Possible to share controller load onto multiple controllers?

[Alan B]

It is a strange situation. At first look, connecting power electronics in parallel would seem to create a situation where a small timing error could cause disaster - a shoot through event where the power devices would simultaneously switch the motor terminal to both the high side and the low side. If this were to occur the FETs would be destroyed in short order.

But the particular way the phase switching occurs makes these timing errors (which are not going to be zero) into fairly benign events.

It is just not something that was studied carefully before.

I was ready to dismiss it as being a bad idea for exactly the same reasons, but when it worked I took a closer look and immediately understood why it worked.

It is good to learn something new every day.

Thanks for the experiment!

There are still risks doing this, and if either controller makes a mistake or misreads a hall it is likely to blow both controllers. The controllers may not share current the way you would like, depending. There are lots of details that affect this. But the thing that was expected - that the inevitable timing errors would immediately cause a shoot-through - that just turns out to not be the case. These small timing errors cause current to be run through 2 of the three coils in a compatible way. It is not the best way to get higher power, but it does look feasible. Controllers that were designed to be paralleled could do this even better.

 

[markz]

The good news is, there are controllers that are el-cheapo-mundo.

 

[Alan B]

Make sure they're phased the same before paralleling them. The risk is real.

 

[macribs]

The good news is, there are controllers that are el-cheapo-mundo.

Also of importance is the possibility to save space using two smaller controllers that can more easily be tucked away rather then one large hard to place controller.
I wonder how the Phaserunners from Justin would work in a dual setup, those are what 8 kw peak controllers. How much power will dual set up give? If they work in dual setup you get a little extra safety as you can limp home with one controller. Those tiny Phaserunners will not take up much real estate, and could easily be using frame as heat sink. And the price is ok too.

 

[Samd]

That's true.


Sent from my iPhone using Tapatalk

 

[t3sla]

Be interesting to see what happen when you roll backwards and when brakes are locked zero start

 

[Samd]

I gotta test that for sure.

 

[fechter]

If the controllers don't have synchronous rectification (like most Xie Chang units), then shoot-through won't happen, but the PWM frequency of the two controller is likely to be slightly different. The difference between the frequencies will create a 'beat frequency' where the outputs are in phase and out of phase. When they are in phase, they will share nicely. When they are out of phase, the effective duty cycle might be twice the in-phase amount. If the beat frequency is high enough, you might not notice it other than a humming sound. If the beat frequency is like 1Hz, you might wind up with a pretty bad surging or pulsating in the motor. If the controller's current limiter responds fast enough, it might compensate for some of this. This problem won't happen when the controllers are at 100% duty cycle and there is no more PWM. It would be most noticeable at low speeds.

It will be interesting to test.

One other thing to watch out for is to make sure both controllers have a very low resistance path between their negative battery connections. A voltage difference caused by IR drop in the wire could mess with the throttle signals. Locating both controllers physically close with a short wire between the negative connections should avoid any issues.

 

[Martin A]

Arlo1 I think posted this video of a Zero with close to double HP due to dual controllers.

[youtube]3eL0EMFPnVw[/youtube]

Looks Awesome. What thread was this in?

 

[Alan B]

If the controllers don't have synchronous rectification (like most Xie Chang units), then shoot-through won't happen, but the PWM frequency of the two controller is likely to be slightly different. The difference between the frequencies will create a 'beat frequency' where the outputs are in phase and out of phase. When they are in phase, they will share nicely. When they are out of phase, the effective duty cycle might be twice the in-phase amount. If the beat frequency is high enough, you might not notice it other than a humming sound. If the beat frequency is like 1Hz, you might wind up with a pretty bad surging or pulsating in the motor. If the controller's current limiter responds fast enough, it might compensate for some of this. This problem won't happen when the controllers are at 100% duty cycle and there is no more PWM. It would be most noticeable at low speeds.

It will be interesting to test.

One other thing to watch out for is to make sure both controllers have a very low resistance path between their negative battery connections. A voltage difference caused by IR drop in the wire could mess with the throttle signals. Locating both controllers physically close with a short wire between the negative connections should avoid any issues.

I think he's already done it with good results (post #4 in this thread, and reported on the Facebook group).

I agree with what you said, and note that at each hall transition the controller PWM will be resynchronized so a pair of the same controllers (seeing the same throttle signal) will tend to stay synced pretty well. Hall transitions occur quite often, especially in a geared or high pole count DD motors.

 

[macribs]

Arlo1 I think posted this video of a Zero with close to double HP due to dual controllers.

Spoiler: DualController

[youtube]3eL0EMFPnVw[/youtube]

Looks Awesome. What thread was this in?

Can't remember, and I didn't bookmark. Maybe someone else reads this and can help you out. I think Arlo1 was working on that bike for a friend iirc.

 

[Alan B]

Arlo's Dyno thread.

 

[Martin A]

Thanks Alan and Amberwolf. Here's the thread for others who may be interested:

https://endless-sphere.com/forums/viewtopic.php?f=1&t=53617&p=1168431&hilit=3eL0EMFPnVw#p1168431

 

[fechter]

I agree with what you said, and note that at each hall transition the controller PWM will be resynchronized so a pair of the same controllers (seeing the same throttle signal) will tend to stay synced pretty well. Hall transitions occur quite often, especially in a geared or high pole count DD motors.

I guess I didn't think the PWM oscillator ever changed phase, but in a digital controller I can see how that might be the case. So in essence, the two controllers are sync'd by the hall signals. It might go weird first starting up from a dead stop, but that certainly won't last long.

 

[Alan B]

Right, I've been studying algorithms for motor controllers lately, and the software is likely to start up a new PWM when triggered by the hall signal. If it was an older one with limited PWM generators it might use hardware to vector one PWM to different FET drivers, but all the modern chips have a PWM generator for each of the six FET output signals. It would be hard to maintain PWM phase from one generator to another and not worthwhile. Now it is possible that on every other hall transition they could leave the PWM running while they switched the commutation side FET, but it would still resynch every other hall transition that way, and if it was my software I would think that shutting off the PWM during the commutation would be better for the FETs though maybe it doesn't matter much. The commutation side FETs don't need to be hot switched, though if 100% PWM is used they might be. Lots of details depend on how they coded the algorithms and used the hardware.

 

[markz]

Using 3 controllers would be pushing it.

Paralleling the hall and phase wires on a matched pair of controllers then what about the throttle, same thing parallel the wires?

 

[Alan B]

This is very experimental. Proceed at your own risk. There are many ways this might go bad.

But, surprisingly to a lot of us, it doesn't instantly blow up, at least with the controllers Samd is experimenting with.

This is what I would do. I have NOT done this myself.

Use two identical controllers.

Parallel connect the motor phases, halls and power inputs. The input battery grounds should be very short, low resistance and connected together as close to the controllers as practical.

The throttle is fully connected to ONE controller, and a single wire carrying the throttle voltage passes over to the second controller.

They should be programmed and calibrated identically, and already set to rotate the motor the proper direction. If anything gets out of sync, well you can guess.

Then it might work. Or it might still die in flames.

Please video the flames.

If one controller goes into regen and the other doesn't, bad things are likely. Regen is probably a bad idea in this setup. Or it might work ok, depending on how they programmed it.

Lots of uncertainty.

That's why it is an experiment.

 

[Samd]

Good summary

Also note that the dual sevcon video above is a bit different - Sevcons and Adappto's allow a master/slave relation to be set up where one CPU is doing all the driving.

My experiments to date have been about (almost)syncronising two controllers each with their own CPU/clock etc.

 

[markz]

Anymore updates?

 

[Samd]

Hahahaha. Too many projects!!!


Sent from my iPhone using Tapatalk