BLDC: where is my BEMF?

righteous · Apr 21, 2018

Gents,

These days I'm trying to identify my BEMF (BEMF=voltage induced in the coil by the moving magnetic field) in my BLDC system, but I can't seem to find it, maybe you can help me?

This is the circuit diagram for your kind reference

This is the scope shot of phase A
ch3 control signal for A(high side) ACTIVE LOW
ch4 control signal for A(low side) ACTIVE HIGH
ch1 and ch2 is the voltage across the coil
WHITE TRACE is ch1 - (minus) ch2

As you can see the calculated white trace resembles somewhat the theory taken from **broken link removed**

BUT... when I put ONE probe ONLY and connect it directly across the coil, the picture is totally different

So... where is my BEMF in the above scope shot?

Easy peasy · Apr 21, 2018

maximum BEMF is clamped by the diodes in the fets and the supply rail, when you turn a leg fully off, and the volts fly up (or down) that is the back emf at that instant. it is not a straight forward question as the BEMF is super posed on the voltages supplied by the other switching legs ...

- - - Updated - - -

worry more about your pulse widths and trying to get the speeds right (change of timing) for acceleration, deceleration, steady state speed control with variable loads ...

KlausST · Apr 22, 2018

Hi,

Seems like a touch on/touch off switch (with a single switch) is difficult without a MCU.

Sorry to have to correct you.
You talk about an inductor's BEMF.
A motor's BEMF (that is caused by the mechanically rotating magnetic field) in you configuration is difficult to measure, be ause it is influenced by the other terminal voltage pluse the BEMF voltage of the other coils.
https://www.st.com/resource/en/application_note/cd00020086.pdf
With a motor in star configuration each terminal truly belongs to one coil....if you now additionally have the star point you are able to measure the voltage of a non_excited coil without influence of the other coils.

Klaus

Easy peasy · Apr 22, 2018

There are 2 elements to instantaneous EMF seen at a switching node, the inductor EMF from energy stored in the air gap and windings and the other more obvious motor EMF caused by the generator aspect of the windings moving in a mag field...

KlausST · Apr 22, 2018

Hi,

True. We know. To avoid even more confusion I think we should focus on that what the OP asked:

BEMF=voltage induced in the coil by the moving magnetic field

This voltage is sinusoidal. But in the delta connection of OP's motor ... it's superimposed by the BEMF voltages of the other coils and the excitation voltages of the other coils.

I don't know how to measure it.
Maybe leave all signal floating (they are still limited to the bus voltage)
wait a little (until the peak has gone)
Then measure the difference voltage across two motor terminals.

Klaus

righteous · Apr 22, 2018

Easy peasy said:
maximum BEMF is clamped by the diodes in the fets and the supply rail, when you turn a leg fully off, and the volts fly up (or down) that is the back emf at that instant. it is not a straight forward question as the BEMF is super posed on the voltages supplied by the other switching legs ...

You may be confusing BEMF with FLYBACK, here is a thread that addresses the misappropriate use of the terms BLDC: Once and for all EMF, BEMF, CEMF or Flyback?

- - - Updated - - -

KlausST said:
With a motor in star configuration each terminal truly belongs to one coil....if you now additionally have the star point you are able to measure the voltage of a non_excited coil without influence of the other coils.

Yes I see that now, in the beginning the delta was chosen for some reasons that made sense then, but let me try and reconfigure it to star and see where that leads me.

KlausST · Apr 22, 2018

Hi,

but let me try and reconfigure it to star and see where that leads me.

Star configuration with star point wire = 4 motor terminals.

Then measure the voltage: not_excited_coil_connection to star_point.

Klaus

Easy peasy · Apr 22, 2018

This voltage is sinusoidal. But in the delta connection of OP's motor ... it's superimposed by the BEMF voltages of the other coils and the excitation voltages of the other coils.

In a typical BLDC motor - BEMF is trapeziodal ...

- - - Updated - - -

the important part is to try and see where the motor back EMF crosses the virtual neutral point, to give a reference for switching, the flyback produced EMF can make this hard to see - as well as any inconstant switching in the other legs ....

The ST app note points out that if you make it easy to see the crossing of the BEMF then you can't run to 100% PWM ...

FvM · Apr 22, 2018

A star configuration doesn't change the fact that the current in one winding is causing voltage drops in the others. But you'll get surely different waveforms, I doubt that you manage better to decode it.

You may be confusing BEMF with FLYBACK, here is a thread that addresses the misappropriate use of the terms.

The terms have a purpose for the intuitive description of DC motor driver circuits, but I think they are inappropriate for analysis of AC machines like BLDC. The usual description is by phasors for rotor emf, inductive and resistive voltage drop. In a square wave switched driver, harmonic currents are additionally complicating the picture.

righteous · Apr 22, 2018

FvM said:
A star configuration doesn't change the fact that the current in one winding is causing voltage drops in the others. But you'll get surely different waveforms, I doubt that you manage better to decode it.

Here is the scope shot of phase A (probe across A and Neutral/Virtual ground) it resembles the scope shot of the delta configuration although a bit clearer

That would still require some serious signal processing to get something useful out of that. I'm surprised it's still such a gang bang of various currents and voltages, even after eliminating the iron core and the mutual induction in the windings. I think I will have to take an other approach.

Kajunbee · Apr 22, 2018

If the OP were to use another motor to spin the bldc motor at the same rpm would this generated voltage be the same as the bemf created when there's and applied voltage.

righteous · Apr 22, 2018

Kajunbee said:
If the OP were to use another motor to spin the bldc motor at the same rpm would this generated voltage be the same as the bemf created when there's and applied voltage.

Yes indeed, but what would be the point though? I mean other than quantifying the bemf voltage?

I wanted to use the bemf for regenerative braking, but since the bemf is lower (I checked by spinning it around by hand) than Vcc it will require additional circuitry.

FvM · Apr 22, 2018

The objective of the thread hasn't be told clearly. If it's estimating the actual rotor phase for sensorless control, detecting rotor emf zero crossing as mentioned by Easy peasy could be a solution. For reconstruction of the complete rotor emf waveform, you need phase current and voltage input for your model.

Kajunbee · Apr 22, 2018

If your looking at regen braking I would think it would mean everything. Regen braking as far as I know I is when there's no power applied to motor. When you disconnect the input the motor will coast to a stop. While it's slowing down it will generate a sinusoidal voltage. This generated voltage is what your using to slow the motor down.
I have a GE X-13 bldc motor that I use as a generator. It's a 230 vac 1075 rpm blower motor. When I spin this motor at 1075 rpm it generates about 115 vac. If your wanting the bemf to equal the input voltage i don't think this is possible at the same rpm.
Maybe I'm misunderstanding the goal. Are you wanting to harvest the emf by regen braking? Or do you believe the generated bemf with and applied voltage is limiting the speed of the motor?

righteous · Apr 22, 2018

FvM said:
The objective of the thread hasn't be told clearly.

That is correct, as I have informed you on a previous occasion, I'm reluctant to disclose the main objective because then this thread would be filled with posts informing me that my undertaking is impossible.

FvM said:
For reconstruction of the complete rotor emf waveform, you need phase current and voltage input for your model.

Could you please elaborate on that, I don't quite understand what you mean.

- - - Updated - - -

Kajunbee said:
Regen braking as far as I know I is when there's no power applied to motor. When you disconnect the input the motor will coast to a stop. While it's slowing down it will generate a sinusoidal voltage. This generated voltage is what your using to slow the motor down.

Yes, that would be more or less what I'm looking for, except I don't want to power it off while it is braking, think of it as a genset.

- - - Updated - - -

Kajunbee said:
f your wanting the bemf to equal the input voltage i don't think this is possible at the same rpm.

I don't think so either, that's why some additional circuitry would be required to bring the rectified bemf above input voltage.

- - - Updated - - -

KlausST said:
A motor's BEMF (that is caused by the mechanically rotating magnetic field) in you configuration is difficult to measure, be ause it is influenced by the other terminal voltage pluse the BEMF voltage of the other coils.
https://www.st.com/resource/en/application_note/cd00020086.pdf

Klaus, in the ST application note Figure 4.

The wave forms () look like my scope shot (if you squeeze your eyes together), so I'm thinking, when I'm putting the scope probe directly across the inductor, maybe I'm measuring current and not voltage? It does say "Current in the inductor"...

FvM · Apr 22, 2018

That is correct, as I have informed you on a previous occasion, I'm reluctant to disclose the main objective because then this thread would be filled with posts informing me that my undertaking is impossible.

I didn't hear a reasoning for isolating rotor emf ("BEMF") in the measurement. The topic of the previous threads was slightly different, I think.

Could you please elaborate on that, I don't quite understand what you mean.

Terminal voltage of a single winding is Vemf + L* dI/dt. You need to substract inductive voltage drop L*dI/dt to get rotor emf. Winding resistance can be considered too for more accurate calculation.

righteous · Apr 22, 2018

FvM said:
I didn't hear a reasoning for isolating rotor emf ("BEMF") in the measurement. The topic of the previous threads was slightly different, I think.

In post #15 I give reasoning to Mr. Kajunbee.

FvM said:
Terminal voltage of a single winding is Vemf + L* dI/dt. You need to substract inductive voltage drop L*dI/dt to get rotor emf. Winding resistance can be considered too for more accurate calculation.

Next question, out of ignorance; Would it be possible to filter the terminal voltage, so that only the bemf remains, and would the linkage between circuitry after the filter and coil still remain intact? E.g. if you draw a current after the filter, will the voltage respond as if were a coil?

FvM · Apr 22, 2018

I don't think so either, that's why some additional circuitry would be required to bring the rectified bemf above input voltage.

All you need is the three phase bridge driver as shown in post #1 and an appropriate switching pattern to perform the regenerative breaking. The previous discussion doesn't consider that the bridge circuit is able to boost the motor voltage. As a prerequisite the momentary rotor angle must be known. So if you don't have it already, you need a circuit to measure the phase of the rotor emf.

Perfect BLDC respectively AC motor control is difficult at low speed, applications requiring it are preferably designed with a resolver.

righteous · Apr 23, 2018

FvM said:
All you need is the three phase bridge driver as shown in post #1 and an appropriate switching pattern to perform the regenerative breaking. The previous discussion doesn't consider that the bridge circuit is able to boost the motor voltage. As a prerequisite the momentary rotor angle must be known. So if you don't have it already, you need a circuit to measure the phase of the rotor emf.

Perfect BLDC respectively AC motor control is difficult at low speed, applications requiring it are preferably designed with a resolver.

You mean like mr. debrajdeb is proposing here https://www.microchip.com/forums/FindPost/431639 ? Except not switching all 3 lower mosfet's but merely the floating phase(s) i.e. the phases that are not energized?

What do you mean by "resolver", something more precise at lower speeds than hall sensors e.g. an angle sensor?

FvM · Apr 23, 2018

The link is describing a simplified regenerative braking method, a subset of the available generator operation modes of an inverter.

The question is what you want to achieve. If you just want to recover some energy, the simplified method should work. Synchronizing to the rotor angle is still necessary.

Welcome to EDAboard.com

BLDC: where is my BEMF?

Full Member level 2

Advanced Member level 6

Advanced Member level 7

Advanced Member level 6

Advanced Member level 7

Full Member level 2

Advanced Member level 7

Advanced Member level 6

Super Moderator

Full Member level 2

Advanced Member level 1

Full Member level 2

Super Moderator

Advanced Member level 1

Full Member level 2

Super Moderator

Full Member level 2

Super Moderator

Full Member level 2

Super Moderator

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor