Its not so slow anymore Options

[wortdog]

I have been chasing down a nasty misfire for the last couple of months(since the weather has cooled off and the air density has gone way up, basically). I changed plugs, wires, fuel pump, hotwired the pump, MAFs, ECUs, power transistor, and the coil pack. Seeing I was pretty high on the duty cycle with the injectors, I swapped to the ones I just bought for the Buick, 750cc. Don't buy GM style injectors for a DSM unless you're willing to hack them in a bit, they definately don't fit 'properly'.

I've recently acquired the means to burn DSM chips, so I burned a chip for a stock, unhacked MAF, 750cc injectors, 8k redline, and a patch that halves the Hz count from the MAF so that values up to 3200 Hz can be logged. This didn't fix the problem at all, although it did get my duty cycles down to 57% at 5k.

The MAF patch came in handy however, as it allowed me to note a disturbing trend in the data. Before 5k rpm, the airflow signal climbed a bit as the RPM went up, which is expected. After 5k, where it peaked at 1940 Hz or so, it started dropping dramatically. By 5200, it was knocking a bit, by 5500 it was completely cutting out. It was down to around 1300Hz right before it droppped dramatically as the car began cutting out. As I had already swapped MAF's before, and I was pretty close to the 2000 Hz point where the 1G MAF starts having issues with over-running, I decided to try a 2G MAF.

I burned a new chip to match the new MAF, along with some duct tape, some sacrificial 1G wiring connectors, and a borrowed 2G MAF, and it was back to datalogging. This time, the data looked a lot better, with airflow tracking RPM for the most part, nice stable 0.89 O2s, and only 8-10 counts of knock at the top of 4th gear. Still had a slight pop to it.

Then I remembered that I had purposely swapped in janky plugs to see if I could change the problem at all. A quick swap back to NGK-R plugs, and the pop went away, along with some of the knock Duty cycle is up to 71% around 6800, which seems a bit high for just a port-matched EVO III 16G. Boost is creeping up to 18psi from a set 15psi on the boost controller, so I don't think I have a major boost leak, but I can't rule that out. I did notice on the way back into town that the outside temperature was only 17 degrees, so that may explain the large fuel thirst.

Big thanks to KOU Motorsports for letting me borrow a sensor to test. I imagine I'll have to give it back fairly soon, and I'll need the injectors for the Buick anyway. My current thought is to get a set of 660s or 720s and run a GM MAF straight to the ECU with no translator. There is a group of people working on this, and we're probably 25% away from doing it. There are also a couple of people working on a speed-density style re-write of the ECU code, so that a 3-bar MAP sensor would replace the MAF entirely ala HKS VPC. Some interesting things in store for the DSM EPROM ECU, assuming they quit exploding left and right. I noticed some cap leakage on mine for the first time tonight

[natedogg]

That's some cool stuff, Eric. It would be great for someone who doesn't want to modify or tune their car beyond what it was when the chip was burned for him. But the moment the car is further modified or the owner decides he want to do some tuning of his own, it presents a problem.

That is the main attraction to the MAFT if you ask me. The fact that you can tune with it, not necessarily the fact that you can run a GM MAF vs. a DSM MAF, although that is certainly a plus as well.

If you could develop a way to tune or flash the chip on the fly by simply hooking a laptop up to the diagnostic port, now that would be incredible. Many of the capabilities of a stand-alone without all the cost or complexity. Something similar to what the 2G guys have in DSMlink. I understand its alot easier to do with 2G OBDII common protocol, but it shouldn't be that hard to crack 1G OBD protocol even if it is Mitsu proprietary. Maybe someone on that website has done it already. I don't know. The next step would be to use Eprom chips that can be written to on the fly, or even just session flashed.

If I had more time I'd look into it myself, but I'm busy enough writing, deciphering, and flashing Cat's proprietary protocol.