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

[MPaulHolmes]

I was actually switching at 10KHz. The good news at least on my end so far is, this layout hasn't had any problems in the DC case, and that's been with current up to 1550amp at maybe 160v. The controllers are identical except that in one case, it's 3 sine waves being formed, and in the other case it's either a constant frequency PWM or variable frequency PWM (to make the rushing water sound rather than a piercing noise). I had set the DC controller's PWM frequency to like 4KHz at one point (on accident), and it was making some of his instrumentation go bonkers. When it was changed to 10KHz (and 9-12kHz in the spread spectrum case), that problem went away.

It may help that the control stuff is several inches away from the high power switching, and the face of the high power switching is a copper sheet rather than a cable. There's almost no loop area between the copper sheets. The sides where the radiation can leak out is small and far from the bulk of the control board. It will be very interesting to see how things go at higher voltage and current. I've only tested it at 20 amps and maybe 120 volts. I hope the results are similar to the DC controller, as the layouts are the same.

[thingstodo]

Originally Posted by MPaulHolmes

I was actually switching at 10KHz.

Sorry - I relied on my faulty memory. Thanks for the correction

The good news at least on my end so far is, this layout hasn't had any problems in the DC case, and that's been with current up to 1550amp at maybe 160v.

The largest motor I can test (at least, I'll try to test) is 400VDC bus (maybe 280 VAC) with 300 amps continuous. I'm not sure what over-current the motor can take for 10 - 30 seconds, but I hope at least 3X. I hope the DC motor that will be coupled to it can put out that sort of power for 10 - 30 seconds .. before it melts

I had set the DC controller's PWM frequency to like 4KHz at one point (on accident), and it was making some of his instrumentation go bonkers. When it was changed to 10KHz (and 9-12kHz in the spread spectrum case), that problem went away.

Good info. My wiring is pretty hideous, especially in testing. So any potential problems with noise *SHOULD* be a problem for me during my testing.

It may help that the control stuff is several inches away from the high power switching, and the face of the high power switching is a copper sheet rather than a cable. There's almost no loop area between the copper sheets. The sides where the radiation can leak out is small and far from the bulk of the control board. It will be very interesting to see how things go at higher voltage and current. I've only tested it at 20 amps and maybe 120 volts. I hope the results are similar to the DC controller, as the layouts are the same.

I remember the discussions - you agonized over a *LOT* of details. I'm not saying that we will have problems. I'm not questioning your design. I'm a paranoid .. I'm actually *TRYING* to have noise problems during testing so that when you wire neatly and follow some rules .. we should never see any issues.

Onward!

[Piotrsko]

Tesseract over on DIY e cars said some choke style filtering was required on the DC-DC converter which seemed to fix the issue of magic smoke emmiting. I'm unimpressed with EVTV.

Once again, Jhmo, make the wiring as clean as possible because interference gremlins are annoying as all get out to get rid of once you get them. Intermittent gremlins otoh..........impossible.

[e*clipse]

I'm probably paranoid too - exspecially about those intermittent gremlins.

I've been looking for a way to make a faraday cage around normal twisted pair wires. This company sells a ton of different wire looms, some designed for insulating, some designed for EMI shielding, some designed for MIL-spec high temperature, some just designed to look cool...

WireCare.com - Your Cable Management SuperStore!

At a minimum, there are some good ideas for wiring issues we all run into w/ EV's.

- E*clipse

Originally Posted by Piotrsko

Tesseract over on DIY e cars said some choke style filtering was required on the DC-DC converter which seemed to fix the issue of magic smoke emmiting. I'm unimpressed with EVTV.

Once again, Jhmo, make the wiring as clean as possible because interference gremlins are annoying as all get out to get rid of once you get them. Intermittent gremlins otoh..........impossible.

[freebeard]

thingstodo -- Thanks for citing EVTV. When I've mentioned them before, it's gone unnoticed.

They point out that clutchless single-sped drivetrains are having problems, even Tesla, probably from shock loads when the power is applied in milliseconds.

Piotrsko -- Why do you say you're unimpressed? Because he snores when he inhales? I think I'll go see if the new episode is up.

MPaulHolmes -- Does your controller allow for ramping up the power?

[MPaulHolmes]

Oh you can definitely ramp up the power. Throttle is proportional to torque. If you want a gentle little nudge, then just give it a teenie tiny kersniggle of throttle.

[freebeard]

But if you floor it can there be there a little lag built in? To protect the hardware?

The new EVTV is about Apple poaching engineers from Telsa and BMW and last 2 hours is about pushing info around on the CANbus. All about byte ordering and offsets. It's less snore inducing on 1.25 speed.

Edit: I woke up at 3am and remembered the numbers. Allegedly, the gas engine loads the drivetrain in 80 milliseconds, while the electric motor takes 5 milliseconds.

[MPaulHolmes]

You can set it to any ramp you want. 80mS would be no problem if you wanted that.

[thingstodo]

Originally Posted by thingstodo

Our instrument supervisor came up through the ranks, so he's the most experienced tech I know. He has explained grounds to me many times .. Obviously I'm a slow learner. I just don't seem to be able to keep it straight.

I just talked to Rob and I need to get this written down before I mess up the details:

- recommended cable is #18 twisted pairs, individual shield on the pairs, and an overall shield. We just moved to a new inventory system ... I'm having trouble finding the 600V rated cable. This cable has the right specs, except for the 600V insulation. I think we use this stuff in panels only? Under $3 per foot. Belden - 9552 0601000 - Cable - Multiconductor - Allied Electronics

- tieing both shields together, on the same end, is OK if you are not in the Mhz or Ghz range. 500 Kbits is just barely into the Mhz range.

- we don't use a braided shield any more. It was used in the 60's and 70's for noisy environments. It no longer appears to be needed (not sure about that part)

- when you have an isolated supply, you can *make* an isolated ground with a large copper buss bar that is isolated from all else by insulators. This seems a bit extreme to me. I didn't ask how to figure out what 'large' is.

- using negative on the power supply as a ground has some issues, particularly for switching power supplies. Adding a low Equivalent Series Resistance (ESR) capacitor across the power supply leads helps if the power supply does a soft start. Otherwise you need a filter so that the capacitor does not short out the power supply on startup. The capacitor looks like a short circuit to high frequency noise.

- grounds come to a common point. NO DAISY CHAINS. Use finely stranded wire to give it low resistance to high frequency. And make the ground cables at least as large as any of the conductors that are tied to it.

That's all my notes and all I remember from the conversation

[e*clipse]

Thank you for looking into this, thingstodo!

FWIW, I did some digging for a cable that would meet the specs you provided, and found something at Mouser: ( I really don't need 1000 ft of the stuff )

6459 OR005 Alpha Wire | Mouser

Unfortunately, they didn't have a data sheet.

According to Mouser's info, It's an 18g shielded single twisted pair wire good for 600V.

I only need 1 pair, so this works out pretty well for me. Also, it's in 100ft spools which only cost about 60 bucks.

***edit*** I looked into it at Alpha wire's website. The wire is also only good for 300V.
However, for a bit more, this one fits the bill: (600V, but the wire gage is 20AWG)
5616B2001 BK005 Alpha Wire | Mouser

- E*clipse