Cagiva Supermoto Bulid (4kW Legal)

[minde28383]

Also, chain is ultra loud, like 2-stroke engine loud :/

It's because small sprocket is between ~12 teeth and rear one ~52 teeth. Front small sprocket spins quickly and the chain is smashing at high rpm at front sprocket teeth, therefore the sound. You can decrease sound increasing front sprocket teeth by few teeth and decrease rear sprocket teeth or leave rear sprocket as it is, change the ratio little bit, but it might have a negative effect on torque and efficiency and speed.

These motors were used in motorcycles with 48v or 52v or 60v don't remember exactly and where doing 100km/h with modest amperage. These motorcycles had ~ 4,8:1 ratio. This motor is quite slow and it is good in a matter if you want to have less sound in chain rear transmission.

eq. if front sprocket would 25teeth and rear 50teeth than it would be quiet, like with hub motor where it is used as mid-drive motor like in vector or new quilbix frames.

Another option to use pulleys and belt instead of the chain but then again propper belt tension becomes problematic and technically challenging.

 

[Bikebrap]

Yeah, was aware of small sprocket problem before bulit, but I didn't know it would be this bad. I aimed for high torque so that's why 13/90 sprockets.
Belt was in plan before chain, but chain was a quicker solution. Now that I have more time and figgured out how to make huge rear belt pulley I might give it a shot.

 

[trazor]

I'm a bit late to this thread. Amazing work!

Caution with tensioners in the chain load side. They can take a big amount of force and deform or brake them. I think that in conversion projects is always better to position motor output shaft on the same point as the original, if possible. Great work!

 

[Bikebrap]

Kelly KLS-7230S just arrived.

72V
300A burst
gonna put huge ass heat sink on it
(btw my phase wires melted and shorted out on old controller. Controller is stil alive It had both side covers oppened and fan blowing air trough. That is why I finally ordered kelly)
aaand.... it's a sine wave so everything should be a lot quiter

 

[minde28383]

This kelly 300A will be ok, but I would preffer 500A or 550A phase for this motor.
How much you payed for this Kelly?

 

[fechter]

The Kelly should be good for the 4kW limit, but that motor looks like it could take way more for bursts.

Too bad the Sevcon programming is such a nightmare. It's a great controller once everything is dialed in but there are only a few people in the world that understand what all the parameters do.

 

[whereswally606]

sorry bikebrap you got mail, sorry I didn't see your post earlier re SEVCONs

 

[minde28383]

I too gave up on Sevcon regarding its programming and sold it and bought Sabvoton. Use it for few times and new Sabvoton controller broke. I hope you will have better luck with Kelly.

 

[Bikebrap]

Didn't have time to update this before so here it is.
Took out batteries from my bicycle. They were welded in three packs when I bought them. I disassembled everyhing, ripped appart all the welds from tabs and sanded them a bit for easier welding afterwards.
View attachment 5
View attachment 1

Printed out formers so they match 20s7p perfectlly.
View attachment 3
Had to glue two parts toghther because whole part is larger than print surface. Glued portion even has greather streangth than the part itself.

View attachment 4

 

[Bikebrap]

Kelly is mounted on the bike.
Motor has sin/cos resolver so I had to make my own hall encoder.

View attachment 1

View attachment 2
Aligning the encoder so it is in phase with the motor phases.

Kelly had no torque at low rpm and slow response even with all current maxed out in settings.
I tought Kp and Ki settings were not for my controller as it was nowhere mentioned that is has any kind of PID regulator so I assumed that setting were for some other controllers.
By playing with Kp Ki and Speed err limit (whatever "speed" err limit" is, could not find good info) I got much better response but still has no torque when standing still or when speed is bellow 0.5 kmh. Seems like I'm nitpicking here but I really want to be able to lauch from stand still, or to have all torque when stuck in mud or sand, or on uphill.
Problem is that the structure of this internal regulator nor good explanation of his parameters is described well anywhere. Manual has minimal info on them, just says Kp is proportional and Ki is integral, like, yeah but of what, what is input, what is regulated, what is output, in which units, nothing.

 

[minde28383]

I too now that Kelly controllers like yours does not have good torque from stand still when used with hubs. Looks like it has similar effect on mid drives too, nevertheless it is less noticeable in mid drives due reduction.

 

[minde28383]

Your sin/cos resolver is very nice. Actually it is amazing...
What printer you are using?
Is it DIY printer from parts or you bought it as 'ready to print'? What's model, make?
I'm very curious about 3d printers, but don't have personal experience using them. Time after time i need to do some little box or nob or lever for my rare diy home projects. 3D printers used to be very expensive, but know I see online some cheap small 3d printers.

my few thoughts about Kelly and Sabvoton controllers, and overall.
I think I now few technically inclined people but they did not know if something can be done to overcome soft start from stand still with specific controller. As of many manufacturers Kelly Software is what it is. Their app don't have much accessible parameters to be set on their application user interface, not open source for sure. By keeping certain parameters locked or within certain manufacturer preserve and prolong their controller working hours. It basically reduces controller fault rate therefore end users would see it as - controllers which tend to brake less. That is how I see manufacturer policy.
Kelly soft start was and still is the main and the only reason why I recommend Sabvoton. I don't want to be too much one one side or another, but for now I will say that Sabvoton controllers are not perfect too. My last and practically new Sabvoton controller is in service center right now, software issue, not blown fets, caps etc, issue with software inside controller logic (inside). Depending on what outcome from service center I think I might write somewhere what ''great'' experience I had with 4 out of 5 Sabvotons I dealt.

 

[Bikebrap]

Kelly controler test
Kp set ultra high
[youtube]8wz2CtQqC8A[/youtube]

 

[Bikebrap]

I too now that Kelly controllers like yours does not have good torque from stand still when used with hubs. Looks like it has similar effect on mid drives too, nevertheless it is less noticeable in mid drives due reduction.

I managed to enhance throttle response by setting Kp and Ky relly high. Because I don't know the structure of actual regulator I don't have a feeling for these parameters, by default they were obviously set toooo low for this motor.

This encoder was printed on friend's ultimaker, but now I got an Ender 3 from Creality for 200$ and the prints are unbelievably good

 

[Arlo1]

Nice work. Where did you get the rear sprocket machined at and what did it cost?

 

[Bikebrap]

Nice work. Where did you get the rear sprocket machined at and what did it cost?

I constructed it in solidworks and sent .dxf to my friend who has all kinds of machines soo.. yeah... paid around 100$ if I remember correctly, together with battery boxes.. for material only

 

[Bikebrap]

Did some "wannabe" aesthetic touch up Need to remove that bulky instrument cluster and mount headlamp on forks, it is now fixed on chassis.
And added a heatsing to kelly controller, will add a fan too but it doesn't look like it needs it

also burst 300A is definned only by "10 seconds". It does not say in which intervals, or is it just some aproximation of thermal limit. Not sure what am I getting because it doesn't feel powerfull as it should.

View attachment 1View attachment 3

 

[Bikebrap]

Motor overheated & demagnetised.

So now that I added a heatsink, contoller was no longer cutting down power. So after a long non-stop ride with regenerative braking enabled the motor overheated and demagnetized. I was riding at constant speed, not too aggresive, around 3/4 throttle. So I suspect that motor was already a bit demagnetised because this should not cause overhating of the motor that is rated for 7kW continous. Another clue is what I mentioned earlier, that the bike did not feel 22kW powerfull when kelly was maxed out.
So I asume it was demagnetised to 60-80% before, what combimed with not ideal aligement of my encoder caused too much heat generation at constant XY speed. All this combined with mud guard in front of the motor that obstructed air flow caused a magnet overheating and demagnetisazion because the motor was very very hot to touch. Now it has like 30% of torque compared to before and is preaty much usless and is gonna sit in the garage for now

Did not monitor motor temperature because I was not pushing it hard and it was very cold outside, nor was it ever too warm when riding.

 

[litespeed]

That’s a bummer for you!

Hope to see you get back on this.

Tom

 

[Bikebrap]

So I did some measurements to make sure what happend. Used old controller to get exactly half throttle/half pwm (kelly has regulators n sht) and messured rpm to get KV.
If rated voltage is 55V and rated rpm 6000 that gives KV of 109. (Heinzman datasheet)
At 50% PMW or 41V (82V Batt) motor spins at 9000 which is KV of 219

BUT.. The same Heinzmann datasheet spcifies Ke of 4.44 [V/1000rpm] which is KV of 227
and if I spin motor manualy at 1950 rpm I get 25Vpp which is KV of 78 for peak to peak voltage between phases.


Here is one of the mesuremens where I mesured both stators. Yelow is one stator and blue is the other on the same phase. Heinzman has two separated stators ,as it is axial flux motor, thay get connected together outside. There is this weird ripple on every mesurement. Probably from one of the magnets. Also on both controllers motor sounds like it is gasoline engine on launch-control when I twist on fullt throttle, rpms oscilate.

[youtube]Hxm0AwPkwN8[/youtube]

In the mean time I'm making a spredsheet that compares motors in one comparable unit, I think that would be (Peak power/mass)/max rpm.. or Torque/mass so from that number you can see who deviates a lot and has that number "too good". The more data the better.

 

[eee291]

Hi, I really like your build.


Could it be that you hall sensor converter was lagging behind at higher rpm?
This would make the Controller apply current at the wrong time causing the Motor to overheat.

I don't know if you mentioned it but what did the registration cost?

 

[Bikebrap]

////

 

[Bikebrap]

Hi, I really like your build.


Could it be that you hall sensor converter was lagging behind at higher rpm?
This would make the Controller apply current at the wrong time causing the Motor to overheat.

I don't know if you mentioned it but what did the registration cost?

It is possible if you refer to arduino converter. It has about 10kHz sampling rate for analogRead if I remember correctly. So the limit is around 6000 rpm (at that speed delay is about equal to 1/6 of phase frequency or one hall signal going up or down.). But that doesn't matter as hall signals in controllers are used only at low speeds for taking off. Later these controllers often switch to zero-crossing detection for phase commuting as they do all the time if you run them sensorless. In that mode they have trouble with taking off but can in fact do it, just not as smooth.

But in the end I was using that converter only for short period. The real hall encoder that was printed afterwards was working fine as those are the same hall sensors used in other motors. I just placed them on other location. And that hall sensors are analog technology so much faster than any processor. Only problem was making sure that everything is in phase with internal magnets and stator.

Registration was around 250$

 

[Bikebrap]

[youtube]Q8J8lCsx4Do[/youtube]
I opened the motor and edge of rotor has beed grinding on the stator so this is where it lost magnetic properties. I don't know any more what happened. As I spent a lot of time and money I bought budget motor just to get it running again QS 3000W mid drive, reported to be able to take much more than what is in the name.

In the mean time I finally took Nextion screen and put together some dashboard for all the data. Now I figuring out how to read data from Kelly to Arduino so that I don't have to install my sensors everywhere when there is already that info in the controller.
[youtube]v-ErIg-Rk_o[/youtube]

 

[fechter]

[youtube]Q8J8lCsx4Do[/youtube]
I opened the motor and edge of rotor has beed grinding on the stator so this is where it lost magnetic properties. I don't know any more what happened. As I spent a lot of time and money I bought budget motor just to get it running again QS 3000W mid drive, reported to be able to take much more than what is in the name.

The rubbing is what caused the magnets to overheat and demagnetize. Seems like a manufacturing defect. Sucks.

It will be interesting to see how the QS motor works.