CroBorg Super Commuter

Twelve FET Controller Temperature

And the answer is... outside controller case near FETs 61 degrees C peak, and that is taking it somewhat easy speed-wise and pedaling some on the uphills. 13 miles, 1200 feet climbing. Peaks of about 45 amps at 68 or so volts, right around 3KW. The 12 FET Infineon controller is on stock settings, probably 40 amps. No cut-outs noted this day, so it has been hotter at times in the past. Ambient about 20C. The controller is in free air inside the Borg covers, probably not much airflow.
 
Alan B said:
The controller was probably quite warm, I didn't know that at the time but have discovered that since, and this was near the hottest point of the ride. The joint was tack-soldered and appeared crystallized and cold. So your hypothesis has some merit, though I don't think it was near solder-melting temperatures.
Click to expand...

Hi Alan,

Great build, watching with interest as you refine your ride. Regarding controller heat, even at 90% efficiency a 6,000 w controller will shed 600 w of heat. Inside a confined space like the battery box of a Greyborg, that quickly leads to some high temps, unless you have some means of shedding the heat.

I know that you want to retain the protection, water-proofing, and aesthetics of an enclosed electronics bay. So, have you considered a thermoelectric (Peltier) cooler, like this example?

10080-04.jpg

You'd need to mount the cold side of the cooler against the controller inside the battery box, with the hot side of the cooler exposed to the outside. Perhaps a small water-sealed window through the box? Since the exposed cooler will be hot when it's shedding heat from the controller, you'll need a guard to cover it to prevent burning your leg. An air scoop to direct air flow over some fins might work well, and hide all the tech from casual eyes.

Of course (for the the electronics hobbyist), there are open source Ardunio projects to run the cooler. You may even be decide to harvest a few watts of electricity from the heat differential to power the scheme. :mrgreen:

Good luck with you mods, and enjoy your Cool Ride! 8)

Cheers,
Holocene
 
Thanks for the comments Holocene, and welcome to my thread.

I don't think the 24 FET controller ran very warm at all (I didn't have a temperature gauge on it, but I did check it at the top of the ride a time or two), but some cooling is advisable for sure. Peltier devices are nice for moving the heat 1/8" (or whatever the thickness of the device is), but they do it at a cost - quite a bit of power consumption. If somewhat lower temperatures are needed they can be quite useful, though their temperature drop is fairly limited. In our case what we probably need is some airflow (and a more efficient/capable controller), preferably some that is rain-resistant. Not that my bike is ridden in the rain, but in theory...

I should get some air moving in the vicinity of the controller, one way or another. I should probably try the 18 FET to see if it is any better while we wait for the 24 FET repair.
 
2014-06-17%252008.31.02s.jpg

Controller Temperature at the end of the inbound commute, it has dropped from 60C during the last mile of downhill at the end of this trip.

I've done two commutes with the temperature readout on the 12 FET Infineon controller case near the FETs. On the inbound trip (13 miles), with more climbing (1200 feet), and more gradual climbing, the controller reaches right around 60C in 20C ambient. On the return trips, with a very steep first mile (15% and most of the 600 total feet of climbing), and more gradual downhill, uphill and fairly level miles (still 13) the controller reaches about 50C in 30C ambient.

Running partial throttle may raise controller temperature slightly, but not much. It is possibly a tradeoff between increased airflow at higher speed and the increased dissipation of higher current played against the PWM dissipation, though the PWM is present even at full throttle most of the time so it is hard to get away from.

The new Cycle Analyst V3 is reading current and power, now I need to get the feedback working and pass the throttle through the CA. I did change the minimum voltage setting on the CA to a higher value to reduce the chance that it will get scrambled (a problem that some have noted). If it is part of the controls it needs to be very reliable. I'll have to see if I have all the connectors and wire for this upgrade.
 
This morning, before heading out to work, I rewired the Borg. The Magura throttle adapter pc board was removed, a plug with a 470 ohm resistor on the +5 side to the pot installed (so an open ground can be sensed and shutdown the power) and the throttle routed to the Cycle Analyst V3. On the other end, an adapter was prepared that intercepted the throttle output from the CA-DP plug and routed it to the Controller's throttle input. This is necessary for motor controllers that have the V2 type diode-coupled override type input (newer controllers have a different arrangement and can accept throttle from the CA-DP cable directly).

I programmed the CAv3 for the input throttle voltage range (0 to 4.5 volts) and the throttle overvoltage (4.7). The output voltage range defaults were about right, so they were not changed and a quick test showed that throttle-through mode worked, and we had basically the same PWM type throttle as before. Then I enabled the current control throttle (at 40 amps) with speed (at twenty-something) limit. I used the default feedback parameters to start with.

I did a couple of wheel-off-the-ground tests, and then, wearing safety gear, did some street testing. All seemed well, so I suited up and headed to work, about an hour late. :)

All I can say is that I should have done this a long time ago. It has been on my list, but early on I didn't want to try beta code on the 24 FET controller. Now I'm on the 12 FET which is a better power level to test at, and the CAv3 code is very stable, the 3.0 prelim 6 release has been out and tested for a long time. So both of my excuses are voided. With good stable code even the 24 FET will not be a problem.

So now I have a very gentle torque-type throttle (battery current is not quite torque, but it is similar in effect, just some scaling differences), and have an easily adjustable speed limit without setting up the laptop to mess with the controller. It is much easier on the wrist (not extra-sensitive to small changes). It is also easier on the motor and dropouts to avoid those surges of acceleration that are fun but hard on equipment. Of course you can reprogram for through-throttle in a moment, and get back to what you had before, so there are lots of options and it puts you in control of how the system responds.

I can feel the throttle feedback loops occasionally, such as when letting off the throttle sensing a very slight delay before power drops or when it hits the speed limit and overshoots slightly, but it is not a problem. Tuning the loop parameters can improve this, but it is not even necessary. They have chosen good default values.

Riding the CroBorg now feels a bit more like the Zero Motorcycle I test drove. Gentle throttle with mild acceleration at low throttle settings, but hidden behind that is a lot of power when the throttle is opened wide. Nice and smooth. Controllable. Can't wait to try it on the 24 FET controller!

If you haven't tried a Cycle Analyst version 3, consider it. It is a nice upgrade. It has a lot more features I'm not even using yet! :)

Good job Justin!!
 
Today I left for work a bit late, so the ambient temperature was warmer. On the long climb portion of my ride the 12 FET controller rose above 62C so I pulled over and opened the Borg covers under the seat (where the controller is). I slid the big tie wrap that was holding the covers closed in between the covers and the frame, forcing a half inch crack along the rear portion of the covers. After the controller cooled ten degrees I resumed the trip, and immediately upon moving the airflow dropped. Even though the controller was not directly in the airflow it cooled considerably and quickly. This confirms that even minor improvements in cooling airflow make a big temperature difference in this setup.

My present thought is to open the covers slightly at the front (cut back the nose portion) to allow air to be forced in, and cut some exit half-holes on the rear of the cases along the frame.

I tested the same cooling setup on the return trip, and the temperatures were reduced by 12-15 Centigrade degrees from the usual 50C peak.
 
One commute run results on the cooling improvement (detailed above) shows a 5-10 degree C improvement (on the controller case) (peaking at 52C instead of 62C). Definitely important for my commute. The improvement doesn't seem to help much until temperatures start to get hot, but once the controller gets above 40C it helps, and more the higher it goes, and you can really see temperatures jump up 2-3C when you stop and the airflow is lost.

When I get some time I'm going to improve the outlet holes into a slot that will roughly double the exit cross section, and make the two covers more equal since I didn't eyeball the rear semicircles very symmetrically. :)

I have my doubts that the cuts in the front are helping much, they may be but looking into them the cross section is packed with batteries and wires so the flow is not very good there. The air can also come in near the crank on one side, and things there are pretty open since the batteries and wires don't block most of the opening.
 
have you tried installing fans? i don't know where they would fit, but the best solution (cooling wise, not optically) would be to cut a 8cm hole just above the controller and let a fan push air inside. even with low rpm (not audible for you) it would dramatically reduce the controller's temperature.
the other possibility would be to install the fans directly at the cut opening, but i don't know if this is doable and if those fans would produce some noticeable air flow.
 
Alan B said:
I have certainly thought about it. Not sure it is needed with the 24 FET controller, and there is less space then. With the 12 FET it would be useful. The difficulty is I don't have a good way to power a fan at the moment. Unless there are fans that work well from 70VDC.
Click to expand...
how do you drive your light? a 12c ac/dc power supply will definately work on 70v. ALL i tried did. and you only need <1a to drive a fan.
you may even think of just making a 6vm hole on top of the controller, and then many small holes on both side plates (the ones which are screwed on) and just push some air through the inside of the case. even the non moving air under the big side cases can take away a lot of heat it MOVED.
 
I have a couple of 12V adapters but haven't tested them yet. 63V at pack min is pretty low for a converter designed for 170VDC, I would expect them to work okay at low current but not at full ratings.

My lights are high voltage or have a separate battery at the moment. Some kind of 12V converter is in the plans.

However at the moment the natural cooling seems to be adequate. It will be interesting to see what temperatures the 24 FET rises to, I've never found it to get warm, but I've only checked a few times. Now with a readout I'll have more data.

Thanks for the comments,
 
The 24 FET controller was repaired and reinstalled. It made the inbound commute as always, but on the way home it popped while starting on the 15% grade after I pushed through the bicycle exit gate at work. It is a steep start. It feels like one phase shorted, I had to unplug the motor to ride back to my office.

Back to the 12 FET for awhile, it is getting hot (51C) but not cutting out with the cooling holes.

Yesterday I decided to try the 18 FET 150V controller that I bought last year for the voltage upgrade that never happened. I installed it after work and made the difficult homeward commute with it. I took it easy, but it worked fine. Good acceleration and regen, too.

I have had about 6K miles from the 24 FET, but it has failed 5 times total. Something about the low resistance of the Cromotor perhaps.
 
Alan, you really need to give up on these rubbish controllers...

I've had experience with popping Lyen 18 fets multiple times, ended up replacing with a very reliable, but still not very nice to use Crystalyte 18fet from a Stealth bomber, but the final solution - Adaptto mini-e was a night and day difference. Smooth, controllable throttle that's substantially nicer than even what the CAv3 can put out, superior display and fully programmable on the bike with no hassles. Better economy, higher top speed, silent acceleration and badass onboard charging with BMS. Not to mention it's tiny! The Cromotor is much better suited to the larger Max-e, it would certainly be under-stressed at your current power levels (rated for 12kw).

The other excellent option now available is the Sabvoton from Zombiess, it's got many of the features of the Adaptto (but definitely not all) for a relatively affordable price.

I cannot stress the difference enough, it's like going from a horse and cart to a magic carpet!
 
I've been watching both the Adaptto and the Sabvoton controllers for awhile now. Both of these are definitely interesting, and it is good to see some choices. I like Lyen's excellent support, I have at least 5 of his controllers and only this one has had problems - the others have all been problem free. I'm tempted to try the Sabvoton since zombiess is supporting them and his support has also been excellent. The Adaptto is difficult to procure and difficult to get support for and expensive so it is a bit lower on my list. Feature wise it is definitely great though.

Today I reduced the the CA V3 AGain by 30% due to some surging in the battery current throttle, and it helped a lot. This was the second commute run up the steep hill, the one that killed the 24 FET controller twice. I didn't give it full throttle all the way yet, but I did work up to full throttle about halfway up the hill and it was fine. The controller heated up to about 43C later in the ride, which is not bad considering the minimal airflow it is receiving.
 
Last weekend I reconfigured the Borg and set up a different controller for testing. While my 24 FET is getting repaired, I'm trying out a Sabvoton sine wave SVM72150 from zombiess.

I could not get this larger unit to fit in the controller space on the Greyborg Warp frame. It is in between the 18 FET and the 24 FET controllers in length and cross-section, but between the combination just didn't want to go in there.

So I moved the lower 1/3 of my battery pack up into the controller space and hard mounted the Sabvoton into the lower battery area using some aluminum angle extrusions. I put the plug and terminal end down so the high voltage terminals are under the lower cover. The five screw terminals are excellent - good and solid connections, though they do need a plastic cover for safety.

At the same time I installed a fuse and external shunt which had been on my list for some time, but is recommended for the Sabvoton. Having an external shunt is nice in that changing out controllers doesn't require recalibration of the Cycle Analyst. The wiring I used is 8 gauge, which is a good match for the dual 12 gauge wires I have on my batteries.

I had one problem with the shunt scaling of the Cycle Analyst, it was reading about half of what it should. Turns out the low value of this 200 amp 75 millivolt shunt takes us out of the CA "Lo" range, and it must be configured for the "Hi" range before it will accept the actual shunt value properly. Now it reads accurately, and in X.Y kW rather than watts, which is actually an improvement display wise.

The computer software has many parameters that can be set and read out. It allows reading the settings that are already in the controller, which is one nice feature. The hall detection is fun, you don't have to worry about the wiring order, just plug all the wires in and it run this mode in the software to detect angle and phasing of the halls. The motor slowly twitches around and it measures the firing angle and order.

I tested field weakening with no load (wheel off the ground), and indicated speed increased from 37 to 49 with 50 amps of field weakening. On the ground I see a peak speed of about 33 with no field weakening, I haven't tried running with it on.

I tried a few settings of battery and phase current, and selected a very modest 80/100A setting to start with. I also set slip regen to 30 amps and turned field weakening off. It has plenty of power, I'm seeing 6kw peaks (18S 72V 20AH). It feels like less power than the 24 FET Lyen, but that is due to the smooth control. I had the 24 FET set for 60A battery so it was actually a bit less peak power, but the low speed current multiplication of the 24 FET is very high causing abrupt acceleration. This is probably what causes the failures, actually, but it does feel impressive.

The Sabvoton feels "in control", with really smooth torque control that keeps rising as the throttle is twisted. Speed slides right up like an electric subway train, with constant acceleration but without the sudden thrusts of the Xie Chang type controllers. I hear that zombiess runs 150/300A with this controller and the Cromotor, but I have not pushed it that far (he reported the wiring does get hot after a few stop/acceleration cycles). The 11 Gauge phase wiring and the motor heating itself will be the limit with this setup. After a commute run to work (with 13 miles and 1200 feet climbing) it soaks out to about 50C. Warm but not hot.

Slip regen is nice, though 30A is more than I'd like for flat ground, and less than I'd like for my 15% grades. This Sabvoton has special software to support a variable ebrake input 0-5V, so I need to make a variable voltage output brake lever. That will be really excellent, and then I'll reduce the slip regen to maybe 10A for a gentle decel on throttle release. Or turn it off, we'll see what feels best.

Due to weather I've only made one full commute and a few local test runs on this new Sabvoton controller, but I'm impressed. It doesn't have all the features of the Adaptto, but at half the cost it provides a major improvement in smooth control over the standard Xie Chang design. I'm still running the throttle through the CAv3, but I've turned all the controls off and increased the ramping rates because the Sabvoton is already smoothing the throttle, and with the CAv3's default rates it was a bit laggy. It is fine now, and I can use the CAv3's limit features for other things like top speed or total power if needed.

Overall energy consumption on commutes is lower by about 5-10% on this one data set, probably due to smoother acceleration and less power wasted oversaturating the motor at low speeds. This gentler acceleration will also lead to better rear tire life (my second moped tire is nearing the wear bars already!).
 
Thanks.

On Regen and Battery Capacity

Right now my pack is capable of 2C charging (Turnigy rated) and is 20AH so that indicates a max regen of 40 amps. I would like to get to 50 amps or perhaps more, so that is yet another reason to increase battery capacity.

If I do some design and fabrication I should be able to get 25AH on the bike. Beyond that is getting difficult.

I wonder how closely we need to stick to the charge limits of these packs for regen use? It is pretty short term, so perhaps we can push a bit past 2C?
 
You must log in or register to reply here.

Similar threads

N
recomendations for wheel tire combo?
Replies
1
Views
930
amberwolf
amberwolf
EVGator
EEB Mid Drive Bike - Complete or Part out
Replies
3
Views
1,910
skeetab5780
skeetab5780
D
Super 73ish Retro Wheelie Bike
Replies
0
Views
1,164
Dogboy1200
D
justin_le
Upcoming Major Changes and Updates to the ES Site
Replies
11
Views
2,422
Two4Seven SuprDan
Two4Seven SuprDan
E-HP
Which electric bike should I buy?
Replies
3
Views
1,195
MarkJohnston
M
Back
Top