JD's BMC V3 Build - in progress

[oatnet]

AAARrrrgghh.

I put a connector on the shunt, but it was ugly. The hall and throttle connectors I did earlier today went smooth as silk, but this one, everything that could go wrong on this simple little task, did. I had to do it twice, and the second time was ugly too. Maybe I am just tired and getting sloppy.

Anyhow, so I hooked the shunt up to the BMC bike, and it workedno problem. Then I went to set the shunt value. Turns out this is one of the Beta-version CycleAnalyst's Justin sent me for evaluation back when he was still fishing for a name for it... and it doesn't have an option to set the Rshunt value, so for all practical purposes it is USELESS. I added a speedo cable to it just for this build, I have it all zip-tied and routed through the channel in the frame, and I hacked up a StandAlone cycleanalyst so I wouldn't have to uninstall this one.

I have few new direct-connect Cycleanalysts lying around, so I grabbed one of them and plugged it in loose. A standalone Cycleanalyst I have on one of my TidalForce bikes shows an Rshunt value of 1.399, so I set this new one from the factory default of 1.000 to 1.399. No load it peaks at 105w. Stalled when hold the front brake and crank the throttle, it goes up to 450w. Riding 10' across my ManPit it runs about 250w, and that is at about the max speed it was running before.

AAAArrrrggghh.

 

[oatnet]

One more step forward, two steps back.

I messed around with all of the settings, and found that by changing the control mode to 'speed' I was able to get it to work. I had initially set it at 'balanced', and later tried 'torque', but the initial setting was 'Speed'. So I cleared a path, took it outside, and it did seem to perform fairly well. It didn't jump off the line, but once it was rolling it seemed to build nicely.

I went about 1/2 mile, down a hill and up the other side. Pulled over and touched the controller, it was hotter than I could hold. The phase wires were hot too, melting the pointless hot glue in the BMC's andersons. I pedalled home. I'll upgrade my Phase wires to 8gauge right up to the motor, but I don't think that is the source of the heat, they were hotter by the controller than by the motor.

I am pleased to see that everything works, that I don't have a bad hall sensor in the motor or something, and I have some practical discoveries to pass on to Kelly.

DD, are you using 'speed' control mode on yours, and have you tried Torque or Balanced?

-JD

 

[dirtdad]

Those come from the factory balanced, I think, like you say, and I have not played with that setting.

 

[oatnet]

Those come from the factory balanced, I think, like you say, and I have not played with that setting.

Mine came from the factory in 'speed' mode, I changed it to balanced right away. If you have not adjusted your settings you are probably in 'speed' mode too. If you get a chance to check it, and try 'balanced' or 'torque' modes, please let me know. I have a feeling it isn't just my controller...

-JD

 

[oatnet]

Response from Kelly:

Sorry for the inconvenience. Our worker don't mark the numbers according to the wiring diagram,to avoid the repetition.Please see the new attached pdf file.

So far the high speed firmware only could offer the speed work mode.
You need to make sure a good radiating condition for the controller,such as putting the fans around.
It is a good idea to limit the max motor current at 70% more or less.Motor isn't efficient with peak current anyway.
A forced air flow is prefered.The BMC V3 motor is rated at 48V 1KW.If it works at the 72V battery under the high speed firmware,I guess the hot phenomenon is a normal occurrence and should be expected.

Very disappointing. Limiting motor current to 35a on a 50a controller is not the answer I was looking for. I have never had a brushless Kelly require any additional cooling, even pulling 100amps at 72v they have stayed cool. The new small form factor does not help if I have to add a cooling plate and a fan, I end up with more mass than their large format controllers. Honestly, with the amount of heat generated in a mere 1/2 mile, while exposed to full airflow, I think it will be difficult to supply enough cooling.

It is also frustrating to have the controller offer 2 modes of operation that don't work, and have the diagram numbered wrong (tonight I'll post the corrected PDF they sent me).

This overheatIng sounds just like the problems other folks were reporting on other controllers. I suspect that this new motor causes controllers to overheat at 72v, and the new Kelly does nothing to address that problem. Sure, it goes up to 70000 electric rpm (70mph in a 26" wheel) but what is the point if the controller melts down long before that.

So, I'll split my 72v pack and try it 36v. I typically expect amps to be the main cause of heat from resistance, but if it stays cool at 36v it will prove out my 72v theory. Then I'll add 4s of headway cells to test at 48v. Somewhere along the way I'll beef up the phase wires.

But tonight, welding class.

-JD

 

[dirtdad]

I have found that with BMCs V2 and V3, 72V just does not get you much more than 48V which seems to be its sweet spot.

I will have to try the speed mode in the next week or two.

 

[nicobie]

I have found that with BMCs V2 and V3, 72V just does not get you much more than 48V which seems to be its sweet spot.

I agree.

My V2 is switchable between 54v and 76v. At 76v I get about 3 more mph and just a bit more acceleration. I never could get an Infineon (old style) to work right with the V2, but a cylte analog works fine.

 

[oatnet]

I have found that with BMCs V2 and V3, 72V just does not get you much more than 48V which seems to be its sweet spot.

Disappointing, I had dreams of taking this to 120v and seeing what it could do. I was expecting overvolting performance similar to x5's, but I guess this is just a 35mph hub motor that is lightweight, not a 60mph screamer. I wonder what the problem is at 72v+... I remember reading somewhere that this V3 was supposed to be 1/4 the resistance of the V2, I'd think that was a good thing. Has anyone had the cover off theirs yet? Fany of Kelly said something about this motor having 80 pole pairs - which I interpreted to mean 16 pole pairs at a 5:1 ratio, I think if she were correct we would not get 35 MPH out of it from a 70000rpm controller.

Today I tried it on the same 1/2 mile course at 36v/100% current, not sure of the actual amps because I can't set the rShunt on my Beta Cycleanalyst, but it must have been significant because my 30ah pack dropped from 40v to 34v under load. I pullied about 23mph but I didn't let it run very far before the hill. Launch torque felt about the same as 72v, hill climbing torque a little bit weaker. The controller was merely hot this time, and the phase wires were very HOT. I need to upgrade my phase wires to 8 gauge for the next test see if the last few inches of 16ga going into the axle can handle 45a continious, and see what controller temps are like when the phase wires aren't heating up. If significant ambient airflow isn't sufficient to cool this controller at it's rated performance, then I don't think forced air flow will help, and I'd have to rate it a fail. I can mount it to an aluminium plate to help grab more airflow, but that will also defeat the point of a tiny controller.

Once I get cooling issues sorted I'll build a 12v booster pack from some spare headways to get a 48v test pack and see what's what with the V3.

-JD

 

[oatnet]

I never could get an Infineon (old style) to work right with the V2, but a cylte analog works fine.

My analog Cyltes were just acting wierd, no power just like 'torque' mode and 'balanced' mode on the Kelly. Was the infineon the one with the upgraded clock chip?

-JD

 

[oatnet]

I will have to try the speed mode in the next week or two.

Judging by Fany's response "So far the high speed firmware only could offer the speed work mode." if your kelly is working you have to be in speed mode right now.

-JD

 

[nicobie]

Was the infineon the one with the upgraded clock chip?

-JD

I don't think so. I tried two different ones from Keywin and both acted the same. the 48v/25a one I bought 6 month ago and the 72v/40a one about 4 months ago. I don't know if they were of the soft start type or not, maybe that was the problem. They worked but it was either almost all on, or all off and most likely was what caused the sprag clutch (freewheel) to pop.

 

[Muad'dib]

I have found that with BMCs V2 and V3, 72V just does not get you much more than 48V which seems to be its sweet spot.

Disappointing, I had dreams of taking this to 120v and seeing what it could do. I was expecting overvolting performance similar to x5's, but I guess this is just a 35mph hub motor that is lightweight, not a 60mph screamer. I wonder what the problem is at 72v+... I remember reading somewhere that this V3 was supposed to be 1/4 the resistance of the V2, I'd think that was a good thing. Has anyone had the cover off theirs yet?

I have opened mine up, trying to upgrade the wires, but all I could fit in there after countless hours of trying was 14 gauge motor wires. The original wires may have been 16...they don't look much different than the 14. You can see the windings for yourself:

 

[oatnet]

Has anyone had the cover off theirs yet?

I have opened mine up, trying to upgrade the wires, but all I could fit in there after countless hours of trying was 14 gauge motor wires. The original wires may have been 16...they don't look much different than the 14. You can see the windings for yourself:

Thanks Muad'dib, good job! Yeah, I suspected there wasn't much room If only we could get wire that is shaped like a slice of pie, in 1/3 of a circle :lol: . I am hoping that 8 gauge, going as close to the motor as I dare, will short the high resistance path enough, and act as enough of a heat sink, to allow high-amp performance.

I count 32 magnets, and 36 stator poles. I see something else I need to learn about motors - I thought there were usually the same number of stator poles as magnets. Fany of Kelly later confirmed that they were seeing 80 electric RPM during testing, but didn't know the breakdown between poles and gearing, but if it has 32 magnets, aka 16 pole pairs, and 80 rpm, then it should be the same 5:1 ratio.

If it is still open, and you remember when you are actually working on it, can you spin the motor by hand (5)rpm and verify that is in fact (1) full wheel rpm?

-JD

 

[Muad'dib]

If it is still open, and you remember when you are actually working on it, can you spin the motor by hand (5)rpm and verify that is in fact (1) full wheel rpm?
-JD

It is still open. I don't understand exactly how to measure the rpm like you asked. Can you explain (as if to a 5 year old!) how you want me to measure it?

 

[TPA]

Probably means revolutions????

In other words is it a 5 to 1 ratio. Turn the motor 5 times to see if the wheel turns once????

 

[oatnet]

If it is still open, and you remember when you are actually working on it, can you spin the motor by hand (5)rpm and verify that is in fact (1) full wheel rpm?
-JD

It is still open. I don't understand exactly how to measure the rpm like you asked. Can you explain (as if to a 5 year old!) how you want me to measure it?

(oatnet smacks stupid self in forehead) Whups, I made no sense :lol: , as TPA said, I meant revolutions not RPM. Can you rotate the stator by hand 5 complete revolutions, and verify that doing so spins the wheel one complete revolution? If so you will verify that the motor has 5:1 gearing.

-JD

 

[oatnet]

I updated the phase wires to fine-stranded 8 gauge today, cutting them off near the axle, and soldering on new ones. The only part of the phase wires that still generates resistance is between the 8 gauage and the windings, and I guess there is enough copper in the 8 gauge to act as a heat sink.

I ran the same test run at 36v and found this eliminated the overheating phase wire problem. The controller was slightly warm. Then I tried the test at 72v. The phase wires were cool (even by the motor), the motor was cool, but again the controller was too hot to touch, showing this is a controller issue.

Overall, I was underwhelmed by performance, I was expecting something that would exceed the kick of a 5305. At both voltages, launch was pretty weak, but after 4-5mph it came on strong and I reckon here is where it is possible it would out-pull a 5305. I got 23mph at 36v and 28mph at 72v; granted the flats were pretty short so I didn't get a chance to let it build speed. Next I'll add a 12v headway booster pack to the 36v pack and try it at 48v, and if it passes the temperature test I'll do some serious performance testing. I also need to ask Justin if there is any way I can set the rShunt value on this beta CycleAnalyst, or I'll swap it out with another one, so I can get some actual power readings.

-JD



View attachment 5

 

[Mark_A_W]

"BMC needs to go to a larger axle size, so it can bore an oval between the flats for thicker wire."


They can't go to a bigger distance between the flats on the axle, or it won't fit in the dropout.


If they go to a bigger axle, with the same flat distance, and an oval hole, they will have to broach the hole, rather than drill. An oval hole is a very difficult thing to make.

Bafang BPM motor is sensorless - means you don't need to squeeze hall wires in as well - more space for phase wires. And it seems to work, instant takeoff...at least on the bench test.

 

[Muad'dib]

If it is still open, and you remember when you are actually working on it, can you spin the motor by hand (5)rpm and verify that is in fact (1) full wheel rpm?
-JD

It is still open. I don't understand exactly how to measure the rpm like you asked. Can you explain (as if to a 5 year old!) how you want me to measure it?

(oatnet smacks stupid self in forehead) Whups, I made no sense :lol: , as TPA said, I meant revolutions not RPM. Can you rotate the stator by hand 5 complete revolutions, and verify that doing so spins the wheel one complete revolution? If so you will verify that the motor has 5:1 gearing.

-JD

Hmm, stator = stationary part of motor = doesn't rotate?? The part of the motor that does turn, attached to the sun gear of course, is inaccessible when the motor is inside the wheel. Sorry if this is just my own misunderstanding.

 

[oatnet]

I was able to set the rShunt value on my 'Beta Version' CycleAnalyst. I have set this parameter a billion times on on the CA, before I emailed Justin, I checked three times in case the button was double-pressing and skipping by it, never saw the option. Justin confirmed it was in the software, and this time I found it on the first try - and it took several more tries to get the value set right because the button was double/triple skipping. Gotta work on that, but now I am able to measure current with moderate accuracy.

So, I started at 36v and ran a course around my neightborhood, including the 2 steep hills that got the controller really hot at 72v

With the Controller set at 100% motor current, aMax was 95.57a, vmin was 28.8v, and I used 43.8 wh/m
With the Controller set at 50% motor current, aMax was 59.19a, vmin was 32.0v, and I used 48.3 wh/m Then I tried a run 'mild' course (same terrain, avoiding the hills and got 42.3 wh/m
With the Controller set at 25% motor current, aMax was 32.87 and I got 37.1 wh/m on the 'mild' course (forget steep hills with Kelly's "speed mode")

The Kelly controller for the BMC V3 only has 'Speed mode', which means it starts off with low amps and offers higher amps as speed increases. At 25% motor current, this meant I launched at 12 amps or so, which was totally lame for going up hills. I think if I could run balanced mode and assuming that at 25% that gives me 32amps all the way through the range, this motor might pull OK.

So then I tried 72v, with the controller still set at 25% motor current, back to the steep course, amax was 30.21a and I got 45.6 wh/m. The good news is that the controller stayed cool even after the hill climb, whereas it was too hot to touch at 100% motor current. The bad news is at slow speeds, the controller only pulled 4 amps, I had to pedal to get moving. Once speed built up, it pulled nicely even at only 30amps.

After beinhg satisfied with many Kelly Controllers, I have to rate this one a FAIL. It did not die, but it didn't perform either, as dialing it down enough to keep it cool means totally inadequate launch torque; again, a balanced mode or torque mode might help here. 40-48wh/m is just abysmal for my test course, I was hoping to be in the 25-30wh/m range to be a little better than my x5. Top speed at 36v was 23.5mph for 25%/50%/100%, and 29.5mph on my 48v test at 100% the other day.

Maybe its the motor, but others have reported much better performance with other controllers, I'll have to try some analog xlytes next and see what I can get out of it. Paired with the mini kelly controller, it is no 5305 slayer. OTOH, I always run my x5s at 72v, and since I haven't been able to run the V3 at 72v very well, maybe it hasn't had a fair shot.

I'd really need to see 25wh/m at 20mph to trade in my x5's. To get rid of the hubmotor weight at the rear and have a lightweight free-wheeling commutor bike, I could live with only 48v and only getting 29.5mph, if I can get up hills too. I'll keep trying to see if this motor is gonna work out for me.

-JD

 

[Toorbough ULL-Zeveigh]

about splicing multiwire cable if it hasn't been mentioned already, something I like to do when there's room is to stagger the splices to spread them out a bit.
it's extra work & u have to strip off more of the outer jacket to allow for the shortest wire on either end sufficient length to slip the inner heatshrink far enuf so it don't shrink while soldering the wires together.
but it makes for a smaller bump in the final product that all the joins don't stack up one on top of the other for a clear look & helps with the strain relief i think.
like i say, there may not have been enuf free length in this case but something u might what to keep in mind if the opportunity lends itself to a staircase splice.

 

[oatnet]

about splicing multiwire cable if it hasn't been mentioned already, something I like to do when there's room is to stagger the splices to spread them out a bit.

Hi T!

The point of the splice on the phase wires was to remove as much of the high-resistance 16ga that was reaching jacket-melting temps, and add as much 8ga wire as possible to leach heat away from the remaining 16ga, so all three Splices were as close to the axle as I dared to cut them.

Also, wouldn't staggering the joints give each of your phase wires slightly different resistance, resulting in each phase getting slightly different watts?


For those of you just tuning in, don't miss my performance results for the mini Kelly controller on the bottom of the last page.

-JD

 

[Toorbough ULL-Zeveigh]

ah yes, the heat would be an over riding factor.
just to clarify, the amount of offset between each wire would be or could be as little as the length of the solder joint.
just enuf where the one splice ends the next one begins, looks like maybe an inch or inch & a half diff in ur case between longest & shortest.
wouldn't think an inch of added resistance should amount to much, depends on the amps i guess.
so it's good for lower power wiring anyways & it's more useful the more individual wires u have.
with just the 4 lines it doesn't look like a snake swallowing an antelope.

 

[RPA]

That really sucks about the Kelly controller. It sounds like it's worse than my little BMC 30A controller. Hopefully this new BMC one will be better.

 

[oatnet]

That really sucks about the Kelly controller. It sounds like it's worse than my little BMC 30A controller. Hopefully this new BMC one will be better.

Yeah, really disappointing. My experience differs so much from what I have seen other people report at 48v, from top speed to effeciency to torque at launch (although I might just be spoiled by 5305s at 88v :lol: ) that I think it has to be the controller. Before the kelly I was having problems getting the V3 to work with my old xlyte controllers, but now that I have soldered the hall connecter to a BMC pinout, I am not eager to tear up the hall connectors on the motors or controllers to try it again. I have a large qty of the 3,4, and 6 pin flat connectors used on xlyte controllers, but I never found bulk stock of the 5-pins to buy.

I just paid Ilia for one of the 72v35a analog xlyte controllers he has been selling with the V2, and one of the new 72v50a xlyte digital controllers that he just got in (today!) and has been told will work with the BMC v3. He says he tested the V3 at 40v on the analog controller, so I'll give it a try at 36v(40v) and 48v(54v) and see how it compares with the Kelly for launch, speed, and wh/m, but hearing other people toasting analog controllers at 72v I'll stop there. Then I'll test the 50a digital at 36v and 48v, and consider 72v if it seems it is doing well.

All I want is a 9 pound motor that will give me 30mph at 25wh/m on the flats and climb hills like they were downhills, is that too much to ask? :lol: Seriously though, my Tidalforce x bikes have great balance, do my daily commute (hills and all) at 28-32 wh/m, with speeds of 25mph on flats to 30+mph with tailwinds. The only think I don't like about TF is the 60+lbs from the battery and the motors, and the size of the motor/front hub battery that screams EBIKE louder than any planetary gear.

Since overvolting isn't working so this won't be the high-speed screamer I originally intended, my next plan is to make it a lightweight stealthy commuter. With the 9b v3, and a 7lb 15s3p a123 pack I'll be adding a mere 16lbs to the frame. The motor's small footprint and low, discreet battery mount should help with the stealth. Although the battery only has 250wh (all I can fit in that space), because the cells are a123 I can blast them with a 14a (2c) charge at lunch and fill them up in 30-40 minutes. All I have to do is to make enough torque to get up hills and bring commuting performance down below 38wh/m for it to work.

-JD