CroBorg Commuter

[liveforphysics]

Aww, my condolences for your controller. It happens.

Hopefully you lucked out an it's just a single bank to replace.

[Alan B]

Seems a bit odd with stock parameters and unmodified, but we'll have to see what's up. I wonder if block time is involved. It was one of those situations where block time might have numbed the normal safeties.

[ZOMGVTEK]

It's nice to carry along a little 6 FET for when the main controller inevitably blows. Surprisingly enough, I never had to use mine, but I beat the crap out of it to test it out and it survived. It's a cheap 48V 6 FET I got off eBay with 4110's and 100V caps. Left the current limit at 20-22A. 2kW is usually enough to get the bike home.

Sure beats pushing a 100lb bike 20 miles...

[Alan B]

I may do that. Was thinking of an 18 fet spare, but that would be hard to carry.

[ZOMGVTEK]

Even the 6 FET is a little big to carry around. A spare 18 FET is a bit overkill. I left the LVC stock, so it works off 12S in the event of some sort of partial battery failure. The only thing that was a bit of an oversight is hall sensor failure. I don't think my controller is capable of running sensorless, which would be a nice touch.

Just buy some super cheap POS controller, mod it for 24S, and tape it somewhere. It's really a good idea for a heavy bike with sufficient capacity to go a substantial distance from home.

[Alan B]

Spare controller ordered.

Methods HVC/LVC boards came in. Very pretty!

[zombiess]

I'm betting the FETs in that controller were not matched. Paralleled FETs that aren't matched for their miller plateau can have very different turn on and off times causing current hogging. That's why I build my own controllers no matter how much I hate doing it. I wouldn't be surprised if my 18 FET 4115 could handle some short bursts of 130A at 125V if it had matched FETs. I'm pretty sure the 36 FET controller I just sold could do some 250A bursts into a small wheel at 125V, but I never got over 120A with it which couldn't even get it warm. I'm trying to build up another 36 FET now and also have a 24 FET I'll probably build with 4110 FETs at some point.

Good luck fixing the controller Allen. They can be a real PITA to work on. I'm desoldering a 36 FET board right now and it's quite a time consuming process.

Since you are taking it apart, you might want to look into my 36FET controller build. I don't know what insulator is used on your controller, it could be kapton with thermal grease or it could be a mystery insulator. I personally like using Wakefield 173-7-220P insulators. I'm also sold on using buss bars to connect the tabs as you've probably seen me post more times than you can count. The devil is in the details with making these things live.

[Alan B]

Thanks for your comments, Jeremy.

I put only about 400 watt hours through it, perhaps 15 miles. It was at default parameters, about 3kw peak if the shunt is 1 milliohm that I set the CA for. Regen on the downhills was up to about 550 watts.

This is the case I shortened to fit, though I kept the board in another room during machining and cleaned the aluminum bits out before reassembly.

Neither the motor nor the controller were warm at all. I had travelled a mile or two around the worksite. I had to push it uphill back to my office. At least it was only a quarter mile or so.

I ordered a backup controller. 12 FET since most of my controllers are 12 FET.

[Alan B]

Update on the CA Beta

I have the CA beta on the Borg. Initially when we hooked things up it was interfering with the throttle (which is wired direct to controller in the usual manner, not through the CA), so we unplugged it for the run on Sunday. Yesterday I plugged it in and went through all the setup, but it still slowed the motor. So I removed the throttle wire from the connector to allow the motor to get to full speed. So the CA was working yesterday. I have not checked the calibration, but it was set for 1 milliohm shunt and showed about 3kw peak which is about right for default settings as far as I know.

This is the first beta code version still in the CA, which is way out of date now.

[The Mighty Volt]

Bad News

Went out for an easy noontime ride around the worksite. Nothing got warm, but on a gentle takeoff from stop the 24 4110 FET controller with stock settings, 62V, about 3000W peak -- shorted.

The motor cogs badly even with no power. Had to unplug the motor from the controller to ride it, and push it back uphill. Seems like shorted FET(s).

Now I'm sweaty.

Why would that happen, just like that???

[Alan B]

Don't know yet why. Could speculate but that will just cloud the story. Everything is new, so part of the shakeout.

[Alan B]

Failed Controller Measurements

Measuring with a meter set to the "diode" mode, looking back into the phase outputs:

When Blue is negative, see 0.44 volts (this is probably the forward drop of the FET diodes).

When any other combinations, see charging, the voltage value rises as the bus capacitors are charged by the meter through the diodes.

This, and the layout of the FETs tells me that one or more of the high side BLUE FET(s) are shorted, and the rest are OK!

Looking inside, everything looks fine. No obvious sign of problems, though it does smell a bit odd.

[zombiess]

The other possibility is sometimes you get a bum fet. It happened to me on a controller lyen sent me. Opened it up and everything looked good but one phase had a short. When the fets were removed only one was dead, but it physically looked perfect. Controller lasted less than 1 mile at stock settings. Ended up having Lyen swap out the entire bank with spares i brought the night before the first grange race i went to. No problems since then and i crank 65 to 105 amps through that same controller now at 125v. Obviosly its my modded 18 fet, but they are not matched. Good practice would be to replace all the fets in that one bank with ones from the same batch. They are more likely to be close in specs to each other since they are from the same lot.

[The Mighty Volt]

I'm betting the FETs in that controller were not matched. Paralleled FETs that aren't matched for their miller plateau can have very different turn on and off times causing current hogging. That's why I build my own controllers no matter how much I hate doing it. I wouldn't be surprised if my 18 FET 4115 could handle some short bursts of 130A at 125V if it had matched FETs. I'm pretty sure the 36 FET controller I just sold could do some 250A bursts into a small wheel at 125V, but I never got over 120A with it which couldn't even get it warm. I'm trying to build up another 36 FET now and also have a 24 FET I'll probably build with 4110 FETs at some point.

Good luck fixing the controller Allen. They can be a real PITA to work on. I'm desoldering a 36 FET board right now and it's quite a time consuming process.

Since you are taking it apart, you might want to look into my 36FET controller build. I don't know what insulator is used on your controller, it could be kapton with thermal grease or it could be a mystery insulator. I personally like using Wakefield 173-7-220P insulators. I'm also sold on using buss bars to connect the tabs as you've probably seen me post more times than you can count. The devil is in the details with making these things live.

Christ. So we need to capacity-match batteries, match the fets.....jaysus what else do we need to match??

[Alan B]

Matching the FETs is a job for the controller manufacturer. Selecting them from the same manufacturing batch should be pretty close. You find out how well they did when the controller survives a few miles. For whatever reason, this one did not. Maybe we will learn more during the repair, or by how the backup 12FET works out, or maybe not.

Looks like the backup 12FET controller is shipping today. Who knows, might have it tomorrow or the day after. Then we'll do some more testing...

I think I will make up a "CA Throttle disconnect" cable. A little short male-female extension for the CA that does not include the throttle control, or has a switch in the line so it can be disabled. I don't want to have to pull the throttle line from both controllers, and it might be awhile before I can update the firmware in the beta CA.

[Alan B]

Fun with Temporary Battery Packs

or, Fun with Bricks

I have a few 8 amp hour Zippy packs. Two 6S and three 5S. At the moment I have a temporary pack made from the three 5S to give 15S. I was thinking I might get another 5S or 6S pack so I could make 18S or 20S. But I could, without buying anything, do 5+6+5+6 for 22S. I am not crazy about 24S due to the lack of voltage margin, but 22S might be a nice spot to try! Also 6+5+6 for 17S is not bad.

[zombiess]

Matching the FETs is a job for the controller manufacturer. Selecting them from the same manufacturing batch should be pretty close.

Should be but I've seen quite a variance that is surprisingly large and others that are all really close.

[Alan B]

Package arrived with 12FET backup controller. Really fast shipping from Lyen! I put connectors on to match the CroBorg.

I also made a Cycle Analyst extension cable that doesn't have the throttle control wire in it. So just by manipulating plugs I can disable that feature since it is not working right yet in my CA. I used the small crimper from Sparkfun. It worked GREAT on the JST connectors used for the CA. It really helps to have the right tool. Makes those connectors a pleasure to work with.

So tomorrow we'll see if the CroBorg works again at 15S... Maybe this weekend I can give a go at 22S.

I sent the folks at Yasusu an email asking them which end should go where. They replied with an apology that some parts were delayed. On well. Clearly not English majors.

I'll just go ahead and try it the other way.

[Alan B]

Testing backup controller. It works but feels like less thrust. Reads peaks of 4kw but shunt set 1mohm so readings 2x high.

[richmpdx]

Alan,
For crimpers, the right tool indeed makes a big difference. Does the SparkFun crimper also work for the JST connectors used for balance wires, which I believe are JST XH? I have a crimper that is OK for the JST SM connectors as used for the cycle analyst. It is also OK for the male JST XH pins, but is real struggle for the JST XH female connectors.
Rich

[Alan B]

I have not tried the crimper on the JST-XH female pins. I hope it does work on them. That's one of the reasons I got it. Turns out that the JST-SM pins used for the Cycle Analyst seemed to fit the largest of the crimp spots, something like 1.9mm. This crimper has about four sites, 1.0, 1.3, 1.6 and 1.9mm if I remember correctly. Which one do you think is needed for JST-XH female?

At lunchtime I tried to swap the shock over, but it won't fit the other way. The swingarm doesn't have enough clearance for the ends of the shock. I had a similar problem at the upper end, I had to file a bunch to get it to fit. There needs to be more clearance to fit these shocks with wide heads and fairly short necks. So I don't really have an option to turn this one over. Fixing that would be difficult as it would require removing metal from some pretty solid material.

[Alan B]

Took a spin around the neighborhood last evening. Great way to meet people, riding around on an unusual looking bike on a warm Saturday evening. Met some neighbors a few blocks away who were quite impressed with the setup.

Today I'm going to try 22S. If that works I'll take a longer ride, the three bears loop around a couple of lakes here, with three increasingly challenging hills. I should get regen going for this. Charging batteries now. It will be 6S+5S+6S+5S at 8AH for 670 watt hours. That will make 22 miles at 30 watt hours per mile, so I'll have to stay under 30 wh/mi to make this 22 mile ride. Or pedal home.

[Alan B]

The Three Bears Loop



I hooked up regen this morning, and made up a 22S 8AH pack of Zippy Lipo. Then I readjusted the Cycle Analyst beta3 to 2 milliohm shunt.

I tested the regen as I rolled out. No dice. I guess the programming is not set for it. This 12 FET is still on factory settings.

With this battery pack and facing a 23 mile ride with plenty of climbing I decided to take it easy and pedal a lot.

22S was a bit over 90 volts to start. Running high voltage makes many changes to the system. It not only goes faster, it dumps more power and the throttle is more sensitive. It is more difficult and tiring to manage the throttle.

But it does dump lots of power. I saw peaks of about 5-6kw. Nothing popped. I tried to keep it down as I needed economy on this run. My average speed was 16 mph, and that included some fast downhill runs. The CA indicates my top speed to be well over 100 mph, but I only remember getting to about 40 mph on a big downhill. The handling at speed was sure-footed and smooth.

This ride is a bit rough in spots, this is very rural pavement. The suspension and cloud9 seat soaked it up and made a comfy ride. A thin layer of gravel on the pavement didn't bother the traction, the tires just felt planted. Very nice feeling. Not like the sketchy feeling one gets with bicycle tires in gravel.

Beautiful day for a nice ride. I noticed that the young cyclists are interested in the ebike and start talking about how they could use one, where the older cyclists say you are cheating. The younger folks are embracing technology to a larger degree and they don't see it in quite the purist way. The electric bicycle is a different vehicle than the road bikes they were enjoying today. And commuting is a different purpose than a Sunday ride. I did enjoy blowing by them at 20 mph on a long uphill, though, and they didn't seem to mind at all.

This Cromotor is really something. It wanted to leap ahead, speed up. I felt as though I was holding it back the whole time. And I was, to make the power budget, which I did make but the voltage was starting to drop at the end, so I think it was close. The CA claims only 6 amp hours were used of 8, but the voltage tells a different story. I suspect the shunt is not calibrated well enough yet.

I checked controller and motor temperatures a few times with the bare hand test. The motor never got above barely warm. The controller got quite warm on the long uphill climbs. Not hot, but almost hot. You could hold your hand on it without discomfort. Based on that I think the 12 FET is okay for a backup but not the best choice for a totally reliable primary controller for the Cromotor at 22s. The 24 FET never got warm, and it felt stronger as it was dialed to higher current.

I didn't try a full speed run, but with 22S it wanted to zip right up to 30mph and was still pulling when I backed off.

All in all it was a great ride. Happy Father's Day! Time to BBQ!

[Alan B]

Charging of the battery took 6.8 amp hours. That is about 550 watt hours. Efficiency is then about 25 watt hours per mile. Not bad at all for a heavy bike with big heavy fat tires and suspension, over a circuit with not-great pavement and some pretty good hills at 16 mph average speed.

[pchen92]

Hi,
I set the shunt value at 1.750mOhm for my 12 fets. It seems accurate.
Tell me if that works for you