Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[MPaulHolmes]

Yes. First it's to be a buck charger and version 2 is to be with PFC. He even has like $15000 or who knows what worth of thunderskys just sitting there waiting for a BMS and charger, to start testing them. Also the car sitting there! With the motor installed! A big dang ADC motor, not like my pile of crap.

[NiHaoMike]

Originally Posted by adamj12b

OK, so the isolation uses a minimum of 4 isolated DC-DC converters. One for the 3 low sides and 1 for each high side. They way this work, is because of the internal configuration of the IPM. The low side of the IGTB is connects to the negative battery terminal, this is also connected to the negative supply terminal for the logic of the IPM.

So a little off track for the High side. The way the bootstrap caps work is they charge when the phase is pulled to the negative side of the pack. Since this is common to the negative supply of the logic, the cap charges with 15VDC. when the IGBT goes high, the logic uses the power stored in the caps to power the gate.

Now the reason for not having a common isolated supply for the whole module has to do with the internal wiring distance, delay and inductance. So when you have 3 IGBT's it is not possible to pick a common point to tie the the supply to, So you use 3 so they are all powered at the same time with the same delay, because they are all symmetrical.

So, with the Ebay IGBT modules, you will need 2 small isolated DC-DC converters to power each one. One for High side and one for Low side. There will be two opto-isolator's, One for High and one for Low. Each will get its power from the respective DC-DC converter.

Thats what I have for now. Some time tomorrow, I will post a schematic showing what I mean.

You only need one DC/DC with multiple isolated outputs. Adding additional outputs to an unregulated push pull converter is just a matter of adding more secondary windings and rectifiers.

Switchmode Power Supply For Car Audio
That's a starting point (start with the simplified version). Just use smaller components, change the turn ratios to get +-15v or so, and add more secondary windings. Leakage inductance will cause the output voltage to rise with no load, so use zener diodes to keep it within the ratings of the components. (The normal output voltages should stay below the zener diode voltages.)

As for signal isolation, Analog Devices makes some very impressive isolators. The ADUM7441 is good for 3 signals in one direction and one the other, such as 3 phase drive one direction and fault indication in the other.
ADUM7441 | 1 kV RMS Quad-Channel Digital Isolator | Digital Isolators | Interface | Analog Devices
I have used another Analog Devices chip that isolates USB and I'm very impressed at its performance.

And although the high voltage circuit would be isolated, you'll want some connection to ground to stop charge buildup and EMI but still keep isolation. I suggest a series string of two or more large resistors (1M total resistance or greater) in series from each side to ground, along with two Y2 capacitors from each side to ground. That will "soft ground" the midpoint as well as act as a discharge resistor for the main capacitors.

[adamj12b]

Originally Posted by NiHaoMike

You only need one DC/DC with multiple isolated outputs. Adding additional outputs to an unregulated push pull converter is just a matter of adding more secondary windings and rectifiers.

The only problem with using one supply with multipal outputs is you would need to make sure all the output leads are the same length to minimize delay from wire inductance. The problem with a single supply is you cant pick a common gnd point between 3 IGBT modules and have everything be the same length.

Originally Posted by NiHaoMike

As for signal isolation, Analog Devices makes some very impressive isolators. The ADUM7441 is good for 3 signals in one direction and one the other, such as 3 phase drive one direction and fault indication in the other.
ADUM7441 | 1 kV RMS Quad-Channel Digital Isolator | Digital Isolators | Interface | Analog Devices
I have used another Analog Devices chip that isolates USB and I'm very impressed at its performance.

The opto's that are in use now in the DC controller and are planed to be used in the AC controller are the HCPL4504. This high speed opto has a 3.75KV isolation rating.

-Adam

[NiHaoMike]

The only problem with using one supply with multipal outputs is you would need to make sure all the output leads are the same length to minimize delay from wire inductance.

The DC/DC supply only outputs 4 separate isolated DC outputs. Bypassing is local to the gate drivers. Delay is irrelevant.

A better idea for signal isolation is to use two processors, one on the low voltage side and the other on the high voltage side. Then only one signal isolation zone is needed to isolate serial communications. (As opposed to having to isolate gate drive, current and voltage sensing, and fault protection signals.) Maybe use a relatively powerful processor for motor control and high voltage side sensing and a simpler processor for I/O multiplexing/demultiplexing and low voltage side sensing. An additional benefit is that isolation delays are not introduced in the PWM loop.

[davidrrrd]

this is a link to a page on EVTV.com with lots of info on thundersky chargeing, this guy did his homework on them!

Electric 1957 Porsche 356 Speedster EV
there all informitive, ect. but dec.18th is your first stop!
enjoy!
david

[thaelin]

Thank you so much david for that link. I watched the part where he figured out what was killing his cells. It just makes so much sense. I understand why my friend lost his pack in less than a year in his prius at over $9K.

More now I hear. I am not into cars much but I am going for a bike. Not sure I want to do a bms now. Think the open charge method will do me fine. A friend over in the UK said no matter what size, I needed a bms for it. Well....

@ Paul:
Many thanks for your venture into the 3 phase arena. I mainly just want to see what makes them tick. A fet controller will most likely do me fine for under 600 watts. Nothing like making it your self tho. And saving a bundle doing it.

thaelin

Originally Posted by davidrrrd

this is a link to a page on EVTV.com with lots of info on thundersky chargeing, this guy did his homework on them!

Electric 1957 Porsche 356 Speedster EV
there all informitive, ect. but dec.18th is your first stop!
enjoy!
david

[etc]

Paul,

I have come across this 3 phase motor. After I test it to see if it works, I could send it your way. Let me know what you think.

Specs:
Lincoln Electric
10HP
230v/460v
26.6/13.3 amps
1750 rpm
10" diameter

Brian

[bennelson]

Just one thought on AC motors...

My friend Tom has a Dodge Neon he converted to AC with a stock salvaged AC drive and motor. It is also a 1750 rpm motor on it.

Although his car is clutchless, I notice that he shifts far more than I do on my DC conversion (also clutchless.) I believe that it is because over-volting the DC motor just gives you lots more speed (RPMs), but AC speed is controlled by frequency, therefore, typical AC motors may have a more limited range of speed than many DC motors, at least in what's available for an inexpensive home-build.

On the other hand, a car like a Solectria Force is AC, and is basically just direct-drive. It has no shifting what-so-ever. I believe the difference is in the design of the motor itself (although I don't know how much difference the the controller makes)

So, just throwing it out there that a homebrew AC conversion might be best off keeping the clutch, to make shifting easier - at least on lower RPM motors.

[jackbauer]

a two pole motor should be capable of twice its base speed. 6000rpm. more for 60hz countries.

[TheSGC]

Originally Posted by bennelson

Just one thought on AC motors...

My friend Tom has a Dodge Neon he converted to AC with a stock salvaged AC drive and motor. It is also a 1750 rpm motor on it.

Although his car is clutchless, I notice that he shifts far more than I do on my DC conversion (also clutchless.) I believe that it is because over-volting the DC motor just gives you lots more speed (RPMs), but AC speed is controlled by frequency, therefore, typical AC motors may have a more limited range of speed than many DC motors, at least in what's available for an inexpensive home-build.

On the other hand, a car like a Solectria Force is AC, and is basically just direct-drive. It has no shifting what-so-ever. I believe the difference is in the design of the motor itself (although I don't know how much difference the the controller makes)

So, just throwing it out there that a homebrew AC conversion might be best off keeping the clutch, to make shifting easier - at least on lower RPM motors.

The Solectria's doe have a specially designed motor, 120-156 volts and up to 12,000 RPMs, but it has full torque up to 4,000 RPM and drops off after that. I have a Solectria ACGTX20 that I will be building a controller for. I am hoping to get Solectria controller guts to learn from.