Motor disassembly?

[fGene]

Hi,

I am working on a mod for MA2 which calls for motor rewinding -- a task which requires a full motor disassembly, including removing the stator plates stack from the base. Currently I am not sure if this is possible at all (depends on what type of glue DJI is fond of), but it's clear that the first step I need to do is to simply remove the bell. Most motors I dealt with in the past had a spring clip holding the shaft by a groove at the end, so that when the clip is removed, the bell along with the shaft can be pulled out. On a MA2 (and I suspect on most other Mavics) there's no spring clip to remove, and it looks like the end of the shaft is permanently pressed onto the rear ball bearing, so that they are impossible to separate without destroying the bearing. So I am kind of stuck at this point, and would like to ask:

Has anyone successfully taken apart a motor on a Mavic?

If yes, please share some tips. Thanks.

 

[WithTheBirds]

Why do you think you will destroy the bearing? Even if there was a circle and washer you would almost certainly need a press to seperate the bearing from the bell and shaft assembly.

 

[fGene]

Why do you think you will destroy the bearing? Even if there was a circle and washer you would almost certainly need a press to seperate the bearing from the bell and shaft assembly.

I can't really tell how the shaft end is secured in the rear ball bearing, I don't see any washers or clips or or anything that would help me understand how they are connected. They might be pressure-welded to each other. If that's the case and I simply apply pressure along the shaft, that may destroy the bearing.

 

[WithTheBirds]

Well if it destroys the bearing (unlikely) you shouldn’t have any problems finding a replacement. For the cost of the motors service (disassembly) probably isn’t intended.

What is the intent of your postponed mods? Do you have a means to alter the ESC drive parameters? What will your modified winding method achieve? Are you changing props also? We should expect a reasonable effort has already been applied in development to tune the propulsion system.

 

[fGene]

Well if it destroys the bearing (unlikely) you shouldn’t have any problems finding a replacement.

Well I want to avoid that, obviously. Currently I don't have any replacements.

For the cost of the motors service (disassembly) probably isn’t intended.

This I know too, strangely enough.

What is the intent of your postponed mods? Do you have a means to alter the ESC drive parameters? What will your modified winding method achieve? Are you changing props also?

I want to experiment with larger props, to achieve quieter flight and possibly longer flight time. For this I need to lower the kv rating of the motors, by rewinding them with higher number of turns per stator tooth. At this point I don't plan to do anything with ESCs.

We should expect a reasonable effort has already been applied in development to tune the propulsion system.

I don't understand your attitude. What do you mean by "we should expect"? I don't feel a slightest bit obliged to meet anybody's expectations, I am just experimenting on a drone that I bought with my own money, (similar to what I did with Mavic MIni), and I fully believe I am within my rights to do so. All I want is some tips regarding motor disassembly. Have you done it? Can you share them?

 

[WithTheBirds]

Well I want to avoid that, obviously. Currently I don't have any replacements.



This I know too, strangely enough.



I want to experiment with larger props, to achieve quieter flight and possibly longer flight time. For this I need to lower the kv rating of the motors, by rewinding them with higher number of turns per stator tooth. At this point I don't plan to do anything with ESCs.



I don't understand your attitude. What do you mean by "we should expect"? I don't feel a slightest bit obliged to meet anybody's expectations, I am just experimenting on a drone that I bought with my own money, (similar to what I did with Mavic MIni), and I fully believe I am within my rights to do so. All I want is some tips regarding motor disassembly. Have you done it? Can you share them?

It seems you have misunderstood- I am certainly not in any way attempting to discourage you or anyone from experimentation.

To clarify all I am saying is it is reasonable to expect that DJI, or those engaged by them to develop the propulsion system, went to great lengths to match the motor parameters and drive circuitry to meet a design criteria (a likely fact that should be, probably is, glaringly obvious). This design exersize almost certainly extended to the algorithms employed in the flight controller.

You will probably get the thing to fly spinning larger props. It may produce a less objectionable sonic signature also. You probably won't find any efficiency gains and you could be taking the FETs in the inverter stages of the motor drivers close to the upper end of their current rating.

The KV mod might render the motors unusable absent the possibility of editing the parameters in the motor driver SOC. I don't know which motor driver implementation DJI are using here, with the TI solution in earlier models the motor inductance parameter and KV values seemed very important to ensuring the FOC worked reliably.

The other point you probably haven't missed is that you might find the altered motor/prop combo will create issues with the flight controller, flight characteristics measured will almost certainly be significantly outside the expected ranges.

Having said all that I am genuinely interested in seeing where you end up with this.

 

[DanMan32]

I don't think it would be necessary to alter the motor KV to accommodate larger props. The flight control is closed loop so when it notices more lift than usual with the larger props, it will reduce the RPM. It may oscillate and encounter more vibration than normal or it may reach its expected lower RPM limit and generate an error. Changing the motor characteristics won't help since the lift will be governed by RPM which the FC would set regardless of the motor KV.

If the motor is built as it was with the P3, the shaft is pressure fit to the bearings as well as being retained by a C/E clip. There are YouTube videos on replacing the bearings for P3 motors.

Come to think of it, there were two P3 designs with the main board matched to the motor type. You couldn't use board revision A with motor revision B. Therefore I don't think the MA2 would work with a different motor type.

 

[Tiny]

What props gene? How much larger are you trying to go?

 

[fGene]

It seems you have misunderstood- I am certainly not in any way attempting to discourage you or anyone from experimentation.

To clarify all I am saying is it is reasonable to expect that DJI, or those engaged by them to develop the propulsion system, went to great lengths to match the motor parameters and drive circuitry to meet a design criteria (a likely fact that should be, probably is, glaringly obvious). This design exersize almost certainly extended to the algorithms employed in the flight controller.

I apologize. By re-reading your previous post, I can now see there was a misunderstanding on my part, likely attributable to my less-than-perfect English (also to the fact my baby was crying while I was typing).

You will probably get the thing to fly spinning larger props. It may produce a less objectionable sonic signature also. You probably won't find any efficiency gains and you could be taking the FETs in the inverter stages of the motor drivers close to the upper end of their current rating.

The KV mod might render the motors unusable absent the possibility of editing the parameters in the motor driver SOC. I don't know which motor driver implementation DJI are using here, with the TI solution in earlier models the motor inductance parameter and KV values seemed very important to ensuring the FOC worked reliably.

This might well be correct. My primary goal with this mod is to lower the noise; any efficiency advantages or disadvantages that come along with it are secondary concern. That being said, I still hope it is possible to increase efficiency with larger props if the power delivered by ESC is kept in check -- hence the need to decrease KV rating. I want FETs to be operating at approximately the same power level, albeit with lower switching frequency due to less RPM. Lower switching frequency should not be a problem at all as far as FETs are concerned, but there could potentially be a problem if DJI monitoring software decides the parameters are way off and blocks the normal operation of the drone.

Well, if this is the case, I think I know how to trick the software: use a gear system. This option has many disadvantages (too numerous to list here) but also too big advantages: 1). doesn't involve motor disassembly 2). naturally allows for arms extension to accommodate larger props. Still, I don't want to go down this route unless it is unavoidable.

The other point you probably haven't missed is that you might find the altered motor/prop combo will create issues with the flight controller, flight characteristics measured will almost certainly be significantly outside the expected ranges.

No, I don't think I missed this point. If you are referring to attitude measurement and corrections (controlled by PID parameters of the FC), yes, they will no longer perfectly match the airframe as the airframe will no longer have the same size and inertia. But the thing is, in DJI drones (according to my experience) the PID system is very forgiving, usually allowing huge changes to airframe and still keeping acceptable flight characteristics -- just look up videos where people put GoPro on a Mini, and it still flies reasonably well. Another example is prop guards which change the airframe very noticeably, yet are a part of the official package. As for measuring speed, position etc. that is done with GPS/GLONASS receiver and has nothing to do with airframe. Same goes for compass.

Having said all that I am genuinely interested in seeing where you end up with this.

Like I said before, I am not obliged to meet expectations But I will certainly let the community know if something comes out of my project. Right now it is simply too early to make any predictions.

 

[WithTheBirds]

I apologize. By re-reading your previous post, I can now see there was a misunderstanding on my part, likely attributable to my less-than-perfect English (also to the fact my baby was crying while I was typing).



This might well be correct. My primary goal with this mod is to lower the noise; any efficiency advantages or disadvantages that come along with it are secondary concern. That being said, I still hope it is possible to increase efficiency with larger props if the power delivered by ESC is kept in check -- hence the need to decrease KV rating. I want FETs to be operating at approximately the same power level, albeit with lower switching frequency due to less RPM. Lower switching frequency should not be a problem at all as far as FETs are concerned, but there could potentially be a problem if DJI monitoring software decides the parameters are way off and blocks the normal operation of the drone.

Well, if this is the case, I think I know how to trick the software: use a gear system. This option has many disadvantages (too numerous to list here) but also too big advantages: 1). doesn't involve motor disassembly 2). naturally allows for arms extension to accommodate larger props. Still, I don't want to go down this route unless it is unavoidable.



No, I don't think I missed this point. If you are referring to attitude measurement and corrections (controlled by PID parameters of the FC), yes, they will no longer perfectly match the airframe as the airframe will no longer have the same size and inertia. But the thing is, in DJI drones (according to my experience) the PID system is very forgiving, usually allowing huge changes to airframe and still keeping acceptable flight characteristics -- just look up videos where people put GoPro on a Mini, and it still flies reasonably well. Another example is prop guards which change the airframe very noticeably, yet are a part of the official package. As for measuring speed, position etc. that is done with GPS/GLONASS receiver and has nothing to do with airframe. Same goes for compass.



Like I said before, I am not obliged to meet expectations But I will certainly let the community know if something comes out of my project. Right now it is simply too early to make any predictions.

I have a high confidence I understand what you are hoping to achieve, increasing the conductor length should allow higher torque within the SOA of the FETs. Thinner wire gauge and more of it will increase the IR losses though as no doubt you are aware. No free lunch.

Instinct tells me finding a way to mount larger motors will be a much better option than reduction gearing.

 

[fGene]

I don't think it would be necessary to alter the motor KV to accommodate larger props. The flight control is closed loop so when it notices more lift than usual with the larger props, it will reduce the RPM. It may oscillate and encounter more vibration than normal or it may reach its expected lower RPM limit and generate an error. Changing the motor characteristics won't help since the lift will be governed by RPM which the FC would set regardless of the motor KV.

I beg to differ. The efficiency of a propulsion system is determined by many factors, but it is very essential to match motor kv to nominal/expected RPM as well as possible. You can't simply put on a larger prop and expect it to work, you might end up with burned ESC in no time, and the efficiency will plummet for sure.

The errors that might be generated are a real concern, yes.

If the motor is built as it was with the P3, the shaft is pressure fit to the bearings as well as being retained by a C/E clip. There are YouTube videos on replacing the bearings for P3 motors.
Come to think of it, there were two P3 designs with the main board matched to the motor type. You couldn't use board revision A with motor revision B. Therefore I don't think the MA2 would work with a different motor type.

Well, unfortunately I don't see any clips on the shaft. The shaft might be pressure-fitted only, or there might be some other hidden retainer. Phantom 3 motors are different in this regard.

 

[fGene]

I have a high confidence I understand what you are hoping to achieve, increasing the conductor length should allow higher torque within the SOA of the FETs. Thinner wire gauge and more of it will increase the IR losses though as no doubt you are aware. No free lunch.

Yes, that's the idea. I rewound many motors in my life, mostly successfully DJI stock motors are some of the best in overall quality that I've seen, but they can still be improved. I closely inspected MA2 motors, and believe it is possible to add more coils per tooth even while keeping the same wire gauge. Or even a lower gauge (to keep resistance the same) -- but that might be too problematic. Can't tell at this point.

Instinct tells me finding a way to mount larger motors will be a much better option than reduction gearing.

Weight is a concern here. Like I said, DJI motors are of very high quality, which I want to keep, and the next size is MPP/M2P/M2Z/M2E motors which are too big and heavy for MA2 in my opinion. The geared reductor should add about 5-6 grams per arm by my estimate and provide the needed extension, which weigh-wise might be preferable to installing a larger motor which will also need extension. We'll see.

 

[fGene]

What props gene? How much larger are you trying to go?

Gonna start with M Platinum - M2 sizes, probably MAS version.

 

[bumper]

The shaft is held in the stator (bottom housing) by a circlip at the bottom of the motor. To get the that the motor needs to be removed from the arm. Remove circlip and gently pry up on the "bell" or rotor housing (I might have used a punch too, can't recall). It will come upward and remove from the stator with the shaft still fixed to the bell.

Both bearings stayed in the lower housing (stator). The shaft fit in those bearings is not an interference fit, but more like a tight slip fit.

 

[Alan_M]

I want FETs to be operating at approximately the same power level, albeit with lower switching frequency due to less RPM.

I think you’ll find that the switching frequency of the FETs is not related to motor RPM. They will still be switching at the PWM frequency.