[Tesseract]

I've seen this recommended for Curtis controllers, but not for any others. I think this is because in the Curtis controllers PWM duty cycle is proportional to throttle with cycle-by-cycle current limiting to protect against overcurrent (this is why Curtis drops the frequency at low motor RPM because then there is little back EMF to oppose the rise in current, so additional inductance will help here).

That said, the ideal type of controller for traction motors (AC or DC) has motor torque (current) proportional to throttle, so current should always be under tight control. A certain minimum level of inductance is necessary for this type of controller to accurately regulate current, but most motors have more than enough inductance so this is a non-issue.

What a coil of wire in the motor circuit will definitely do, though, is radiate lots of electromagnetic noise. This could interfere with other electronic devices, even in other cars on the road. The FCC tends to frown upon this, though, I'll admit, people rarely complain about this sort of thing so probably not something to worry about.

 

[Tesseract]

In theory a well designed inductor should also improve the motor efficiency.
....
If only someone had a dyno setup.......

Sure...with an emphasis on "well-designed", though.

I have measured the ripple current through one of the WarP9s on our dyno and the results where somewhat surprising... there is some saturation effects as current increases, but even at 400A the AC ripple component was only 14%, or 56App. The ideal AC ripple component is usually set to 40% in a switching power supply, which would be 160App in this case! So this implies a relatively high inductance in the WarP9 - even at 400A - of over 100uH.

If the Curtis is struggling with this amount of inductance then you will need a pretty big coil of wire to make a meaningful difference - something on the order of at least 25% of the motor's inductance, but preferably more like 50% to 100%. Plugging some random numbers into my handy single-layer inductance calculator spreadsheet, it seems that a close-wound coil of 20 turns and 200mm diameter of 2/0 wire (OD: 13mm) will have an inductance of ~43uH but require 13.6M of cable!?! That's almost 50' of wire which at a current price of around $3 per foot is a pricey bit of kit, and that's not even considering the resistive losses which will assuredly greatly outweigh the iron losses in the motor housing!

Anyway, while you are technically correct that there are some iron losses in the motor housing, and that adding external inductance will reduce those particular losses, the unfortunate reality is that you will be adding far more in losses from cable resistance.

Think it could help a noisy motor?
...
I have a few air core inductors and one iron core one that was made with an old transformer.

Well, Pete, if the noise is the motor "singing" at the controller's switching frequency then extra inductance might quiet it down, but most motors don't seem to act as particular good "speakers" until the frequency drops below 6kHz or so. Unfortunately, most motors do sing pretty loudly at 1.5kHz, which is the fallback frequency for Curtis.

As for making your own inductor with old transformer cores, keep in mind that the ripple frequency is the switching frequency so a 60Hz transformer isn't going to cope very well here (it's going to have extremely high losses - much higher than the motor will).

 

[Tesseract]

... I used an iron core made of very thin laminations...

There's a huge amount of dc bias here, though, so you really need to use a core with an air gap (preferably a distributed one) otherwise it is just going to saturate. Powdered iron or MPP are good choices.

14m of 2/0 wire has only 3mOhm of resistance according to my calculations, at 250A (~rated cont motor current) that's just 200W of losses out 20kW to propel the car, 1% significant but not that much.

Sure, at 250A the I²R loss in the 14m of cable isn't too bad - and the AC losses are probably negligible - but at 400A (the example datum I used before) the conduction loss will be 480W. That exceeds the conduction loss of the IGBTs in the Soliton1 (~0.9V at that current).

I seem to recall that you guys measured about ~10% lower motor efficiency than the manufacturer stated, I wonder how much of that can be attributed to iron losses from ripple current.

It's been awhile since I did that test, but at 200A - the current the WarP 9 is rated to carry for 1 hour - the efficiency was the same as what NetGain claims, or about 85%. It was at much higher currents that the efficiency plummeted. I don't recall the exact number, but at 1000A and 105V (1900 RPM on our dyno) the efficiency was closer to 60%.

Kostov has been working on some new motor designs for EVs - including a ~10"/250mm model - that we'll be testing out on our dyno soon. I'll try to take better notes this time since the goal before was endurance testing of the controller prototype, not seeing what the WarP 9 could do.

As for the AC loss component in the motor frame, I'd guess that 56App of ripple (from the previous example) would maybe result in 56W of additional losses, with almost all of those in the back iron (or shell) as the field poles in the WarP 9 are laminated.