ERPM limits of controllers, an observation

bearing

10 kW
Joined
Dec 23, 2008
Messages
640
This may be known, but I haven't read it anywhere, so I thought I would put it out here to see what you say.

I've seen that controllers have a limit on drive frequency to the motor, the so called Electrical RPM(ERPM) limit. Since microcontrollers are fast, it seems odd to me that the microcontroller would cause the limit. After looking at some numbers it seems that it could actually be the PWM frequency limiting the ERPM.

Since a BLDC is commutated 6 times per cycle, and there is 60 ERPM per cycle, the ERPM to commutation frequency has a simple relationship of 10:1. (I hope I'm correct?)

http://www.kellycontroller.com writes:
Up to 40,000 electric RPM standard. Optional high speed 70,000 ERPM, and ultra high speed 100,000 ERPM. (Electric RPM = mechanical RPM * motor pole pairs).

Optional Ultra High Speed to 100,000 ERPM
Price: $99.00

1. Dual 32 bit micros, with 33.3kHz PWM
2. Up to 100,000 Electric RPM

100.000 ERPM is a commutation frequency of 10kHz. Since the PWM frequency is 33kHz, it seems Kelly limits the ERPM to a about 3 PWM periods per commutation period.

Hobbyking writes:
Max Speed;
2 Pole: 210,000rpm
6 Pole: 70,000rpm
12 pole: 35,000rpm
http://www.hobbyking.com/hobbyking/store/__8921__TURNIGY_K_Force_120A_HV_OPTO_5_12S_Brushless_ESC.html

210 kERPM is a commutation frequency of 21kHz. The PWM frequency is 24kHz, which makes it limit to about 1 PWM-period per commutation period.

So, to me it seems the physical limit of a controller using PWM at 20kHz is 200,000 kERPM, and the only way to increase the ERPM is to increase the PWM frequency (which increases losses in transistors), or changing the strategy of driving the motor away from the PWM:ing 3-phase bridge.
 
I see that also the Adaptto also has ERPM limits of 70,000 ERPM.

Would that mean the Adatto maxi-e would be able to run an Astro motor with stock Adaptto firmware?
Astro 3220 6T has 105 Rpm per volt. So at lets say 20s hot off the charger @ 83v (4,15v max) would work just fine as long as Halls are added to the astro? That gives rpm of 8715. So pretty close to 70K erpm or to be correct 69,720 ERPM.

Astro 3220 6T has 8 poles x 8715 motor rpm = 69720 ERPM.

As far as I can tell there should be no need to run RC firmware on the Adaptto in order for it to work with the Astro 3220 6T.

Then one would have 6 kw continuous and burst up to 14 kw according to Matt.

The fact that I have not seen any run this setup makes me think either my math is off the chart and completely wrong or there is more to it then just ERPM to determine if a controller will work with a certain motor?

I can easily make room for a reduction gear/jackshaft to get the optimum gear ratio for the wheel. But lets first focus on the motor/controller choice. Will this work for me, Astro 3220 motor run by an Adaptto maxi-e controller with stock firmware?
 
You might be right but then Astros website have wrong information. On their website is says 8 poles.

But if you are correct that would mean it would be even lower ERPM. It would slice my calculation in half.
So the astro 3220 6T would have an ERPM of 35,000. Well within the limits of the Adaptto.

So why is no one using the Adaptto for RC motors? It seems doable, yes?
 
macribs said:
You might be right but then Astros website have wrong information. On their website is says 8 poles.
Unless the poles are segmented (they aren't on the Astro), one magnet is one pole. So, 8 magnets is 8 poles or 4 pole pairs.
 
Ok so my math still stands and my first count seem correct.
But how does this go in the real world? Have anyone run high powered RC motors like astro 3220 with a conventional controller like the Adaptto maxi-e? As far as I can see it should work. But will it work you think?

And if it will not work will it work if Adaptto is flashed with RC firmware? Not even sure if the RC firmware that is floating around does work for the Adaptto - it seems the "RC firmware" was originally made for the Sabvoton.
 
macribs said:
Ok so my math still stands and my first count seem correct.
No, because, as riba said, the calculation is based on pole pairs and not poles.

rpm x n. pole pairs gives you the erpm

rpm / 60 * n. pole pairs gives you the commutation frequency in hertz (cycle per second)
 
Well wouldn't that just mean that the ERPM would be half of what I first calculated?
If it is pair of poles and that makes 4. It should then be motor RPM x 4 pole pair, correct?
And that would mean astro 3220 has ERPM of around 35,000.

Look I am no engineer or mathematician so if if I am doing something fundamentally wrong here let me know.
But as I read I think on adaptto site ERPM = motor rpm x number of poles.

If that is correct it seems math will turn out to either about 70,000 ERPM or half that 35,000 E-rpm.
 
macribs said:
Look I am no engineer or mathematician so if if I am doing something fundamentally wrong here let me know.
But as I read I think on adaptto site ERPM = motor rpm x number of poles.
macribs said:
AdapttoWebsiteIndexPage said:
Ability to use in conjunction with high-speed engines to 70000 ERPM (1ERPM = rev / min motor * number of pole pairs)
As you quoted in the other thread, it's pole pairs, not poles.
 
OK. Well in that case I guess even the 4T motor would work well within limits.
No gain from over volting those astros seems to be consensus so I will drop that.
 
bearing said:
100.000 ERPM is a commutation frequency of 10kHz. Since the PWM frequency is 33kHz, it seems Kelly limits the ERPM to a about 3 PWM periods per commutation period.

The units of ERPM is cycles-per-minute, whereas Herz is cycles-per-second. So:

100,000 ERPM / 60 = 1667Hz commutation frequency

The motor speed being (assuming 10 pole pairs) 100,000 / 10 = 10,000 rpm

Or

10,000rpm / 60 = 167 rev/sec
 
Punx0r said:
bearing said:
100.000 ERPM is a commutation frequency of 10kHz. Since the PWM frequency is 33kHz, it seems Kelly limits the ERPM to a about 3 PWM periods per commutation period.

The units of ERPM is cycles-per-minute, whereas Herz is cycles-per-second. So:

100,000 ERPM / 60 = 1667Hz commutation frequency

The motor speed being (assuming 10 pole pairs) 100,000 / 10 = 10,000 rpm

Or

10,000rpm / 60 = 167 rev/sec


It was a long time since I wrote that, but I can't see anything wrong with my (simple) math, even though it might be a bit obsolete with regards to vector control (which is being discussed in the thread).

100000 {ERPM] / 60 [Hz/ERPM] = 1667 Hz motor frequency (not commutation frequency)
6 commutation steps per electrical cycle-> 10000 commutation steps per second. Easier to just divide ERPM by 10 to get commutation frequency.

Vector controllers doesn't necessarily use discrete communication steps in it's control software, but rather a continuous transfer function between the current vector measured and the voltage vector (PWM) produced. But the vector controllers have a similar upper motor speed limit of course, because if the PWM frequency is low, the resolution of the produced sine-like PWM gets coarse at higher motor frequencies (ERPM), which makes it hard for the controller to do it's work / maintain sync.
 
I am confused by your post, although I could be wrong as much of this is new to me.

What are you saying is the difference between motor frequency and commutation frequency? I have not heard before of the term "motor frequency". I know only rotational speed and commutation frequency - assuming that the controller must deliver one pulse-per-pole-per-revolution, so if 10 poles = ten pulses per 1 revolution of the motor.

The part I don't understand is how you got from 100,000 ERPM to 10KHz ;)
 
The 6-step trapezoidal controller, which my original post was about, will commutate 6 times per electrical cycle.

http://www.pmdcorp.com/images/2013/january_2013_deep_dive/F5_Commutation%20Schemes%20-%20final.jpg

That's why I multiplied the motor frequency by 6 to get the commutation frequency. 1667Hz * 6 = 10kHz.


I don't know in detail how these controllers work. When the hall sensors says that the controller needs to commutate, will the controller stop the current PWM period instantly, and start a fresh PWM period? or will it wait until the current PWM period ends until it commutates?
I have always suspected the latter.

From the page where I found the PIC above:
"Trapezoidal commutation generally works well, but has discontinuities at the Hall transition points. This increases torque ripple and vibration"
http://www.pmdcorp.com/news/articles/html/motion_control_amplifier_deep_dive.cfm

This quote indicates the same as I have suspected. That the controller commutates at the start of a PWM period. This means that it will be a delay until commutation of somewhere between 0 and 50microseconds (with 20kHz PWM frequency). At low speeds, where the commutation period is several milliseconds, this jitter is negliable. At high speed, say 10kHz commutation frequency, which is a commutation period of 100 microseconds, this jitter is very significant.
 
Back
Top