My new Colossus Water Cooled motor

[Bluefang]

Love the bike and the way its been running with the Kelly and Colossus.

Whats the peak power you expect to get out of that motor now that its water cooled? Have you considered going to a sine wave controller if you decide to upgrade the controller? I personally just changed from a kelly to a sabvoton controller, I lost some peak power but just the clean control of the motor i get from the controller have allowed better accelleration even with a lower peak power. The kelly would constantly hit some limit and cut power, this was the same for 4 different kellys.

Looking at golden motor they have brought out some sine wave controllers, one with 700amp current. Would be interesting to actually see the power that motor can make before it overloads the water cooling and starts getting hot.

By the way with the new sine wave controllers the hubmonster seems even happier where it used to get to 100 degrees C its now sitting at 80 or below. Wicked fun been dead silent too

 

[SplinterOz]

Bluefang thanks.

Peak won't change with cooling. Peak is related to the saturation point of the stator and core, how many amps and therefore peak torque multiplied by the max rpm of the motor. Remember the efficiency drops off significantly as the backemf increases with rpm.

Cooling helps with continuous in the sense that you can shed more waste heat therefore run longer at any given power output.
That is probably oversimplifying it.

The problem with most other controllers for this motor is they can't handle the rpm multiplied by the pole count. This value (Kelly calls the erpm) is the rate the controller has to switch the phases on and off. Sevconn does not switch that fast... not sure about the others.

 

[circuit]

Bluefang thanks.

Peak won't change with cooling. Peak is related to the saturation point of the stator and core, how many amps and therefore peak torque multiplied by the max rpm of the motor. Remember the efficiency drops off significantly as the backemf increases with rpm.

Cooling helps with continuous in the sense that you can shed more waste heat therefore run longer at any given power output.
That is probably oversimplifying it.

The problem with most other controllers for this motor is they can't handle the rpm multiplied by the pole count. This value (Kelly calls the erpm) is the rate the controller has to switch the phases on and off. Sevconn does not switch that fast... not sure about the others.

Wow...

efficiency has absolutely nothing to do with amplitude of BEMF, no mater how you put it.

Controllers can't handle such motors not because 'erpm', but because their typical 16kHz PWM switching creates high current ripple. Kelly offers a optional firmware with increased switching frequency, proportionally lowering current ripple and enabling control of crappy motors. Of course that increases switching losses in controller. That is a poor man's workaround solution.

 

[Bluefang]

I understand the saturation points for current and the extra heat if you go beyond the motors rpm limits but i would guess that it would be highly unlikely that you ever took the motor to those levels. Specially as a daily ridden bike it would seem more likely that actually pushing more controlled power into the motor would limit the amount of actual waste heat generated by slow accelleration. Been only air cooled untill now you have to average out all the power and relize that there is no cooling when you are stopped, now you have changed to water cooling your average power handling of the motor has increased a huge amount specially for the areas that would normally really hurt in terms of heat. The dreaded stop start traffic.

My hub monster runs cooler with a higher phase amp setup then lower, in stop start traffic it gets hot quickly with only ~300amps. Same motor and essentially the same conditions it ends up running cooler at ~600amps, but a shit load more fun.

 

[Arlo1]

Controllers can't handle such motors not because 'erpm',

Actualy Circuit this is true for this motor. The one I have has 20 magnets for a 10 poll pair count. Which means 1 mechanical revolution is 10 e-revolutions.
Max frequency for a sevcon is 500hz which would allow a max of 30,000 erpm or 3000 mechanical rpm.

 

[John in CR]

I expect to get my bike back to the bikeshow next year and win another most unique award.

I'd certainly give you another one. I bet they can't get over how such a small motor can push a full size bike. I know a bit about it and I can't.

I'm more into acceleration than speed, though having both is certainly fun for the rare occasions I get on the highway. To get a feel for your acceleration, since it's so hard to gauge anything from videos, can you tell us peak battery current the Kelly is delivering and the total weight your Colossus is pushing?

Now that you have more than one, what's the next project, or are you having a hard time getting motivated with such a quality e-ride for daily transport?

John

 

[circuit]

Controllers can't handle such motors not because 'erpm',

Actualy Circuit this is true for this motor. The one I have has 20 magnets for a 10 poll pair count. Which means 1 mechanical revolution is 10 e-revolutions.
Max frequency for a sevcon is 500hz which would allow a max of 30,000 erpm or 3000 mechanical rpm.

This is only part of the story... Yes, erpm is also important, because it is very hard to do any voltage regulation when commutation frequency gets close to PWM frequency. And because of this, some controllers are programmed to simply limit the erpm. So it is only a product of too slow PWM. For this motor, high PWM frequency is needed both to keeep up with erpm and, more importantly, low inductance.

 

[MitchJi]

Hi,

Max frequency for a sevcon is 500hz which would allow a max of 30,000 erpm or 3000 mechanical rpm.

How about a sabvoton?

 

[SplinterOz]

Bluefang thanks.

Peak won't change with cooling. Peak is related to the saturation point of the stator and core, how many amps and therefore peak torque multiplied by the max rpm of the motor. Remember the efficiency drops off significantly as the backemf increases with rpm.

Cooling helps with continuous in the sense that you can shed more waste heat therefore run longer at any given power output.
That is probably oversimplifying it.

The problem with most other controllers for this motor is they can't handle the rpm multiplied by the pole count. This value (Kelly calls the erpm) is the rate the controller has to switch the phases on and off. Sevconn does not switch that fast... not sure about the others.

Wow...

efficiency has absolutely nothing to do with amplitude of BEMF, no mater how you put it.

Controllers can't handle such motors not because 'erpm', but because their typical 16kHz PWM switching creates high current ripple. Kelly offers a optional firmware with increased switching frequency, proportionally lowering current ripple and enabling control of crappy motors. Of course that increases switching losses in controller. That is a poor man's workaround solution.

Sorry I didn't get the BEMF right. It does limit RPM, http://m.machinedesign.com/motorsdrives/difference-between-ac-induction-permanent-magnet-and-servomotor-technologies section on limitations and challanges, but I couldn't find a reference to efficency.

As for the eRPM I did not say it was the whole story but you have to start with a controller that at least has the capability to spin the motor to the correct RPM.

 

[SplinterOz]

Hi,

Max frequency for a sevcon is 500hz which would allow a max of 30,000 erpm or 3000 mechanical rpm.

How about a sabvoton?

They seem to only go up to a 6kw version. I have dyno'd at the wheel to 14kw.

Edit... just saw their 96 volt which should peak at 19kw. Maybe.

 

[querlenker]

I can say that the big watercooled MGM Compro controller work pretty well. 14s 200A no problem, but when you decrease the pwm frequency to get the controller temperature down and go close to the amp limit, the mosfets blow in 3s.. Kelly does the job but the torque is real low, phase amps seams to be limit at lot. And if you look at the dimensions and weight of the Kelly there is no advantage to the Sevcon. I would never go back from Sevcon..
If you have the same motor that I had in my hands, I would carefully adjust the position of the halls with a scope Tony. Also measure the angle between the halls, I need to bent them and glued them to get them correct positioned. If you switch to Sevcon in the future, I can suggest to buy the sin/cos encoder "RMB29 ROTARY MAGNETIC MODULE" from rls .si.

 

[liveforphysics]

Sevcon gen4 wouldn't be able to spin the motor faster than 3,000rpm. It also likely would struggle with current control faults.

 

[Hillhater]

I never said Kelly will not run this. I did say you need to select the right Kelly to run it. I also do not believe Kelly is a great controller but if it works for you congrats.

I keep reading threads where after much trouble and problem solving, the issue is eventually traced to the Kelly controllers which are found to be Unworkable for one reason or another (EG , Oatnets buggy,....Rippertons Mira, etc) , such that I wonder why you guys don't avoid using them ?
Is it just that they are cheap ?

 

[Arlo1]

I never said Kelly will not run this. I did say you need to select the right Kelly to run it. I also do not believe Kelly is a great controller but if it works for you congrats.

I keep reading threads where after much trouble and problem solving, the issue is eventually traced to the Kelly controllers which are found to be Unworkable for one reason or another (EG , Oatnets buggy,....Rippertons Mira, etc) , such that I wonder why you guys don't avoid using them ?
Is it just that they are cheap ?

I tried one Kelly it blew up in less then ten minutes of spinning the wheel gently in the air while figureing out hall combos. I will never spend a penny on a Kelly Ever! I also seen all the threads from other people who really understand and helped me learn a lot of what I know about this stuff and who show how pour of a job at building a controller they are.

 

[SplinterOz]

can we keep on topic.

 

[SplinterOz]

I never said Kelly will not run this. I did say you need to select the right Kelly to run it. I also do not believe Kelly is a great controller but if it works for you congrats.

I keep reading threads where after much trouble and problem solving, the issue is eventually traced to the Kelly controllers which are found to be Unworkable for one reason or another (EG , Oatnets buggy,....Rippertons Mira, etc) , such that I wonder why you guys don't avoid using them ?
Is it just that they are cheap ?

Ripperton is still using kellys on his race bike. The Mira had issues due to two motors, two controllers and timing glitches.
http://www.diyelectriccar.com/forums/showpost.php?p=510178&postcount=1051

 

[SplinterOz]

Video or it didn't happen.
[youtube]bHU5nu6zXl8[/youtube]

Note the battery on the bike at the moment is 5.8ah of lipo (20s) so all the power levels are wound down.
I will be putting in 20ah of LiPo in on the weekend for more comprehensive testing, I also have to mount the temp sensor and a few other bits before I push the power levels.

The LiPos will be replaced with 60ah of Kokams before the end of the year.

Thanks

 

[SplinterOz]

... but because their typical 16kHz PWM switching creates high current ripple. Kelly offers a optional firmware with increased switching frequency, proportionally lowering current ripple and enabling control of crappy motors. Of course that increases switching losses in controller. That is a poor man's workaround solution.

Looks like you understand why the Kelly can run this motor. The PWM on the (ultra) high speed firmware is around 33.3khz, this does increase switching losses however the gain in efficiency on the motor easily balances this. As a whole the system is quite efficient.
The newer Kelly controllers (with the E designation at the end of the part number) have a better heat and current path allowing them to handle the extra losses if you use a slightly overspec controller. The motor is good for around 350 to 400 amps peak and if you use a high speed 500amp Kelly controller it handles the switching losses quite well.

I will be testing this motor with a 72 volt 300 amp Kelly controller over the next week or so. Based on the bahaviour I have seen before it will work but be down on torque as the current limiting circuits tend to be more agressive.

This "poor man's workaround" is better than the one required for Ultramotive Carbon motor (http://www.varleygroup.com/site/electric-products) which is to add 15kg of inductors on the phase leads.

 

[John in CR]

The PWM on the (ultra) high speed firmware is around 33.3khz, this does increase switching losses however the gain in efficiency on the motor easily balances this.

Why would increasing the PWM rate increase motor efficiency?

 

[Arlo1]

The PWM on the (ultra) high speed firmware is around 33.3khz, this does increase switching losses however the gain in efficiency on the motor easily balances this.

Why would increasing the PWM rate increase motor efficiency?

Helps keep the current ripple smother which reduces the over currents.

 

[Hillhater]

I never said Kelly will not run this. I did say you need to select the right Kelly to run it. I also do not believe Kelly is a great controller but if it works for you congrats.

I keep reading threads where after much trouble and problem solving, the issue is eventually traced to the Kelly controllers which are found to be Unworkable for one reason or another (EG , Oatnets buggy,....Rippertons Mira, etc) , such that I wonder why you guys don't avoid using them ?
Is it just that they are cheap ?

Ripperton is still using kellys on his race bike. The Mira had issues due to two motors, two controllers and timing glitches.
http://www.diyelectriccar.com/forums/showpost.php?p=510178&postcount=1051

Sure, but Rip has had multiple failures and tech glitches on the bike, and the Mira saga was frustrating too.
What about Oatnet's saga with the Buggy ?
It just seems the problems and failures are disproportionate to the number used, and the tech support is often less than useful or just not there.
Maybe there are not too many alternatives available ?

 

[SplinterOz]

Ok so back from holidays and some more work.
Thinks I have completed:
- temp sensor in motor
- dash constructed with BMS
- completed the 12v wiring
- programmed the controller
- programmed the fuse box (yep I know lookup m-Unit http://motogadget.com/en/electrics/...m-unit-digit-tastersteuerung-u-sicherung.html.
- mounted the 20ah LiPo pack in the stainless steel box.

Took it for a quick spin round the block and it is all good. Temp cutout works (set it to 60 degrees to confirm cutoff) and the bike pulls really well.

Notes on the motor.
It is quieter than the original prototype, this is likely due to the better balancing on the can and the hall sensors being spaced accurately.
It is getting a little warm and I think I may have the timing a little retarded ( I forgot to note which direction I was spinning the motor when we positioned the halls)
It cools much faster than the prototype.

Note I have not yet connected the water cooling, hopefully a job for next weekend.

 

[larsb]

@SplinterOZ: How's it going with the motor ? Based on the previous problems for others with colossus motor i'd guess you eventually switched to something else but a happy story about 30 000 kms of uneventful riding would be so much nicer to hear

 

[tomarth]

Hi SplinterOz

I'm building a lightweight motorcycle, would you still recommend this motor in 2019? The last documented build with this motor I could find was from 2015: featuring the midget Go Kart and the Kelly KSL-D 400 amp controller (20S/10ah/lipo).

I know other users had problems with this motor, but i was very impressed reading through your builds!

I was originally looking at the QS 3000 138 motor, which is a powerful motor, but it's twice the size and weight of the 120/100!

Do you now recommend the Kelly KSL-D, as apposed to the KBL72501B? Does this mean that 120/100 supports sinusoidal mode and can it be be powered by a controller in FOC mode? And does this improve the efficiency of the motor?

(I'm looking at the Alien Power Water cooled 120-100, which I'm guessing is very similar to your 12kw colossus)

 

[SplinterOz]

On the original motor, I was running the bit for about a year. My biggest issue was the battery pack, the way I had connected the headways in the middle caused no end of trouble. I did burn out the original prototype motor on one phase (connect a temp sensor).

I acquired two more of the production run and placed one in the bike with a different battery pack, ran that for a few months. But again the little frame meant the battery was always the issue. We then used the same motor in a go-kart for a while.

Life has caught up with me and I have started multiple projects but not quite completed them. I am now riding a 2nd hand Zero SR while the next project is on hold.

I wouldn't use this for a street legal motorbike now, mainly because I want more power. the lightweight go-kart was great fun. I have a friend that used one on an off-road mountain bike and that was awesome.