Video proof - myths of 120% throtle setting and more!

There are a lot of rumors surround these controllers so I decided to dispel some of them and post hard evidence in the form of oscilloscope videos demonstrating what the controller does in each throttle mode.

Test setup is one of my custom built EB224-AS-1 IRFB4115 24 FET controllers running an unloaded Cromotor V0 with a 16" moped tire. I have no real way to load the system down besides the brake and really didn't have enough free hands to do that kind of testing right now, so these are all unloaded tests, but the PWM that occurs will hold true no matter what the load. I'm sure as more current flows there will be some spikes and possibly ringing on the FETs but what I saw here was minimal. I originally wanted to do some of these tests because of some of the conversations that Arlo1 and I have had about overshoot, he was curious to see what happens on these controllers. Hopefully I caught some of the info for you buddy.

As for the other topics... On to the myths!

Myth 1: The controllers never get to 100% throttle and is always in PWM - 100% FALSE!
I have read many times on here that these controllers never go to 100% throttle. This is not true at all, if you program the controller for 100% throttle on one of the switch settings you will drop out of PWM mode and end up at full on for the commutation.

Myth 2: The controllers stay in PWM when programmed for 99% throttle which is good for low inductance motors - 100% True
I've read that running these controllers at 99% throttle vs 100% or higher will help them drive low inductance motors with less failures. This should be true, because when I program my controller for 99% throttle it will NEVER drop out of PWM. This is a good thing if you have a hard to drive motor.

Myth 3: The controllers advance the hall timing to increase the speed when programmed over 100% - False?
I know that there has been a lot of guessing how the Xie Chang controllers increase the speed of a motor when set to over 100% throttle such as 105% all the way up to 120%. Some say it advances timing, some say it changes the throttle input, etc. Watch the video and see for yourself. What occurs is the normal trapezoidal wave form ends up going into a square wave with a little bit of PWM and is extended by about 10% past it's normal commutation cycle. My best guess without measuring one of the other phases simultaneously is that the PWM seen in this mode over 100% throttle inversely coincides with what is happening on the other phases thus preventing two phases being on simultaneously. In other words, phase A is on with no PWM, but as it starts to overlap phase B, they add in PWM that is the opposite of the next phase. In other words, phase A overlapping PWM is high, phase B PWM is low. This allows the over all time power is applied to the motor to be extended by around 10% providing additional time the motor is under power by eliminating the trapezoidal part of the wave form. At least this is my best guess as to how it's operating. It's certainly strange and I'm no expert, so motor experts, please add your thoughts.

Here is the video (it's uploading right now, give it about an hour from this post and it should be available.

[youtube]3tIt_FMTa6I[/youtube]

zombiess said:

What occurs is the normal trapezoidal wave form ends up going into a square wave with a little bit of PWM and is extended by about 10% past it's normal commutation cycle.

Click to expand...

I noticed this too. In this video the motor is in 100% mode at the beginning, @0:26 it's in 120%. Notice the change in the waveform.
[youtube]_WbRy3ZkAh8[/youtube]

Can you explain how that is working. Sorry my vid isn't finished uploading yet.

zombiess said:

Can you explain how that is working.

Click to expand...

The top wave is Hall signal, the other one is phase.
Halls weren't placed at neutral timing slot from memory.
Looks like the controller goes from trapezoidal to sq. wave envelope.

full-throttle said:

zombiess said:

Can you explain how that is working.

Click to expand...

The top wave is Hall signal, the other one is phase.
Halls weren't placed at neutral timing slot from memory.
Looks like the controller goes from trapezoidal to sq. wave envelope.

Click to expand...

I mean the 120% throttle. Do you think my guess of operating is on the right track?

Great work Zombiess. I love when myths get busted, since the truth is better than fiction.

I need help with Myth #2 though. I believe that it keeps the controller in PWM, but I don't see how that helps protect the controller from a low inductance motor. In fact, that's the part that sounds like myth to me, since going to full duty reduces switching and its losses. Are those who put forward the 99% speed setting idea trying to say that the controller goes to full duty at full throttle even at low rpm, so the current protection of PWM is lost? If that's what is meant, then I don't think a no load test is going to cut it, because you need to see what happens during acceleration. I believe what you'll find is that it doesn't end PWM and go to full duty until somewhere around peak power (around 50% of top speed) where rpms are sufficient for BEMF to make the controller safe.

As support of my view I offer 2 factoids. Note that I always use a max speed setting of 100%, never 99% and never greater than 100%.:
1. I've blown a lot of controllers, and every one failed at fairly low speed, and almost all at partial throttle.
2. In the course of tuning I have seen an unexplained surge of power during acceleration. This was without the CA's throttle override wire connected, so the surge couldn't be CA caused. What happened was that during hard acceleration, at about half of top speed the power surged significantly with a sudden thrust of additional acceleration. This effect was noticeable from peak input of about 15kw up to about 25kw, and was most noticeable in the low 20's.

Your observation about 100% throttle causing full duty and an end of PWM seems to be a plausible explanation of the surge in power that I've experienced. It makes sense to me at such high power, that the increase in RMS current as the controller goes from PWM to full duty would be significant enough to feel.

John

John. I have blown a lot of controllers on colossus too. I always had a 100% 0r higher setting. 99% would have helped save them from blowing in some cases.
I had a few dyno runs and a few test rides where I was full throttle from 0-about 50km (might be where the PWM stops) and the controller popped at these speeds.
I can also feel this burst of power when it transitions from pwm to full block communication on my X5 powered bmx. I won't be trying any more testing with these controllers on colossus unless I get the controller for free and I have something to gain from it. I truly feel there is 2 reasons these china controllers fail with low inductance motors and that is the transfer to full block communication and the fact they only watch battery amps. The way I see it is they program the controller with calculations based on X motor inductance and then they know that at a certain rpm and certain battery current the motor current would be a certain number as well. But when you replace the motor with one with lower inductance they don't know and allow to much current flow to the motor and this is the current the fets see causing them to pop.

Zombies how hi can you get the spike over source voltage? Keep scoping Its time to bump up the scope to a 100mhz or higher big moose says you need 100mhz or higher as well. I did not see the inductive spike I have been working out until I upgraded my scope from 50-100mhz so its possible you will see more with a faster scope.

That discussion is becoming really interesting!, great Work Zombiess!!

We had alot of problem with the Kelly 144V 600A BLDC $$$ chineese controller and the dual 602 enertrac motor.. ( just like connecting two arallel 5302... :lol: ) and the controller is always seeing surge and pop alot of error if we crank to omuch the throttle... we will test with 99% throttle max setting for fun..

Doc

Jeremy You kinda showed what I though was going on. I origanly though it was advancing timing but then others pointed out the full block thing. SO my guess was it was widening the block. I think the added pwm is to keep the current within a reasonable level. You should mesure the time from the start of the pwm to the hall sensor and try mesureing at 99% 100% 105% 110% and 120% then do the same on the end of the cycle to the end of the hall.

Arlo1 said:

John. I have blown a lot of controllers on colossus too. I always had a 100% 0r higher setting. 99% would have helped save them from blowing in some cases.
I had a few dyno runs and a few test rides where I was full throttle from 0-about 50km (might be where the PWM stops) and the controller popped at these speeds.
I can also feel this burst of power when it transitions from pwm to full block communication on my X5 powered bmx. I won't be trying any more testing with these controllers on colossus unless I get the controller for free and I have something to gain from it. I truly feel there is 2 reasons these china controllers fail with low inductance motors and that is the transfer to full block communication and the fact they only watch battery amps. The way I see it is they program the controller with calculations based on X motor inductance and then they know that at a certain rpm and certain battery current the motor current would be a certain number as well. But when you replace the motor with one with lower inductance they don't know and allow to much current flow to the motor and this is the current the fets see causing them to pop.

Click to expand...

I guess I can see the current surge being enough to push one over the edge in some conditions, but other than that I find it baffling that you guys think switching saves anything. I see switching as the cause of problems not the solution. I want my controller working at full duty. I always visualize current in terms of water flow, and with a low inductance motor turning that faucet off and on turns it into a water hammer.

What's the worst thing you can do with a low inductance motor?...Partial throttle up a hill. While high throttle like 99% isn't where I had issues 99% is still partial throttle

Colossus has such a low inductance that you probably didn't get enough ride time to determine which conditions are ok and which aren't. Colossus is just crazy hard to drive for controllers, so why not just add some more turns and increase the voltage? I see it more as a motor problem than a controller problem.

What kind of phase/battery limit ratio were you running? Keeping that ratio low is what I attribute my success at high power with single controller rigs. Now with the dual controllers for my 6 phase motors high power is simple because the split load doesn't stress any part of the system ?

Another monkey wrench in the mix is that I've found that the Wuxi controllers have an easier time driving the low inductance motors than the XieChang controllers. Jeremy Harris suggested that Wuxi uses a lower PWM rate.

Doc,
Why not run the double stator Enertrac with 2 controllers? It doesn't help with the inductance, but phase resistance doubles and current is cut in half, and it's much easier to find 2 controllers each pushing half the power.
John

John 3:1 2.5:1and 2:1. You have to understand in 22useconds of on time the fets will be flowing enough current to kill them selfs in a 24 get 4110 controller running colossus. It's not a motor problem because as you approach performance resistance and inductance drop. I have rewound the motor to 1/2 the kv and 4 x the inductance but this doesn't help as muchbecause the needed higher voltage to run the motor becomes very expensive and hard to controll. The OEM motor wants 130-170 volts to make some great power how hi of voltage do you think I need to run? Ever priced out caps for 300+ volts? 170v dc is scarry as hell....
Also look at the amount of time between phases at a given rpm if you feed any motor with full block communication you will either blow the fets or put unessacary heat into the motor. Pwm is the right way to run a motor but there is now better ways of utilizing the pwm. Sine with Foc and center aligned time base is going to change it all. Think of it as putting as much amparage in at the best spot possible so you get the most torque you can with the least amount of wasted power.

John in CR said:

I need help with Myth #2 though. I believe that it keeps the controller in PWM, but I don't see how that helps protect the controller from a low inductance motor. In fact, that's the part that sounds like myth to me, since going to full duty reduces switching and its losses. Are those who put forward the 99% speed setting idea trying to say that the controller goes to full duty at full throttle even at low rpm, so the current protection of PWM is lost? If that's what is meant, then I don't think a no load test is going to cut it, because you need to see what happens during acceleration. I believe what you'll find is that it doesn't end PWM and go to full duty until somewhere around peak power (around 50% of top speed) where rpms are sufficient for BEMF to make the controller safe.

As support of my view I offer 2 factoids. Note that I always use a max speed setting of 100%, never 99% and never greater than 100%.:
1. I've blown a lot of controllers, and every one failed at fairly low speed, and almost all at partial throttle.
2. In the course of tuning I have seen an unexplained surge of power during acceleration. This was without the CA's throttle override wire connected, so the surge couldn't be CA caused. What happened was that during hard acceleration, at about half of top speed the power surged significantly with a sudden thrust of additional acceleration. This effect was noticeable from peak input of about 15kw up to about 25kw, and was most noticeable in the low 20's.

Click to expand...

Read up on how inductors function, it makes more sense to me after I read this article and its example with an inductor in parallel with a light bulb.
http://electronics.howstuffworks.com/inductor.htm

The PWM (switching on and off) constantly creates a resistance to the flow of current as the field builds. How long that field takes to build depends on the inductance of the motor, higher inductance, the slower the field builds/collapses and low inductance means field builds fast/collapses fast. So if you think about a motor and PWM, you are constantly causing a field in the coils to build and collapse which acts as a resistance to the flow of current and therefore causes the FETs in the controller to flow less current. If you remove PWM and go to 100% duty the field builds, but if the period is too long for the amount of inductance the motor is at, then it acts as a dead short to the FETs. The amount of time the field takes to build/collapse is determined by the inductance of the motor which changes with RPM. Keeping a lower inductance motor in PWM causes the field to switch on / off multiple times per commutation vs just once like it does at 100% PWM.

As for load, the controller has no idea about load and from what I've seen produces 100% PWM if commanded to 100% PWM by the throttle. I have previously tried loading a wheel with the rear brake and it didn't make any difference. I did this a around a year ago, but I could try it again if I break out the test setup again for more video. I can block the rear wheel and just crank the controller WOT into a stalled motor, it should go to 100% PWM and then into protect mode and shut down.

One of the more experience members might be able to come up with a better example. I do know form reading literature that keeping a controller in PWM helps drive low inductance motors more reliably, but at the cost of increased switching losses.

When you had your failures where they low side or high side FETs? The low side never goes into PWM under any conditions, it is always 100%.

On thing to consider is the pole count when understanding if full block communication is a good thing or not. Ie crowmotor has 23 poll pairs right Jeremy? Colossus has 10. Giving full block at 1 rpm that's 23 erpm for crowmotor and 10 erpm for colossus. So that's .043 min per erevalutions for crowmotor and .1 min per erevalutions for colossus. Which is .007245 min per block section for crowmotor and .016666 min per block for colossus. Which meens if you were to run crowmotor at full block at 1rpm you would have mosfets on for .4347 seconds at a time at 1rpm. And full block for colossus at 1rpm would meen the fets are on for .999999 seconds at a time. You can work the math from there for different motors. And if you don't find the spread sheet bigmoose posted on my inductance thread so you can see how fast the phase amps build I will later.

Not load, but rpm. I don't think there's any way it's going to full duty on the launch. That combined with my use of low ratios of phase:battery current limits is why my bikes never reach peak power input until the 25-30mph range. Full duty without the significant rpm would mean no current limiting just because you went to full throttle.

Yes, I understand that that inductors slow current. That's why low inductance motors are more difficult to drive because what's coming back at them from the motor has more jagged and sharp edges each time it's turned on and off, and a higher inductance softens and rounds the edges.

I always defer to you guys about electronics, but I feel like you may be looking at it from the wrong direction. I've always been under the impression that it's the switching of the high current that is more of a problem than the current level itself. Sure the surge in RMS current that Arlo1 and I have seen as the controller goes to full duty could push fets past their limit, but I've only seen that surge on one motor and the controller never blew and that surge had to be at least 50A. During current limiting is where I've blown almost every controller, and I think it's almost always the high side fets that go (though I could be wrong about that, since I stopped even bothering to open them up).

I understand Arlo1's point about the Chinese controllers not having fine control over current, but to me that's not relative to this 99%/100% thing.

As another example of how partial duty is like a saw blade to our controllers, I have blown 2 controllers on flat ground at less than 5mph. Both were at very low throttle positions where it was impossible not to pulse the throttle. OTOH, I have never blown any controller at significant speed and WOT. The most common situation of blown controllers has been going up hills at partial throttle where traffic didn't allow me to reach what I consider a minimum safe speed.

I'm a hell of a lot more concerned about my controllers' safety going over the hill near my house using the route with the 5-6% grade, than the highway route over the same hill that 7-8% grade. That's because I can go faster, often even WOT up the highway route, but curves and traffic keep me at partial throttle on the other route. I've blown more controllers on that 5-6% hill than anywhere else, and I've never blown one on the highway route, despite the much higher current required for the much higher speed.

I want my controllers going to full duty.

John

i gota see these videos when i get home..

I've been talking with zombies via pm on some issues i've had driving my x602... and have discussed these topics thanks to his R&D

I will post some of his pm's .. hope you don't mind zombies.. but they shed some very helpful info IMO..

i'll post when i can.. gota sort em out..

thanks
-steveo

John in CR said:

Doc,
Why not run the double stator Enertrac with 2 controllers? It doesn't help with the inductance, but phase resistance doubles and current is cut in half, and it's much easier to find 2 controllers each pushing half the power.
John

Click to expand...

yeah.. at 2500$ each with shipping and customs.. there is no way we should have to use 5000$ of controller!!! Damn! .. We had to take the kelly because it was the only one BLDC controler that can take up to 180V and 600A... but... it ws not cheap!

If only sevcon would be availlable at higher voltage and that would be as easy as the kelly to program and no need to buy the 1000$ programmer kit....too...

Otherwise we would have choose the sevcon as well...

Doc

John in CR said:

(...) and I think it's almost always the high side fets that go (though I could be wrong about that, since I stopped even bothering to open them up).

Click to expand...

Can make sense if those controllers don't do synchronous rectification and let the current re-circulate through the the top MOSFET's internal diodes (which causes a lot of heat dissipation).

John in CR said:

As another example of how partial duty is like a saw blade to our controllers, I have blown 2 controllers on flat ground at less than 5mph. Both were at very low throttle positions where it was impossible not to pulse the throttle. OTOH, I have never blown any controller at significant speed and WOT. The most common situation of blown controllers has been going up hills at partial throttle where traffic didn't allow me to reach what I consider a minimum safe speed.

Click to expand...

As the motor spins faster, back EMF builds up and this reduces the current in the motor, this can be the cause. Most current is consumed during acceleration, in particular from a stop.

So can anyone disprove or prove that keeping a low inductance motor in pwm will make a controller less likely to fail. Something beyond anecdotal observations. I believe I understand why it works, I just like to have confirmation so I don't go around spreading inaccurate info.

This is awesome man! Thanks!

It isn't really advancing the timing, but it kind of is, since it can PWM a little bit at the beginning of the cycle. Advancing the timing would really be a misnomer though as you've mentioned. This makes alot more sense after seeing the scope plots.

+10 life points 8)

zombiess said:

So can anyone disprove or prove that keeping a low inductance motor in pwm will make a controller less likely to fail. Something beyond anecdotal observations. I believe I understand why it works, I just like to have confirmation so I don't go around spreading inaccurate info.

Click to expand...

Can you get your wife to hold the brake or something, to demonstrate that a controller can go to full duty at low rpm? That's my biggest problem with the concept, because I believe once you get to an rpm above current limiting that BEMF protects the controller and you're home free, and at rpms below that point full duty cycle isn't possible. If full duty happens at relatively low rpm, then there would be no control over current and insufficient BEMF to protect the controller, so I'd agree with you 100%. That would explain many of my blown controllers.

Has anyone ever blown a controller at high speed that couldn't be easily explained by a lack of cooling air flow or a phase wire short?

John in CR said:

zombiess said:

So can anyone disprove or prove that keeping a low inductance motor in pwm will make a controller less likely to fail. Something beyond anecdotal observations. I believe I understand why it works, I just like to have confirmation so I don't go around spreading inaccurate info.

Click to expand...

Can you get your wife to hold the brake or something, to demonstrate that a controller can go to full duty at low rpm? That's my biggest problem with the concept, because I believe once you get to an rpm above current limiting that BEMF protects the controller and you're home free, and at rpms below that point full duty cycle isn't possible. If full duty happens at relatively low rpm, then there would be no control over current and insufficient BEMF to protect the controller, so I'd agree with you 100%. That would explain many of my blown controllers.

Has anyone ever blown a controller at high speed that couldn't be easily explained by a lack of cooling air flow or a phase wire short?

Click to expand...

Sure I can, and it sounds like a good test to try out. I'll give it a shot next time I break out the test setup which should be in the next week or two since I have some controller builds going on right now that need to be tested. Just picked up some 4awg solid copper wire to build up the traces.. mmm beefy.

I have a question for you John. How would the controller know when or when it is not in BEMF since it doesn't even know exactly many phase amps it's outputting? These controllers have no way of knowing what kind of load they are driving (it would need to monitor the phase wires to know what's occurring). They are not very intelligent so I would be surprised to see them trying to protect themselves by preventing 100% PWM at even 0 speed (but I can't be sure without testing). Block time is a great example of how dumb they can be. They allow for the controller to continue powering the motor even if the wheel is locked up for over a second unless you program it to a lower setting.

John in CR said:

Has anyone ever blown a controller at high speed that couldn't be easily explained by a lack of cooling air flow or a phase wire short?

Click to expand...

Yes on colossus I eased up to almost max rpm with part throttle then BAM at full throttle at close to max rpm it blew to shit more then once this is because back emf doesnt protect these controllers on low inductance motors uless you are basicly at max rpm anything below even 1% will cause to much current. This is also because they are trap wave controllers and if you try to keep the fets on even at max rpm the whole time you will cause to much current to flow through the fets.

Arlo1 said:

John in CR said:

Has anyone ever blown a controller at high speed that couldn't be easily explained by a lack of cooling air flow or a phase wire short?

Click to expand...

Yes on colossus I eased up to almost max rpm with part throttle then BAM at full throttle at close to max rpm it blew to shit more then once this is because back emf doesnt protect these controllers on low inductance motors uless you are basicly at max rpm anything below even 1% will cause to much current. This is also because they are trap wave controllers and if you try to keep the fets on even at max rpm the whole time you will cause to much current to flow through the fets.

Click to expand...

Arlo1,
You're essentially in current limiting the entire time, because you didn't have enough controller to feed the motor what it wants. I'd bet the 99% isn't going to make any difference, because it never reaches full duty. You had to build a special controller to run the thing.

zombiess said:

I have a question for you John. How would the controller know when or when it is not in BEMF since it doesn't even know exactly many phase amps it's outputting? These controllers have no way of knowing what kind of load they are driving (it would need to monitor the phase wires to know what's occurring). They are not very intelligent so I would be surprised to see them trying to protect themselves by preventing 100% PWM at even 0 speed (but I can't be sure without testing). Block time is a great example of how dumb they can be. They allow for the controller to continue powering the motor even if the wheel is locked up for over a second unless you program it to a lower setting.

Click to expand...

The controller knows battery current, and the controller knows the duty cycle, so that's how it estimates phase current....Battery current / duty cycle = phase current. Current limiting is done by it's only means of control, which is PWM. My guess is that a painfully low inductance motor like Colossus blows the Chinese controllers like popcorn because they can't react quickly enough. ie by the time the controller realizes that the current overshot the program limit it's already too late and it's overshot it by so far that it already cratered.

Of course duty has to be limited by more than just throttle position. That's why when I get on my bike and take off it doesn't matter if I twist the throttle to 50% or to 100% the early stages of acceleration are exactly the same with our velocity throttles.

If someone has a more accurate description for how these things work I'm all ears, but the controllers have to be able to at least crudely sense and limit phase current. Otherwise there's no way one could survive my 1.5 turn motor that's wound with something like 7 gauge worth of copper for those 1.5 turns. Thank goodness mine has more inductance than Colossus, because no way do I have the patience to go thru want Arlo1 has done to get a motor running.