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

[freebeard]

thingstodo --

Originally Posted by explainthatstuff!

Page not found! ("404")

In the VW world an input shaft is a standard tool to align the clutch to the flywheel. They're even available in plastic. Not that that would take the torque. If you can cut the shaft you can also make 4 more cuts and leave a 3/4" square stub standing. Pop you torque wrench on that; no welding.

what is a common failure mode for a 3 phase motor that has had too much current?

I was going to say the magic smoke comes out, but thingstodo covered it.

it turned out to be a debugging array index that went out of bounds, and modified a couple variables.

Buffer overrun? Sanitize your inputs or you'll have hackerz in there.

Cyruscosmo, CrazyRick -- Back on page 118, I pointed to this video;

Field Oriented Control of Permanent Magnet Motors

..and it has the most lucid explanation of FOC I've found. Astro looked at it and said it made "a crazy sort of sense".

What do you think?

[cts_casemod]

Without going trough the complete thread, just wanting to know if Paul's inverter is already going.

I am making efforts designing one myself. I mean, redesign the one that served me for the last two years.

My build thread can be found by googling "vw polo ac induction" although the details for the new inverter are yet to be published

Regards

[MPaulHolmes]

no its not ready yet. I discovered the problem though. I'm using 6 surface mount drivers, and just noticed that I only soldered one side of one of them. haha. It was making intermittent contact.

clearly the motor must be broken, because it could never be something I did! haha

So, I have an ingenious plan to fix it. I'll solder it.

in fairness to me, they never mentioned that requirement in the datasheet.

[cts_casemod]

Originally Posted by MPaulHolmes

no its not ready yet. I discovered the problem though. I'm using 6 surface mount drivers, and just noticed that I only soldered one side of one of them. haha. It was making intermittent contact.

clearly the motor must be broken, because it could never be something I did! haha

So, I have an ingenious plan to fix it. I'll solder it.

Ah! Great to know!
Paul, what approach are you using for short circuit detection?

I am a bit undecided if I should use VCE sat or hall sensors with a comparator. The first should have the advantage of not introducing any extra inductance on the DC-Link, but its bit of a pain to implement.

[MPaulHolmes]

I'm using both. desat would be a horrible pain, if not for an awesome little chip that is both a driver and desat detector (and isolated supply undervoltage protector!), all for $4 each. its only for 1200v 150amp igbts, but I added a current boost on the output using an npn pnp totem pole that was recommended in their datasheet. I think its the fod8316. I'm on my cell phone, and its a little hard to check. it drives 600v 600amp igbts no problem at 10khz.

[e*clipse]

Originally Posted by MPaulHolmes

no its not ready yet. I discovered the problem though. I'm using 6 surface mount drivers, and just noticed that I only soldered one side of one of them. haha. It was making intermittent contact.

clearly the motor must be broken, because it could never be something I did! haha

So, I have an ingenious plan to fix it. I'll solder it.

in fairness to me, they never mentioned that requirement in the datasheet.

LOL! I guess that *sorta* points to the design's robustness. That's a good thing!

Freebeard - yes, I've seen the video. It was a good review. This whole discussion I was attempting to bring up was trying to get folks to step back from the problem and **really** understand what was physically happening. I mean feel it. Unfortunately, I'm not that good at communicating, so the point was missed. I'm happy to see a couple folks understood my point, but they already knew what was going on...

[e*clipse]

Originally Posted by thingstodo

Running out of voltage, and using the available voltage and current to best advantage, just makes sense.

I think I get it

thingstodo,

Just wanted to say, I have a huge amount of respect for your knowledge and experience with this. The last thing I ever wanted was to have a dissagreement with you about this.

Thanks for sticking with it; I'll try to be more clear in what I'm attempting to communicate.

[freebeard]

Imagine what it's like trying to learn from the discussion.

What I see from my perspective is people trying to agree on a common vocabulary. Ferinstance the cited video never uses the term 'slip angle'. I suspect it's lurking in there somewhere, possibly as .... I guess I'll have to watch it again.

[cts_casemod]

Originally Posted by MPaulHolmes

I'm using both. desat would be a horrible pain, if not for an awesome little chip that is both a driver and desat detector (and isolated supply undervoltage protector!), all for $4 each. its only for 1200v 150amp igbts, but I added a current boost on the output using an npn pnp totem pole that was recommended in their datasheet. I think its the fod8316. I'm on my cell phone, and its a little hard to check. it drives 600v 600amp igbts no problem at 10khz.

Oh, the criminal that was not doing his job properly, I can guess

I cant stock those here in the UK, but a quick look shows the TI ISO5500DW and the Avago ACPL-339J-000E as possible replacements.

Do you have some unpopulated boards you could sell for these? looks like a fairly decent solution.

Regards

[e*clipse]

I'm nearing completion of the resolver board layout.

Brief statistics: Board dimentions: 2.8" X 1.225"

87 components... I've managed to fit all this on a two layer board, with all the components on top for a toaster-oven surface mount solder job. Hopefully there isn't a noise interaction problem between this board and the main controller. If there is, I guess it could have a 4 layer board for some shielding, but they cost twice as much.

The main problem I'm having is how to solidly attach this board to the main board, in a vibrating environment.

There is a little space for something like a 3mm non-conductive screw & spacer in the gap between the logic and driver sections on the main controller board.

There also appears to be a bit of space on either side of the QEI connector for perhaps a latching connector or something like that??

I have found some 5 pin connectors that have latches, but no female sockets to go with them.

I'd like to have something more solid than one bolt holding this critical part to the control board. I guess I could machine something out of nylon or something... Or maybe simply use a pin header and solder both boards to the header. It would just be a RPIA to fix anything if something went wrong with either board.

Does anyone have some real creative and small ideas for fastening this thing? Maybe there are some svelt little clips?