Turnigy CA120-70 8600W Brushless Outrunner (100c eq)

[dave.com]

Finally got motivated to hook up the motor and controller and see if it all works. The controller is a Kelly KEB72801X with the ultra high speed firmware.
http://www.youtube.com/watch?v=xU2pruxPJoU

I tried tuning the bracket position with the motor running slowly, but couldn't really see any big reductions in no-load current. Maybe the motor needs to be under load, with the controller set to constant speed mode.

The bracket works ok, but there are issues. I allowed 0.5mm clearance between the motor can and bracket. The concentricity of the can and protruding screw heads make this a bad idea. 1mm would be better. The entire bracket can rotate to change hall timing, as well as each individual hall sensor. I was thinking I could fine tune the individual hall positions using a scope, but have yet to try it.

Next step is to break out the 20S3P worth of nano tech packs...

If anyone is interested in the Solidworks files of the bracket, pm me. I also have a model of the motor shell.

Dave

 

[liveforphysics]

Dave! You rock!

We should meet up!

 

[Arlo1]

Yeh dave very cool and even smaller then my bike! But I think you will find internal halls will work boatloads better!

 

[magudaman]

So what is with the high current no load? Again that seem very bad for the speed it is spinning?

Anyone have this mounted on a vehicle yet?

 

[dave.com]

Ok, so I've been blindly dorking around with this motor and the hall placement. I don't have immediate access to a dso, so I used my Protools LE DAW to record the outputs of a single hall and motor phase output as audio signals. Here's what it looks like http://davedesign.smugmug.com/Machines/Turnigy-CA120-70/17641606_qp2HBV#1363715478_5VCBbCX

The "scope trace" is the 4th image in the gallery, motor phase on top, hall on the bottom of the image (note the phase of the hall is 180 out from the motor). The hall output and phase are roughly aligned, at least they seem to be, but the duration of the hall is longer than the motor phase. I can't really draw any other conclusions from this "experiment". Should the hall fire precisely at the zero crossing of the phase, or should it straddle it (center of hall "on" pulse aligned to peak of motor phase)? I admit I'm completely clueless about hall timing. In the last hour or so of googling, I've found all sorts of interesting white papers and app notes on hall commutation of brushless motors, but can't seem to find a simple hall timing diagram I remember looking at a while ago. One of the app notes did mention that it's essentially impossible to maintain the perfect controller phase angle with only halls. Something about the ideal being 90 degrees, but hall commutated controllers in the real world varying from 60 to 120 degrees.

Aha! I found the image I remembered

And hey Luke, you're always welcome at the dave lab, I owe you several beers at least for the plasma burns and magic smoke your posts have saved me from. I'm near "The Plant" shopping center in SJ.

 

[Biff]

Ok, so I've been blindly dorking around with this motor and the hall placement. I don't have immediate access to a dso, so I used my Protools LE DAW to record the outputs of a single hall and motor phase output as audio signals. Here's what it looks like http://davedesign.smugmug.com/Machines/Turnigy-CA120-70/17641606_qp2HBV#1363715478_5VCBbCX

The "scope trace" is the 4th image in the gallery, motor phase on top, hall on the bottom of the image (note the phase of the hall is 180 out from the motor). The hall output and phase are roughly aligned, at least they seem to be, but the duration of the hall is longer than the motor phase. I can't really draw any other conclusions from this "experiment". Should the hall fire precisely at the zero crossing of the phase, or should it straddle it (center of hall "on" pulse aligned to peak of motor phase)? I admit I'm completely clueless about hall timing. In the last hour or so of googling, I've found all sorts of interesting white papers and app notes on hall commutation of brushless motors, but can't seem to find a simple hall timing diagram I remember looking at a while ago. One of the app notes did mention that it's essentially impossible to maintain the perfect controller phase angle with only halls. Something about the ideal being 90 degrees, but hall commutated controllers in the real world varying from 60 to 120 degrees.

Aha! I found the image I remembered

And hey Luke, you're always welcome at the dave lab, I owe you several beers at least for the plasma burns and magic smoke your posts have saved me from. I'm near "The Plant" shopping center in SJ.

If you are hooking up to a Kelly that graph is slightly wrong (If my memory serves me right) you want H2 and H3 swapped. So your H1 should be a square representation of M3-M1 (which it is) but your H2 should be a square representation of M1-M2 and your H3 is a square representation of M2-M3. (The kelly uses ABC rather than 123) For those of you using sevcon Gen 4, your hall timing should be Hall U = square representation of M2-M1, Hall V = square representation of M3-M2, Hall W = square representation of M1-M3 (in the sevcon manual it actually says you want the hall A waveform to be the inverse square representation of M1-M2, but I think that is more confusing)

-ryan

 

[liveforphysics]

And hey Luke, you're always welcome at the dave lab, I owe you several beers at least for the plasma burns and magic smoke your posts have saved me from. I'm near "The Plant" shopping center in SJ.

I don't drink, but I'm certainly down for meeting up sometime.

If you're just in San Jose, I could throw an extra couple kwhr's of LiPo in a backpack and ride over hwy17 on my ebike to visit.

 

[Arlo1]

I don't drink,

I don't think Im the only one who knows thats proly a very good thing! And I meen that in many ways.........

 

[dave.com]

And hey Luke, you're always welcome at the dave lab, I owe you several beers at least for the plasma burns and magic smoke your posts have saved me from. I'm near "The Plant" shopping center in SJ.

I don't drink, but I'm certainly down for meeting up sometime.

If you're just in San Jose, I could throw an extra couple kwhr's of LiPo in a backpack and ride over hwy17 on my ebike to visit.

Re drink, yeah, I could probably figure out this hall timing if I hadn't killed so many brain cells.

Another good reason to ride over the hill, the De Anza electronics flea market is next saturday. It's held in the college parking lot, and there is always lots of cool tech kit there. Starts early am. There is a remote chance I'll have that motor in something by then and try to ride it over. In any case, I'll be there.

17 on a bicycle based ebike?

 

[dave.com]

17 on a bicycle based ebike?

Duuuh, nevermind. I just watched LFP's laguna seca video.

Er, what I meant was, watch out for the CHP radar .

Back to halls, trying to find enough clip leads to hook up all 3 motor phases and halls to my digi 002r to get a better idea of precise hall timing. I'm pretty sure the hall signals on 2 phases are being distorted by the motor can.

 

[dave.com]

More on halls. Again, ES has saved me about another 500 fruitless hours machining parts that won't work and trying to set my external hall sensor timing while the sensors are overcome with the stator's magnetic field. I've been looking into alternate commutation sensor schemes. A lot of good info in Burtie's thread here http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=28966. In addition to Burtie's clever optical design, It seems there are several products that are essentially packaged encoders with commutation outputs. 14 pole devices, which is what the 120-70 needs, are a little hard to find, however. The RMC RMC22 is available with 14 poles, but being made by a Renishaw division, is probably not available unless we can do a group buy for 10,000 and are Siemens. Ok, how about this one http://www.anaheimautomation.com/pr...rotary-item.php?sID=372&pt=i&tID=1063&cID=422 There is also one with a 3/8" shaft (and a much higher resolution encoder) for a few bucks more. Another supplier, but still not available through their e-commerce site http://www.wolfautomation.com/produ...ic-encoder-high-resolution-outputbrtimken-m15

And after all that, I finally arrive at Timken's e-commerce site http://www.timkenencoders.com/servlet/the-1/Magnetic-Encoder,-Encoder,-Modular/Detail, and they don't offer the 14 pole option on their configurator. I would bet it's just firmware, they are just dividing down the encoder pulses to generate the commutation signals.

So what's left, make a little 12/14 timing belt drive? Maybe I'll call these guys tomorrow and see if they will make the 14 pole configuration.

 

[bigmoose]

Dave, don't give up on the RLS components. I was successful securing a sample kit from them about a year ago.

My biggest design problem was I couldn't come up with an elegant and sturdy way to mount a shaft end sensor on the outrunner topology motor. I just didn't package well, in fact, I couldn't package it at all.

 

[dave.com]

And yet another shaft encoder that has commutation outputs. They might be convinced to make a 14 pole device for us, though they only list 4, 6 and 8. http://usdigital.com/products/encoders/incremental/rotary/kit/ec35

 

[dave.com]

Dave, don't give up on the RLS components. I was successful securing a sample kit from them about a year ago.

My biggest design problem was I couldn't come up with an elegant and sturdy way to mount a shaft end sensor on the outrunner topology motor. I just didn't package well, in fact, I couldn't package it at all.

Well, I'm willing to do almost anything to avoid milling the stator, so I may give them a call. The parent company, Renishaw is actually pretty good to deal with, once you get over their pricing (how 'bout $1k for twelve kilobytes of software - Inspection Plus probing software for CNC machines).

I'm hoping US Digital or Timken can come up with something, because I'm pretty sure they will be cheaper.

 

[olaf-lampe]

How about using an epoxidisk mounted on the shaft with 14 tiny magnets in it? The statorfield wouldn't disturb there IMHO, or at least the field would be in the wrong direction.

 

[Vadim]

What about an optical encoder instead? Just white and black paint on the outside of the can.
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=425-2044-5-ND
Digital output (.4v low, 4.5v on high)