PDA

View Full Version : Anyone Liquid Cooling Their Controller?



Dale
29 April 2014, 1432
I put my EBR back together over the weekend with an improved motor mount but I didn't achieve the objective of relocating my controller to an area with more direct airflow. The controller is located in the area formerly occupied by the battery, which is under the seat directly under my butt.

Currently teeing up a Thunderstruck liquid cooled heatsink and got to wondering if anyone else was doing liquid cooling and what you were using for the other system components, especially the pump and radiator.

I found a pump on eBay for solar applications (http://bit.ly/1fuLZVU) and thought it might work since it's designed to handle liquid up to the boiling point.

Any comments or suggestions?

Thanks,

Dale

Nuts & Volts
29 April 2014, 1554
Ive used a lot of PC cooling components. Koolance is a good place to start.

Sent from my SCH-I605 using Tapatalk

Athlon
29 April 2014, 1701
in the project I do for my customer I use Mes-Dea pump but for myself I use mostly computer related stuff.

Liquid cooling the controller is a good thing because you also add thermal mass and this is useful for peak power. Water have one of the higes specific heat so every kg of water (4,18) you carry around works better than similar wheight of alluminium (0,89) ( http://en.wikipedia.org/wiki/Heat_capacity) .

Liquid cooling is more complex and have more moving parts , is more expensive , is harder to make and the pump consume energy so is usefull only when you don't have any other chooice.

jonescg
29 April 2014, 1800
The other frustrating thing about liquid cooled inverters is they ask for flow rates of 8-15 l/m, but then they go and put pissy little 3/8 fittings on their coolant loop. It means running much higher pressures for the same flow rate, so choose your pumps wisely. Oh, and all the pumps come with 3/4 fittings :roll:

podolefsky
29 April 2014, 2051
Netgain makes liquid cooling kits that include the pump, radiator, fans. Basically everything except the chill plate for your controller. They make a standard version that should work for most controllers, and a performance version for very high power controllers. =

You can get them from a number of EV parts sellers.

http://www.evolveelectrics.com/Netgain%20Warp%20Drive%20Controller.html

What controller do you have? You might get away with a heatsink and a fan to move air through. Might see if you actually have overheating issues before investing in liquid cooling. My Curtis is under the "tank" with a good size heatsink. It gets air but not all that direct. I don't have any heat issues even at highway speeds.

Dale
30 April 2014, 0635
Thanks all for those references. I stumbled across the PC cooling components, too, and was wondering if those might work.

@podolefsky - I have the Curtis 1238. Since I have no sink at all and it's under the seat, it runs close to the thermal cutoff limit every time out, not pushing hard and still in cool weather. I thought I might prefer to transfer the heat somewhere else. That kit looks like the ticket, thanks!

Hugues
30 April 2014, 0956
Thanks all for those references. I stumbled across the PC cooling components, too, and was wondering if those might work.

@podolefsky - I have the Curtis 1238. Since I have no sink at all and it's under the seat, it runs close to the thermal cutoff limit every time out, not pushing hard and still in cool weather. I thought I might prefer to transfer the heat somewhere else. That kit looks like the ticket, thanks!

had the same problem a few weeks ago, now solved (probably) by increasing my ratio from 4:1 to 6:1, if at all possible for you of course. Now motor temp is below 80C, and when going uphill, bike is 300 kgs

Athlon
30 April 2014, 1044
I have the Curtis 1238......have no sink

It surprise me that your controller is even running witout any heatsink , even a small finned aluminium sink will be a HUGE improvment if now you are running almost ok without any sink.

Before stepping to liquid I think a 20-30mm thick finned sink can completly solve all your problem , yust allow some air to get to the fins and everything will run ok

Athlon
30 April 2014, 1102
something like this one can really improve your performace
http://www.tecnoal.it/k245-profili-serie-k-dissipatori-a-pettine_ric0_130_ita.asp and is only 25 mm thick

if you have a good air flow ( either with fans or dinamic ) you can use sink with more finn like this one
http://www.tecnoal.it/kf240-profili-serie-k-dissipatori-a-pettine_ric0_392_ita.asp

as you cans see with forced air the C/W is just 0,043 so it can handle the Curtis even in the worst case

podolefsky
30 April 2014, 1458
I agree, a heatsink will make a huge difference.

I've used this one from heatsinkusa.com. 11" length is a very good fit on a Curtis 1238. Only about $35, just have drill holes.

http://www.heatsinkusa.com/10-000-wide-extruded-aluminum-heatsink/

Dale
30 April 2014, 2127
So I ordered up one of these and I'll try that first. If that does it, then I'll just return the cooling plate. Thanks for that link - I had run across their site before and then lost it.

podolefsky
30 April 2014, 2155
Hope it works for you.

You probably know this, but just in case make sure to use a heatsink compound. Helps a lot with thermal contact. I like GC Type Z9. It works well and a 1 oz tube is about $6, plenty for the Curtis heatsink. Hardware or electronics stores will have it, if not then Amazon or ebay.

Hugues
01 May 2014, 1100
I agree, a heatsink will make a huge difference.

I've used this one from heatsinkusa.com. 11" length is a very good fit on a Curtis 1238. Only about $35, just have drill holes.

http://www.heatsinkusa.com/10-000-wide-extruded-aluminum-heatsink/

I'm struggling a bit to fit what you are saying with my, somewhat limited, experience so far with my bike (AC-20/96V, Curtis 1238-7601). Let me explain:

- My controller sits in a poorly vented area, on an aluminum plate without fins, and poorly vented as well.
- I measure the controller temperature with the temperature sensor of the Orion BMS, glued to the base of the controller. Also one on the side of my motor.
- I have checked all my logs so far, the max controller temperature i see is always in the low 30C range, once at 39C.
- I'm not sure under which conditions i could bring the controller temperature much higher, as its my motor temperature that is the bottleneck. This one would reach 100C while my controller would barely be at 40C. I presume 40C is far from the max temp of the Curtis controller, although i could not find this in their specs.

So which controller temperature do you actually reach on your controller so that you need a heatsink ? and then what is the corresponding motor temperature ?

Just trying to understand.

podolefsky
01 May 2014, 1120
The controller temperature needs to be measured inside, at the transistors that heat up. The outside plate will be cooler. The Curtis cuts back at 80 C and shuts down at 90 C.

If you have the Curtis display, you can read the internal temperature. It's probably higher than 40C, but maybe not by much. If your controller is on an aluminum plate, that's probably helping even without fins.

Athlon
01 May 2014, 1144
motor and controller temperature MUST be sampled on the inside.

Curtis have an internal temperature sensor near the transistor , a 40 on the outside mean at least 60-65 on the inside.

Also motor temperature must be sampled by the internal sensor and the internal sensor must be connected to the controller so the controller can cut back if the motor overheat.Temperature difference between the outside of a motor and the copper windings can be very high , sometime even more than 100 so a barely warm outside motor at 40 is in reality very near to selfdestruction with windings at 140 .

p.s Curtis will start cut back at 80 internal temperature and stops at 90.

temperature setting for the motor depends on the type of insutation used on the windings , some will fail at low as 100 other can resist up to 180-200 .

Hugues
01 May 2014, 1236
The controller temperature needs to be measured inside, at the transistors that heat up. The outside plate will be cooler. The Curtis cuts back at 80 C and shuts down at 90 C.

If you have the Curtis display, you can read the internal temperature. It's probably higher than 40C, but maybe not by much. If your controller is on an aluminum plate, that's probably helping even without fins.

I agree.

But your heat sink is outside the base plate and that's where i measure the 30C. The heat sink in this case is only seeing a temperature gradient between 30C and the outside temperature. So it's not shedding so much heat.

Have you actually seen high temperature on your controller before adding the heat sink ? And then seen a major improvement after adding it ? That's what i'm trying to understand really.

Hugues
01 May 2014, 1238
...

Also motor temperature must be sampled by the internal sensor and the internal sensor must be connected to the controller so the controller can cut back if the motor overheat.Temperature difference between the outside of a motor and the copper windings can be very high , sometime even more than 100 so a barely warm outside motor at 40 is in reality very near to selfdestruction with windings at 140 .

p.s Curtis will start cut back at 80 internal temperature and stops at 90.
....

Thanks, that's how I'm setup too.

podolefsky
01 May 2014, 1426
I agree.

But your heat sink is outside the base plate and that's where i measure the 30C. The heat sink in this case is only seeing a temperature gradient between 30C and the outside temperature. So it's not shedding so much heat.

Have you actually seen high temperature on your controller before adding the heat sink ? And then seen a major improvement after adding it ? That's what i'm trying to understand really.


The heatsink doesn't need to be that far above ambient temperature to shed heat. It's the gradient from air to the heat source (inside the controller) that is most important.

Yes, I have had a Curtis AC controller mounted to a metal plate overheat, hitting 90C and shutting down. Add heatsink, controller now stays under 65C with the same load.

podolefsky
01 May 2014, 1429
The HPEVS systems have the controller and motor temperature sensors built in. Controller is built into the board, so it's there by default. I'm pretty sure if you don't hook up the motor temperature sensor it will fault. It's also part of the harness that includes the speed sensor, so it's very unlikely you'd be running without the motor temperature cutback in place.

Dale
01 May 2014, 2123
Yeah, I guess that raises a good point - I was taking that reading off of the Spyglass that came with the motor and controller. I presumed that to be an internal temp that it was reporting.

Noah, thanks for the compound reference. I knew I needed some, appreciate the specific recommendation.

d.

Hugues
01 May 2014, 2234
Yeah, I guess that raises a good point - I was taking that reading off of the Spyglass that came with the motor and controller. I presumed that to be an internal temp that it was reporting.

Noah, thanks for the compound reference. I knew I needed some, appreciate the specific recommendation.

d.

i'm not so sure anymore what the spyglass temp is showing, internal controller or internal motor ? I did not find it in the specs

frodus
01 May 2014, 2340
I'd Ask the person who programmed your controller what value they're posting to the spyglass.... I think you said you had it programmed by arc, right? Sure he'd know.

The stock HPEVS firmware shows both IIRC.

podolefsky
02 May 2014, 0555
The older HPEVS firmware doesn't show controller temp, at least not by default. If it says "MTemp" that is motor temperature. The newer versions (5.00 and above I think) can display controller temp. It will read "CTemp". I'm not sure if it shows CTemp by default, or if you have to turn it on.

Instructions are here, go to programming instructions > on road vehicle. Then find your firmware version.

http://hpevs.com/wire-schema-trouble%202.htm

ARC EV Racing
02 May 2014, 1133
I'd Ask the person who programmed your controller what value they're posting to the spyglass.... I think you said you had it programmed by arc, right? Sure he'd know.

The stock HPEVS firmware shows both IIRC.

I'll have a look later but I'm pretty sure it shows motor temp. Hugues if you want controller temp too we can just put it in with the next throttle code mod, it's very easy.

Sent from my AZ210A using Tapatalk

ARC EV Racing
02 May 2014, 1139
Regarding 1238 controller cooling, we make a cooler plate which we developed for our race bike. Before we got it working internal temp was getting towards 60c, afterwords it was more like 30c (ambient +10) depending on the circuit. I thought that controller was hard coded to cut back around 65? Cant remember now but the specs are right at the end of the manual.

Sent from my AZ210A using Tapatalk

podolefsky
02 May 2014, 1156
Linear cutback starts at 85C, shuts down at 95C.

Hugues
02 May 2014, 1218
Thanks Matt. If I had a x'mas wish list, I'd go for a nice CAN port with both temperatures, motor amps, rpm, and all the goodies I could log with Torque. Maybe next winter when you are less busy LOL.

Is it battery amps or motor amps that make the controller temperature go up ?

Athlon
02 May 2014, 1233
Motor Amps is heating both the controller and the motor.

Battery amps just heat the controller partially.

changing the gear ratio lowered the motor amps and you see it how much this changed the situation.

podolefsky
02 May 2014, 1448
I understand why changing the gear ratio would help with motor heating. I'm not so sure about controller. Could be, but are you sure that with the same power the controller will run cooler? I asked HPEVS the same question and they said no.

ARC EV Racing
02 May 2014, 1511
Thanks Matt. If I had a x'mas wish list, I'd go for a nice CAN port with both temperatures, motor amps, rpm, and all the goodies I could log with Torque. Maybe next winter when you are less busy LOL.

Hugues I checked and the spyglass output cycles though RPM (motor RPM), V (keyswitch voltage), Ah (amp-hours used) and BDI (battery state of charge as a %). It's in the release notes but I checked the code and it agrees; no temps are sent to the spyglass. We have CAN data output on our TT race bike so that should be properly developed by the end of the season but at the moment I don't plan to work on OBDII message output. It's actually going to send telemetry back to our servers over a cellular link.

I don't want to hijack this thread but drop me an email if you like.

ARC EV Racing
02 May 2014, 1516
I understand why changing the gear ratio would help with motor heating. I'm not so sure about controller. Could be, but are you sure that with the same power the controller will run cooler? I asked HPEVS the same question and they said no.

I agree, I can't see how the gear ratio would make a difference to controller heating if it's delivering the same power. Unless using more revs = more volts and less current, hence less I^2R loss?

Athlon
02 May 2014, 1550
heating in the controller is proportional to the current flowing across the mosfet (or IGBT).

Even when turned on mosfet resistance is not null , typical mosfet have a votage dopout of about 0,7 volt so heating in Watt is current times 0,7.

Inside the controller I2R losses are high because mofet pins are very thin and R goes up.

When you are not at 100% duty ( the controller is trading voltage for amps) you also add switching loss , when the mosfet is switching and for a very short time is not in the full on or full off states but is working as a resistor heating up very quikly.

Also for the same MECHANICAL power if you are not in the best RPM range motor efficiency is less so you have to use more ELECTRICAL power.

once you go above the rated speed of the motor the controller start to work in the field weakening region increasing the current again to remove some magnetization from the rotor to increase RPM.

the best ratio is the one that allows RPM to be closest to the base speed of the motor but without going in the field weakening region , any other ratio for the same mechanical power will be les efficient so controller will heat up more.