PowerVelocity controller review

[Addy]

So run that AOT290 at 77F and 500 watts and 42 amps and you get 11.9 volts. Who is going to run at 12 volts? Obviously no one and there are really 4 mosfets, not 1 running at a time in a phaserunner.

You're making the same mistake again! This is not how mosfet dissipation works. You have to remember that some of the power is going to the load (motor). What use would a controller be if 100% of the power going into it was dissipated in the mosfets?

500W of dissipation in the AOT290 with the typical RDS of 2.7mOhm:

I = sqrt( 500W / 2.7mOhm)
I = 430A

Mosfet voltage:
V = 430A * 2.7mOhm
V = 1.16V

Does that mean the system voltage is 1.16V? No! It means that the mosfet has a voltage drop across it of 1.16V. The rest of the voltage is across the wiring and motor.

 

[ElectricGod]

Sorry, but that math is not accurate. The FET device rated dissipation is a rating of the heat path out of the die, and has little to do with the FET power handling capability in switched mode, which is the way they are operated in these motor controllers. Doubling the number of FETs does NOT double the power handling capability of a controller due to mismatch and current imbalance. The more FETs you add, the less benefit you get from each additional FET due to the uneven current division.

The numbers I mentioned for the PhaseRunner are actual user measurements from real tests, not theoretical calculations or pie in the sky specs. In my own experiments I have run 25 amps at 80+ volts and the PhaseRunner did not even get warm. It was not close to any limits. Most controllers don't drive their FETs well enough to reach full device capability, and they use the inherent body diode which generates more heat than the FET itself, limiting the system performance (as well as other limiting design flaws). The PhaseRunner doesn't have these faults, so it has more capability than a 6 FET controller generally has (and has a higher cost due in part to the components [and engineering] required to make this happen). It is in a different class.

The PowerVelocity Sinusiodal controller provides some of the features at a lower price point and with less tuning complexity. They are different products at different price points. I don't really view them as directly competing. If I need small size, waterproof, FOC, torque throttle, real time variable regen/braking, or extreme field weakening then the ASI/Grin PhaseRunner is appropriate to consider (or the larger ASI units if more power is needed). If I need something that is lower in cost and still sinewave quiet then the PowerVelocity is a good one to consider, if real-time variable ebraking is not needed. I'm not sure but I think the PowerVelocity doesn't have torque throttle either. When I wanted high power, FOC sinewave control, torque throttle and smooth full range electric braking I ended up with the Sabvoton. Different devices for different requirements.

It would be better to stick to actual experience rather than theoretical bashing of other products in this review thread. I responded due to incorrect theory and presented some actual data about the PhaseRunner since it was being discussed. The PhaseRunner data (and more) can be found in it's own (long) thread.

This thread is about Power Volecity controllers not controller A vs controller B. If you wish to do that sort of a commentary, please take it elsewhere. Notice the title for this thread "PowerVelocity controller review". Also I did say that the phaserunner is an excellent controller. So no diminishing of anything happened. Nor did I attack you. I presented real numbers based on real component specs...nothing more and nothing less. 500 watts x 4 mosfets = 2000 watts. This math is incorrect for calculating phase amps, but for calculating what your watt meter sees...it's dead on.

Regarding your comments about where the PV controllers fit into the grander scheme of things...I agree with you and have never said otherwise. I really have no idea why you are hanging on this. Well before you ever posted in this thread, I clearly stated where I saw the 3232 MCU fitting in. I'm guessing you never read my review of the 3232 MCU. Go back and read that post. We ARE on the same page here I even said as much.

I have an idea...how about post in this thread how to actually calculate real world wattage rather than say "Sorry, but that math is not accurate". How about show the real math if I'm over simplifying things or flat wrong. I have no ego to bruise here. If I'm wrong, show it to me. I've been wrong before...it's no big deal to me. If/when you do show me how to calculate real wattage, then I will look it over, ask questions, possibly challenge you on a few things, google on it further and then if I see where my mistake is, I'll thank you for the information and for setting me straight. Like I said, being wrong is no big deal to me. Show me the correct way to calculate for the wattage that the watt meter would see and what the maximums for the mosfets would be.

On a side note about AOT290's specifically. I've experienced similar results with this mosfet. Run them at or near their limits and they don't get warm. I've always assumed it was the Rds or possibly their fairly fast switching time or some combination of things like that. It's one of the reasons I really like this mosfet. I don't see any controller I have upgraded to this mosfet get anywhere near as warm as it did with whatever it used to have in it.

I thought I would point out a detail that you said...
"In my own experiments I have run 25 amps at 80+ volts and the PhaseRunner did not even get warm."
Given: 4 AOT290 mosfets at 500 watts each = 2000 watts.
2000 watts/80 volts = 25 amps. 2000 watts/25 amps = 80 volts...yup it all looks good to me.

Thank you for proving what I said and that my math is in fact correct. Where did I go wrong here? I'm not seeing it. The numbers work out exactly. Keep in mind and I have said this previously that I'm not referring to phase amps, but rather what the watt meter between the pack and the controller sees. From that point of reference, what you quoted were readings from a watt meter or CA. This is exactly what I have been referring to all along and have been clearly saying as much. A 6 fet controller that uses 500 watt mosfets ought to be a 2000 watt controller...plain and simple.

 

[ElectricGod]

So run that AOT290 at 77F and 500 watts and 42 amps and you get 11.9 volts. Who is going to run at 12 volts? Obviously no one and there are really 4 mosfets, not 1 running at a time in a phaserunner.

You're making the same mistake again! This is not how mosfet dissipation works. You have to remember that some of the power is going to the load (motor). What use would a controller be if 100% of the power going into it was dissipated in the mosfets?

500W of dissipation in the AOT290 with the typical RDS of 2.7mOhm:

I = sqrt( 500W / 2.7mOhm)
I = 430A

Mosfet voltage:
V = 430A * 2.7mOhm
V = 1.16V

Does that mean the system voltage is 1.16V? No! It means that the mosfet has a voltage drop across it of 1.16V. The rest of the voltage is across the wiring and motor.

Hi!

Thanks for posting about what is happening in the phases, but not about what the watt meter sees. I have no problem with this math. However, I have deliberately NOT mentioned anything about what is happening in the phases...just what the watt meter is going to see. I think people need to read what I actually wrote. If they did, then there would be no controversy here. Also, keep in mind that 500 watts is a maximum, not a continuous. Go ahead and run your car at the rev limiter all the time and see how long before it throws a rod. This is no different. Your calculations for maximum phase amps are like running your car at 6000 rpm all the time. Things will give out soon.

Thank you again for posting how to calculate for what is happening in the phases. This is good information and helpful.

I distilled this down into it's generic form...

Calculated phase numbers:
Givens: mosfet wattage (Wm)
Mosfet Rds (Rds)

Max phase amps (Ip):
Ip=sqrt(Wm/Rds)

Mosfet voltage drop (Vm):
Vm = Ip x Rds

Voltage across motor phases (Vp):
Vp = Battery voltage - Vm


BTW...no one rates a controllers wattage based on phase amps. If they did then specs would be wildly stratospheric. There's already so much fudging and faking and incomplete specs as it is. LOL!

 

[Addy]

However, I have deliberately NOT mentioned anything about what is happening in the phases...just what the watt meter is going to see.

I mentioned what is happening in the phases because this is what you need to know to calculate mosfet dissipation. Knowing that the controller is consuming 2000W from the battery tells you nothing about how much of that is turning into heat in the mosfets. Because of this, you can't assume that 4 mosfets rated for 500W result in a controller limited to 2000W. On top of that, the 500W rating of the AOT290 would only hold true in a very ideal world. It would be very difficult to actually transfer 500W of heat out of any mosfet in a TO-220 package.

Aside from that, it's very important how the controller is designed. The mosfets must be driven quite well to keep their losses low and there are more losses than just the simple RDS losses I showed how to calculate. The main point here is that knowing which mosfets a controller uses, doesn't give you any accurate information about how powerful the controller really is.

I'm glad you appreciate the information I posted.

 

[ElectricGod]

However, I have deliberately NOT mentioned anything about what is happening in the phases...just what the watt meter is going to see.

I mentioned what is happening in the phases because this is what you need to know to calculate mosfet dissipation. Knowing that the controller is consuming 2000W from the battery tells you nothing about how much of that is turning into heat in the mosfets. Because of this, you can't assume that 4 mosfets rated for 500W result in a controller limited to 2000W. On top of that, the 500W rating of the AOT290 would only hold true in a very ideal world. It would be very difficult to actually transfer 500W of heat out of any mosfet in a TO-220 package.

Aside from that, it's very important how the controller is designed. The mosfets must be driven quite well to keep their losses low and there are more losses than just the simple RDS losses I showed how to calculate. The main point here is that knowing which mosfets a controller uses, doesn't give you any accurate information about how powerful the controller really is.

I'm glad you appreciate the information I posted.

Yup...I agree...there are mitigating circumstances that happen in the phases. BUT what the watt meter sees is not going to change hardly at all. Would it have helped if I had put it in big bold red letters "THIS IS WHAT YOUR WATT METER SHOULD NOT EXCEED"? We are in agreement here. Where is the issue? I know...I'll go back and add that so there is no confusion.

 

[Addy]

Yup...I agree...there are mitigating circumstances that happen in the phases. BUT what the watt meter sees is not going to change hardly at all. Would it have helped if I had put it in big bold red letters "THIS IS WHAT YOUR WATT METER SHOULD NOT EXCEED"? We are in agreement here. Where is the issue?

A 6 fet controller that uses 500 watt mosfets ought to be a 2000 watt controller...plain and simple.

This and the related things you have been saying is the issue. I just explained that you can't determine the controller power limit this way

 

[ElectricGod]

Yup...I agree...there are mitigating circumstances that happen in the phases. BUT what the watt meter sees is not going to change hardly at all. Would it have helped if I had put it in big bold red letters "THIS IS WHAT YOUR WATT METER SHOULD NOT EXCEED"? We are in agreement here. Where is the issue?

A 6 fet controller that uses 500 watt mosfets ought to be a 2000 watt controller...plain and simple.

This and the related things you have been saying is the issue. I just explained that you can't determine the controller power limit this way

OK...I'm clearly not getting what your point is. I've always referenced what the watt meter is going to see. How is that incorrect? The other things happen in the non-linear side of the controller and in there, better MCU better FOC, motor winding, iron losses and the list goes on are going to play into what happens on the non-linear side. In this side of things those simple calculations aren't really accurate, just estimates that are close enough.

 

[Addy]

OK...I'm clearly not getting what your point is. I've always referenced what the watt meter is going to see. How is that incorrect?

So if your watt meter reads 2000W, how do you know what the controller limit is? You can't tell just from that. Generally you should follow the controller manufacturer's specified limit, unless you understand how much heat the controller is going to dissipate. This depends on a lot of things, you can't just calculate it using a mosfet's specified maximum dissipation

12 fet (IRF4110): 370 watts x 8 mosfets = 2960 watts maximum
12 fet (AOT290): 500 watts x 8 mosfets = 4000 watts maximum
18 fet (IRF4110): 370 watts x 12 mosfets = 4440 watts maximum
18 fet (AOT290): 500 watts x 12 mosfets = 6000 watts maximum
24 fet (IRF4115): 375 watts x 18 mosfets = 6750 watts maximum

This is wrong. That's what I'm trying to say

 

[ElectricGod]

OK...I'm clearly not getting what your point is. I've always referenced what the watt meter is going to see. How is that incorrect?

So if your watt meter reads 2000W, how do you know what the controller limit is? You can't tell just from that. Generally you should follow the controller manufacturer's specified limit, unless you understand how much heat the controller is going to dissipate. This depends on a lot of things, you can't just calculate it using a mosfet's specified maximum dissipation

12 fet (IRF4110): 370 watts x 8 mosfets = 2960 watts maximum
12 fet (AOT290): 500 watts x 8 mosfets = 4000 watts maximum
18 fet (IRF4110): 370 watts x 12 mosfets = 4440 watts maximum
18 fet (AOT290): 500 watts x 12 mosfets = 6000 watts maximum
24 fet (IRF4115): 375 watts x 18 mosfets = 6750 watts maximum

This is wrong. That's what I'm trying to say

It's not wrong from the perspective of what the watt meter sees which I have been very clear and specific about.

So lets run a scenario here to simplify things. We have a 2000 watt controller for this scenario and two motors. One motor is poorly designed and is 40% efficient. The other is well designed and 90% efficient. The 40% motor will waste 60% of the energy going into it probably as heat leaving 40% to do actual work. The 90% motor will waste 10% as heat and 90% will do work. Lets say our controller draws 40 amps and is running at 50 volts or exactly 2000 watts. With either motor, the watt meter between the controller and battery pack is going to read 40 amps, BUT one motor is going to get very hot and do very little work and the other is going o do a lot of work and get maybe a little warm. Now take two identical 90% efficient motors and a cheap Chinese trapezoidal controller and the latest and greatest in FOC controllers. Both controllers have identical mosfets, are running at 50 volts and are current limited to 40 amps or 2000 watts. In this scenario, the trapezoidal controller is going to control the motor less effectively than the FOC controller. You will get more work out of the motor on the FOC controller than you will out of the motor on the trapezoidal controller on the exact same voltage and current in. Despite the exact same motor, wattage in and mosfets, the better MCU is going to do a better and more efficient job of running the motor and cause it to do more work as a result. The watt meter however will always measure 40 amps and 50 volts. I think this is what you are referring to as "not correct"...one controller is capable of more work than the other is despite the exact same amount of supplied energy, exact same motor and mosfets.

So then, how does one calculate for those kinds of variations one controller/MCU to another? The best I know how to do is to look at the mosfets specs and use that for maximums I should see at the watt meter. Anything more than that is completely variable from MCU to MCU and the settings found in that specific controller. I probably sound absolute in my watt meter numbers. They are not going to fail you. They are going to be solid numbers you can depend on and NOT blow mosfets. Anything more than that is variable at best. The same can be said on the non-linear side. Take the specs for the mosfets, calculate the values you get and those numbers are not going to fail you. The specifics of the MCU and what it adds to the equation is a completely separate subject which is not easily calculated.

IE: The best I can do is say X mosfet is good for Y watts and that means a controller will safely do Z watts at the watt meter. Same for the non-linear side, but with different math. Everything else is variable based on your settings and your MCU and the firmware implemented in it. I want to be clear that I never implied that the mosfets are consuming all the wattage. That would be silly and would mean the mosfets were 100% inefficient and therefore completely useless as anything other than as heaters. We obviously want them passing as much current as possible onto the motor phases with the least amount of losses.

 

[Addy]

IE: The best I can do is say X mosfet is good for Y watts and that means a controller will safely do Z watts at the watt meter.

Ok, here's an example. Let's say you use a good 6 mosfet controller with 500W mosfets. Let's assume the motor is 90% efficient and the remaining 10% wasted energy is lost in the mosfets. If, like you assume, 4 mosfets are working at any given time, and they're dissipating their maximum of 500W each, so 2000W total, then the power delivered to the motor is:
2kW * (90% / 10%) = 18kW!

So your controller with 500W mosfets would be measuring 20kW at the watt meter.

Of course, this is unrealistic, but do you see why mosfet ratings can't be used to determine the controller's power limit?

 

[Alan B]

Let's take a look at various current levels and see how much heat is generated in the FETs of an actual controller.

Some AOT290L Specs:
Drain Current Max 140A (110A at 100C) continuous
Max drain resistance 3.5 mOhms at 25C (5.7 at 125C)
Diode forward voltage max 1V (at 20 amps)

Calculations
The calculations are fairly simple: I squared times R for heat generated in the FET by current when the FET is on. When the FET is off the heat is zero (unless current is flowing in the body diode, see below). During the transition between states the power dissipated is large, but the time is so short (in a good design) that the heat generated is small compared to the conduction heat at high current.

This current is passing through two of the FETs. (Actually it is passing through one fully, and divided between two others as the vector rotates, but to keep it simple we'll call it two for now). So we multiply by two to get total conductive heat in the heatsink. The following table shows this for 25 amps at 25C, and for 25 to 70 amps at 125C.

FOC Controllers like the PhaseRunner use "synchronous rectification" where they turn on all the FETs involved in conduction rather than allowing some of them to operate in the "diode" mode. The control algorithms and timing are more complex with synchronous rectification so it takes more CPU power and more software.

For controllers that use the inherent body diode, the power dissipation is E * I, or current times 1 volt, the forward diode voltage drop. I don't know if the PowerVelocity is doing this, but most trapezoidal controllers do. In this case the above calculation of heat applies to only one FET, the other uses the current times voltage drop calculation, which would be from 25 to 70 watts in this case, more power dissipated in the diode conduction mode than in the FET on state.

As you can see in the table below, the heat in the FETs and heatsink is very low even at 40 amps. Above 40 amps it begins become more significant due to the squared term in the equation. However this is well above 3000 watts in the load. Users have reported being able to run the PhaseRunner at the 5KW level with a sufficiently large heatsink, at least for some modest period of time.

I added the right hand column to the table below to show what happens in the common controllers that use the body diode to handle the current. This results in a significant increase in heat production, especially at modest currents.


In Conclusion
The heat generated in the FETs has nothing to do with the FET's Power Dissipation rating. FET heat is produced primarily by current flow through the FET and the resistance of the FET. The system voltage has little to do with FET heating. When the FET sees full voltage it is turned off and there is no current flow or heat. The system voltage has a lot to do with power delivered to the load. To get the most power through a PhaseRunner requires operating toward the higher voltages and using a motor that uses the voltage with lower current (often called torque type motors). A so called "speed" winding that uses lower voltage and higher currents will not be able to get as much power from a PhaseRunner, or any controller with modest current capability. For those you'll want to use a high current controller. Choose the motor to match the controller for best results.

Observation
The same capability should apply to the six FET PowerVelocity controller if they drive the FETs as well and if they avoid using the body diode of the second FET. If they use the body diode, then the ratings of the 12 FET model would be approximately equivalent to the 6 FET PhaseRunner due to the extra heat generated in the body diodes. That may be why we see the somewhat similar capability of the 12 FET PV vs the 6 FET PhaseRunner, but I don't know for certain how the PV controller handles the FETs.

 

[ElectricGod]

Let's take a look at various current levels and see how much heat is generated in the FETs of an actual controller.

Some AOT290L Specs:
Drain Current Max 140A (110A at 100C) continuous
Max drain resistance 3.5 mOhms at 25C (5.7 at 125C)
Diode forward voltage max 1V (at 20 amps)

Calculations
The calculations are fairly simple: I squared times R for heat generated in the FET by current when the FET is on. When the FET is off the heat is zero (unless current is flowing in the body diode, see below). During the transition between states the power dissipated is large, but the time is so short (in a good design) that the heat generated is small compared to the conduction heat at high current.

This current is passing through two of the FETs. (Actually it is passing through one fully, and divided between two others as the vector rotates, but to keep it simple we'll call it two for now). So we multiply by two to get total conductive heat in the heatsink. The following table shows this for 25 amps at 25C, and for 25 to 70 amps at 125C.

FOC Controllers like the PhaseRunner use "synchronous rectification" where they turn on all the FETs involved in conduction rather than allowing some of them to operate in the "diode" mode. The control algorithms and timing are more complex with synchronous rectification so it takes more CPU power and more software.

For controllers that use the inherent body diode, the power dissipation is E * I, or current times 1 volt, the forward diode voltage drop. I don't know if the PowerVelocity is doing this, but most trapezoidal controllers do. In this case the above calculation of heat applies to only one FET, the other uses the current times voltage drop calculation, which would be from 25 to 70 watts in this case, more power dissipated in the diode conduction mode than in the FET on state.

As you can see in the table below, the heat in the FETs and heatsink is very low even at 40 amps. Above 40 amps it begins become more significant due to the squared term in the equation. However this is well above 3000 watts in the load. Users have reported being able to run the PhaseRunner at the 5KW level with a sufficiently large heatsink, at least for some modest period of time.

I added the right hand column to the table below to show what happens in the common controllers that use the body diode to handle the current. This results in a significant increase in heat production, especially at modest currents.


In Conclusion
The heat generated in the FETs has nothing to do with the FET's Power Dissipation rating. FET heat is produced primarily by current flow through the FET and the resistance of the FET. The system voltage has little to do with FET heating. When the FET sees full voltage it is turned off and there is no current flow or heat. The system voltage has a lot to do with power delivered to the load. To get the most power through a PhaseRunner requires operating toward the higher voltages and using a motor that uses the voltage with lower current (often called torque type motors). A so called "speed" winding that uses lower voltage and higher currents will not be able to get as much power from a PhaseRunner, or any controller with modest current capability. For those you'll want to use a high current controller. Choose the motor to match the controller for best results.

Observation
The same capability should apply to the six FET PowerVelocity controller if they drive the FETs as well and if they avoid using the body diode of the second FET. If they use the body diode, then the ratings of the 12 FET model would be approximately equivalent to the 6 FET PhaseRunner due to the extra heat generated in the body diodes. That may be why we see the somewhat similar capability of the 12 FET PV vs the 6 FET PhaseRunner, but I don't know for certain how the PV controller handles the FETs.

Thank you posting this! Very interesting stuff. Some I've read before and some I haven't. For example the body diode bit is new information. I assumed it was there to protect the mosfet from bEMF spikes like how you put a reverse biased diode across a relay coil. I'm guessing it can serve dual roles in that case. It sounds like using the diode is a bit of a cheat or rather poor timing in the MCU. Is there an article about this specifically that you can recommend?

I have no idea how the 3232 MCU handles the transition states. I have 3 o-scopes. I suppose if I connect probes to the various mosfet gates in the H-bridges, I'll see some interesting wave forms. What should I look for to determine if the body diode is used or not? Let's prove this thing out! What a great opportunity to learn something I haven't messed with before.

Regarding that 5kw you mentioned. Is that phase watts? I can't imagine how it could be anything else. There is a point where exceeding certain limits is just magical nonsense. 5kw split among 4 mosfets is 1250 watts each or 2.5X more than their rated wattage. This seems highly unlikely.

By "transition dissipated power",are you referring to the mosfet shutting down and as a result its increased resistance from source to drain? This would build up a good bit of heat for a short period of time. Of course fast mosfets such as the AOT290 are going to go from conducting to not conducting much faster than say the IRF4110. So by it's very nature the AOT290 on the same MCU is going to run cooler for that reason alone vs IRF4110's. Never mind that the Rds is a bit lower as well.

I have a ASI BAC2000 controller. It uses the same MCU as the phaserunner and is a 12 fet controller that uses the AOT290. Based on that 5kw you mentioned (To be honest, it sounds like absolute nonsense to me), then my BAC2000 is really capable of 10kw (More nonsense in my mind). It wouldn't be hard to implement my BAC2000 on my Currie scooter. It currently is running a 12 fet PV with AOT290's. This would be similar to my scenario with a trapezoidal and FOC controller, except it would be a basic FOC controller vs an excellent FOC controller. This my kind of challenge. The battery pack is capable of 100 amps continuous at 66 volts so there will be no problems with power delivery. I'm really curious to see how that works out! Right now I'm operating in 100% disbelief that I will see more than a small improvement. The motor is an 80kv C80100 which according to Alien Power is rated for 7kw. That too is laughable nonsense to me. Most 80mm outrunners with the same height stator stack are more like 4800 watts so I'm skeptical of that 7kw. But if what you say is true, then the first thing I'm going to notice is a large increase in motor torque. I will report my findings as I get the controller swap out going in my BAC2000 review thread. Right now that scooter has seen a single 6 mile ride and a couple of really short rides. I haven't tuned the controller settings beyond going with what will definitely work. I'm sure I can turn up the phase and battery amps more. That 6 mile ride got the controller just a little warmer than air temperature. We will see! I have no problem with being wrong.

What were you referring to "torque type motors"? At first I thought you were referring to motor Kv, but I don't think that's the case.

You've mentioned mosfet heat several times in various posts. Since the AOT290 switches states quickly and has a low Rds is will build up less heat regardless how it's used. However electronics all have limits. Super cool the mosfet to 0 kelvin and it's still not going to conduct 500 amps even if it's legs could handle it. Heat is not the only factor in junction failure. It's a good one no doubt! But every transistor junction has it's limits where it just melts down. If you could put a perfect heat sink at O kelvin on the mosfet, it's junctions would fail never the less at something like...I don't know 160 amps.

This brings me to something else. The TO-220 package. The 3 legs on anything in this package are .77mm x .48mm in cross section. This is a very tiny conductor for much amperage to flow through. It would be brilliant if that was the case because all my phase wires would be made of 20 awg wire. It's just not realistic to assume that those spindly legs are going to conduct 50 amps, never mind 140 amps. They would be glowing red hot in a second if they did. Why do I bother with 10 awg phase wires and 5.5mm bullets, when 20 awg wire and a 2mm bullet would be overkill? 20 awg is about 2X the cross sectional area of a TO-220 leg. Realistically, The TO-220 package in a 500 watt mosfet is going to see at 48 volts more like 10 amps and the legs can handle that just fine.

 

[Alan B]

These issues have been covered here on ES many times in threads on controller designs and builds. Including the inherent body diode (which is there because of the way the FET is made, but using it or not is a design choice, and it is quite lossy).

Watts are watts. DC or three phase. You can't generate 5kW of power in the motor without drawing 5kw plus losses from the battery.

The speed of switching of the FETs is mostly dependent on their gate drive system pushing a lot of current into their large gate capacitance. These FETs can all operate at many Megahertz, the devices themselves are not generally the limiting factor in the speed of response.

There is power dissipated in the FET during the transition. One way to look at that is at the switching midpoint. A simple view would be to consider the moment where the FET is halfway off. At this point assume it sees about half voltage and half current (the worst case). So in this case 40V and 20 amps (for the rated 80V 40A case). So the power being developed in the FET is about 800 watts peak during the transition. Ouch! But if this lasts for only a microsecond, then multiply 800 watts by 0.000001 seconds and the resulting power consumed is not of much consequence (even done many times per second). But you can see that the gate drive is very important, if a cheap controller has the same condition for a millisecond it will be dissipating 1000 times as much on the transition. These are the types of things where a quality design that doesn't cut costs too much can be more efficient, and why gate drive chips handle several amps.

Keep in mind this is switch mode operation of the FETs. They are operated open or (nearly) shorted, so their dissipation is a tiny fraction of the power flowing through them. They don't see the power. They control it. This is not analog electronics. It is like the audio systems in phones and laptops, they use switchmode audio amplifiers and produce lots of audio power with almost no heat or energy waste to make the batteries last longer, and keep the systems small, lightweight and cool running. See Class D amplifiers for more info. This is also how switching power supplies produce power with low losses and run cool and efficient. These systems are Digital, not analog. The design principles and equations are different. This FET can handle 500 amp pulses. Switching quickly and precisely changes everything. The Rds becomes a very if not the most important feature, especially at higher current since I is squared in the power relationship, so R needs to be kept very, very small.

I would expect the ASI FOC 12 FET controller to outperform the PowerVelocity 12 FET by some margin (it's not a fair comparison). It is rated for up to 75 amps continuous and 150 amps peak motor current (whereas the PhaseRunner is 40 continuous and 90 peak rated). Commercial motor controller ratings tend to be conservative, the commercial market is mostly that way. But you'll have to try it and see. Those values seem reasonable, doubling the FET count produces a bit less than double the current capability. Think in terms of current. If you run it at 80 volts and 75 amps average / 150 amps peak that's 6kw continuous and 12kw peak. If the motor winding matches the voltage and currents of the controller well it should make that. Kv is another (and better) way of differentiating "torque" and "speed" windings. But if the motor wants 15 volts and 400 amps for full speed and torque, that isn't a good fit for the ASI 12 FET. Think of it as impedance matching between the motor and controller.

To reach the level of current in the leads that these devices are rated for one must pull heat out of the legs. If they are soldered or clamped to copper bus bars they can apparently get to those levels, but those are not practical any more than running 500 watts of dissipation is. Like any engineering, one must back off to a reasonable level, the specs must be properly understood. But as you see from the table above, running almost 6kw output power only requires about 10 watts of heat dissipation in each FET. That requires a really good heatsink to keep the FETs cool, but it is not impossible.

There are many other things in the controller that can limit the current capacity (and thus power) besides the FET itself. Most controllers don't have heavy enough traces, and all the lead free tin solder and solder braid hacked on isn't really equal to a good bus bar, bandaids don't always work that well. The wires to the battery and motor are rarely sufficient either. Weak links in the chain. A company like ASI puts a lot of thought into choosing each component and conductor, and they know that it will cost them in many ways if they fail. They have engineering modelling tools that will tell them where the heat buildup will be an issue and how much copper to commit where, and what will melt or fail under what conditions.

Be careful with scope measurements. It can be tricky to measure the FETs because one moment they're nearly shorted with low voltage drops, and the next they have full battery voltage across them plus spikes. Introducing the scope's ground in the wrong way can damage the controller.. This has also been discussed on some ES threads where folks have made DIY controllers. Special differential probes are generally advised.

 

[Addy]

Regarding that 5kw you mentioned. Is that phase watts? I can't imagine how it could be anything else.

It sounds like it's 5kW in the load (motor):

However this is well above 3000 watts in the load. Users have reported being able to run the PhaseRunner at the 5KW level with a sufficiently large heatsink, at least for some modest period of time.

5kw split among 4 mosfets is 1250 watts each or 2.5X more than their rated wattage. This seems highly unlikely.

I think you're still misunderstanding this. A mosfet with a maximum dissipation of 500W could drive a load with many times that wattage.

 

[liveforphysics]

You've summed the peak continous thermal shedding Watts of waste heat from the parts, not the amount of power they transfer through them, which is many orders of magnitude greater, but also split between high side and low side switches.

Using today's best mosfets (nothing starting with IRF), it's possible to control about 1kW per 100v capable mosfet, but each phase requires high side and low side switches.

This would means a 6fet done well on today's best Ti-Nextfets should be ~3kW continously with a real heat path. Depending on how it's bussed and its current sensor capabilities, it should also be capable of 9kW+ bursts.

A modern FET in a 24fet controller sharing current well and using real gate drive can achieve 12kW continuously and 30kW+bursts of output power and remain >98%efficient at 30kW means just 600W of heating for the specific heat value of the aluminum to absorb for a few seconds during the burst.

 

[madin88]

I've never personally tried a phase runner, but they are a 6 fet controller based on the AOT290 so the wattage they are capable of is limited to a maximum of 2000 watts. I have no idea if you can get the full capabilities out of the mosfets in that controller or not.

Continuous power mainly depends on heat dissipation, and as already mentioned the thing with the "500W per FET" is not true.

Do PV controllers have temperature control/monitoring, or do they feature protection against overheating?

Regarding that 5kw you mentioned. Is that phase watts? I can't imagine how it could be anything else. There is a point where exceeding certain limits is just magical nonsense. 5kw split among 4 mosfets is 1250 watts each or 2.5X more than their rated wattage. This seems highly unlikely.

I assume it was battery watts, so the power that was going into the controller.
But why you still make the mistake of summing the watts of the FET's?

If the controller was working at 80V, it would mean that 5000W / 80V = 62,5A where flowing through the FET's (lets assume voltage and amp's measured RMS). I would say it is fact and not nonsense that Phaserunner can do this :wink:

 

[ElectricGod]

Hi everyone!

Thanks so much for your posts!

So I think the first thing here is that it appears that I've been interpreting the wattage spec for X mosfet as the limits of the mosfet. IE: 500 watts can be 100 volts at 5 amps or 50 volts and 10 amps or whatever. Instead what you guys appear to be saying is that the mosfet can dissipate 500 watts of heat, the wattage passing through the mosfet can be significantly higher than this.

Ok..then...I stand corrected and that's totally cool with me. I have no problem with getting things wrong and learning from my mistakes.

BUT I have a new question then. Lets use the AOT290...does that mean I can really push 140 amps through it at 100 volts as long as the mosfet itself is always dissipating less than 500 watts of heat? That's the specs from the data sheet. And then that would play directly into that switching/transition time and keeping it micro-short so that heat dissipation is minimized as much as possible. That leaves only the back diode as a controllable design option and heat source. So as long as the transition times are uber short and the back diode isn't used, then that leaves the Rds as the limiting source for mosfet heating and the only thing that can't be controllered by design changes. Under perfectly ideal situations the thing that the mosfet needs to dissipate 500 watts of heat for is Rds. Did I get that right? That would mean that a single AOT290 mosfet can pass 1400 watts through itself as long as it never dissipates more than 500 watts of heat. Mind blown!

I realize that "ideal" is exactly that...not real world. So in reality, some transition time is always going to happen and perfect heat sinks don't exist and the back diode sees some level of activity so margins for error have to accounted for. So something more like 1000-1200 watts can pass through the mosfet...still mind blown. LOL!


Regarding putting a scope on a mosfet...
I was intending to monitor the gate junction of mosfets. In that case, batt- and ground on the scope ought to be able to be the same connection. Vgs is going to be a few volts so that's not a big concern.

Transition time question...
In a truly square wave output from the MCU, then the mosfets are either on or off since all you are doing is biasing the gate fully or not at all. However sinusoidal controllers don't do that. They move towards fully biased on and off in a sinusoid. That means the transition from fully on to fully off and back again is directly defined by the frequency of the sine wave signal that is biasing the gate. How then does that transition time get to be super short? It can only be as fast as the half of a sine wave signal biasing the mosfet gate.

Thanks guys this has been great!

 

[ElectricGod]

I've never personally tried a phase runner, but they are a 6 fet controller based on the AOT290 so the wattage they are capable of is limited to a maximum of 2000 watts. I have no idea if you can get the full capabilities out of the mosfets in that controller or not.

Continuous power mainly depends on heat dissipation, and as already mentioned the thing with the "500W per FET" is not true.

Do PV controllers have temperature control/monitoring, or do they feature protection against overheating?

Regarding that 5kw you mentioned. Is that phase watts? I can't imagine how it could be anything else. There is a point where exceeding certain limits is just magical nonsense. 5kw split among 4 mosfets is 1250 watts each or 2.5X more than their rated wattage. This seems highly unlikely.

I assume it was battery watts, so the power that was going into the controller.
But why you still make the mistake of summing the watts of the FET's?

If the controller was working at 80V, it would mean that 5000W / 80V = 62,5A where flowing through the FET's (lets assume voltage and amp's measured RMS). I would say it is fact and not nonsense that Phaserunner can do this :wink:

These controllers don't have internal heat sensors by default. Vadym is adding that to the latest iteration of the monitoring module and it will likely see the light of day soon. My latest module comes with a heat sensor so I know it's close. It requires coding on the module and in the phone app to make it all work still. This iteration of the monitoring module will probably just read controller temps, but not do anything to limit the controller. Something that is future efforts is to incorporate the throttle into the module. At that point throttle curves and other very interesting things are possible...including limiting the controller based on temperature thresholds. Every time we talk about these modules, they get more and more stuff crammed into their future development. Some of them might happen such as built in security, remote disabling the controller, having the EV get too far from your phone and auto-disabling, remote lock up of the motor, variable regen, motor temp sensing, BMS and cell monitoring and lots of other ideas. They all require some hardware and software to implement them. There's lots of potential here and who knows how far it will go. The form factor for the module is pretty tight too. It's about the size of a 9 volt battery so that it can always fit inside the controller shell.

 

[ElectricGod]

Last night I powered up the latest 12 fet I got and ran it on 2 outrunners that I knew would exceed the MCU's eRPM. I was curious to see if sensored or sensorless would get me more motor RPM's.

I have a CA80-160kv Turnigy motor that I added halls to. It has 9 poles. I also have an Alien Power 12090 at 50 kv and 14 poles.

I was running at 18S or 74 volts. That works out to 106560 eRPM for the CA80-160 and 51800 eRPM for the 12090.

Sensored or sensorless, both motors spun up to about 50,000 eRPM and then the controller lost sync. For the CA80-160, that was about 20% throttle. For the 12090, I was just shy of WOT.

 

[Alan B]

Typically one needs a large margin between operational values and absolute max values. The goal with a controller is NOT to dissipate power in the FET, so some of the other limits will likely be reached first. Ultimately the cooling system will control most of the limiting.

In many cases the absolute max values occur at unreachable device temperatures, like the die being 25C. That would require refrigeration, not just a heatsink.

I think the PhaseRunner data has already answered the question. 40 amps is easy. 70 amps is sustainable for some period of time, probably not continuously. That is spread across 3 devices. So 70 amps (average) works if the device gets to rest 2/3 of the time. 140 Amps or 500 watts dissipation per device would require Liquid Nitrogen cooling, perhaps.

In a sinusoidal controller the FET is NOT operated in the linear region. It is switched fully on and fully off much more often than the period of the sine wave, but the on/off ratio is varied such that the resulting average current waveform is sinusoidal. Research class D amplifiers to see how this is done in every cellphone and laptop and low end stereo.

 

[ElectricGod]

Typically one needs a large margin between operational values and absolute max values. The goal with a controller is NOT to dissipate power in the FET, so some of the other limits will likely be reached first. Ultimately the cooling system will control most of the limiting.

In many cases the absolute max values occur at unreachable device temperatures, like the die being 25C. That would require refrigeration, not just a heatsink.

I think the PhaseRunner data has already answered the question. 40 amps is easy. 70 amps is sustainable for some period of time, probably not continuously. That is spread across 3 devices. So 70 amps (average) works if the device gets to rest 2/3 of the time. 140 Amps or 500 watts dissipation per device would require Liquid Nitrogen cooling, perhaps.

In a sinusoidal controller the FET is NOT operated in the linear region. It is switched fully on and fully off much more often than the period of the sine wave, but the on/off ratio is varied such that the resulting average current waveform is sinusoidal. Research class D amplifiers to see how this is done in every cellphone and laptop and low end stereo.

Is there a PWM or something like it associated with the sine wave control signal? IE: You turn on and off the mosfet quickly while also drawing a sine wave. Something like this...

 

[Alan B]

Yes, PWM is used. When the FETs are fully on (to plus battery and ground) the current increases in the inductive motor winding. When the FETs are shorting the windings the current circulates and decreases. So the current deviates slightly from a true sinewave, but it does not drop to zero between PWM pulses (as your sketch suggests). The FET gates are driven with square wave PWM of about 10-15 volts for ON and shorted with respect to the source (so zero volts drive) for OFF.

The low side gates are near ground and not hard to scope, the high side FET gates are relative to the FET sources which are tied to motor windings and varying over a wide range of voltage from ground to near battery level and much harder to look at.

The inductance of the motor carries stored energy in the magnetic field, and this continues the motor current during the PWM OFF period. The motor essentially "filters" the PWM signal into a sinewave.

During the PWM OFF period the inherent body diodes carry the circulating current, or with synchronous rectification then the FETs that would be conducting via their diodes are turned ON to reduce the voltage drop and dissipation.

 

[Powervelocity.com]

Gotta subscribe to this.
On the subject of heat monitoring, it's standard in the latest iteration of the telemetry module. Two temp inputs: one is wired to a thermistor attached directly to one of the mosfets and the other one is wired outside to be used for motor temp probe, battery or just ambient temperature readings. The module is set to shut off MCU if temperature on mosfets gets too high (I am thinking 100C would be a good threshold). It's also possible to do dynamic rollback of the current based on the temperature but that's a bit more involving on the programming and testing side. May come a bit later.

I've never personally tried a phase runner, but they are a 6 fet controller based on the AOT290 so the wattage they are capable of is limited to a maximum of 2000 watts. I have no idea if you can get the full capabilities out of the mosfets in that controller or not.

Continuous power mainly depends on heat dissipation, and as already mentioned the thing with the "500W per FET" is not true.

Do PV controllers have temperature control/monitoring, or do they feature protection against overheating?

Regarding that 5kw you mentioned. Is that phase watts? I can't imagine how it could be anything else. There is a point where exceeding certain limits is just magical nonsense. 5kw split among 4 mosfets is 1250 watts each or 2.5X more than their rated wattage. This seems highly unlikely.

I assume it was battery watts, so the power that was going into the controller.
But why you still make the mistake of summing the watts of the FET's?

If the controller was working at 80V, it would mean that 5000W / 80V = 62,5A where flowing through the FET's (lets assume voltage and amp's measured RMS). I would say it is fact and not nonsense that Phaserunner can do this :wink:

 

[ElectricGod]

Gotta subscribe to this.
On the subject of heat monitoring, it's standard in the latest iteration of the telemetry module. Two temp inputs: one is wired to a thermistor attached directly to one of the mosfets and the other one is wired outside to be used for motor temp probe, battery or just ambient temperature readings. The module is set to shut off MCU if temperature on mosfets gets too high (I am thinking 100C would be a good threshold). It's also possible to do dynamic rollback of the current based on the temperature but that's a bit more involving on the programming and testing side. May come a bit later.

Hey Vadym!

I've been getting a great education form a bunch of smart people on this thread.

Thanks to...Alan B, madin88, liveforphysics and Addy.

 

[ElectricGod]

Yes, PWM is used. When the FETs are fully on (to plus battery and ground) the current increases in the inductive motor winding. When the FETs are shorting the windings the current circulates and decreases. So the current deviates slightly from a true sinewave, but it does not drop to zero between PWM pulses (as your sketch suggests). The FET gates are driven with square wave PWM of about 10-15 volts for ON and shorted with respect to the source (so zero volts drive) for OFF.

The low side gates are near ground and not hard to scope, the high side FET gates are relative to the FET sources which are tied to motor windings and varying over a wide range of voltage from ground to near battery level and much harder to look at.

The inductance of the motor carries stored energy in the magnetic field, and this continues the motor current during the PWM OFF period. The motor essentially "filters" the PWM signal into a sinewave.

During the PWM OFF period the inherent body diodes carry the circulating current, or with synchronous rectification then the FETs that would be conducting via their diodes are turned ON to reduce the voltage drop and dissipation.

I've got some reading to do here. I thought I understood how the mosfets were controlled, but maybe not. I'm still unclear how a square wave creates a sinusoid in the phases. I plan to read about class D amplifiers. It's been since 1987 since I took amplifiers in college for my EE and haven't really thought about it since.