Goin' Megasquirt (My progress thread)

[Lex]

How's the tune? Are you getting the miss at 3000 rpm?

[photoeye0]

I haven't had the car on a dyno yet, but I think the tune is good (for a street tune). My fuel map still needs to be fine tuned, I idle around 11.5 to 14.0, it's around 11 under boost, but my part throttle needs the little work it sits around 11 but I have my freeway 3000-4000 rpm's worked out to a nice 14.5 to 15.8. The miss happens around 2000-4000 rpm's and it's totally random. I can start the car up drive it to work and not a damn thing, but some days it's so bad I'll just bring it home and drive the wagon. It feels vary similar to someone doing a data burn as your driving. And I did notice that it's not as bad with brand new plugs (less than 24 hours of run time on them), but it's still there. And another thing I see when it happens my tach jumps up about 500 rpms which leads me to believe that it's noise from the CAS causing a "false trigger".

I may be able to shed some light on this-- Running different firmware (MS1E 029y4) and the 4G63/Miata 4+2 wheel I was having the exact same issue. Most prevalent at light level cruise conditions between 2k-4k rpm, maybe under the slightest bit of throttle. I put the car on the dyno and started scoping signals under these exact conditions.

First thing I noticed was the 12v signal into the ECU got really noisy when revving the car or putting any load on it-- noisy alt probably. Fixed that with a 3300uf cap between 12v and GND. That helped a bit but was not the real root cause. The size of that cap is probably overkill, but it did smooth out the 12v input better than a 1000uf cap did so we went with the higher value part and left it there. Don't stop here though.

The real resolution for my issue at least, was noise on the 2nd trigger input signal coming from the CAS. You're need a really high resolution scope to capture the noise, the noise blips I captures were roughly 4us (4 millions of a second) wide, but it was enough to present an additional, or in some cases simply a mis-timed rising/falling edge to the CPU which triggered it either at the wrong time or an additional time causing a misfire. I could see this clear as day in the scope traces once I cranked the sample rate up high enough to get a look at it.

Long story short we needed to filter the noise out without filtering the signal out or negatively effecting it. Math happened and we started trying a few capacitor values and landed at .1uf between the processor input and GND. A much smaller value seemed to solve the issue, but the .1uf did as well and provided some leeway in case the noise ever worsened, and it had no negative effect on the signal at any RPM this engine will ever see.

Now for some pics...




Blue = CMP/2nd Trigger signal
Red = ECU Ignition signal out to Ignitor
**Both taken at the processor**
Sample Rate 1.6kS/s (slow)

Here you can see the symptom-- at this slow sample rate the 2nd Trigger signal looks noise free and fine, but the ignition signal just drops out for a single event and then hums along just perfectly after that, until the next occurrence that is.







Blue = CMP/2nd Trigger signal
Red = Debug signal indicating 2nd trigger has just triggered
**Both taken at the processor**
Sample Rate 1MS/s (much faster)

This image shows three normal trigger events with a 'bonus event' in the middle that shouldn't be there. Due to the high sample rate the trace looks a bit noisier overall and some of this is erroneous (and due to a not so great chassis ground probably), but the big negative spike on that middle trigger event just before it's falling edge is real and is the cause of the extra trigger event that can be seen on the Red trace.




Blue = CMP/2nd Trigger signal
Red = Debug signal indicating 2nd trigger has just triggered
**Both taken at the processor**
Sample Rate 1MS/s (much faster)

This is the same trace taken just above, zoomed in much closer and with the noise measured at 4.15us.


Note that sometimes the noise would be on the high side of the signal in the form of a negative spike, and sometimes it would be on the low side of the signal in the form of a positive spike. Either way lead to a poorly time extra trigger event that would leave the coil with inadequate charge time for the next normal event can cause a misfire. Note also that there were more of these events happening that what could be felt-- I could sometimes see them in the scope trace without it being noticeable on the car just due to the timing of the event not having as much effect.



And the last image... the simple little filter circuit we used to fix the problem. Normally you'd need a cap and a resistor for a low-pass filter like this, but the resistor was already in circuit in our application, we just needed the cap in the right spot... We initially tried the cap right at the input of the ECU (more or less at the connector where the 2nd trigger came in) and it didn't help. It needs to be between the resistor and the CPU, one leg there and the other to ground. Here's a pic of what didn't work on the left, and what did on the right:





BIG thanks to James, Bruce and our own Matt Cramer for helping us nail this, and I'll glad to share in hopes it will help others-- MD95 it sounds like it may be exactly what you're seeing from what I can tell.

Hope this helps!

[Lex]

I have had misses because the electrical lines were not clean. This happenned when I was running an inverter off the cig lighter for the laptop. Can you capture the miss in a log?

[photoeye0]

Well I have an inverter plugged in to run my lap top, cuz the battery is shot. I got the log viewer working but it did show up in my 1st log, but I'm going to try again. I'll also try unplugging the inverter see if that helps. I would have never thought that the inverter could be an issue.

[Lex]

If it's a power related issue you should see all the signals in the MS go crazy at the same time. There will be an obvious spike or dip in all signals.

[photoeye0]

I didn't see anything like that in my 1st log, and I haven't had a chance to look over my second log. But even with the inverter unplugged, heater and radio off it still had the miss.

And now there's more lovely news...I might have smoked my turbo. But I'm hoping I just blew out a gasket. Everything sound fine but over 5 psi or so I get this high pitch buzz/whistle and I might give another pound out of it. I'm going to try to pull the turbo tomorrow (if I can get shop time) and see how it all looks.

[FE3-323]

which turbo??

[Lex]

You probably lost a bolt out of the exhaust housing. There are 4 of them and often get loose on the VJ20. Check those out first.

[photoeye0]

I'm going to put th turbo out today and check all the bolt on it and the mani and hope.

[FE3-323]

double nut the studs from the manifold with SS m8x1.25 found @ ur local hardware store & never worry again

[photoeye0]

Well it wasn't a leaky turbo or mani. I puled out the turbo made sure everything was tight, put a new mani gasket, and gave all the gasket a coat of ultra copper. I also gave the turbo a good look over, no cracks, no odd colors, no oil, and a vary small bit of shaft play. The pissed off cat noise went away, but I'm still only making 5 psi. I'm going to try unbolting the DP from the rest of the exhaust and taking it for a spin, maybe I nuked the cat.

And I didn't make it to radioshack, so I'll have to try the cap later. I'm more concerned with the turbo BS right now.

[FE3-323]

it may sound sketchy, but after you've taken care of the cat problem. take the line off of your wategate & see if it passes 5psi..... that could be your issue there.

[photoeye0]

Well pullin the cat fixed it!!!
Now I have to try out that cap, but it's going to be -40 wind chill tomorrow...so I might not make it out.

[smoke187]

i feel your pain needs to get warm out and stop snowing

[FE3-323]

for snow & cold