If your Mavic falls out of the sky...

[Brojon]

Excuse my ignorance but going to throw something in and you can correct me if I'm way off the mark.

Am I right in thinking the mavic motors are vector controlled AC motors?

I usually work with 20kw+ induction motors and when power is lost the voltage decays and shifts a phase as it coasts to a stop.

Could this decay in motor voltage explain the drop you see?

They're steppers.

 

[Brojon]

Before you guys get too far off in the weeds. Looking at the motorVolts has proven to be a good way to determine if either the battery came loose or there was a battery failure.

The assertion I'm making is that given the sample rate the voltage would be there one sample, gone the next since (again my assertion) there's not enough capacitance possible to store enough energy to result in an ability to see the voltage drop.
All conjecture based on my experience in electronics and control systems.

 

[sar104]

I wouldn't think so since the voltage is converted into pseudo sinusoidal waveforms for the motor windings. The drivers are usually high power stepper driver chips containing power MOSFETs which will suppress any EMF from the motors. The chips have feedback (usually via serial) corresponding to the integral of the power delivered - not the actual voltage since it varies by a sine approximation.
Here's a good overview of a typical ESC.

Interesting description but it primarily addresses the ESCs, which doesn't change the fact that the motor windings are inductive. There is stored energy there that has to dissipate somewhere (which takes time) in order for the voltage at the motors to drop. I'm not following where you think that energy is going. I'll try to do a test this evening; hand catch a Mavic, pop the battery out with the motors still producing approximately hover thrust and look at the 50 Hz motor voltage data in the DAT file.

 

[Callum]

They're steppers.

Are you sure about that? Specs state FOC control.

 

[Brojon]

Interesting description but it primarily addresses the ESCs, which doesn't change the fact that the motor windings are inductive. There is stored energy there that has to dissipate somewhere (which takes time) in order for the voltage at the motors to drop. I'm not following where you think that energy is going. I'll try to do a test this evening; hand catch a Mavic, pop the battery out with the motors still producing approximately hover thrust and look at the 50 Hz motor voltage data in the DAT file.

What I'm trying to say - apparently poorly - is that the voltage reported isn't an actual A/D measuring the voltage at the motor windings.
It is an average reported by the ESC back to the flight controller via a I2C serial bus. The motor windings are isolated by the actual drivers which are almost 100% guaranteed to be a series of power MOSFETs. Any back EMF is dissipated by these or even a suppression diode used to shunt the EMF voltage to protect the ESC circuits. They do this because "inductive kickback", as it's known, is typically a very short duration spike potentially of much greater amplitude than the applied voltage. Even if they didn't short the EMF out it still isn't what's being measured.

 

[sar104]

What I'm trying to say - apparently poorly - is that the voltage reported isn't an actual A/D measuring the voltage at the motor windings.
It is an average reported by the ESC back to the flight controller via a I2C serial bus. The motor windings are isolated by the actual drivers which are almost 100% guaranteed to be a series of power MOSFETs. Any back EMF is dissipated by these or even a suppression diode used to shunt the EMF voltage to protect the ESC circuits. They do this because "inductive kickback", as it's known, is typically a very short duration spike potentially of much greater amplitude than the applied voltage. Even if they didn't short the EMF out it still isn't what's being measured.

I was under the impression that the ESC included the power transistors, so why do you think that it is not reporting actual RMS voltage to the motor windings? Is that specified somewhere?

 

[Brojon]

I was under the impression that the ESC included the power transistors, so why do you think that it is not reporting actual RMS voltage to the motor windings? Is that specified somewhere?

It does - MOSFETs. RMS means Root Mean Square and is essentially the integral of the varying voltage that is applied to the motor phases.
If we carried your assertion to the extreme then what you assert would be that each motor phase is reported since each phase receives a different voltage waveform. I think we can agree that isn't the case so we must assume that the ESC is averaging the voltage it's applying to all three independent phases. I think we can also agree that it would be unwieldy to have three separate voltage measuring circuits on each leg going back to a circuit that sums all legs and applies the RMS integral to it to give us a value. In my experience the controller simply reports the voltage as a calculated value from the current being drawn since that is a summation of all activity on all three legs of the motor. It is isolated from the actual motor windings since it is measured going into the driver not out - makes for a much simpler and more robust design.

 

[Brojon]

Are you sure about that? Specs state FOC control.

You're correct - I was thinking about steppers in relation to operation (how many steps in a rotation) and you know how the brain loves to fill in

 

[Callum]

It does - MOSFETs. RMS means Root Mean Square and is essentially the integral of the varying voltage that is applied to the motor phases.
If we carried your assertion to the extreme then what you assert would be that each motor phase is reported since each phase receives a different voltage waveform. I think we can agree that isn't the case so we must assume that the ESC is averaging the voltage it's applying to all three independent phases. I think we can also agree that it would be unwieldy to have three separate voltage measuring circuits on each leg going back to a circuit that sums all legs and applies the RMS integral to it to give us a value. In my experience the controller simply reports the voltage as a calculated value from the current being drawn since that is a summation of all activity on all three legs of the motor. It is isolated from the actual motor windings since it is measured going into the driver not out - makes for a much simpler and more robust design.

That makes sense but I'm guessing current is constantly monitored on each of the phases so ESC can vary phase angle to reduce flux current and increase efficiency.

Is the reported voltage an average derived from the current at the windings?

Im just curious as have only ever dealt with large 3 phase induction motors. It's strange to see so much reference to motor voltage when frequency is more relevant.

 

[sar104]

It does - MOSFETs. RMS means Root Mean Square and is essentially the integral of the varying voltage that is applied to the motor phases.
If we carried your assertion to the extreme then what you assert would be that each motor phase is reported since each phase receives a different voltage waveform. I think we can agree that isn't the case so we must assume that the ESC is averaging the voltage it's applying to all three independent phases. I think we can also agree that it would be unwieldy to have three separate voltage measuring circuits on each leg going back to a circuit that sums all legs and applies the RMS integral to it to give us a value. In my experience the controller simply reports the voltage as a calculated value from the current being drawn since that is a summation of all activity on all three legs of the motor. It is isolated from the actual motor windings since it is measured going into the driver not out - makes for a much simpler and more robust design.

You misunderstood my question perhaps. I'm a physicist - I know what a MOSFET is and I understand RMS which, if you want to be precise, is the square root of the time-averaged mean of the square of the voltage calculated from a rolling integral of the square of the voltage divided by the rolling window interval, not the time-integral of the voltage. I was questioning your statement that the motor windings were isolated from the ESC by the MOSFETs - your answer that you are assuming that the measurement is upstream of the MOSFETs explains what you meant.

Anyway - I'm not trying to assert anything - I was just questioning specifically why you think that the motor voltages reported are not either the RMS of one phase or the mean of the RMS of all the phases. And my fundamental point remains - if the supply is disconnected then it appears to me that there is both capacitatively and inductively stored energy that will delay the voltage decay. By how much I don't know because I don't have a full circuit description, but I'm not convinced by hand-waving arguments based on assumptions on what is being measured and where it is being measured.

We clearly need data.

 

[sar104]

That makes sense but I'm guessing current is constantly monitored on each of the phases so ESC can vary phase angle to reduce flux current and increase efficiency.

Is the reported voltage an average derived from the current at the windings?

Im just curious as have only ever dealt with large 3 phase induction motors. It's strange to see so much reference to motor voltage when frequency is more relevant.

I've also wondered why they report motor voltage, but we have used it because it is the fastest sampling rate voltage data available.

 

[Brojon]

You misunderstood my question perhaps. I'm a physicist - I know what a MOSFET is and I understand RMS which, if you want to be precise, is the square root of the time-averaged mean of the square of the voltage calculated from a rolling integral of the square of the voltage divided by the rolling window interval, not the time-integral of the voltage. I was questioning your statement that the motor windings were isolated from the ESC by the MOSFETs - your answer that you are assuming that the measurement is upstream of the MOSFETs explains what you meant.

Anyway - I'm not trying to assert anything - I was just questioning specifically why you think that the motor voltages reported are not either the RMS of one phase or the mean of the RMS of all the phases. And my fundamental point remains - if the supply is disconnected then it appears to me that there is both capacitatively and inductively stored energy that will delay the voltage decay. By how much I don't know because I don't have a full circuit description, but I'm not convinced by hand-waving arguments based on assumptions on what is being measured and where it is being measured.

We clearly need data.

Precision isn't necessary for a speculative discussion. Integral and "average" are good enough. Heck the useful approximation techs use is .707 x peak voltage.
I seem to recall I've detailed my technical background in another thread.
As far as hand waving - look at a picture of the ESC board. See any large caps on there? Me neither. I can state with a a certainty that the circuit is protected against back emf - it would be reckless for an engineer not to - especially since a suppression diode is usually built in to most power MOSFETs for motor control. Refer to page 8 I provided the math for a resistive/capacitive network based on your assertion there was some capacitor holding the charge. Now it appears you are trying to assert the inductive collapse alone would prolong the voltage. In fact I can practically write a check on the premise the power outputs of the ESC are diode protected - they dump the energy which is an even faster discharge than involving an imaginary big capacitor. I never said the values reported weren't RMS - I specifically said integral keeping it simple since that is in fact a mostly true statement sufficient to make a point. Really - why is it you go to such extraordinary lengths to refute almost everything I post on this board and reduce it to an exercise in minutiae?

 

[Brojon]

That makes sense but I'm guessing current is constantly monitored on each of the phases so ESC can vary phase angle to reduce flux current and increase efficiency.
Is the reported voltage an average derived from the current at the windings?
Im just curious as have only ever dealt with large 3 phase induction motors. It's strange to see so much reference to motor voltage when frequency is more relevant.

Nah - think more in terms of a VFD.
You know that little rotor dance that happens when you fire it up? That's the ESC firing the phases to see which one moves so it knows where to "start". After that it pretty much acts like a stepper motor.

 

[Callum]

Nah - think more in terms of a VFD.
You know that little rotor dance that happens when you fire it up? That's the ESC firing the phases to see which one moves so it knows where to "start". After that it pretty much acts like a stepper motor.

It surely can't be FOC (vector) then as stated in specs.

I thought sensorless FOC needs feedback from at least 2 windings to reduce flux current and increase torque current.

If it was operating on a scalar the efficiency would drop.

Look at any FOC block diagram and it has current feedback betweenMOSFETs. And motor.

 

[lisadoc]

Nerd fight!

Wow... this whole discussion is like my Mavic - way over my head.

Thank heavens for engineers who can figure all this stuff out. You all lost me at "the voltage of the motors... blah... blah..." After that my head exploded.

Continue on... (without me interjecting mind you). I'm impressed that there are such technically minded people on this forum that have the desire to dig through such things. Did you guys really say "he has better organized records than I do."? Amazing... Are you all really keeping organized records of other people's crashes?

 

[sar104]

Precision isn't necessary for a speculative discussion. Integral and "average" are good enough. Heck the useful approximation techs use is .707 x peak voltage.
I seem to recall I've detailed my technical background in another thread.
As far as hand waving - look at a picture of the ESC board. See any large caps on there? Me neither. I can state with a a certainty that the circuit is protected against back emf - it would be reckless for an engineer not to - especially since a suppression diode is usually built in to most power MOSFETs for motor control. Refer to page 8 I provided the math for a resistive/capacitive network based on your assertion there was some capacitor holding the charge. Now it appears you are trying to assert the inductive collapse alone would prolong the voltage. In fact I can practically write a check on the premise the power outputs of the ESC are diode protected - they dump the energy which is an even faster discharge than involving an imaginary big capacitor. I never said the values reported weren't RMS - I specifically said integral keeping it simple since that is in fact a mostly true statement sufficient to make a point. Really - why is it you go to such extraordinary lengths to refute almost everything I post on this board and reduce it to an exercise in minutiae?

I'm not trying to refute everything you post; I'm trying to have a discussion. It was you who offhandedly dismissed the observation that voltage decay had been observed in several cases that were identified for other reasons as battery disconnects. And you cannot expect to lecture me or anyone else with inaccurate descriptions of such things as the meaning of RMS and then be offended when I point out that your description is wrong. Perhaps if you were not to jump to conclusions as to the technical competence of the audience you might provoke less testy responses.

Yes - the initial suggestion was that capacitatively-stored energy was responsible for the delayed decay and I, for one, had not run the back of the envelope calculations to test that hypothesis. You did, and showed that it was not a reasonable explanation. Hence my suggestion that the inductance of the motor windings might be responsible - that was also a hypothesis, not an assertion, since we have data to account for. I plan to test it with some simple experiments, while you seem content just to continue to assert that it is wrong. Which it may be, but then how about suggesting an alternative explanation for the existing data instead of just ignoring it?

 

[sar104]

Nerd fight!

Wow... this whole discussion is like my Mavic - way over my head.

Thank heavens for engineers who can figure all this stuff out. You all lost me at "the voltage of the motors... blah... blah..." After that my head exploded.

Continue on... (without me interjecting mind you). I'm impressed that there are such technically minded people on this forum that have the desire to dig through such things. Did you guys really say "he has better organized records than I do."? Amazing... Are you all really keeping organized records of other people's crashes?

It is a nerdy discussion I'll admit, but that's what scientists do. Our difference here appears to be that I think it is a valuable discussion while @Brojon appears to be offended by anyone challenging his assertions. I've no idea why - I'm certainly here to be challenged.

@BudWalker certainly has an organized database of log files and events - whereas I do not. I save the data and the analyses but not with sufficiently detailed notes that I can easily pull up data categorized by event.

 

[BudWalker]

I'm not trying to refute everything you post; I'm trying to have a discussion. It was you who offhandedly dismissed the observation that voltage decay had been observed in several cases that were identified for other reasons as battery disconnects. And you cannot expect to lecture me or anyone else with inaccurate descriptions of such things as the meaning of RMS and then be offended when I point out that your description is wrong. Perhaps if you were not to jump to conclusions as to the technical competence of the audience you might provoke less testy responses.

Yes - the initial suggestion was that capacitatively-stored energy was responsible for the delayed decay and I, for one, had not run the back of the envelope calculations to test that hypothesis. You did, and showed that it was not a reasonable explanation. Hence my suggestion that the inductance of the motor windings might be responsible - that was also a hypothesis, not an assertion, since we have data to account for. I plan to test it with some simple experiments, while you seem content just to continue to assert that it is wrong. Which it may be, but then how about suggesting an alternative explanation for the existing data instead of just ignoring it?

I did an experiment with my Mavic. No props, ran the motor speeds up and then removed the battery. It pretty much looks like all the other abrupt power loss incidents I've looke at.

@Brojon

 

[BudWalker]

......
@BudWalker certainly has an organized database of log files and events - whereas I do not. I save the data and the analyses but not with sufficiently detailed notes that I can easily pull up data categorized by event.

I'm so misunderstood. I don't have a database and I'm certainly not organized enough to have notes. I usually end up searching the forum for my own posts until I find what I'm looking for. Gimme a break you guys.

 

[sar104]

I did an experiment with my Mavic. No props, ran the motor speeds up and then removed the battery. It pretty much looks like all the other abrupt power loss incidents I've looke at.
View attachment 24954
@Brojon

Nice work - you beat me to it - I was just about to set up the test. That is good evidence that a battery disconnect gives that signature in the motor voltage. Now it would be nice if we could find a solid physics-based explanation. With a spare motor, a switched voltage source and an oscilloscope it should be fairly easy to measure motor inductance. Anyone interested?