Uh-oh... What did I kill...

[Petrovich]

I just got my Microsquirt project motorcycle back to my house (if I have to push this thing for 3 miles again, I'll put pedals on it). Now, for the past couple of weekends, we've been trying to get it started, and had a hard time getting a tach signal. So today I take the MS home, and it smells like a droid corpse. Sure enough, I open it up, and the board around ignition drivers and the heat sink has little black bubbles - looks like the whole thingie got overheated.

So my question is, a) how do I find out what's broke, and b) how do I keep this from happening again? My dwell settings were left at default 3 ms, I was driving a Subaru dual coil on a wasted-spark 4-cylinder bike. Would all this cooking affect my VR sensor input? As far as I know, everything else was working fine.

[SQLGUY]

Well... it sounds like the VB921's are blown, though it's hard to say for sure without taking a multimeter to them. AFAIK, they are no-longer available, but there are some Bosch parts that are now being used in their place... pretty much a drop-in except that you have to add a heat sink insulator (and maybe some silicone grease, depending on the type of insulator you use).

This shouldn't affect the VR input.

For the dwell, did you measure the coil inductance and resistance, or scope them to adjust dwell, or just leave the default in place? If you left the default in place, it's possible that you were overdwelling them and blew the 921's that way.

How is your ignition trigger system set up? You mentioned VR, but nothing else. What make/model is this bike? Is it crank wheel triggered or dual tachs?

Funny that you're using the Subaru dual coil pack. I considered one of those before going with my Saturn coils. I liked the layout and compactness of the Subaru coils, but I was worried about difficulty in finding terminals that would fit them.

[Petrovich]

Ignition trigger is a spare Ford EDIS VR sensor that I had, one wire is routed to signal, and the other one is attached to the shielding. It is triggered by a plastic drum with 7 bolts attached so that the heads form a 8-1 trigger wheel. I did leave the dwell as is, since I have no inductance meter or an osciloscope. Can I just set the dwell to a very low value, and then raise it up until I get spark?

I don't have a problem finding a replacement for VB921's since I have a bunch of NOS ones. Hell, they might even work, still - we've been getting spark every time we would have any sort of a tach signal. But the whole volcanic ash that I have around them kinda worries me.

Oh, and this is all on an '82 Suzuki GS650g. I don't have stock coils, stock ignition, or stock anything, so it's all made out of whatever I could find on a local junkyard or in a parts store.

[SQLGUY]

If the VB921's still seem to be working, I'd leave the dwell alone until you get the bike running reliably, then try reducing it while monitoring coil and driver temperature, and spark intensity, until you get to what seems a good compromise (as cool as possible with still a good spark). Or get an LCR meter and measure the coil. Or check to see whether someone might have already posted characteristics of that coil somewhere.

I'm running an 8-1 "wheel" myself on my '82 Yamaha 750.



With mine I had problems getting initial tach synch. If I tapped the VR pickup with a metal object, I'd hear the IAC valve actuate, but the fuel pump wouldn't run. I had the same behavior when cranking, most of the time. One time in a dozen or so I'd get tach synch, spark, and the fuel pump would run, and, once the bike started, it would run great. Reversing the VR pickup polarity (switching which wires went to pins 32 and 33) helped this issue a lot.

Cheers,
Paul

[Petrovich]

Ok, I got some very steady signal by drastically shortening the sensor wire. Unfortunately, VB921's are actually dead - they were grounding the ignition at all times, causing the whole thing to overheat as soon as I would connect power to the coil.

I pulled them out, and I'm now trying to replace those by bringing them outside of the board. Unfortunately, my tracing abilities suck - I can't find a location where I can pull a signal from. Can anyone give me a pointer as to where I should take IGN1 and IGN2 signals? I got the ground and output lines fine, but the signal thing drives me crazy.

[Petrovich]

SAVE ME SUPERMAN!!!

I have regained ignition on one of the ignitors by inserting new VB921's and re-flowing the whole thing. The second replacement VB921 (the one further from the connector) landed well, and has ground and output showing fine, but there is no input signal going to it (the pin that's farthest from the connector).

Can sombody please take a picture of their GOOD microsquirt, with detailed VB921 shots from both sides of the board? I need to know what trace I need to connect to in order to get proper signal. Otherwise, I'm almost there - crank signal is strong, and I guessed dwell much better this time around, the other igniter doesn't even get warm. Better yet, I tested the fuel pump before trying to diagnose the second igniter, and that piece of junk ran for a good half a minute on two cylinders and residual fuel pressure, just to rub it in. Getting taunted by a motorcycle isn't fun at all

[SQLGUY]

The Microsquirt manual has schematics. They show the VB921's being driven nearly directly from the CPU's pins 7 and 8. The signal from CPU pin 7 runs through 100 Ohm resistor R39 for output 2, and 8 runs through 100 Ohm resistor R38 for output 1.

If you find where the working one is connected, the other pin should be for the non-working one, and you can then probably run some 30 gauge wire wrap wire to make up the damaged trace.

[Petrovich]

I'm getting there! The diagram is kinda conflicting with the label on the back... According to the label on Microsquirt, my ignition drivers have opposing wiring, such that Ign2 is wired to resistor R38 and Ign1 (the one I had trouble with) had R39 left, which is what I hooked it up to (there's a nice trace ending with a drilled hole coming down from R39, sitting right next to the injector drivers, that's easy to tap into). Now I'm getting signal on both, although with the current dwell (2.1ms, random guess) Ign1 misfires a lot more often than Ign2, at least according to the timing light. One problem I ran into was that Ign1 had its ground made of ash, so I looped its ground pin to Ign2. Is there a better way to ground ignition drivers? Ign2's ground is already rather fragile, I don't want it to burn up one day, as it also happens to hold the transistor in place.

[SQLGUY]

So what coils are you using?

[Petrovich]

The coilpack is still the same Subaru unit that I pulled off of a junkyard car, together with plug wires and connector pigtail. It came off of an early 90's Legacy, I think. The plugs are NGK's with stock (really small) gap.

It looks like people used ~2.5ms dwell on those coils, but I couldn't tell if those were the coils like mine, or newer COP style ones.