with this 9-FET and use the 12-FET for my everyday commuting.

hydro-one   10 kW

10 kW
Posts: 866
Joined: Dec 19 2008 2:02pm
Location: winipeg Canada

Re: Fan cooling controller

Post by hydro-one » Nov 29 2010 10:13am

one thing is to make sure the fets are not touching the heatsink. check with a multimeter. Its been said-- but it can be very tricky to get right when replacing multiple fets. use methods guide when repairing/modding controllers.
Pro ebiker
28s hk lipo
3ooA kelly/cromotor 17"
sana cruz bullit

Post Reply