Solution: cable shielding Edit

Feb23. After a bit of cable movement and rerouting, it turns out that my signal cable is picking up interference from the MagMotor brushes. The most convenient routing of the cable from the CPU to the OSMC runs right near the rear housing of the Mag. If I slide it far from the Mag, all is good, as I move it closer, things start to get wonky. I should have known better to have the cable so close.

As for your EMI, don't use ali foil. Use something you can solder to, which aluminum foil isn't. You could also use a bit of bent metal screwed to a grounded fixture (Foil might be a bit flimsy for that). If you situate it physically between the brushes and the cable, you'll get some "Faraday cage" type benefit. That'll be an empirical fix though - just see what works. By far the easiest, IMHO, and the most reliable and repeatable, is to remake your signal lead with a grounded sleeve, such as a length of AV coax even. The sleeve needs to be connected to ground, preferably at both ends. If it's a loom of several conductors, you could even try something like shielded CAT5, which should be easy to source. It doesn't sound like what you've got would be too long for that.

Twist the power cable tightly (twisted pair) and do the same with the motor cables, twisted and separate from the power cables.

I scrounged up a shielded Cat6 cable, rewired the OSMC and microcontroller with that, and all is a smooth as butter. It's amazing what a bit of shielding can do.

Ah, I think I see the confusion... I didn't explain that it's polypropylene CLOTH/FABRIC that I use (woven or non-woven) not solid pieces of PP. This is found in, eg, cloth grocery bags, but the most economical source is "landscaping fabric". (which isn't identified on the packages but it's PP cloth.) I rip it into ~~6" x 6" pieces for stuffing into the molds. And I use epoxied PP strapping or 'webbing' (strips of very heavy cloth) to hold magnets on rotors. I confess I was initially surprised to see how well the PP fabric wets with epoxy (and apparently with polyester) resin, knowing how little sticks to the solid pieces. Strength: Think of how strong those yellow PP ropes are, or the straps. And there are lawnmowers with solid PP bodies.

Pete Covert Edit

saticoy I'm building a magmotor based skateboard, and am using an OSMC to power things. I've done this successfully in the past, although I was using an MC1 as a motor controller instead. With the new version I thought I'd try an OSMC due to size and weight reasons. I've blown the OSMC up twice now trying to spin the MagMotor with no load. The first time I had some buggy code that may have been slamming the throttle backward and forward, and I was also dropping AHI and BHI to zero during "coasting" scenarios. I spent about a month reading up on this group. There's a lot of info archived away here, sifting through it takes some serious time. I changed my control scheme to always hold AHI and BHI high, and PWM either ALI or BLI while holding the other line low.

I read up on common repairs, tested which components I could, and ended up replacing the HIP chip and 12 of the 16 FETs (all except Q13 through Q16) with matching FB3207Zs. After debugging my control code, and verifying that the OSMC would power on without popping any FETs, I tried testing with a much smaller motor. The test program would ramp from 0 to 100% in on direction, pause 5 seconds, ramp back down to 0%, pause 5 seconds, ramp to 100% in the other direction, pause 5 seconds, and so on.The small motor worked fine. No strange noises, no odd behavior. I then tried the MagMotor.Low duty cycle seemed fine. Once I got near the 80 to 100% mark, the motor would pulse very noticeably. I killed power, checked for heat on the FETs, feeling none, I tested a few more times. Each time I cut power, the FETs were cool.

I let the code cycle from forward to reverse a few times without killing power. Suddenly I truly got 100% power, as the motor ran full speed with no pulsing. About mid shrug, Q1-4 (AH) and Q9-12(AL) exploded.I can continue sifting through the archives, but I'm hoping someone can pipe in and say "oh, obviously! component xyz is blown as well, and needs to be replaced". Otherwise I foresee a lot of trial and error and trips to DigiKey in my future.

Feb 20 Edit

david.coones What value are your gate resistors?, And do you have diodes in the gate lines to facilitate turn off of the FETs? It looks there might be an issue with ringing on your gates, which is partially switching them back on via the miller capacitance. This would be because the inductance of your system is of the same order as the lead inductances local to the FETs. By increasing the inductance in the line, you soften the current ramp and the resonance through the gate cap on the FETs goes away.

I'm getting the impression that the acoustic noise could be similar to what you'd get if you switch both the high sides and the low sides in the bridge rather than just the top or bottom legs (leaving the other permanently on). This could be consistent with low impedance on the motor lines and higher impedance on the gate FETs, particularly at switch off.

This particular mechanism is a resonance thing, but won't be affected too much by switching frequency. And lowering the local impedance to the batteries by increasing the size of the reservoir caps, although it's a good idea to do it, might not make the problem go away. Incidentally, what do your battery cables look like? Sorry if I'm confused, but the cables/connections you've been discussing are between the OSMC and the motor, but not the OSMC and the battery, correct? Inductive battery leads (long/thin) and low impedance motor connections could contribute to a resonance problem. Also, getting the FET leads as short as possible, if they're not already, might help.

Try changing the gate resistor values (down) and see what happens. Just make sure its not so low that it over drives the HIP. I can't remember what you said the arrangement was, two or four FETs in each leg, but a resistance for the parallel gate in each leg of the bridge shouldn't go below about 15Ohms for the HIP.

Hope this changes what's happening. It might not be for the better, but at least it would confirm there's a resonance issue involving the gate impedances.

other Edit