Why is it that IGBTs specifically were a problem? I don't know of anything special about them that would cause a problem. Other than turn-on/off time. If the turn-on/off time was too fast it would cause HV spikes, but that's a function of the drive circuitry, not the IGBT itself.

The rise time of an IGBT is much quicker than a bipolar switch (such as a GTO) and that is one part of the problem. However, the total problem must take into account both the "Intrinsic" variable (the effect of the rise time of the IGBT itself) and the "Extensive" variable (the switching rate). One pulse cycle alone will not cause an immediate problem, but the effect is cumulative and millions or billions of cycles will eventually puncture the insulation.

By the way, the development of the IGBT was largely based on research by Japanese elevator companies who wanted a solid state alternative to generator field control which was the world standard for elevators and the largest application of variable speed drives. In fact, most Japanese elevator companies are also subsidiaries of electronic device (and also magnetic material) manufacturers such as Mitsubishi, Hitachi, Toshiba, etc.

In the 1980s, the editor of one of the elevator trade and engineering journals in the U.S. conducted an interview with the lead engineer of one of Japan's elevator manufacturers on the future of research and development in elevators. He made a rather astounding statement that "Innovation in the field of elevators must now begin at the molecular level". His statement emphasized that the problem of making improvements in elevators was a parallel to the problem of improving computers. For elevators, this meant that motor controls would require drastically improved power electronic devices such as SCRs, Mosfets, and ultimately IGBTs, all of which the parent company would develop in its semiconductor division.

To provide a mechanical platform for their electronic research, Westinghouse licensed it's elevator technology to several Japanese firms who developed prototypes of solid state controls that were used by Westinghouse in the U.S.

 

Kermit2 has a good idea. You might consider a heating element as a field weakening resistor too.

strantor and Glenn, I haven’t kept up on drive tech, please correct me if I am wrong.

In a PWM device........there is no throttling. In other words, it’s not like an inverter, where we try to have a sine output.

The device is either off, or full on. The throttling is done with time. Not bias.

This throttling time is called duty cycle. I believe we can control that duty cycle with only one sensor input.

Carriage displacement. We can use this RATE to control the duty cycle.

If you plot acceleration, you will see a abrupt change in the slope when you lose traction.

In processor time, this is not abrupt, it’s a curve. I believe it is the tire unwrinkled curve.

I believe you will see a tire winkle curve earlier in the slope.

Anyhow, the idea would be to use duty cycle of PWM to maintain wanted slope on accelerometer.

If you use gears, each gear will have it’s own slope and rate.

There are other strategies. If the front tires don’t leave the surface, and you don’t have an accelerometer, you could use front wheel displacement VS motor rpm for PWM control.

Also, I might be wrong on this, but I think the armature current phase would see the slope change.......but not the wrinkle curves, and therefore would not know the change in slope is coming. We want to back off on duty cycle so as not hit that new slope.

Did I thoroughly confuse everyone? Sorry, I wasn’t trying too.

Am I all wet about this?

I would want to keep that first wrinkle on the tire, thru-out the entire run.

 

Thanks to everyone for your comments and input. The discussion has been very helpful to start dusting off my electronics skills.

I am going to find a small series motor to test with and built up a low power circuit as a proof of concept. I will probably do a simple resistor in parallel with the field coil to get base data from, and then work on replacing that with a basic circuit to replicate the resistor results.

This is a project that I am going to work on over the winter. Driving / racing season will be over for me soon as the weather cools off here in Maine.

This will be one of several winter projects / modifications I have in mind to further beef up performance and reliability of my car. One of the other ones, which is way off topic for this forum, is to replace the external transmission fluid pump with a hydraulic accumulator, which should be able to yield higher clutch pressure for low gear on launch.

I will plan to update this thread when I have questions or something to share with results.... or failures :-O

Cheers!
Wayne

 

You should start researching some industrial fork lift manufacturers. They have already solved this problem. Field weakening is very inefficient
and the best way is to use PWM and a SEPEX motor. In our industry we use a power transistor for armature control and an H-Bridge set up to control the field circuit. Both circuits are activated to achieve maximum torque for start up and when the encoder senses sufficient speed, the system then start to decrease the current in the field circuit, thereby allowing the motor to spin faster for speed. Just my two cents...thanks for letting me share.

 

You should start researching some industrial fork lift manufacturers. They have already solved this problem. Field weakening is very inefficient
and the best way is to use PWM and a SEPEX motor. In our industry we use a power transistor for armature control and an H-Bridge set up to control the field circuit. Both circuits are activated to achieve maximum torque for start up and when the encoder senses sufficient speed, the system then start to decrease the current in the field circuit, thereby allowing the motor to spin faster for speed. Just my two cents...thanks for letting me share.

 

From what I know, there are no 2,000 amp 200 volt sep ex controllers.

I have ridden in Dennis Berube's S10 and when the timers hit, your head hits the head rest. I have also ridden in a Tango, but it uses series / parallel shifting of two motors to overcome back emf.

Wayne's idea of a programmable field weakening device isn't a bad idea, if it can be done. It would prolong acceleration in a smooth curve, rather than dangerously snapping a resistor onto the stator. It would also increase top speed.

I don't know electronics well enough to know how one would change the resistance attached to the stator in a controllable fashion other than use a raspberry to switch an array of contactors with 400kW resistors attached to them. I would like to hear more about the buck boost idea.

Wayne has about $10k invested in his motor and controller. He wants a way to get more out of that investment rather than start over. Can this be done?

 

This company built the DC traction motors and controllers for our trolley buses in San Francisco:

http://www.skoda.cz/en/products/

You might send them a description of your application.

 

The motor was originally a 48V, 200A GE forklift motor. Dennis Berube (look up Current Eliminator electric dragster) completely rebuilt the motor specifically for drag racing, and it now can handle 200V and at least 3000 amps. He has been rebuilding and tuning motors for drag racing for decades, and has held world records for as long. Even the brushes are his own chemistry. It is his advice that I am following to add shunt resistance across the field coils. I am certain that it will work and work well, I just want to modernize the resistor method with a controllable resistive load.

Yes, you start finding the speed to enable it, by measuring field voltage and amperage during a run and find the peak voltage.

The motor controller can dump data 10 times a second and has a nice set of data to use. The attached chart is a sample of what it provides. It shows a run made at about 65% power. The transmission was clearly slipping on the shift to high gear, and you can see the power trailing off as the rpm climbs.

Do you know the field and armature resistance?

 

A milliohm meter bounces around when trying to get a resistance reading. I suspect that inductance is the cause. However, generally speaking, the ranges of resistance values I see shows about a 20:1 ratio of resistance for the armature:field coils. Also, running at 200 motor volts, the field coils will only drop about 10V of that.

As a guess, the motor will be well below 0.1 ohm, and the field coils would be maybe 5 milliohm?

I will try again tonight to get some better measurements and at least provide the ranges I'm seeing.

Do you know the field and armature resistance?

 

I think that with the low voltage drop across the field coils (10V or less) and the amount of current that would need to be shunted, maybe a maximum of 500A, is only a 5 kW load. Also given that it will only be used for 4-5 seconds at a time, a power circuit that is capable of 5kW should be able to handle it. The armature consumes the other ~395kW.

From what I know, there are no 2,000 amp 200 volt sep ex controllers.

I have ridden in Dennis Berube's S10 and when the timers hit, your head hits the head rest. I have also ridden in a Tango, but it uses series / parallel shifting of two motors to overcome back emf.

Wayne's idea of a programmable field weakening device isn't a bad idea, if it can be done. It would prolong acceleration in a smooth curve, rather than dangerously snapping a resistor onto the stator. It would also increase top speed.

I don't know electronics well enough to know how one would change the resistance attached to the stator in a controllable fashion other than use a raspberry to switch an array of contactors with 400kW resistors attached to them. I would like to hear more about the buck boost idea.

Wayne has about $10k invested in his motor and controller. He wants a way to get more out of that investment rather than start over. Can this be done?

 

A milliohm meter bounces around when trying to get a resistance reading. I suspect that inductance is the cause. However, generally speaking, the ranges of resistance values I see shows about a 20:1 ratio of resistance for the armature:field coils. Also, running at 200 motor volts, the field coils will only drop about 10V of that.

As a guess, the motor will be well below 0.1 ohm, and the field coils would be maybe 5 milliohm?

I will try again tonight to get some better measurements and at least provide the ranges I'm seeing.

Looking at the curves that looks pretty close. When I did the curves I got 220 volts at about 1500 amps at stall (start up) so .15 ohms for motor cables etc. So for the sake of discussion maybe the field is .01 ohms. So if we were to shift it when the current dropped to 100 amps that is only 100 watts in the field coil. To make it simple you could probably just shunt it with a FET with an on resistance of .01 ohms or so - no linear stuff involved and only 50 watts in the FET. You would need to be careful about what the inductive kick looks like from the PWM , worry about any regenerative braking and reverse if you have it. Could be easy enough you could have several steps.
Oops, forgot about the current / 10. So it's a few thousand watts. Makes it more difficult.