Test Equipment Anonymous (TEA) group therapy thread

[Specmaster]

   

Guys....look at that photo from Tekwiki and compare it with how I initially set up mine. IDENTICAL. And guess what......set up that way IT PUSHES AIR OUT THE FRONT.   Edit, I found the direction arrow on mine. SAME. Spins clockwise when viewed from inside case. And I verified that anyway.

med, both of these fans are configured to PULL. Which is how it should be, as the filter is in front of the fan.

   

I've annotated these two pics to show how the fan blade and motor direction SHOULD BE for PULL or PUSH configuration.  Med... that fan blade is on in PUSH configuration in the 2nd pic. You are right; you originally had the fan blade in the correct orientation for PULL, but the motor is still running in the wrong direction for PUSH configuration.

Essentially, you've tried it every way but the way it should be... and you still aren't happy with the noise. Whether that's due to your your noise level tolerance, or endemic to the fan or due to the fact it's not assembled correctly... with a filter in front of it... none of us (including you) will know unless you actually assemble it correctly.

I've just recently spent days fighting down this very demon of noisy blowers on my Dell server PSU project... the truth is you simply cannot say how noisy it is going to be until it is assembled as intended, with fan going correct way, and filter installed. Even having the scope on top of the thing will make a huge difference in how it's supposed to work, how the air flows, and how much noise it makes.

Cheers,

mnem
 

Thanks for producing these drawings, it makes it graphically easy to understand, besides why would you want to drag hot air past a motor that is itself a source of heat, albeit a small one when if that motor overheats and fails, the extra cooling for the scope has gone 

[med6753]

Look guys....I'm DONE discussing this fucking fan. It's coming out and something of my own design is going in.

Give it a rest. Next subject. 

[Specmaster]

<SNIP>

Apart from the obvious idiocy here, the failure is not electrical. The lack of a load has no effect on the speed of a BLDC, jut the current. A BLDC is not a  I don't think overheating or bad commutaion was the cause of the failure either.
The aluminium rotor bell failed under the G force / fatigue. This resulted in los of the magnets. No electrical failure so far. The loss of the magnets significantly reduces the back EMF which is the main thing limiting the current. Current shoots up and windings fail through I2R heaing.
They are lucky a magnet did not hit someone.[/quote]

No; the fire was caused by a shorted FET(s). That ESC is toast. The cause of the shorted FET was loss of sync caused by loss of commutation feedback; period. It's endemic to the breed; something that happens all the time.

The guy wants to think his motor was running fast enough to grenade from centrifugal force; much more likely is that the magnets came loose and contacted the stator. This doesn't often happen so uniformly; but it does happen that way occasionally. I've done it.

Well, it's not impossible it grenaded from CF. In our quest for ever-lighter motors with ever-higher torque per gram, obviously manufacturers are going to have to shave the weight from somewhere. Preferably in the moving mass, as fast changes in prop RPM are the core principle of how these things can fly and maneuver so aggressively. That means that the CNC aluminum armature and the ferrous gauss concentration rings (magnet rings) are going to be shaved down to just a hair more than they absolutely need to be in order to not fly apart under normal usage.

This guy is deliberately abusing the motor by running it with no load. That is why this video is sheer stupidity.

mnem
 
[/quote]

There is NO evidence that the ESC has failed or there was a shorted FET. At 100V applied and that power level a FET failure would result in obvious damage.The stator fried because the current went up due to loss of back EMF. High end ESCs like th one in use have protection against commutation errors. If you watch the slow motion part of the video you can clearly see the open end of the rotor bell starting to expand just before it lets go. The rotor pole piece has also bell-mouthed due to centrifugal forces.

[/quote]
But what about the 2 bearings? There has to be one at each end of the rotor, and they would be the first item to fail surely?

[bd139]

Shit go bang. Don't do that shit again  

[Specmaster]

@med6753, As far as I'm concerned, whats been driving onwards with this, you say it is blowing air out the front, it is against all logic unless we either some misinformation or we have misunderstood what you have told us so far and it is that mystery that I'm primarily concerned about here. Something here is flying in the face of engineering theory, you drop a statement and expect the enquiring minds to just forget about it 

[Specmaster]

Shit go bang. Don't do that shit again  

Ok, I'll buy it, what happened?

[Robert763]

<SNIP>

Apart from the obvious idiocy here, the failure is not electrical. The lack of a load has no effect on the speed of a BLDC, jut the current. A BLDC is not a  I don't think overheating or bad commutaion was the cause of the failure either.
The aluminium rotor bell failed under the G force / fatigue. This resulted in los of the magnets. No electrical failure so far. The loss of the magnets significantly reduces the back EMF which is the main thing limiting the current. Current shoots up and windings fail through I2R heaing.
They are lucky a magnet did not hit someone.

No; the fire was caused by a shorted FET(s). That ESC is toast. The cause of the shorted FET was loss of sync caused by loss of commutation feedback; period. It's endemic to the breed; something that happens all the time.

The guy wants to think his motor was running fast enough to grenade from centrifugal force; much more likely is that the magnets came loose and contacted the stator. This doesn't often happen so uniformly; but it does happen that way occasionally. I've done it.

Well, it's not impossible it grenaded from CF. In our quest for ever-lighter motors with ever-higher torque per gram, obviously manufacturers are going to have to shave the weight from somewhere. Preferably in the moving mass, as fast changes in prop RPM are the core principle of how these things can fly and maneuver so aggressively. That means that the CNC aluminum armature and the ferrous gauss concentration rings (magnet rings) are going to be shaved down to just a hair more than they absolutely need to be in order to not fly apart under normal usage.

This guy is deliberately abusing the motor by running it with no load. That is why this video is sheer stupidity.

mnem
 
[/quote]

There is NO evidence that the ESC has failed or there was a shorted FET. At 100V applied and that power level a FET failure would result in obvious damage.The stator fried because the current went up due to loss of back EMF. High end ESCs like th one in use have protection against commutation errors. If you watch the slow motion part of the video you can clearly see the open end of the rotor bell starting to expand just before it lets go. The rotor pole piece has also bell-mouthed due to centrifugal forces.

[/quote]
But what about the 2 bearings? There has to be one at each end of the rotor, and they would be the first item to fail surely?
[/quote]

As I said look at the slow motion video it is clear the failure started with the rotor bell expanding.
Bearing failure wold cause vibration and eventually allow the rotor to touch the stator. in this style of motor the bearngs are much smaller diameter than the rotor bell so centrifugal force is much less. Also as there is no propeller thar is no thrust load which is the major load on the bearing in normal operation.
I've been playing and working with motors for about 50 years. Started playing with brushless DC motors over 20 years  ago. That includes ones alot bigger than fitted to quad copters.
The guys in the video were deliberately going for overspeed failure.

[mnementh]

Also, shame on you Spec, for bringing that vid in here with our conversation aboot med's blower; the two are aboot as closely related as an ox is to the cart it pulls. 

I have half a mind to give you 40 lashes with a wet noodle.

med's motor is an induction/repulsion motor; it works by the electrically induced magnetism in a laminated metal core pushing against the field which created that magnetism. By nature, these motors will attempt to reach an rpm that is dictated by the number of poles in the rotor vs those in the stator and the fixed AC frequency applied; essentially a small-scale 2-phase AC industrial motor. It will tend to run at that speed over a very broad voltage range; as a result, "speed control" by varying voltage doesn't really work; we need to change speed via PWM, etc.

A modern high-KV (K-RPM/V) quadcopter motor is essentially a PWM-switched 3-phase permanent-magnet DC repulsion motor operating at high frequency; its speed is governed by the applied DC voltage and the frequency/duty cycle of its PWM commutation cycle.

As we began making motors capable of developing higher RPM, we hit up against the wall of primitive slow fixed-commutation ESCs. This wall was broken by the development of ESCs which actually read the reverse-EMF pulses created by a magnet passing over a pole, and alter the timing of the next commutation cycle accordingly. These ESCs then developed to the point where speed of the computer running them and speed of the FETs used for switching were the limiting factor.

Think the difference between fixed ignition timing on a old Briggs lawnmower engine vs dynamic ignition timing on larger industrial, motorcycle and automobile engines.

Then began the race for ever-faster ESCs and motors, with each generation of motor and ESC forcing further evolution of the other. These machines are forcing the development of BL motors in all sorts of other applications; it really has been an exciting time to be involved in them.

mnem
Except, of course, when the FAA had to stick their dicks in it.

[AVGresponding]

Finally finished our refurb of a council resource centre this week (it's a short term care home for elderly persons recently released from hospital, in this case), about a month short of 2 years...
Even allowing for the 5 months or so we were not allowed on site at the start of the lockdown, it went a fair way over schedule and budget, as these things do. Poor communication between client and project management, and poor management generally are exclusively to blame for this, so don't have a go at me about your taxes being wasted, you'd be preaching to the choir.

In any case, the new fire alarm and door systems are now completely* operational, and so I had the fun task of removing the old panels from the office they were located in.

The old FAP (Fire alarm panel) is a Menvier (now part of Cooper Lighting and Controls, which in turn is now part of Eaton) of late 1980s vintage, so original to the building which was built in 1989 I believe.



It was a right old mare's nest in there, what with additions to trigger a dialler, close fire doors, and disable the PAC, all done by messrs. Bodgett & Leggitt as far as I can tell.
The new Notifier ID3000 will do a much better job**, with phased evacuation and localised sounders meaning less disturbance for service users, many of whom have dementia and don't react well to loud noises.

The door mimic panel was clearly a custom build for the site, internally as well as the display itself:



Check out those perf-boards!
It wasn't working very well before it was powered off, and looking at all those cooked resistors soldered to the bi-colour LEDs I'm not surprised. Someone calculated their power dissipation wrong... At least it wasn't me this time!   


I forgot to take pics of the emergency lighting central battery unit, was too busy wrestling the fucker out the door with the help of the apprentice and a labourer. Thing weighed a couple of hundred kilos, and that was after we stripped all the gubbins out (4x 12V 38Ah agm batteries (in my car), 1.5KVA trafo (also in my car), and charger boards).
The eagle eyed will have spotted that those are very small batteries. That's because they were replaced at a point when only around 30-40% of the system was still in use, otherwise they would have been 100-110Ah ones.

I did save the 10x38 cartridge fuses out of it, also a 24VDC 4PCO LPCB certified relay, and I saved the boards from the old fire panel for similar reasons; LPCB (Loss Prevention and Certification Board) certified relays. Oh, and the 12V 12Ah batteries from the FAP too...


*At least in theory. In practice, there seems to be a bug in the matrix, and some devices are randomly becoming invisible to the panel, or going into fault (and then later back out). Officially though it's Not My Problem. I just installed the cabling and devices (correctly).
Another contractor is responsible for programming the thing.   
**See above...

[mnementh]

No; the fire was caused by a shorted FET(s). That ESC is toast. The cause of the shorted FET was loss of sync caused by loss of commutation feedback; period. It's endemic to the breed; something that happens all the time.

The guy wants to think his motor was running fast enough to grenade from centrifugal force; much more likely is that the magnets came loose and contacted the stator. This doesn't often happen so uniformly; but it does happen that way occasionally. I've done it.

Well, it's not impossible it grenaded from CF. In our quest for ever-lighter motors with ever-higher torque per gram, obviously manufacturers are going to have to shave the weight from somewhere. Preferably in the moving mass, as fast changes in prop RPM are the core principle of how these things can fly and maneuver so aggressively. That means that the CNC aluminum armature and the ferrous gauss concentration rings (magnet rings) are going to be shaved down to just a hair more than they absolutely need to be in order to not fly apart under normal usage.

This guy is deliberately abusing the motor by running it with no load. That is why this video is sheer stupidity.

mnem
 

There is NO evidence that the ESC has failed or there was a shorted FET. At 100V applied and that power level a FET failure would result in obvious damage.The stator fried because the current went up due to loss of back EMF. High end ESCs like th one in use have protection against commutation errors. If you watch the slow motion part of the video you can clearly see the open end of the rotor bell starting to expand just before it lets go. The rotor pole piece has also bell-mouthed due to centrifugal forces. As I said look at the slow motion video it is clear the failure started with the rotor bell expanding.

Bearing failure wold cause vibration and eventually allow the rotor to touch the stator. in this style of motor the bearngs are much smaller diameter than the rotor bell so centrifugal force is much less. Also as there is no propeller thar is no thrust load which is the major load on the bearing in normal operation. I've been playing and working with motors for about 50 years. Started playing with brushless DC motors over 20 years  ago. That includes ones alot bigger than fitted to quad copters. The guys in the video were deliberately going for overspeed failure.

That has nothing to do with the fire. The fire is caused by a shorted FET(s). PERIOD. Anybody who actually flies these things knows this. What caused the shorted FET is loss of sync because the magnets went away. Whether that was caused by the magnets raising away from the stator due to CF or the bell grenading is irrelevant; I already stated it was possible, tho not likely, that the bell exploded due to CF.

FFS, man. 

Cheers,

mnem
 

[bd139]

Shit go bang. Don't do that shit again  

Ok, I'll buy it, what happened?

Well eventually the Joule–Lenz law was demonstrated spectacularly.

But I hypothesize that the motor was run far far far past its RPM limit causing it to mechanically disintegrate first. The enabler for this was the lack of any load on it plus driving it hard through high voltage and high speed signalling. Once the stator has buggered off into little pieces then the inductance of the coils decreases due to the permeability drop which causes current to ramp up quicker and reach saturation quickly at which point the coils become the aforementioned spectacular demonstration of the Joule-Lenz law. Fire comes out. Idiots giggle at their destruction and everyone goes home safely with bits of powdered magnets entering their soft tissue...

Key thing here is feedback. These are cheap Chinese shit which are sensorless AFAIK (lacking hall / VR sensors for speed measurement) so are cost cut completely. The controller doesn't know when something is amiss so it just lets the fire continue. As long as the MOSFETs don't go outside their SOA, then the ESC will be fine. Now googling the resistance of one of the winding comes up at 0.63 ohms and considering the old light bulb V-I curve which is what this is being non linear, as the coil gets hotter and catches fire it'll increase in resistance and power dissipated. The MOSFETs will stay fairly low PD during conduction due to Rds being really damn low compared to the load R on most of the MOSETs so it'd take a direct short to pop 'em before whatever was connected blew up. Either that or using the wrong MOSFETs or lurking close to the SOA.

Only a badly designed ESC or BLDC controller without any feedback should explode in your face in theory.

On that note the Tektronix 2235 preregulator is a textbook example of what a shit show it can be if you do a terrible job of this. Scared me turning that bastard on it did after the first one blew up.

[med6753]

@med6753, As far as I'm concerned, whats been driving onwards with this, you say it is blowing air out the front, it is against all logic unless we either some misinformation or we have misunderstood what you have told us so far and it is that mystery that I'm primarily concerned about here. Something here is flying in the face of engineering theory, you drop a statement and expect the enquiring minds to just forget about it 

Gee....do you think I don't know or can't tell which way air flows? I showed it, demonstrated it, prove it. What more do you need? At this point if the motor is ass backwards, the fan blades borked, or I'm stuck in a black hole it is what it is. Dunno what else to say other than the whole mess is getting shit canned. 

And now, for our musical selection. 



 

[mnementh]

Shit go bang. Don't do that shit again  

Ok, I'll buy it, what happened?

Well eventually the Joule–Lenz law was demonstrated spectacularly.

But I hypothesize that the motor was run far far far past its RPM limit causing it to mechanically disintegrate first. The enabler for this was the lack of any load on it plus driving it hard through high voltage and high speed signalling. Once the stator has buggered off into little pieces then the inductance of the coils decreases due to the permeability drop which causes current to ramp up quicker and reach saturation quickly at which point the coils become the aforementioned spectacular demonstration of the Joule-Lenz law. Fire comes out. Idiots giggle at their destruction and everyone goes home safely with bits of powdered magnets entering their soft tissue...

Key thing here is feedback. These are cheap Chinese shit which are sensorless AFAIK (lacking hall / VR sensors for speed measurement) so are cost cut completely. The controller doesn't know when something is amiss so it just lets the fire continue. As long as the MOSFETs don't go outside their SOA, then the ESC will be fine. Now googling the resistance of one of the winding comes up at 0.63 ohms and considering the old light bulb V-I curve which is what this is being non linear, as the coil gets hotter and catches fire it'll increase in resistance and power dissipated. The MOSFETs will stay fairly low PD during conduction due to Rds being really damn low compared to the load R on most of the MOSETs so it'd take a direct short to pop 'em before whatever was connected blew up. Either that or using the wrong MOSFETs or lurking close to the SOA.

Only a badly designed ESC or BLDC controller without any feedback should explode in your face in theory.

On that note the Tektronix 2235 preregulator is a textbook example of what a shit show it can be if you do a terrible job of this. Scared me turning that bastard on it did after the first one blew up.

We are literally running these ESCs so fast that yes, they can exceed the switching capability of the FETs used on them due to the combination of loading vs gate capacitance. Whole generations of ESCs have been obsoleted by this very limitation. Same with processors. We started out with Atmega8s. Now we've outstripped even what STM32s can synthesize PWM signals reliably, and had to go with BusyBees because hardware PWM.

No lie dude... it's a whole 'nuther world in here.

mnem
 

[bd139]

Yeah MOSFETs can sit in the linear region for quite a while when switching. Don't forget I've been using them for RF stuff incorrectly and the fuckers do explode. But that's really using the wrong tool for the wrong job (other than the economy factor ). I'd expect better for BLDC controllers. Although I'm failing to see how some of the sensorless feedback mechanisms would service failure conditions accurately which is possibly the issue.

[Cerebus]

The controller doesn't know when something is amiss so it just lets the fire continue.

Yeah, it's waiting for another back EMF pulse to tell it to turn off the current phase and turn on the next. Pulse doesn't come, current half bridge configuration is kept hard on feeding 97V across a pair of windings designed for, what, 24V across them in total, no changing flux to generate back EMF, DC laws take over  - poof!

Only a badly designed ESC or BLDC controller without any feedback should explode in your face in theory.

Or a well designed one that has been fed lies or told to turn off its various limit features. A good back EMF observer algorithm that ran a decent model of the motor would know that something was wrong when it didn't see commutation happen in some given timeout and go into safe mode.

[Specmaster]

Well if nothing else, we have all been highly entertained via the discussions going surrounding these 2 conundrums today and now, its time for me to disappear shopping at the ideal time when there is less chance of finding anyone controlling the number inside the shops and therefore no queueing which suits me greatly with my current leg problem.   

[Cerebus]

Gee....do you think I don't know or can't tell which way air flows? I showed it, demonstrated it, prove it. What more do you need? At this point if the motor is ass backwards, the fan blades borked, or I'm stuck in a black hole it is what it is. Dunno what else to say other than the whole mess is getting shit canned. 

"You are in a maze of twisty passages, all the same."
> REVERSE MOTOR
"You are in a maze of twisty passages, all the same."
> REVERSE BLADES
"You are in a maze of twisty passages, all the same."
> HAMMER MOTOR
"You are in a maze of twisty passages, all the same."
> PLOVER
"You are in a maze of twisty passages, all the same."

[mnementh]

Finally finished our refurb of a council resource centre this week (it's a short term care home for elderly persons recently released from hospital, in this case), about a month short of 2 years...
Even allowing for the 5 months or so we were not allowed on site at the start of the lockdown, it went a fair way over schedule and budget, as these things do. Poor communication between client and project management, and poor management generally are exclusively to blame for this, so don't have a go at me about your taxes being wasted, you'd be preaching to the choir.

In any case, the new fire alarm and door systems are now completely* operational, and so I had the fun task of removing the old panels from the office they were located in.

The old FAP (Fire alarm panel) is a Menvier (now part of Cooper Lighting and Controls, which in turn is now part of Eaton) of late 1980s vintage, so original to the building which was built in 1989 I believe.



It was a right old mare's nest in there, what with additions to trigger a dialler, close fire doors, and disable the PAC, all done by messrs. Bodgett & Leggitt as far as I can tell.
The new Notifier ID3000 will do a much better job**, with phased evacuation and localised sounders meaning less disturbance for service users, many of whom have dementia and don't react well to loud noises.

The door mimic panel was clearly a custom build for the site, internally as well as the display itself:

   

Check out those perf-boards!
It wasn't working very well before it was powered off, and looking at all those cooked resistors soldered to the bi-colour LEDs I'm not surprised. Someone calculated their power dissipation wrong... At least it wasn't me this time!   


I forgot to take pics of the emergency lighting central battery unit, was too busy wrestling the fucker out the door with the help of the apprentice and a labourer. Thing weighed a couple of hundred kilos, and that was after we stripped all the gubbins out (4x 12V 38Ah agm batteries (in my car), 1.5KVA trafo (also in my car), and charger boards).
The eagle eyed will have spotted that those are very small batteries. That's because they were replaced at a point when only around 30-40% of the system was still in use, otherwise they would have been 100-110Ah ones.

I did save the 10x38 cartridge fuses out of it, also a 24VDC 4PCO LPCB certified relay, and I saved the boards from the old fire panel for similar reasons; LPCB (Loss Prevention and Certification Board) certified relays. Oh, and the 12V 12Ah batteries from the FAP too...


*At least in theory. In practice, there seems to be a bug in the matrix, and some devices are randomly becoming invisible to the panel, or going into fault (and then later back out). Officially though it's Not My Problem. I just installed the cabling and devices (correctly).
Another contractor is responsible for programming the thing.   
**See above...

Reminds me of the XKCD for good programming... tho I suppose the "scope creep" aspect is pretty much identical no matter what theater, real or virtual...

mnem
I donn' wanna... I donn' wanna drag my sorry ass outside and work on that damn shed...  *pout*

[bd139]

The controller doesn't know when something is amiss so it just lets the fire continue.

Yeah, it's waiting for another back EMF pulse to tell it to turn off the current phase and turn on the next. Pulse doesn't come, current half bridge configuration is kept hard on feeding 97V across a pair of windings designed for, what, 24V across them in total, no changing flux to generate back EMF, DC laws take over  - poof!

Only a badly designed ESC or BLDC controller without any feedback should explode in your face in theory.

Or a well designed one that has been fed lies or told to turn off its various limit features. A good back EMF observer algorithm that ran a decent model of the motor would know that something was wrong when it didn't see commutation happen in some given timeout and go into safe mode.

Ahha that all makes sense. I can see a thousand failure modes appearing in my head now and they all make that unforgettable MOSFET explosion noise, usually followed by religious exclamations like "spaghetti fucking monster!".

"no officer it's not a remote controlled incendiary, it's a Chinese drone"

[med6753]

Well if nothing else, we have all been highly entertained via the discussions going surrounding these 2 conundrums today and now, its time for me to disappear shopping at the ideal time when there is less chance of finding anyone controlling the number inside the shops and therefore no queueing which suits me greatly with my current leg problem. 

Don't get stuck in any fans. 

[mnementh]

The controller doesn't know when something is amiss so it just lets the fire continue.

Yeah, it's waiting for another back EMF pulse to tell it to turn off the current phase and turn on the next. Pulse doesn't come, current half bridge configuration is kept hard on feeding 97V across a pair of windings designed for, what, 24V across them in total, no changing flux to generate back EMF, DC laws take over  - poof!

Only a badly designed ESC or BLDC controller without any feedback should explode in your face in theory.

Or a well designed one that has been fed lies or told to turn off its various limit features. A good back EMF observer algorithm that ran a decent model of the motor would know that something was wrong when it didn't see commutation happen in some given timeout and go into safe mode.

And now you're getting into limitations of A) processors cheap enough to put in a $20-80 ESC along with oodles of expensive MMLCCs, high-current FETs and telemetry features (yes, really) on 4oz copper (or more) and 2) firmware development that is an ongoing open-source concern of a few hundred people working in independent cells primarily as hobbyists for the last decade or so.

mnem
 

[bd139]

And now you're getting into limitations of A) processors cheap enough to put in a $20-80 ESC along with oodles of expensive MMLCCs, high-current FETs and telemetry features (yes, really) on 4oz copper (or more) and 2) firmware development that is an ongoing open-source concern of a few hundred people working in independent cells primarily as hobbyists for the last decade or so.

Sort of like my T12 explosion 

[Robert763]

Also, shame on you Spec, for bringing that vid in here with our conversation aboot med's blower; the two are aboot as closely related as an ox is to the cart it pulls. 

I have half a mind to give you 40 lashes with a wet noodle.

med's motor is an induction/repulsion motor; it works by the electrically induced magnetism in a laminated metal core pushing against the field which created that magnetism. By nature, these motors will attempt to reach an rpm that is dictated by the number of poles in the rotor vs those in the stator and the fixed AC frequency applied; essentially a small-scale 2-phase AC industrial motor. It will tend to run at that speed over a very broad voltage range; as a result, "speed control" by varying voltage doesn't really work; we need to change speed via PWM, etc.

A modern high-KV (K-RPM/V) quadcopter motor is essentially a PWM-switched 3-phase permanent-magnet DC repulsion motor operating at high frequency; its speed is governed by the applied DC voltage and the frequency/duty cycle of its PWM commutation cycle.

As we began making motors capable of developing higher RPM, we hit up against the wall of primitive slow fixed-commutation ESCs. This wall was broken by the development of ESCs which actually read the reverse-EMF pulses created by a magnet passing over a pole, and alter the timing of the next commutation cycle accordingly. These ESCs then developed to the point where speed of the computer running them and speed of the FETs used for switching were the limiting factor.

Think the difference between fixed ignition timing on a old Briggs lawnmower engine vs dynamic ignition timing on larger industrial, motorcycle and automobile engines.

Then began the race for ever-faster ESCs and motors, with each generation of motor and ESC forcing further evolution of the other. These machines are forcing the development of BL motors in all sorts of other applications; it really has been an exciting time to be involved in them.

mnem
Except, of course, when the FAA had to stick their dicks in it.

You are now showing your ignorance of motor operation.
1/ You cannot change the speed of an AC motor by PWM, you change it by changing the frequency of the supply. The only exception to this is crude power reduction that causes increased slip and even pole skipping in the motor This is just about OK for fans  but not much else.
2/ The speed of a brushless DC (BLDC) motor is in NO WAY governed by the applied voltage. A BLDC is basically a a 3 phase AC motor with a permanent magnet rotor. As with all 3 phase AC motors the speed is determined by the frequency and number of poles. The K-V factor just tells you what the back EMF is at a given speed. Obviously you need a supply higher than this to drive any current (producing torque and power), but the speed is still set by the frequency.
3/ PWM Duty cycle does not set th speed, it controls the average power (current)
4/ The sensorless motor / controller designs were developed to reduce cost and weight, not because the sensors were not fast enough. They do have the potential advantage of being able to compensate for field distortion / torque reaction but not all controllers do this. 

[mnementh]

And now you're getting into limitations of A) processors cheap enough to put in a $20-80 ESC along with oodles of expensive MMLCCs, high-current FETs and telemetry features (yes, really) on 4oz copper (or more) and 2) firmware development that is an ongoing open-source concern of a few hundred people working in independent cells primarily as hobbyists for the last decade or so.

Sort of like my T12 explosion 

Remember too... these are hardcore race machines.

While funny-car bracket racers may be willing to go along for the ride in a computer-controlled rail, the AA-level nutcases are not going to put up with nannyware.

They're 110% operating on a Checkered Flag or Crash mentality. Every bit that dies is just another sacrifice upon the altar of SPEED.   

mnem
 

[mnementh]

Also, shame on you Spec, for bringing that vid in here with our conversation aboot med's blower; the two are aboot as closely related as an ox is to the cart it pulls. 

I have half a mind to give you 40 lashes with a wet noodle.

med's motor is an induction/repulsion motor; it works by the electrically induced magnetism in a laminated metal core pushing against the field which created that magnetism. By nature, these motors will attempt to reach an rpm that is dictated by the number of poles in the rotor vs those in the stator and the fixed AC frequency applied; essentially a small-scale 2-phase AC industrial motor. It will tend to run at that speed over a very broad voltage range; as a result, "speed control" by varying voltage doesn't really work; we need to change speed via PWM, etc.

A modern high-KV (K-RPM/V) quadcopter motor is essentially a PWM-switched 3-phase permanent-magnet DC repulsion motor operating at high frequency; its speed is governed by the applied DC voltage and the frequency/duty cycle of its PWM commutation cycle.

As we began making motors capable of developing higher RPM, we hit up against the wall of primitive slow fixed-commutation ESCs. This wall was broken by the development of ESCs which actually read the reverse-EMF pulses created by a magnet passing over a pole, and alter the timing of the next commutation cycle accordingly. These ESCs then developed to the point where speed of the computer running them and speed of the FETs used for switching were the limiting factor.

Think the difference between fixed ignition timing on a old Briggs lawnmower engine vs dynamic ignition timing on larger industrial, motorcycle and automobile engines.

Then began the race for ever-faster ESCs and motors, with each generation of motor and ESC forcing further evolution of the other. These machines are forcing the development of BL motors in all sorts of other applications; it really has been an exciting time to be involved in them.

mnem
Except, of course, when the FAA had to stick their dicks in it.

You are now showing your ignorance of motor operation.
1/ You cannot change the speed of an AC motor by PWM, you change it by changing the frequency of the supply. The only exception to this is crude power reduction that causes increased slip and even pole skipping in the motor This is just about OK for fans  but not much else.
2/ The speed of a brushless DC (BLDC) motor is in NO WAY governed by the applied voltage. A BLDC is basically a a 3 phase AC motor with a permanent magnet rotor. As with all 3 phase AC motors the speed is determined by the frequency and number of poles. The K-V factor just tells you what the back EMF is at a given speed. Obviously you need a supply higher than this to drive any current (producing torque and power), but the speed is still set by the frequency.
3/ PWM Duty cycle does not set th speed, it controls the average power (current)
4/ The sensorless motor / controller designs were developed to reduce cost and weight, not because the sensors were not fast enough. They do have the potential advantage of being able to compensate for field distortion / torque reaction but not all controllers do this.

Yes, hence the etc part of PWM etc, which is what is used to control the 3-phase H-gate controllers commonly in use today.

Jeebus dude... I love a good argument as much as the next guy, but fuck... this is not a good argument.

[EDIT] You are trying to apply motor control principles designed for reliability and fixed-RPM efficiency to motors & ESCs intended for the exact opposite: to switch as much current as possible and to be able to change speeds as quickly as possible in order to effect attitude control of a small aircraft. Literally a apples and ocelots comparison.

mnem