Paul & Sabrina's cheap DIY 144v motor controller

[jedsmd]

Originally Posted by MPaulHolmes

OK, I need a plan for the wiring now. If I got orange 2gauge cable, then it would be "SAE" compliant I think. Here's my plan (so I can test each controller):
3 cables coming out of the controller will be around 3 feet long (from the point of connection inside the controller). They will be orange. All 3 must connect to the Motor Plus.

Okay my ADC motor uses 3/8" studs so I will need lugs with 3/8" holes -- never mind I will just drill yours out.

Here's the fuzzy part (but I'll sound assertive. Feel free to tell me to do something else):
3 cables will come out of the controller, and will all need to be connected to battery pack plus (maybe connect all 3 of them to a big fat 4/0 cable that runs to the battery pack plus). The lugs will protrude around 3-4 inches past the end cap, so I can use metal end caps with cable glands.

Please make my B+ leads 20" long. I will trim and add lugs as required.

3 cables will come out of the controller , and will all need to be connected to battery pack NEGATIVE (maybe connect all 3 of them to a big fat 4/0 cable that runs to the battery pack NEGATIVE). The lugs will protrude around 3-4 inches past the end cap, so I can use metal end caps with cable glands.

Please make my B- leads 20" long. I will trim and add lugs as required.

I am assuming that all cables come out one end of the controller, is that true?
It would be helpful if they did, but not essential.

[MPaulHolmes]

No problem! So let it be written, so let it be done. 3/8 studs is not a problem. I'll just get lugs with that hole size.

[MPaulHolmes]

OK now I need help finding a terminal block that has 10 positions. Maybe screws on the inside, so the screw head clamps a u shaped terminal attached to the wire, and similar for the outside. Is that confusing?! All the terminal blocks I've found only show one side, and I have no idea what's on the other side! anger! Or they just sandwich a wire down by screwing a screw, but I've found those to be very unreliable with stranded wire. It comes loose all the time on my milling machine setup.

Or I could solder one side, and just a screw terminal on the other side. Oh the humanity!

EDIT: NEVER MIND!!! I found something I think will be OK.

[Astro]

Originally Posted by MPaulHolmes

OK now I need help finding a terminal block that has 10 positions. Maybe screws on the inside, so the screw head clamps a u shaped terminal attached to the wire, and similar for the outside. Is that confusing?! All the terminal blocks I've found only show one side, and I have no idea what's on the other side! anger! Or they just sandwich a wire down by screwing a screw, but I've found those to be very unreliable with stranded wire. It comes loose all the time on my milling machine setup.

Or I could solder one side, and just a screw terminal on the other side. Oh the humanity!

EDIT: NEVER MIND!!! I found something I think will be OK.

Paul,

How about something like this. Cheap (<$5) and easy to mount.
10 PIN Aviation Amphenol Cable Connector Plug 19mm 20mm | eBay

[thingstodo]

Originally Posted by MPaulHolmes

OK now I need help finding a terminal block that has 10 positions. Maybe screws on the inside, so the screw head clamps a u shaped terminal attached to the wire, and similar for the outside. Is that confusing?! All the terminal blocks I've found only show one side, and I have no idea what's on the other side! anger! Or they just sandwich a wire down by screwing a screw, but I've found those to be very unreliable with stranded wire. It comes loose all the time on my milling machine setup.

Or I could solder one side, and just a screw terminal on the other side. Oh the humanity!

EDIT: NEVER MIND!!! I found something I think will be OK.

The terminals that we use (day job) in a vibration environment (like a car?) have springs and jaws. The wires push into the jaws and are pressed against the terminal jaws. Some 120V receptacles use this same method. Our electricians don't like these terminals. They are difficult to take apart, and the bases are a bit cheesy - the clips break off pretty easily.

Stranded cable 'relaxes' when used in screw-down terminals. We normally re-torque terminals about a month after commissioning new equipment (if we don't use the spring clamp terminals). Then they are re-torqued about 6 months after that. And then we have a PM to check the torque every year.

What gauge cable do you need the terminals to accept? I can look up the vendor and the part numbers from our warehouse.

If the cables are large enough to require lugs, nordloc washers on the bolts work pretty well but the they still need to be re-torqued .. a lot less often.

[MPaulHolmes]

I found a really nice thing that lets you connect wires through an enclosure, and it was only 50 cents each. Each piece lets 5 wires pass through, so the controller will need 2 of them. I should have taken a picture of it, since they all arrived already. Oh well, here are some other pictures:
https://www.dropbox.com/s/90xm29bsju...2011.32.01.jpg
Female spade connectors soldered into control board, push onto the male spade connectors on the IGBTs (I didn't push all the way down, since it goes on really solidly):
https://www.dropbox.com/s/au2noihb23...2011.52.26.jpg

Besides the spade connection points, there are 4 #8 screw holes, and nylon 1/2" wide standoffs (at least I think the standoffs were 0.5" wide. Oh they could have been 3/8").

Capacitor mounts to igbts and enclosure (I didn't actually have the capacitor mounted to the enclosure for the picture. it's just sitting on the igbts. Heck, it's not even screwed into the IGBTs at the moment either! haha. But there are no screws that are difficult to access. It all goes together really pretty easy.):
https://www.dropbox.com/s/bo1gjy0019...2011.36.04.jpg

Notice that the nice and calm control section (left half of the control board) is physically far far away from the very naughty, noisy, high current section! And it's a 4 layer board, with almost no exposed area to allow for interference to sneak in.
https://www.dropbox.com/s/xkwb6a88xy...2011.52.59.jpg

It's using 3.3Ohm gate resistors. I"m waiting for the npn and pnp current buffers (15 amp peak) to arrive, probably by monday. Those are the open, silvery spots in the TO-220 package laying down.

[MPaulHolmes]

All 3 controllers now have IGBTs with freewheel diode matched voltage drop of +/- 0.001v under a load. I had some nichrome wire and batteries, and it made for a good constant current test. It's supposed to be within 0.05 according to an application note on paralleling them. The differences in drop from the IGBTs turning on should be small, but it won't matter, since each M+ cable will correct for that difference.

[flores]

Originally Posted by MPaulHolmes

All 3 controllers now have IGBTs with freewheel diode matched voltage drop of +/- 0.001v under a load. I had some nichrome wire and batteries, and it made for a good constant current test. It's supposed to be within 0.05 according to an application note on paralleling them. The differences in drop from the IGBTs turning on should be small, but it won't matter, since each M+ cable will correct for that difference.

Does this mean they are almost done ?

[flores]

Originally Posted by MPaulHolmes

The differences in drop from the IGBTs turning on should be small, but it won't matter, since each M+ cable will correct for that difference.

Paul, please help me to understand.. how do the separate cables help with this? are they acting like some sort of buffer? how does this work?
And why does a controller like the evnetics doesn't have this?

[MPaulHolmes]

Lots of debuggIng this afternoon after work. It's all almost good to go as far as the software and circuit board goes.

The soliton people have 10's of thousands of dollars of igbts (we keep buying from them on ebay. haha) that they match for turn-on point, diode voltage drop, etc... We basically had one shot, with a couple orders of 6 igbts. There's just no way to guarantee current sharing in the IGBTs with a sample size that small. So, let's say igbt1 has a voltage drop of 1v when its on. And IGBT2 has a drop of 0.9v when its on. We weren't able to do anything about this due to our gross poverty. LOL. Let's pretend that, as a result, IGBT2 is passing 600amp while IGBT1 is passing 400amp.

Well, that's a shame. Now we need to lower the current for everyone so that IGBT2 is in safe territory. Now IGBT1 is only passing like 200amp. What a waste!

PLAN B: With 3 feet of 2 gauge cable, you get a voltage drop of 0.193v at 400amp, and a drop of 0.289v at 600amp. Basically, that's a difference of 0.1v. Now, it's like we were the big boys! IGBT2 has a voltage drop (counting its cable) of 1.189v, and IGBT1 has a voltage drop (counting its cable) of 1.193v. Now, because they have the same voltage drop, they would both be sucking the same amount of current. So, it wouldn't be 600amp and 400amp anymore. They will be driven closer to each other. Being forced to share current better. But you can always just dump that idea and stick a big bus bar across it. But right now, each of the 3 igbts has its own current sensor, and its own hardware overcurrent circuit, so there's no way that one can go out of bounds.

i2 - i1 = (v1 - v2)/R. Where i2 is igbt2 current, and v2 is igbt voltage drop, and R is resistance from the cable.