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

[thingstodo]

Originally Posted by ricemodder

This is neat. I took apart a 2017 Kia inverter and took a peek at the layout.

It has 2 resolver inputs to a 10/12 bit digital input to a MPC5500 NXP processor.

I can`t read the chips. What are they using for isolation?

[ricemodder]

Originally Posted by thingstodo

I can`t read the chips. What are they using for isolation?

The high and low side drivers are ACPL-332J-000E then fed to push pull transistors for more power.

[thingstodo]

I now have 3 of the 4 Ranger EV's at home and have spent some time looking through the various connectors and parts of the vehicles.

The motor is a Permanent Magnet AC motor (what is the different between PMAC and Brushless DC or BLDC?), but I have no ratings on voltage or max speed.

The Sevcon Gen 4 AC controllers are rated for 650A and have a 48V DC pack (somewhere around 43V is low voltage cutout and the original Lead-acid charger floats at 57.6V). The docs don't mention a boost circuit and I cannot locate an inductor so I expect that the motor is driven directly by PWM of the pack voltage. Speed is limited to 25 kph but I am confident that is a setting in the controller and not a limitation of the motor.

There is encoder feedback although I won't know what voltage until I get a service manual.

Paul - driving this motor (eventually) with my AC controller - would I need to load different firmware or is that a setting change? I have a lot of work to do getting everything else working first, before I venture into replacing the controller and running it at higher top end speed.

I think it would make an interesting video to start from 'here is a motor mounted in a vehicle drive train' and progressing through 'there is not much info on the motor but we entered what we had', 'here is how we did the initial tune' and finally 'this is how it runs after we played with the tuning some more'. It will likely be at least a year before I get to that point though.

What do you think?

[MPaulHolmes]

Hello! I don't think there's a difference between PMAC and brushless DC. Well, I think sometimes PMAC wants a pure AC waveform for best results, and brushless DC wants trapezoidal? But I'm not sure. Also, both motors work fine with an AC waveform.

If it's a permanent maget motor, and you have an encoder, you shouldn't need different firmware. You might just have to set the number of pole pairs, motor type (permanent magnet with encoder) etc.

That would be absolutely awesome to go through the whole tuning process and everything. That's something that I should have done and never got around to doing.

[thingstodo]

Originally Posted by MPaulHolmes

If it's a permanent maget motor, and you have an encoder, you shouldn't need different firmware. You might just have to set the number of pole pairs, motor type (permanent magnet with encoder) etc

Thanks Paul,

Some time ago there was a question about lowering the cost of the AC Controllers for a production run by using parts that were rated for lower voltage, less capacitance, etc.

Did you ever look into the cost savings? I remember thinking (and I think I posted) that the cost savings would not be significant compared to the PITA of tracking all of the extra components.

The SEVCON controller is *TINY*. 1.5 inches thick, about the length and width of a sheet of paper, air cooled, and not much for fins. I have trouble believing that it will do 650A for more than a few seconds before it overheats. The back of the controller is maybe 3/16 inch aluminum and is just bolted to the frame. No thermal grease even! ... or none that is left after the guys at the mine did their service, anyway.

The caps inside that SEVCON must be very small and have incredibly low ESR.

After I get one of these things running I expect that I will be adding a *LOT* of instrumentation to see the temperature the controller is at, the battery amps and motor amps, the motor temps, etc. I'm also looking at monitoring the ESTOP circuit. The interlocks include a safety lockout switch, the key switch, the parking brake, the 'gear' selector (F, N, R), the speed selector (L, M, H), charger plugged in, feedback that the contactors closed ... and perhaps other stuff? Troubleshooting the thing when it stops is not going to be fun without some monitoring and some sort of display ... maybe even a single LED that flashes morse code? How Geeky would THAT be?

[MPaulHolmes]

I know they have low power 6 pack chips that can be very affordable, with a maximum current of maybe 10 amps or something. But they are not isolated. I"ve never messed around with those, but I would like to eventually. Those are rated for 600v though, so the sevcon probably uses some low voltage mosfets and boot-strapping (however the heck that works). The 6 isolated supply way of doing it is the sledge hammer approach.

I love the idea of the LED morse code!!

[thingstodo]

Originally Posted by MPaulHolmes

I know they have low power 6 pack chips that can be very affordable, with a maximum current of maybe 10 amps or something. But they are not isolated. I"ve never messed around with those, but I would like to eventually. Those are rated for 600v though, so the sevcon probably uses some low voltage mosfets and boot-strapping (however the heck that works). The 6 isolated supply way of doing it is the sledge hammer approach.

I love the idea of the LED morse code!!

If the controller is very close to the motor, and it pretty much HAS to be on a vehicle ... then the reflected wave stuff is minimized (since the impedance of the cable is quite different from the impedance of the motor, any waves or pulses will 'reflect' from the motor and you can see a bit over 2X buss voltage at the motor terminals and the controller terminals). A 48V (57V) buss with full reflection is only 96V (115V). Add some safety factor ... 200V (230V) or so? That's still well within the MOSFET range. But driving MOSFETs to switch both the high side and low side takes a bit more circuitry and effort .. so I read. The EVTV episode where they take apart a blown-up TESLA controller showed a *BUNCH* of what looked like TO22 MOSFETs clamped to water-cooled tubes ... not sure if they mixed P channel for switching the positive side with N channel for the negative side?

Mouser lists 300V 400A IGBTs at $35 each, you'd need 6, plus the low esr caps .. anything else that would change based on the lower voltage? I know that you have 6 isolated supplies - would that change with the lower voltage? Somehow I doubt it.

400A does not match the 650A SEVCON I have but SEVCON have a 450A model available for the Ranger EV as well. I'm not actually sure what they would do continuous either. If you are pulling 650A for more than a few seconds, you would melt the cables, the motor, or the batteries even if the controller COULD put out that sort of power.

It's interesting to think about, anyway

[ScottyWarpNine]

I'm building up your controller right now and I have a few stupid questions:

1. In the BOM you list a fancy circular capacitor. I've never seen anything like that before now. Is there any reason why I couldn't just use some big electrolytics?

2. I know the pic code has been evolving, but has the ATtiny code for the DC-DC converters still the same as in the github library you linked to in the instructable?

3. I'm on the fence about buying a Pickit 2 or 3. Would either work? Which would you recommend? 3 is a bit more expensive.

Thanks for the help! I picked up a 1hp motor and I hope to put it in a dirtbike frame soon.

[thingstodo]

Originally Posted by ScottyWarpNine

I'm building up your controller right now and I have a few stupid questions:

1. In the BOM you list a fancy circular capacitor. I've never seen anything like that before now. Is there any reason why I couldn't just use some big electrolytics?

2. I know the pic code has been evolving, but has the ATtiny code for the DC-DC converters still the same as in the github library you linked to in the instructable?

3. I'm on the fence about buying a Pickit 2 or 3. Would either work? Which would you recommend? 3 is a bit more expensive.

Thanks for the help! I picked up a 1hp motor and I hope to put it in a dirtbike frame soon.

I will take a shot at answering the questions (so Paul can correct me):

1 - you can use electrolytics or any other type of capacitors you like. The ring capacitor has some interesting specs. The only one that I remember is Low ESR (equivalent series resistance). And since it has many contacts on the positive and negative plates, it somehow minimizes the noise induced by the carrier frequency. Something about minimal enclosed area? But other capacitors will work. Perhaps not quite the performance, but they will work. I have taken apart industrial controllers from several vendors. NONE of them use these ring capacitors.

2 - I didn't see the code in the instructable. But I know that this code has not changed for quite a while. There was no update (that Paul mentioned)
between the original Cougar controller, the IGBT DC Controller and the AC Controller. I would guess that code is the same. I think I have the hex file in my archives somewhere .. I can send you a copy if I locate it.

3 - I bought the PicKit 3. It is newer and has current libraries for more of the PIC family. It is also what Paul is using, so this should minimize the problems. Paul favors an older version of the C compilers .. and I have some problems building his new code. MPLab X is a re-write from scratch of the IDE, sort of a fork of a different series of IDE that Microchip purchased/is using ... I likely messed up that story. But it is not much like the older version.

Paul has been pretty busy lately - hopefully he is still checking the group

[ScottyWarpNine]

Cool, I just got an SOIC8 clip on programmer so I can program the ATtiny. The Pic can be programmed once it is soldered on the board via the ISP header, correct?