Possible to share controller load onto multiple controllers?

macribs

10 MW
Joined
Jul 22, 2014
Messages
3,702
Would it be possible to share controller load onto multiple controllers? We have seen those 6 phase hubmonsters running dual controllers and share the load.
But would it be possible to share the load onto two or more controllers on a regular 3 phase motor? Each controller would run lighter load with less chance of melt down.

Could this be done? And if it does work, would it work for sensorless only?


This might be basic know how for those of you with EE background or those of you not sitting in the back row in school daydreaming. I have no clue so hence this Q. :oops:
 
No. Not stock. Not for a 3 phase motor.
The controllers would need to be in perfect sync every time they fired on each coil. But that won't ever happen, since the two devices have separate internal clocking, and separate internal programing to respond to different thresholds and inputs. Even if you could get them in sync, they wouldn't stay that way as timing will never stay perfect, and even if the programing is identical, the variance in different systems would mean one controller might decide to cut back power 1 clock cycle earlier than the other, or ramp up power on a different curve, or pull back power differently due to small temp changes.

You could modify the devices to work off the same chip, timing, halls, shunt, etc, etc, etc, but then all you would really be doing is turning 2 controllers into 1.
 
So I gotta confess I tried this on two Ku63 controllers. And a 29mm Mxus V2. Works both sensored and sensorless.
Mounted motor in a wheelstand and applied as much resistance as I could with my wife's oven mitts. Sorry honey!
I need 36v 50a for this motor in a really fast wind, so I hope it all goes ok when I mount them on a bike for further tests.

Tonight I'm gonna try it with some 48-84v 45a Greentimes if time permits. Because I can.

https://www.youtube.com/watch?v=TFYn2Hk3W68
 
What does it take then to beef up an existing controller. I see on the back of my Lyen controller there are some massive amounts of solder, so then would soldering in a copper rod be ideal. Then of course the shunt would need to be modified right, need to beef that up. That can't be all can it.
 
Depends on the fets and caps etc. But sometimes it can be.

The reason I am doing this is that after a point (about 5kw) controllers get damn big, bulky, and damn expensive. Smaller controllers sell more so they are cheaper per battery watt. And it's appealing to me to be able to limp home at half power if I blow a controller.

Actually there's an idea. Running a 250w controller in parallel and enabling it only with a flick of a switch. Hello compliance. :mrgreen:
 
Well twin greentimes work great in sensorless. 84v 90 a.
Need to try sensored next.
http://youtu.be/v537OGkb8_k


Sent from my iPhone using Tapatalk
 
So cool Sam. Any thoughts how it will work under load? Controllers should get an easier job, what about the motor? Total Amp increase will increase motor temp, or what will be the net result you think? Will you do a real world test with load also? Like mounted to a bike riding uphill? Really would love to see twin controller setup tested in Justin's heat thread. That would be cold hard facts, we could see how much power increase output, efficiency and if there are losses somewhere uncounted for. Any chance you got access to a dyno?

Will there be any problems you think with ground loops using one battery? Maybe there is also a need to run dual batteries, and that would maybe counterproductive for the winnings of dual controllers.


Look forward to see how sensored mode works out. That is some awesome Pioneer work done there Sam.
 
I've been applying as much load as possible using my wife's oven mitts held hard onto the spoke flanges. She now has a new set of gloves as I burnt the old ones.

I'll burn the new ones this weekend.

Yes double amps means double torque.

I'm excited that it might work for a few reasons. One is that smaller controllers are cheaper. 10kw controllers are stupidly expensive and too big to fit on bikes.

I'm also excited that a second controller should shed heat over more area in many instances.

And in the event that you blow one controller you get a second to get you home.


Sent from my iPhone using Tapatalk
 
Literally just parallel up the phase, battery and throttle wires.

I havent tried sensorless yet. Hope to try soon on my QS273 scooter.

I've left the 3 speed alone.

I put ignition switches on both to test each one spins in the same direction before enabling both. That may sound cray cray but greentime often use different coloured wires inside and I don't want the phases to be unmatched.
 
Lebowski said:
It'll go wrong practically immediately. On one motor phase one controller will switch on the high side, the other the low side and that will be the end of both controllers...
I was thoroughly chastised for even asking about the technical reasons why this wouldn't work on a brushed controller!
https://endless-sphere.com/forums/viewtopic.php?f=10&t=12292&hilit=brushed+controller#p183091
The second layer of 3 phase switching adds even more things to go wrong.

My prediction is that, at best, putting two identical controllers in parallel will result in a very slight, probably imperceptible, increase in current to the motor. It might share the heat across the two controllers.

I don't think it will double the amps to the motor because during the PWM off the motor's inductance causes BEMF already, so the most the other controller will be doing is 'topping off' that BEMF.
 
If one controller commutates just slightly before the other one, there is a good chance you'll have 'shoot through' where one is fighting directly against the other, resulting in destruction. Depending on the commutation timing, there is a little dead band between high and low states, so you might have a little tolerance.

To make it really work reliably, I think you'd need to synchronize the output stages. This could be done by making one controller the 'master' and disconnecting the gate driver inputs on the other unit(s) and use 6 small wires to carry the gate drive signals from the master to all the secondaries. This way all the FETs will switch at exactly the same time.

You would still have some issues with current sharing and measurement, so don't expect two controllers in parallel to be exactly twice the rating, but it might come close.
 
Arlo1 I think posted this video of a Zero with close to double HP due to dual controllers.

[youtube]3eL0EMFPnVw[/youtube]
 
Samd said:
So I gotta confess I tried this on two Ku63 controllers. And a 29mm Mxus V2. Works both sensored and sensorless.
Mounted motor in a wheelstand and applied as much resistance as I could with my wife's oven mitts. Sorry honey!
I need 36v 50a for this motor in a really fast wind, so I hope it all goes ok when I mount them on a bike for further tests.

Tonight I'm gonna try it with some 48-84v 45a Greentimes if time permits. Because I can.

https://www.youtube.com/watch?v=TFYn2Hk3W68


This is a surprising result. I also expected timing errors to cause failures. But the truth is strange, if you look at the commutation table, it is a gray code type sequence, and only one value changes at each transition. Since the other coil MUST be the changing element, the direction in/out of the common coil terminal cannot change. So there can never be a shoot-through event type conflict. Paralleling controllers only demands fairly good timing, not precision timing.

A->B
C->B
C->A
B->A
B->C
A->C

Strange but true. So timing errors may cause both coils to be energized for a moment, but sinewave controllers do that all the time without problems.

Sensorless mode does the detection of the transition well before the switching event (by monitoring the voltage on the floating terminal), so it is not disturbed by the parallel controllers timing errors (driving the terminal it is monitoring).

If a sensorless controller loses sync it might cause a shoot through event, so I think this is more dangerous with sensorless controllers. It depends on how they handle error conditions and how quickly they recognize problems.

Very interesting, indeed.
 
Alan B said:
As I recall the zero motor had two sets of windings so not quite the same test. :)

Two sets of windings, like a 6 phase motor, like those hubmonstermotors for John in CR? Those monstermotors from John has two sets of windings, so kind of acting like two separate motors in one motorshell. Seems strange that Zero didn't take full advantage of such windings, running the motor with only one controller.
 
fechter said:
If one controller commutates just slightly before the other one, there is a good chance you'll have 'shoot through' where one is fighting directly against the other, resulting in destruction. Depending on the commutation timing, there is a little dead band between high and low states, so you might have a little tolerance.

To make it really work reliably, I think you'd need to synchronize the output stages. This could be done by making one controller the 'master' and disconnecting the gate driver inputs on the other unit(s) and use 6 small wires to carry the gate drive signals from the master to all the secondaries. This way all the FETs will switch at exactly the same time.

You would still have some issues with current sharing and measurement, so don't expect two controllers in parallel to be exactly twice the rating, but it might come close.
When the output stages (power FETS) aren't synchronized, one controller can feed the other's flywheel diode instead of the motor coil? So, even if Samd's paralleling of two unsynchronized controllers does work, it might be heating up the flywheel diodes wastefully?
 
gogo said:
fechter said:
If one controller commutates just slightly before the other one, there is a good chance you'll have 'shoot through' where one is fighting directly against the other, resulting in destruction. Depending on the commutation timing, there is a little dead band between high and low states, so you might have a little tolerance.

To make it really work reliably, I think you'd need to synchronize the output stages. This could be done by making one controller the 'master' and disconnecting the gate driver inputs on the other unit(s) and use 6 small wires to carry the gate drive signals from the master to all the secondaries. This way all the FETs will switch at exactly the same time.

You would still have some issues with current sharing and measurement, so don't expect two controllers in parallel to be exactly twice the rating, but it might come close.
When the output stages (power FETS) aren't synchronized, one controller can feed the other's flywheel diode instead of the motor coil? So, even if Samd's paralleling of two unsynchronized controllers does work, it might be heating up the flywheel diodes wastefully?

Funny you should mention this. As I was having lunch I was thinking about this. I don't think it is a problem with flywheel diodes, they only act when the voltage gets beyond battery plus and minus, but if it was a synchronous controller the active FET they use, instead of the freewheel diode, would turn on and create a shoot-through situation with two controllers in parallel. So essentially the paralleling requires the freewheeling diode to make it work.

The controllers are synchronized by the hall signals. Each regulates current via PWM and supplies as much current as it thinks it can as measured by its shunt.

Here is a phase sequence:

A->B
C->B
C->A
B->A
B->C
A->C

During the transition one controller will move to the next coil a bit ahead of the other, but it won't matter. Take the first and second phase combos in the above chart for an example. Both controllers are feeding current into the A terminal and grounding the B terminal. One of the controllers reacts to the halls first and removes its A drive and begins driving C. It maintains the B connection to ground. A moment later the second controller follows suit. No short circuit.

There may be some funny problems that crop up in doing this. For example the ground paths through the two controllers, and the two shunts may split the current in unintended ways. Especially if the controllers are different sizes. It could be a problem. Then again, the currents might divide properly, depending on the details.
 
macribs said:
markz said:
What does sensorless mean?
Not using the halls?

Correct, sensorless get the timing from BEMF.

Thanks
I cant get my new halls on the mxus3kw v2 to operate properly.
Can the xie chang / infineon / lyen modfied controllers be used sensorless?
I think the sunwin can go sensorless because it has the "learn" mode.
 
Can you use a greentime? That'd be my choice.


Sent from my iPhone using Tapatalk
 
I'm still waiting to do some more testing guys but for now its looking like all may go well.


Sent from my iPhone using Tapatalk
 
I find it odd a few seasoned e-veterans dismiss the possibility of sharing controllers on a 3 phase motor, yet you seem to be doing the unthinkable. Its neither here or there, just saying is all.
 
Back
Top