Motor comparisons [CA120 et al.]

[h0tr0d]

Me thinks it's parasitic inductance which causes ringing. A peak in the ringing then overvolts the FET and blows it. Therefore my preference for snubbers.

Instantly thought of you Lebowski...
[youtube]8S-C4eclB9Q[/youtube]

 

[Lebowski]

Yep, that's it

 

[toolman2]

dam you circuit, you are right about the phase current ripple at no load.

i just ran up a ca120 on an 80v pack and the losses at 1000-2000rpm are massive, it uses around 500w instead of around 100w that it used to, and the change is because of the high battery voltage, something changes pretty drastically

i allways thought (and allways found in practice to, till this ) that you can say double the battery volts and and with fixed phase amps that nothing really changes, and you get the same watts nearly the same efficiency (at a low fixed speed) it just uses half the amps and double the volts at the battery and the controller chops it all up with pwm to give the same end result either way, a bit more loss compared to say running very low voltage and 100% duty, as there a few tens of watts lost to fet switching etc but same enough in watts regardless of battery voltage -phase currents being equal.

so at 45-55v that i have mostly run on it was barely an issue and could hide away unnoticed, but at 80v i cant even get a 1000rpm no load idle without a 100degc motor, its ridiculous -although it precisely matches what circuit had been banging on about, so yes a round of 'i told you so' is clearly in order. :|

it now has about 50a of phase current with no load fairly similar regardless of throttle, its just a 500w+ power sap. it dropped to 8a with a 26uh inductor on each phase and that kinda fixes it (but would not be cool for 200A loaded up) so i though id best have a look with the scope and i find the pwm on the kelly keb high speed is only 16.6khz when in sure i was told it was a much improved 33khz, not overly impressed with the whole situation i spoke to fany and to my disbelief the kellys run the low 16khz pwm for low motor speed and then switch to the 33khz at around 30000erpm and above. i then whinged and moaned at him about this for quite a while untill he agreed to get the engineers to write new firmware to hold 33k allways. he reckoned he might be able to send the new firmware through to try out in a few days. so im real keen to try it, and ill let you all know how it goes..

aaand i dont spose folks have any pointers on building a 3 phase inductor? or how about pwm'ing the battery down in voltage to always be just a few more volts than the motor needs? -maby a bit silly. and im not wrong in saying we need 33khz more at low rpm that high, am i?

fany said that the kellys stay at 16khz at low rpm to save on switching losses, but i have my doubts they stood to lose half a kw.. so im sorry fany but thats why i had such a big cry, and anyway, it felt great to just 'let it all out'...

 

[Martini]

Hello ,

I just receive from the manufactor of the CA motor this new model CA 154-100

Basic parameter::
1、Silicon steel：dia.139mm，laminating thickness: 50mm.，36 slots
2、Magnet：8.5*4.5*50 N42SH
3、Dimension：φ154X104mm, excluding motor mount
4、Weight：7kg
5、Power：max. 30kw
6、Working voltage：＜100V

 

[Njay]

Regarding the video, even with the snubber, that system still has a problem, the very high voltage peak when the MOSFET turns off (roughly 2x Vbat!). The snubber is damping the ringing energy but not preventing the initial "kick" that puts the LC resonating. Not sure what's the cause of the kick on that circuit, seems like if the "free wheeling diode" is taking too long to act - that's what the guy should have tried to fix.

 

[Arlo1]

dam you circuit, you are right about the phase current ripple at no load.

i just ran up a ca120 on an 80v pack and the losses at 1000-2000rpm are massive, it uses around 500w instead of around 100w that it used to, and the change is because of the high battery voltage, something changes pretty drastically

i allways thought (and allways found in practice to, till this ) that you can say double the battery volts and and with fixed phase amps that nothing really changes, and you get the same watts nearly the same efficiency (at a low fixed speed) it just uses half the amps and double the volts at the battery and the controller chops it all up with pwm to give the same end result either way, a bit more loss compared to say running very low voltage and 100% duty, as there a few tens of watts lost to fet switching etc but same enough in watts regardless of battery voltage -phase currents being equal.

so at 45-55v that i have mostly run on it was barely an issue and could hide away unnoticed, but at 80v i cant even get a 1000rpm no load idle without a 100degc motor, its ridiculous -although it precisely matches what circuit had been banging on about, so yes a round of 'i told you so' is clearly in order. :|

Interesting stuff.

it now has about 50a of phase current with no load fairly similar regardless of throttle, its just a 500w+ power sap. it dropped to 8a with a 26uh inductor on each phase and that kinda fixes it (but would not be cool for 200A loaded up) so i though id best have a look with the scope and i find the pwm on the kelly keb high speed is only 16.6khz when in sure i was told it was a much improved 33khz, not overly impressed with the whole situation i spoke to fany and to my disbelief the kellys run the low 16khz pwm for low motor speed and then switch to the 33khz at around 30000erpm and above. i then whinged and moaned at him about this for quite a while untill he agreed to get the engineers to write new firmware to hold 33k allways. he reckoned he might be able to send the new firmware through to try out in a few days. so im real keen to try it, and ill let you all know how it goes..

Im really supprised they did this. I know people think they want the oposite. Hi pwm freqency at low rpm and then reduce the frequency at higher rpm but I want to do more studdiying and testing on this to decide what is best. I think haveing the same pwm frequency is the best way to go.

aaand i dont spose folks have any pointers on building a 3 phase inductor? or how about pwm'ing the battery down in voltage to always be just a few more volts than the motor needs? -maby a bit silly. and im not wrong in saying we need 33khz more at low rpm that high, am i?

I think the way to do it is 3 separate inductors. I have a few to play with. I got a set of hybrid air/ferrite core 27uh inductors with a spec of 20 uf at 200 amps I think it was and then I also made 2 sets of air core inductors. This is how I was first able to ride colossus. Just google inductor calculator. You will also find all inductors in series add up so if you have one with every phase you will have you motor 26uh plus the 2 inductors in series with the phase you are powering so I would be 3 inductors of ~30uh would work great for you.
I used air core because you don't loose as much energy in heating the metal in the center. If you find a multi layer calculator you will find the multilayers help make it more compact and more inductance with less copper.

fany said that the kellys stay at 16khz at low rpm to save on switching losses, but i have my doubts they stood to lose half a kw.. so im sorry fany but thats why i had such a big cry, and anyway, it felt great to just 'let it all out'...

Its really good they are working with you on this though its not often you get this privilege.

 

[Njay]

Im really supprised they did this. I know people think they want the oposite. Hi pwm freqency at low rpm and then reduce the frequency at higher rpm but I want to do more studdiying and testing on this to decide what is best. I think haveing the same pwm frequency is the best way to go.

It's at low rpm that you have higher currents (and typically less cooling). Therefore, the less times you switch per time unit, the lower the losses (MOSFET inductive switching loss is also proportional to current), and you let more time to cool between swithing (and heat losses are proportional to the square of the current). I've eared other controllers do it, like a Curtis or something that goes into 1.5KHz PWM at low speed.

 

[h0tr0d]

This a multiple variable problem that involves 2 pretty complex parts, motor and controller.

On the controller side if you have low PWM freq and low e-rpm (therefore low BEMF), the applied voltage to the motor inductor will be high (battery voltage - BEMF).
Using BigMoose's formula I = (V/R)*(1-exp(-R/L)t).
You'll have low switching losses and very high current ripple. So in a controller it's good to have low PWM freq to balance heat losses from conduction vs switching.
your typical "controller guru" will want that.

On the motor side, with the above scenario, high current ripple will have a tremendous impact, power losses vary with the square of amps. Example for 100A mean phase amps:
No current ripple -> 100^2 =10'000
High current ripple (oscillating between 200 and 0A= mean phase amps also 100A)-> (200^2 + 0^2)/2= 20'000 (That's double the losses with the same mean phase amps!!!!!!!!!)

Toolman experienced this first hand. That why I say less then 100kHz PWM it's no good for a low inductance motor and Circuit says no less then 200kHz!

Another practical example is Thud's rewound CA120. Instead of 3 turn in delta (150kv), he's done 5turn in wye (50kv and almost double the copper weight). So 5 turn is increment of 66.6% (5t/3t) in inductance times the delta to wye transition that multiplies inductance by 3. That's 5x more inductance!!!
Apply that to the above formula you'll have 5x LESS current ripple, therefore losses.

I have re-wound the CA-120 a few times....you can easily get the kv down in the 50 range....no load draws on 12cells drop to less than 1 amp.

from 13A no load losses to less then 1A with a "different" motor, same controller. 13x times less losses no load, not only 5x as shown above because we didn't calc eddy current losses, that vary with the square of rpm (a 50kv motor spins 3x times slower then a 150kv with the same Volts, right? ) not to mention hysteresis losses and other stuff...

 

[zappy]

and now has 1/3 the output as well!

 

[h0tr0d]

and now has 1/3 the output as well!

Of course, "there are no free lunches"!
BUT since Thud put double the copper, he can put double the amps... So 2/3 of the output (probably more because less eddy/hysteresis/mechanical losses).
EDIT: With the big advantage of driving that with a "normal" controller!

I wanted to explain that your typical chinese 16kHz controller only works "decently" at high voltage with a "high" inductance motor (like a hub motor with loads of wire turn per phase). Low inductance motors = fast controllers. Period! At least if you want decent efficiency at low e-rpm.

 

[Arlo1]

and now has 1/3 the output as well!

Of course, "there are no free lunches"!
BUT since Thud put double the copper, he can put double the amps... So 2/3 of the output (probably more because less eddy/hysteresis/mechanical losses).

Post edited I am sorry h0tr0d I didn't read what you said thoroughly.

 

[thepronghorn]

and now has 1/3 the output as well!

Of course, "there are no free lunches"!
BUT since Thud put double the copper, he can put double the amps... So 2/3 of the output (probably more because less eddy/hysteresis/mechanical losses).

Other way around to get the 1/3 kv he needs 3 times more turns per tooth. So that meens 3 times less copper in parallel.

Arlo you might be confusing yourself a little.

Remember that rewinding a motor does not necessarily change its torque potential, so [all things being equal] a motor with 3 times more turns per tooth will spin 3 times slower on the same voltage but make the same torque as before, resulting in 1/3 the power.

However, if one doubled the copper fill in the motor, the torque potential will be twice what it was before, resulting in a motor that spins 3 times slower but with twice the torque making 2/3 the power it previously did.

 

[h0tr0d]

Remember that rewinding a motor does not necessarily change its torque potential, so [all things being equal] a motor with 3 times more turns per tooth will spin 3 times slower on the same voltage but make the same torque as before, resulting in 1/3 the power.

3x more turns = 3x thinner copper (for the same copper fill) so 3x less amps BUT the same torque because 3x km.
So 3x times slower = 3x less power. If then you put 3x more volts (= same base rpm) = same original power.

But again, 3x more volts and 3x more inductance (due to 3x more turns) equals the same current ripple problems described before.
Again, no free lunches.

 

[Arlo1]

Remember that rewinding a motor does not necessarily change its torque potential, so [all things being equal] a motor with 3 times more turns per tooth will spin 3 times slower on the same voltage but make the same torque as before, resulting in 1/3 the power.

3x more turns = 3x thinner copper (for the same copper fill) so 3x less amps BUT the same torque because 3x km.
So 3x times slower = 3x less power. If then you put 3x more volts (= same base rpm) = same original power.

But again, 3x more volts and 3x more inductance (due to 3x more turns) equals the same current ripple problems described before.
Again, no free lunches.

Torque ripple does stay the same acording to circuit (I have to re read what he told me) but inductance goes up at a square of the amount of turns. 2x more turns = 4 times more inductance!

 

[Arlo1]

Arlo you might be confusing yourself a little.

Remember that rewinding a motor does not necessarily change its torque potential, so [all things being equal] a motor with 3 times more turns per tooth will spin 3 times slower on the same voltage but make the same torque as before, resulting in 1/3 the power.

However, if one doubled the copper fill in the motor, the torque potential will be twice what it was before, resulting in a motor that spins 3 times slower but with twice the torque making 2/3 the power it previously did.

I didn't confuse my self I just didn't read the whole post sorry I missed the fact Thud had 2x the copper fill. Post edited.

 

[h0tr0d]

Remember that rewinding a motor does not necessarily change its torque potential, so [all things being equal] a motor with 3 times more turns per tooth will spin 3 times slower on the same voltage but make the same torque as before, resulting in 1/3 the power.

3x more turns = 3x thinner copper (for the same copper fill) so 3x less amps BUT the same torque because 3x km.
So 3x times slower = 3x less power. If then you put 3x more volts (= same base rpm) = same original power.

But again, 3x more volts and 3x more inductance (due to 3x more turns) equals the same current ripple problems described before.
Again, no free lunches.

Torque ripple does stay the same acording to circuit (I have to re read what he told me) but inductance goes up at a square of the amount of turns. 2x more turns = 4 times more inductance!

This is one those times I can say "I'm glad I was proven wrong". Just read Duane Hanselman's book and indeed it varies with the square of turns. Sweet!

So a 3T delta c12070 ~20uH can be turned into a 6t Wye double copper 240uH! Same power, 4 times lower current ripple!

 

[Lebowski]

What stays the same though is the ripple current as a % of the phase current, and since this is what counts, no difference....

 

[toolman2]

fany said that the kellys stay at 16khz at low rpm to save on switching losses, but i have my doubts they stood to lose half a kw.. so im sorry fany but thats why i had such a big cry, and anyway, it felt great to just 'let it all out'...

Its really good they are working with you on this though its not often you get this privilege.

yep, agreed, im stoked they can change stuff that actually makes things work better, and thankfully they seem to be keen on the same motors as us these days. :wink:

 

[toolman2]

yep, i think it all works out the same there in the end, and the motor in question is the ex-lazair plane rewind done by dale (kilo1) with maxed out copper fill, its the same 3t as std, 84kv and wye, but its 9.4mohlms. its around 11uh and i was in a reasonable sweet spot with the 45v pack, prolly like thuds on more volts and less amps.

also, would it surprise anyone to know that this bike always came it with say 7ah showing on the ca and it took 9ah to full them back up, aaand when i dynod it on 4.6kw battery and it outputted 5.3kw.
so theres also some sort of high frequency dramas and or shunt inductance giving the ca errors? -so yep a few little things to deal with..

 

[circuit]

I told you so
Just for the record.

 

[SplinterOz]

I told you so
Just for the record.

You took your time..

 

[circuit]

I told you so
Just for the record.

You took your time..

There's no rush, people will talk about this for a yer or so.

 

[SplinterOz]

I told you so
Just for the record.

You took your time..

There's no rush, people will talk about this for a yer or so.

That should be in the es quotes.

 

[toolman2]

There's no rush, people will talk about this for a yer or so.

who are these people? and where is all this talk going on? :? best for anyone interested to chime in we can have that chat right here.

heres some scope shots of the goings on trying to capture the worst case of 80V and around 1800rpm no load, measuring volts of the phase vs batt neg on with the yellow and current via an inductance free 1mohlm shunt (and a fast diode to keep the waveform simple) on the same phase.

let me know if anyone wants better res or more info,
what do people see? -and what else, or what other way should i measure?

 

[h0tr0d]

What stays the same though is the ripple current as a % of the phase current, and since this is what counts, no difference....

True. Time-constant being the same right (L/R)?
BUT di/dt is half also, so electronics have double the time to put a term to a current spike, most relevant in your typical slow a** chinese controller.
Also 4x more inductance you have half ripple current (same percentage), so half the capacitor bank for ex.

Since we were talking about a double copper motor, current ripple losses are half (double time constant).