12 kw rc motor

[Arlo1]

Ok So Today I worked all the numbers and here is my verdict. From what I have researched and somewhat tested you need ~100uH a phase for inductance for a controller to live. Stock colossus has 8uh! What we need is to wind it to 35kv and terminate it in WYE! I worked all the numbers and it will work out to be 10 turns a tooth and 106uH inductance split into 2 separate stators just look at it as a bottom 1/2 and a top 1/2 for 9 stator teeth a section each section will have 3 for phase A 3 for B and 3 for C Then the bottom 1/2 will need to be completely separate from the top but can still be fired off the same set of halls. So for the controller we will need 6 H bridges 2 groups of 3 then the mosfets will be working with the same kv but 2x the inductance causing less heat and better control.

I have one of the 24 fet boards left I will try to cut the pad in 1/2 for each phase and build one up for this weekend to show you all!
Cheers.
Arlin.

[zEEz]

I have one of the 24 fet boards left I will try to cut the pad in 1/2 for each phase and build one up for this weekend to show you all!
Cheers.
Arlin.

Arlo, I think 12 fet for each set of 3 winds is too little if you plan to apply any kind of load to the rotor ...
IT WILL BLOW AND IT IS A PITY ...
I use a 36 fet controller with the ME0907 motor that is 120uH ... would say 24 mosfet is the minimum ...
Don't want to bring you bad luck, just telling you to wait to have the proper tools ...
... I know the desire to see something working is tempting ... but I would avoid to look for troubles ...

have fun!

[Lebowski]

Ok So Today I worked all the numbers and here is my verdict. From what I have researched and somewhat tested you need ~100uH a phase for inductance for a controller to live. Stock colossus has 8uh!

I don't see what the problem is with 8uH and why it would need te be 100uH. As far as I know the current in an inductor follows:

V = L * delta_I/delta_t

with 100V and 8uH the delta_I/delta_t is 12.5 M Amps/sec (assuming the motor is not rotating yet and back-emf = 0V)

If you're running 20 kHz PWM with let's say 10% dutycycle (at motorstart) this means:
20 kHz => this is 50 usec, at 10% -> the current is allowed to increase for 5 usec -> the current increases by 62 Amps !
And this for only 10% dutycycle ! No wonder stuff goes up in smoke

An extra 100uH decreases the current increase to around 5 Amps. This is survivable but it's not an elegant fix.

I would say the real fix is not an external inductance but an increase of the PWM to a few 100 kHz to 1 MHz.
The main issue then is that you need to beef up the mosfet drivers to be able to handle the switching rate
but this should not be a big issue. In class D audio amplifiers (which drive the same mosfet output stage)
a few 100 kHz is normal...

[Arlo1]

Ok So Today I worked all the numbers and here is my verdict. From what I have researched and somewhat tested you need ~100uH a phase for inductance for a controller to live. Stock colossus has 8uh!

I don't see what the problem is with 8uH and why it would need te be 100uH. As far as I know the current in an inductor follows:

V = L * delta_I/delta_t

with 100V and 8uH the delta_I/delta_t is 12.5 M Amps/sec (assuming the motor is not rotating yet and back-emf = 0V)

If you're running 20 kHz PWM with let's say 10% dutycycle (at motorstart) this means:
20 kHz => this is 50 usec, at 10% -> the current is allowed to increase for 5 usec -> the current increases by 62 Amps !
And this for only 10% dutycycle ! No wonder stuff goes up in smoke

An extra 100uH decreases the current increase to around 5 Amps. This is survivable but it's not an elegant fix.

I would say the real fix is not an external inductance but an increase of the PWM to a few 100 kHz to 1 MHz.
The main issue then is that you need to beef up the mosfet drivers to be able to handle the switching rate
but this should not be a big issue. In class D audio amplifiers (which drive the same mosfet output stage)
a few 100 kHz is normal...

Im just starting with 100uH because of other prople in the world have had to aim for that humber or more to keep their controllers happy.

Having a uber low inductance like 8uH will cause significant switching losses. When mosefets or igbts come out that have shorter on time and off times the switching losses will be reduced then it will be easier to run a motor with less inductance. My 24 4110 mosfet controller blows a fet stage in 20-25uS with a combinded on and off time of 155nS on the irfb4110 the fets are getting a lot of heat just from switching. So if they even turned on and off for a max on time of say 20 uS they will still get a crazy amount of switching heat and be destroyed.

I am going at this will a few different angles of attack. But by summer next year I will have pushed colossus past its limits.
I am still waiting for splinter to run his at hi power and see how the kelly holds up!

[zEEz]

Having a uber low inductance like 8uH will cause significant switching losses. When mosefets or igbts come out that have shorter on time and off times the switching losses will be reduced then it will be easier to run a motor with less inductance. My 24 4110 mosfet controller blows a fet stage in 20-25uS with a combinded on and off time of 155nS on the irfb4110 the fets are getting a lot of heat just from switching. So if they even turned on and off for a max on time of say 20 uS they will still get a crazy amount of switching heat and be destroyed.

Lebowsky, in this Arlo is right ... a D-class audio amplifier is working reasonably at hundreds of kHz pwm
frequency just because the load has a resistance bigger than 2 ohm and a fairly high inductance ...
When they get to kW amplifications, it is always multichannel and multiways ... so the power is limited ...

Here we want 20kW on one output, so it is a problem for the output stage ...

have fun!

[Arlo1]

I have one of the 24 fet boards left I will try to cut the pad in 1/2 for each phase and build one up for this weekend to show you all!
Cheers.
Arlin.

Arlo, I think 12 fet for each set of 3 winds is too little if you plan to apply any kind of load to the rotor ...
IT WILL BLOW AND IT IS A PITY ...
I use a 36 fet controller with the ME0907 motor that is 120uH ... would say 24 mosfet is the minimum ...
Don't want to bring you bad luck, just telling you to wait to have the proper tools ...
... I know the desire to see something working is tempting ... but I would avoid to look for troubles ...

have fun!

Thanks im building a better controller. But as a test i want to do this i know 12 fets a motor is not a lot but it will be about the same as the small colossus in terms of how hard it is on fets!

[Lebowski]

Arlo can you explain what you mean by 'combined on/off time of 155 nsec' ? And by '20 usec on time' ?
20usec seems awfully long, current will shoot up by feww hunderd Amps in that time...

[Arlo1]

Arlo can you explain what you mean by 'combined on/off time of 155 nsec' ? And by '20 usec on time' ?
20usec seems awfully long, current will shoot up by few hundred Amps in that time...

My data sheets for the irfp4110gpbf show an rise time of 67nS and a fall time of 88nS Then added together that's 155nS of switching time. Every time you send an on pulse through one of those fets its going to spend 67nS turing on and 88nS turning off for a total of 155nS in transition per switched on event which during the switching time it is going to produce far more heat inside the mosfet then the rdson value.
The 20uS was a calc from a spread sheet bigmoose gave me to play with it is posted on my inductance thread. The thing is if the fet can only be on for ~20uS in perfect world conditions before its destroyed then its going to have to switch on and off more frequently which causes more heat (and stress) inside the fet it self!

Remember I am looking to build a 50+KW controller!

Here is a screen shot motor A is colossus in stock form note how at the 25uS mark the amps are to hi for a 24 fet controller with 4110 mosfets!

[bearing]

Stock colossus has 8uh! What we need is to wind it to 35kv and terminate it in WYE! I worked all the numbers and it will work out to be 10 turns a tooth and 106uH inductance

Rewinding the motor to get higher inductance will not help, unless you plan to keep the same battery voltage as before the rewind (and get less power in the process).

If you double the amount of winding turns, the inductance will get 4 times higher, and the kV will be halfed. 4 times more inductance will lower rise rate in windings by a factor of 4, and I guess that is what you seek. However, when you double battery voltage to get the same RPM as before the rewind, then you double the rise rate, which will put you at half of what the rise rate initially was. Further, with a halfed kV, the peak current will be halved. So, if the peak current is halfed, and the rise rate is halfed, then the rise time will be the same as it originally was.

If you want to lower the rise time (EDIT, I mean increase rise time / lower rise rate), you could just lower battery voltage, it will have the exact same effect as rewinding the motor without changing battery.

Adding external inductance will not help you get to 50kW either IMO, because that added inductance will prevent currents to rise fast enough to get much torque at high RPM.

I once read about a controller for an ironless motor. It had little different approach compared to the common BLDC controllers. It used the 3-phase bridge only to commutate the machine. Then there was also a PWM stage similar to a brushed DC controller. And in between them there was an inductor. So, it was a high power Buck-converter followed by a 3-phase bridge.

I think the normal hardware approach will do though. I think it's a matter of software. My thought is that the software needs to drive the motor with a waveform which corresponds to the EMF of the motor. If the motor is driven with a constant PWM, then I think the current will spike near the end of every commutation step. When the EMF passes the peak and starts to fall, then the difference between EMF and battery voltage will rise, and the current will then rise too. This is just a theory, and I have no experience in making brushless controllers, but I do have some experience with switched inductors.

[Lebowski]

Arlo can you explain what you mean by 'combined on/off time of 155 nsec' ? And by '20 usec on time' ?
20usec seems awfully long, current will shoot up by few hundred Amps in that time...

My data sheets for the irfp4110gpbf show an rise time of 67nS and a fall time of 88nS

Do you realise that to switch on 4 parallel 4110's (like in a 24 fet controller) in 80 nsec you need a 9 (!) Amps gate drive circuit ? And negligible gate series resistance ?

[Lebowski]

Stock colossus has 8uh! What we need is to wind it to 35kv and terminate it in WYE! I worked all the numbers and it will work out to be 10 turns a tooth and 106uH inductance

If you want to lower the rise time, you could just lower battery voltage, it will have the exact same effect as rewinding the motor without changing battery.

Adding external inductance will not help you get to 50kW either IMO, because that added inductance will prevent currents to rise fast enough to get much torque at high RPM.
...

. My thought is that the software needs to drive the motor with a waveform which corresponds to the EMF of the motor. If the motor is driven with a constant PWM, then I think the current will spike near the end of every commutation step. When the EMF passes the peak and starts to fall, then the difference between EMF and battery voltage will rise, and the current will then rise too. This is just a theory, and I have no experience in making brushless controllers, but I do have some experience with switched inductors.

About the rise time: for this you need real time motor current measurement so that you can switch on the voltage earlier than necessary, such
that the delayed current is actually exactly on time. This, and driving the motor with it's own EMF voltage, is all included in my 30F4011
controller build (see other threads)

[Arlo1]

Stock colossus has 8uh! What we need is to wind it to 35kv and terminate it in WYE! I worked all the numbers and it will work out to be 10 turns a tooth and 106uH inductance

Rewinding the motor to get higher inductance will not help, unless you plan to keep the same battery voltage as before the rewind (and get less power in the process).

If you double the amount of winding turns, the inductance will get 4 times higher, and the kV will be halfed. 4 times more inductance will lower rise rate in windings by a factor of 4, and I guess that is what you seek. However, when you double battery voltage to get the same RPM as before the rewind, then you double the rise rate, which will put you at half of what the rise rate initially was. Further, with a halfed kV, the peak current will be halved. So, if the peak current is halfed, and the rise rate is halfed, then the rise time will be the same as it originally was.

If you want to lower the rise time (EDIT, I mean increase rise time / lower rise rate), you could just lower battery voltage, it will have the exact same effect as rewinding the motor without changing battery.

Adding external inductance will not help you get to 50kW either IMO, because that added inductance will prevent currents to rise fast enough to get much torque at high RPM.

I once read about a controller for an ironless motor. It had little different approach compared to the common BLDC controllers. It used the 3-phase bridge only to commutate the machine. Then there was also a PWM stage similar to a brushed DC controller. And in between them there was an inductor. So, it was a high power Buck-converter followed by a 3-phase bridge.

I think the normal hardware approach will do though. I think it's a matter of software. My thought is that the software needs to drive the motor with a waveform which corresponds to the EMF of the motor. If the motor is driven with a constant PWM, then I think the current will spike near the end of every commutation step. When the EMF passes the peak and starts to fall, then the difference between EMF and battery voltage will rise, and the current will then rise too. This is just a theory, and I have no experience in making brushless controllers, but I do have some experience with switched inductors.

You are right but in this case you are wrong because its to save the controller so its not a fact of getting to the limits of the motor!

[Arlo1]

Arlo can you explain what you mean by 'combined on/off time of 155 nsec' ? And by '20 usec on time' ?
20usec seems awfully long, current will shoot up by few hundred Amps in that time...

My data sheets for the irfp4110gpbf show an rise time of 67nS and a fall time of 88nS

Do you realise that to switch on 4 parallel 4110's (like in a 24 fet controller) in 80 nsec you need a 9 (!) Amps gate drive circuit ? And negligible gate series resistance ?

Yup this is already in place on the china controller.

[Arlo1]

Stock colossus has 8uh! What we need is to wind it to 35kv and terminate it in WYE! I worked all the numbers and it will work out to be 10 turns a tooth and 106uH inductance

If you want to lower the rise time, you could just lower battery voltage, it will have the exact same effect as rewinding the motor without changing battery.

Adding external inductance will not help you get to 50kW either IMO, because that added inductance will prevent currents to rise fast enough to get much torque at high RPM.
...

. My thought is that the software needs to drive the motor with a waveform which corresponds to the EMF of the motor. If the motor is driven with a constant PWM, then I think the current will spike near the end of every commutation step. When the EMF passes the peak and starts to fall, then the difference between EMF and battery voltage will rise, and the current will then rise too. This is just a theory, and I have no experience in making brushless controllers, but I do have some experience with switched inductors.

About the rise time: for this you need real time motor current measurement so that you can switch on the voltage earlier than necessary, such
that the delayed current is actually exactly on time. This, and driving the motor with it's own EMF voltage, is all included in my 30F4011
controller build (see other threads)

Yup thats where im heading. With the not simple brushless controller!

[Lebowski]

Yup thats where im heading. With the not simple brushless controller!

How are you getting along with the programming and stuff ?

[bearing]

real time motor current measurement so that you can switch on the voltage earlier than necessary, such
that the delayed current is actually exactly on time. This, and driving the motor with it's own EMF voltage, is all included in my 30F4011
controller build (see other threads)

Thats impressive, sounds pretty ahead of the competition. Does any of the affordable commercial controllers do that?

I was skipping through your post, but didn't find any related to this subject. Could you give me a link?

[Lebowski]

real time motor current measurement so that you can switch on the voltage earlier than necessary, such
that the delayed current is actually exactly on time. This, and driving the motor with it's own EMF voltage, is all included in my 30F4011
controller build (see other threads)

Thats impressive, sounds pretty ahead of the competition. Does any of the affordable commercial controllers do that?

I was skipping through your post, but didn't find any related to this subject. Could you give me a link?

I posted some stuff on the 'not simple bldc controller' thread, think about page 16 onwards. Once I get the motor
running I think I'll start a new thread about it. There's also a picture here:

viewtopic.php?f=3&t=33403&start=30

showing some of the output you can get from my controller during the calibration phase

[Arlo1]

Yup thats where im heading. With the not simple brushless controller!

How are you getting along with the programming and stuff ?

Just finished my pic kit lessons, finished an led iluminated work bench for my basement, and just finsihed a couple other projects that had to be done. So I will be back on it very soon and its my 1 winter goal to get it working.

[Nuts&Volts]

Just hoping that this motor is still alive and will some day be produced ... I realize some of the engineers are busy with the Rimac EV project

I don't mean to bring back a dead thread, but is there any progress with this motor?

-Kyle

[Arlo1]

Just hoping that this motor is still alive and will some day be produced ... I realize some of the engineers are busy with the Rimac EV project

I don't mean to bring back a dead thread, but is there any progress with this motor?

-Kyle

The motor is basicaly good to go there is just not much for controllers to run it. I anyone wants to help train me with C programin and dspic chips it will help for sure! Marko called me and they do have something going on behind cosed doors! The thing is right now having a motor like this is like having a yaught with a bath tub to use it in! Its. Totaly worthless till a cheep but good controller comes along!!! I plan to have mine done soon!

[Nuts&Volts]

Arlo, thanks for the heads up about this motor still being alive. Don't worry I have been keeping up on the colossus happenings. I want one of these in anticipation of your's (Arlo's), Ricky's or Leb's DIY 3-phase controllers. Or a 144V Kelly if Splinter's Kelly holds up to this beast of a motor.

[tostino]

I've PM'd Marko a few times trying to get in line to buy one, since I am starting up on my ninja 250 conversion project again. I am building two 36 fet infinion controllers to power the colossus I hope to buy.

Still not received any reply from him. If anyone else has info on how to get one of these, i'd really appreciate it.

[Arlo1]

I've PM'd Marko a few times trying to get in line to buy one, since I am starting up on my ninja 250 conversion project again. I am building two 36 fet infinion controllers to power the colossus I hope to buy.

Still not received any reply from him. If anyone else has info on how to get one of these, i'd really appreciate it.

For this motor I'd recomend a controller with ~40khz pwm and phase amp monitering. So even two 36fet controllers may not work well. I have experimented with splitting a 36 fet controller in two but... Anyways me ricky and lebowski have great controller projects on the go!

[tostino]

Eh, i'm not too worried. If I blow them up, I blow em up.
If nothing else, by the time I get everything together and blow the controllers up, there will be better options I can buy available, because right now as far as I know there isn't.

Do you know anything about the motor availability though?

[Arlo1]

Eh, i'm not too worried. If I blow them up, I blow em up.
If nothing else, by the time I get everything together and blow the controllers up, there will be better options I can buy available, because right now as far as I know there isn't.

Do you know anything about the motor availability though?

This thread should get locked till it gets sorted out! We cant dyno or test colossus under hi load untill the controller thing is sorted out!!!! So they will not be for sale till afte that! I have a couple prototypes and i have blown up a lot of shit testing them!! Remember you may turn on your two controllers and turn the throttle slowly to hear a small pop then nothing!!! That has happened to me over ten times alone!