Vibrations because of BEMF Harmonics?

hias9

1 kW
Joined
Jul 11, 2018
Messages
422
On a 23 pole pair 51 slot direct drive hub motor I have vibrations at around 650rpm (250Hz), but only if phase current is still above approximately 300 phase amps at that moment.
If I only accelerate with 50% or 70% throttle for example, there are no vibrations because phase current is lower.

I am using a FOC controller. In FOC control mode, the vibrations are stronger, but only present over a smaller rpm range (maybe 630-670rpm).
Tuning the proportional and integral gains of the phase-locked loop has some influence on the vibrations, but they cannot be eliminated by tuning the gains.
In hall based control mode the vibrations are softer, but present over a bigger rpm range (maybe 600-700rpm).

What do you think could be the cause of these vibrations?
 
I'm using a typical 12/14 Brushless 6384 motor and get some very interesting harmonics too.

And the behavior changes with torque load and rpm as well as battery voltage and the temperature of all aspects.

In my scenario everything works best at full load, full power and a hot day because I gradually and incrementally made choices that got me to that point but it was entirely by intuition and luck.

Seriously hardcore people get an oscilloscope and actually get deep inside what is going on.

For me as long as it works when I need it most I'm content.
 
If you think they are electrical harmonics you can get a meter to measure them. Tell us the value, and the order. Measure them. The presence of harmonics on the AC output of an inverter may indicate a fault within the inverter. THD should not be greater than 5%. Yes, the presence of a large amount of total harmonic distortion might present the case for a poorly running. inefficient motor.

I use a Seaward Solar power analyzer.

Probably not electric harmonics but a critical rpm resonance. Maybe, though.
 
The motor is running great and efficient. I have fine tuned all proportional and integral gains.
It's not an inverter fault because changing the inverter did not change it.
I guess I would need to measure the waveform while riding (because the vibration only occur at high load around 75kph) ?
I thought about buying a raspberry pi and mcc118/128 and measuring the waveform while riding. But what would it be good for? By tuning the gains it's not possible to eliminate the vibrations.
 
hias9 said:
I guess I would need to measure the waveform while riding (because the vibration only occur at high load around 75kph) ?

I dont know. You could put it on a stand and apply brake and throttle at the same time with a old spare rotor and a set of old pads. Load it up under test. You know? it could be a million things. You must start measuring things.
 
I tuned some low power motors like this, but for over 300 phase amps vs. bicycle brake pads at 75kph+, I will probably need many sets of brake pads. We are talking about around 20 horsepowers of mechanical power against bicycle brake pads.
I would also need to buy measuring equipment.

What exactly do you think causes these vibrations?
What exactly should I measure? Phase current waveform?
 
On the very high level you need to comprehend what is going on.

We tend to think of electricity as particles which are electrons and on some level that works in that in absolute numbers we can say a certain amount of energy passes at any point in time.

But in reality the actual way things works is closer to Tesla and the idea if the Aether.

So wave energy is real in that pressure of the Aether runs things but you need to think of all of this as a sort of elastic thing that vibrates as it propogates.

We are dealing with something where resonances are huge and define performance.

It''s like Two Stroke technology long ago where the tuned exhaust recognizes hidden realities that the "profane" are unaware of.

An oscillascope would be your ideal way of watching what is going on.

Let me put it in blunt terms... you will not know much without better equipment to measure.
 
So you mean it is probably caused by the magnetostriction of the stator laminations?

I was thinking about buying a mcc118 or 128 and a raspberry pi.
 
…ok..

I think the aether and elastic properties could be placed in the same spot as hobbits and trolls when it comes to your issue.

If you want to know the issue then you could probably get a spectrum analyzer and try to find which frequencies/harmonics the drive has. One or several of them resonate at the rpm you notice.

I’d guess the harmonics of the motor vs the controller switching frequency or commutation frequency has an interaction
 
hias9 said:
What exactly should I measure? Phase current waveform?

I dont mean like hang on the brake for an hour at full throttle. Ten seconds should do ya. Grease the sacrificial pads of the dyno if you want to test longer. two or three runs through the trouble zone. Looking for voltage or current peaks, ect. Abnormalities. Or just try the " change one variable and see how it effects the readings" method.

I check every motor and every inverter installation with a Power Analyzer. I use a Seaward Solar Power analyzer. I just look for anomalies in the current harmonics at the highest voltage measured at the motor (RPM) speed. Measure for THD with the machine. I dont know how it does it, it just does. Tells me a percentage THD. And then you can look for specific orders of harmonic if they are present. The electric company will fine you if you mess up your neighbors power on the same transformer, from heavy loads with prevalent harmonics: (rattling phase converters), and poor quality inverted power will run appliances like trash... so to do solar installation troubleshooting is what my little clamp meter is made for. This is primarily why I have it. However, it, can measure upwards of 99kHz. So it is useful for DC PWM currents.

Quote from the wiki on Harmonics.
Motors
Electric motors experience losses due to hysteresis and eddy currents set up in the iron core of the motor. These are proportional to the frequency of the current. Since the harmonics are at higher frequencies, they produce higher core losses in a motor than the power frequency would. This results in increased heating of the motor core, which (if excessive) can shorten the life of the motor. The 5th harmonic causes a CEMF (counter electromotive force) in large motors which acts in the opposite direction of rotation. The CEMF is not large enough to counteract the rotation; however it does play a small role in the resulting rotating speed of the motor.
https://www.google.com/search?client=firefox-b-1-d&q=Power+Analyzer

https://www.hbm.com/en/8099/what-is-a-power-analyzer-definition/

https://en.wikipedia.org/wiki/Harmonics_(electrical_power)

https://www.seaward-groupusa.com/products/solar-pv-testing-tools/solar-power-clamp
 
larsb said:
I think the aether and elastic properties could be placed in the same spot as hobbits and trolls when it comes to your issue.

Let me bring you up to speed on what ACTUALLY happened to Tesla and his fellow real scientists of the late 1800's.

First you need to know the larger picture where the "Robber Barons" of the time (but today now go to Davos and the World Economc Forum) were gaining more power over the world. These people took over the scientific community.

A big controversy existed in the scientific community which came about over a philosophy called "Logical Positivism". (1920's)

To make this short the philosophy said:

"If it cannot be measured and narrowly defined it does NOT exist."

So in essence in the 1920's we returned to a kind of "Greek Atomism" with an idea of "particles". (Quantify... Quantum)

Understand that Tesla did NOT envision electricity as particles and none of the other major thinkers of the time did either.

What this means is we sort of fell down into a deep pit in the 1920's and have not yet gotten out.

----------------------------

As for things like harmonics in a complex system it gets to a point where trying to imagine things as particles seems a bit silly.

Little electron particles are not flying around like billiard balls... I mean it's all kind of a joke if you think deeply on it.

What we deal with in these harmonics is pressures of electronic potentials experiencing hysteresis.

Hysteresis is the Aether we might say.

The universe has Hysteresis everywhere... the Speed of Light limitation is simply the Hysteresis caused by the Permittivity of Free Space which is the background lowest Aether quantity. (it's even written in the definitions)

Light when highly compressed becomes Matter and if we get a nuclear war we will see Matter returned as Light.

And fairly recently in the lab they used some high powered lasers acting in a resonance pattern to actually create Matter out of Light.

So we are at the stage where we can convert in both directions.
 
Today I limited power and phase current to 80% of what I usually set it to and had it at full throttle while pulling the brake lever hard.
After about 3 seconds I released the brake lever. The brake was not able to get speed below 75kph and speed already started to rise after 2 seconds.
After releasing the brake, there was smoke coming from the brake and now (half an hour later) it is still smelling bad.
This method however works fine for tuning low power bikes.
Brake is a Magura Gustav by the way, probably one of the strongest bicycle brakes ever built.
I ordered an oscilloscope and have some current sensors here. Maybe I will try again for 2 seconds when the oscilloscope has arrived.
Or mount it at the handlebar :D

While riding I also noticed that lowering the switching frequency moves the vibrations to lower rpms.
 
Today I mounted a small digital oscilloscope on the handlebar connected to a current sensor that measures phase current.
The waveform was a normal sine wave also while the vibrations occurred.

Earlier I mentioned that lowering switching frequency makes the vibrations occur at less rpm.

But if it is an interaction of motor harmonics vs. switching frequency, why do they only occur when phase current is higher than approximately 300A?
 
What FOC controller? Are you using VESC?

This kind of behaviour is frequently caused by oddities and gremlins in the code. The way different controllers deal with things like saturation of the stator, motor harmonics etc varies. Can also be caused by noise on hall sensors, but if it happens with hall and sensorless is most likely a current control loop issue.

Without specific actual data, no one will really be able to guess much.

Try a different controller?
 
No, I am using an ASI controller.

Some vibrations I had on other motors can be tuned away by tuning current regulator loop or phase-locked loop.
In this case however, tuning current regulator loop has no effect at all on these vibrations.
Tuning phase-locked loop has some effect. It changes the quality of the vibrations a bit, but they cannot be tuned away.
I only use the hall sensors up to about 2kph, above that the motor runs sensorless. Vibrations are around 75kph (but only when phase current is higher than approximately 300 phase amps at that moment). So it's not noise from the hall sensors.
When using hall based control instead, the vibrations at 75kph are softer, but over a bigger rpm range (maybe 600-700rpm instead of 625-675rpm).

Changing to a different controller is not really an option because I am not aware of any FOC controller that has similar or higher power output at the same size and is compatible with the Eggrider display.
The nucular 24F has similar power output to the BAC 4000, but is not compatible with the Eggrider display. Any controller with bigger dimensions than a BAC 4000 will probably not fit.

Also I noticed that lowering switching frequency also moves the vibrations to lower rpms.

Could it be winding harmonics resonating with switching frequency?
What would I need to measure to verify that?
What specific actual data do you need?
 
Do you have a lot of winding harmonics? If you have a scope, check the back EMF with the inverter disconnected and spinning the wheel by hand.

Number of possibilities...

Driving the motor into saturation so the inductance drops and the current spikes

Nasty BEMF harmonics which inject current wave forms into the controllers sinusoïdal output

Does the vibration stop below a certain current threshold?

You might still not have the PID tuned properly. PID tuning is deterministic, you shouldn't need to fiddle. Search pole zero cancellation or consider the amount of voltage response that could cause correction of a current error in one pwm cycle (for P) and then the I term is determined by the voltage required to drive against the resistance of the motor.

There's a huge space to tune PID, you may not be anywhere near the right answer... Or maybe you are and it's not that.
 
The controller uses only PI gains, not PID gains.
Yes, I tuned current regulator loop and phase-locked loop gains properly. These vibrations cannot be tuned away however.
The rpm range where those vibrations happen depends on the switching frequency.
Lowering the switching frequency by 15% makes them occur at about 15% lower rpms, so I guess something is resonating with the switching frequency.

They occur both in hall based and sensorless control mode. In hall based mode the vibrations are softer, but over a larger rpm range than in sensorless mode.

The vibrations are only there if phase current is higher than about 300 phase amps at that moment.
Yesterday I noticed that no load current draw is oscillating when within that rpm range. Above a certain rpm level it stops oscillating. Lowering switching frequency by 15% lowers this rpm level by 15%.
Increasing current regulator Kp lowers/eliminates the no load current oscillation, but has no effect on the vibrations under load (when over approximately 300 phase amps) and has other negative effects.

Very probably something is resonating with the switching frequency. Maybe winding harmonics? What do I need to measure to verify this theory?
 
Scope the open circuit backemf to find the harmonics. Some scopes have an FFT.

Then scope the exact switching frequency.

Can you measure the current and pwmb waveforms for it when vibrating? You said it oscillates at no load at the speed.

What frequency are the vibrations? You say the wheel is about 650rpm, so 15kerpm. 250hz. Is that the vibration frequency? A multiple of this (odd harmonic)?
 
I have a simple digital one channel scope. I made it portable and had it mounted at the handlebar while riding. It was connected to a current sensor on one phase wire. Waveform looked like a normal sine wave also while the vibrations occurred.

How can I scope bemf voltage at 75kph? When I turn the wheel with the controller disconnected it’s a perfect sine wave, but I can only measure at low speeds when I turn the wheel by hand.

What means pwmb waveform? How do I measure it?
I know phase current oscillates at that speed no load because of the values measured by the current sensors inside the controller.
Increasing current regulator Kp eliminates the no load oscillation, but has no effect on the vibrations under load (and has other negative effects).

The vibrations are at around 220-240Hz. Switching frequency is 16kHz. Lowering switching frequency moves the vibrations to a lower frequency.
 
hias9 said:
I have a simple digital one channel scope. I made it portable and had it mounted at the handlebar while riding. It was connected to a current sensor on one phase wire. Waveform looked like a normal sine wave also while the vibrations occurred.

How can I scope bemf voltage at 75kph? When I turn the wheel with the controller disconnected it’s a perfect sine wave, but I can only measure at low speeds when I turn the wheel by hand.

What means pwmb waveform? How do I measure it?
I know phase current oscillates at that speed no load because of the values measured by the current sensors inside the controller.
Increasing current regulator Kp eliminates the no load oscillation, but has no effect on the vibrations under load (and has other negative effects).

The vibrations are at around 220-240Hz. Switching frequency is 16kHz. Lowering switching frequency moves the vibrations to a lower frequency.

You don't need to scope it at 75kmph. Just turning by hand enough.

You seem very reluctant to post any actual information. I don't understand why. If you want help, try actually posting information. Like... Screenshots/ settings/waveforms/videos.

Otherwise you not going to get any useful help.
 
I keep posting old information multiple times because most people do not read the whole thread.
As already mentioned, scoping bemf voltage open circuit gives a perfect sine waveform when turning the motor by hand. No harmonics noticeable.

What should I measure and take a picture/video of?
 
Does the noise go from silent motor operation to squealing loudly at an instant transition from quiet to loud?

If so, I've experienced this from poor current control loop tuning.
 
No. Current regulator loop and PLL loop has been properly tuned. Current regulator Kp/Ki does not affect the vibrations at all. PLL loop Kp/Ki has some effect on the vibrations, but the vibrations cannot be tuned away.

When using hall based mode instead of sensorless FOC, the vibrations feel softer, but over a bigger rpm range (maybe 600-650 instead of 615-635).
When feeding the motor with less than 330 phase amps at that moment, vibrations are not noticeable, at 400 they are clearly noticeable, at 430 they are very annoying, and acceleration is affected (not smooth at all any more) when inside that rpm range.
Could this be related to saturation?
 
My guess would be a mechanical resonance in the stator. I have a motor that does this. There is also some magnetostriction happening in the stator iron, so the stator can ring at the right RPM.
 
@fechter: Does your motor that does this also have 46 poles 51 slots and do you only have those vibrations at high phase amps?
If it's because of magnetostriction, should using laminations meant for baking help?
 
Back
Top