TESLA high performance tuning discusion

[Hillhater]

Certainly would be interesting to see how far the stock Tesla components can be pushed .
Most of the <10 sec DC powered runners seem to have switched over to proven competition transmission, rear end , chassis etc for one reason or another.

 

[Punx0r]

How hard would it be for a manufacturer like Zilla to make a ~2000A brushless controller? In my head I figure they've already done the hard work in developing a reliable switched output stage and "just" need to apply those component/layout lessons to a multi-phase topology. Or have I completely underestimated the difference between a brushed and brushless controller?

 

[riba2233]

I would say it's a rather large difference, but I could be wrong

 

[Arlo1]

How hard would it be for a manufacturer like Zilla to make a ~2000A brushless controller? In my head I figure they've already done the hard work in developing a reliable switched output stage and "just" need to apply those component/layout lessons to a multi-phase topology. Or have I completely underestimated the difference between a brushed and brushless controller?

Brushed controllers are a walk in the park compared to brushless.
Sure they will have some of the switching methods down but they will need to learn how to tack the motor and how to make sure the right switches are being turned on Its about 4x the work or more to make a AC controller and the code is no longer a simple PWM that might change a bit with rpm....

 

[Punx0r]

Is the complexity of the control side not the same regardless of the size of the controller? Aren't these functions increasingly available as off-the-shelf I.C.s? I thought the hard part of high powered controllers was the power stage: that you can't just scale up a small design by adding more FETs in parallel. I thought the trick was in getting current sharing between devices and correct and proper gate drive. From that perspective I wondered if a brushed and brushless controller are that different - a switched PWM output stage that can handle very high current without blowing itself to pieces.

Again, I could well be mistaken as I'm only a bystander to power electronics design, but would like to know more.

 

[Arlo1]

Is the complexity of the control side not the same regardless of the size of the controller? Aren't these functions increasingly available as off-the-shelf I.C.s? I thought the hard part of high powered controllers was the power stage: that you can't just scale up a small design by adding more FETs in parallel. I thought the trick was in getting current sharing between devices and correct and proper gate drive. From that perspective I wondered if a brushed and brushless controller are that different - a switched PWM output stage that can handle very high current without blowing itself to pieces.

Again, I could well be mistaken as I'm only a bystander to power electronics design, but would like to know more.

Your also right adding more fets or in the case of real power IGBTs is hard and then the inductance of the the layout becomes an issue when you are adding 6 power switches at a time instead of 1 power switch in a brushed controller making it much harder to keep things tight with a low inductance layout.

The control becomes more complex with bigger power you need to watch a lot more things to keep it safe.
Also as you up the power you typically need more voltage IE its not realistic to make 1000hp with a 20v nominal system. But 5hp with 20v is ok... As you up the voltage the complexity increases and the safety becomes something to deal with as well both driving time and cost to manufacture.

 

[MrDude_1]

Also as you up the power you typically need more voltage IE its not realistic to make 1000hp with a 20v nominal system. But 5hp with 20v is ok... As you up the voltage the complexity increases and the safety becomes something to deal with as well both driving time and cost to manufacture.

So what you're saying is 190amps or so is the max you see as realistic at lower voltages?

 

[Arlo1]

Also as you up the power you typically need more voltage IE its not realistic to make 1000hp with a 20v nominal system. But 5hp with 20v is ok... As you up the voltage the complexity increases and the safety becomes something to deal with as well both driving time and cost to manufacture.

So what you're saying is 190amps or so is the max you see as realistic at lower voltages?

No I should have not put numbers in there.... I was just using examples... I am saying 10,000 amps at 20v is very in realistic.

 

[MrDude_1]

Also as you up the power you typically need more voltage IE its not realistic to make 1000hp with a 20v nominal system. But 5hp with 20v is ok... As you up the voltage the complexity increases and the safety becomes something to deal with as well both driving time and cost to manufacture.

So what you're saying is 190amps or so is the max you see as realistic at lower voltages?

No I should have not put numbers in there.... I was just using examples... I am saying 10,000 amps at 20v is very in realistic.

ah. K. I was wondering if there was some limit I was unaware of...

I know from reading about Justins Cycle Satiator that the UL limit for low DC voltage is about 60v.
We know that a motor is driven the same way, regardless of its size... however designing the power side is harder at higher levels.
so assuming you kept the full 60, the limit is amperage.
As you said earlier, bigger power is harder to do.... but if you had some form of mass produced, common IGBT that could become a psudo-standard... its amp/power limit at the 60v level would be all thats stopping you from making a mass produced commercial gokart or offroad bike.
Does this IGBT exist right now, or are we still waiting for something like this to come along? If its here, what is its limit? I would imagine somewhere in the 200 to 400 amp range.

 

[Arlo1]

Below 250v you would want MOSFETs. The limit is in the package style and leg limits. A T0264 has leg limits of 160amp continuous as does the T0247 but the size of the T0264 allows a bigger internal area for the die and better surface in the back for cooling. There is a few T0264 MOSFETs and igbts that can do 160 amps RMS but MOSFETs are lower voltage so to get the power densitry increase you would move to igbts and that's why I said what I did in terms of igbts being what you want for big power. I have a few 650v rated igbts and some will pulse 300+ amps you can get this with a MOSFET but its limited to the lower voltage of something between 100v-200v before they can't do the big currents anymore. So if you want power with the same size of controller you would need to up the voltage and rewind the motor to work with higher voltages...

 

[Hillhater]

Presumably, Tesla are already able to handle 500+kW , (1200 A). from their inverter...fairly reliably.
So pushing that up to 1800-2000A for a few sec should not be impossible ?
.....for someone who knows how !

 

[Arlo1]

Presumably, Tesla are already able to handle 500+kW , (1200 A). from their inverter...fairly reliably.
So pushing that up to 1800-2000A for a few sec should not be impossible ?
.....for someone who knows how !

That's hard to say. They do have a great design and the system is controlling a induction motor which is a different ball game then a PMAC or BLDC....
They have been up to 1500 amps or something like that DC so on the AC side its likely a decent amount above that....

 

[okashira]

What is harder on the igbts?
phase current or battery current? combination?

 

[okashira]

Presumably, Tesla are already able to handle 500+kW , (1200 A). from their inverter...fairly reliably.
So pushing that up to 1800-2000A for a few sec should not be impossible ?
.....for someone who knows how !

That's hard to say. They do have a great design and the system is controlling a induction motor which is a different ball game then a PMAC or BLDC....
They have been up to 1500 amps or something like that DC so on the AC side its likely a decent amount above that....

I think the power stage should be exactly the same for induction vs bldc vs pmac.
The only difference is the angles you introduce the current: magnetizing current is needed on the D axis and measurement of the rotor flux angle is the same theory too.
I guess they may not even use an encoder? since they can set the rotor flux angle when the car is stopped.

The magic is in their brain of course, Tesla probably has the best induction motor logic out there. Torque sleep (their term for controlling rotor flux strength) is not easy to implement for EV application because of the throttle lag it would create if the rotor flux is too low.

 

[liveforphysics]

How hard would it be for a manufacturer like Zilla to make a ~2000A brushless controller? In my head I figure they've already done the hard work in developing a reliable switched output stage and "just" need to apply those component/layout lessons to a multi-phase topology. Or have I completely underestimated the difference between a brushed and brushless controller?

The difference in build difficulty between making a high powered brushed and high-power brushless controller is >1,000x difficulty factor.

Making a big brushed controller is really not difficult, maybe a 2-5 day project if you already had an appropriate pile of parts to work with. To make the same power controller brushless you're looking at many months of painful struggles if you ever even get close (I never got past just exploding every FET stage I made from shoot-through and noise induced glitches).

 

[Arlo1]

What is harder on the igbts?
phase current or battery current? combination?

Your power switches and motor windings only see phase current.

 

[Punx0r]

Arlo, Luke, thanks for the insights - I now appreciate the differences in topology and the challenges that come with it.

 

[riba2233]

If I had free time, I would be right on developing some crazy 3 phase controller, at least 400 v and 2000 A :lol:

 

[MrDude_1]

I think the power stage should be exactly the same for induction vs bldc vs pmac.
The only difference is the angles you introduce the current: magnetizing current is needed on the D axis and measurement of the rotor flux angle is the same theory too.
I guess they may not even use an encoder? since they can set the rotor flux angle when the car is stopped.

The magic is in their brain of course, Tesla probably has the best induction motor logic out there. Torque sleep (their term for controlling rotor flux strength) is not easy to implement for EV application because of the throttle lag it would create if the rotor flux is too low.

The Tesla motor has an encoder.

I guess most of you have not read this... http://ecarproject.ch/
This guy not only has the Tesla motor running on its own inverter.. he can run any of the major car companies inverters, on most of the other motors.. you could turn a model S motor with a chevy volt inverter for instance... or a prius drivetrain.. etc.

 

[okashira]

I think the power stage should be exactly the same for induction vs bldc vs pmac.
The only difference is the angles you introduce the current: magnetizing current is needed on the D axis and measurement of the rotor flux angle is the same theory too.
I guess they may not even use an encoder? since they can set the rotor flux angle when the car is stopped.

The magic is in their brain of course, Tesla probably has the best induction motor logic out there. Torque sleep (their term for controlling rotor flux strength) is not easy to implement for EV application because of the throttle lag it would create if the rotor flux is too low.

The Tesla motor has an encoder.

I guess most of you have not read this... http://ecarproject.ch/
This guy not only has the Tesla motor running on its own inverter.. he can run any of the major car companies inverters, on most of the other motors.. you could turn a model S motor with a chevy volt inverter for instance... or a prius drivetrain.. etc.

Of course, because the power stages are basically the same

I am sure his ACIM logic isn't as good as Tesla's

The prius motor is IPM, which you could run... with any BLDC controller but to run it right the ideal phase angle is different and is a function of torque and speed and needs to be tuned for the specific motor.

 

[MrDude_1]

I think the power stage should be exactly the same for induction vs bldc vs pmac.
The only difference is the angles you introduce the current: magnetizing current is needed on the D axis and measurement of the rotor flux angle is the same theory too.
I guess they may not even use an encoder? since they can set the rotor flux angle when the car is stopped.

The magic is in their brain of course, Tesla probably has the best induction motor logic out there. Torque sleep (their term for controlling rotor flux strength) is not easy to implement for EV application because of the throttle lag it would create if the rotor flux is too low.

The Tesla motor has an encoder.

I guess most of you have not read this... http://ecarproject.ch/
This guy not only has the Tesla motor running on its own inverter.. he can run any of the major car companies inverters, on most of the other motors.. you could turn a model S motor with a chevy volt inverter for instance... or a prius drivetrain.. etc.

Of course, because the power stages are basically the same

I am sure his ACIM logic isn't as good as Tesla's

The prius motor is IPM, which you could run... with any BLDC controller but to run it right the ideal phase angle is different and is a function of torque and speed and needs to be tuned for the specific motor.

meh. just a matter of tuning. Ive been dicking around with tuning my 80-100 with the Phaserunner controller... its just like tuning a car with a laptop.. boring, tedious, methodical, and requires gathering data to do anything meaningful.
so you do it until its close enough for you to live with, and then you move on with your life. Trying to eek that last 1% efficiency takes all the fun out of it anyway.

 

[Arlo1]

Nope not even close. You have to write software to run and control the powerstage and if you get it wrong you will have a ball of plasma.

 

[MrDude_1]

Nope not even close. You have to write software to run and control the powerstage and if you get it wrong you will have a ball of plasma.

uhh.. thats what that guys board does for you. it makes it much simpler. I suppose you could do something stupid to still make a plasma ball, but its not as hard as starting from scratch.

http://www.diyelectriccar.com/forums/showthread.php?t=152234

 

[Arlo1]

Oh I see what you are saying.. If you want to go that route you can also use what the guys at EVTV have done to crack the can on the Tesla...

 

[okashira]

Yes that's what I'd use if I was using tesla motor. Might as well take advantage of their nice ACIM logic.
I'm more interested in using the tesla powerstage on a better motor

On that note we just got the motor out an hour ago.