25 Oct

Finally drove the car to see how it deals with a straight “16-lb” wastegate spring and no ECU control. The figure’s in quotes because the manufacturer states that the rating assumes that intake and exhaust gas pressure are equal. If they aren’t, the actual pressure may be higher or lower. Mine has always been a bit higher (I think that’s a good thing), with it actually running 230 kpa, or 19 pounds; not a big deal but interesting. What is a big deal is that after many acceleration runs, there hasn’t been one MAP drop. This seems to show that the problem is either wastegate float due to using a weaker spring, or the ECU isn’t controlling boost properly (I suspect the former).

The next step is to reduce spring pressure to 14 psi (the maximum I’m willing to run on 91 octane gasoline), then use the ECU and boost control valve to adjust boost higher for ethanol. So – changing only one thing at a time – the spring will be changed but still have no ECU involvement. If there’s no MAP drop, then the boost control valve will be plumbed back in and we’ll go from there.

Separately, I’ve noticed that under hard acceleration, the air/fuel ratio gets a bit lower/richer than it should. Closed-loop lambda control is enabled, but even though the richest target mixture at the high end is 0.78 lambda, it’s dropping as low as 0.69. Anyhow, a question regarding lambda PID has be posted to the ECU forum. Somewhat related, injector duty cycle is maxing out at 89%, right on the edge for being undersized. Once the mixture is leaned out slightly, that should improve injector margin (and/or I can increase fuel pressure).

My brother has been pestering me about doing a trackday event – it has been a long time. The choices are Autoclub Speedway, about 90 minutes away. Further away are Willow Springs, Streets of Willow, Buttonwillow, and Laguna Seca, in that order. I’d like to get a rematch with Willow Springs, but almost as good is Streets of Willow, located next door, which is very twisty, favoring light cars which handle. There’s also a large skidpad there; I want to take Midlana out and do some traction-limit/drifting turns to get an idea of its threshold balance. It’s simply been too dangerous/foolhardy to do so on the street.

I’ve been to Buttonwillow once before, but for whatever reason don’t really like it – not sure why, maybe it’s just a lack of familiarity. Autoclub Speedway is the closest and the cheapest, but it’s also the highest speed course with scary concrete walls, though going off in the dirt at Willow can end as badly if the car hooks and rolls. Than of course there’s Laguna Seca, sort of the holy grail of tracks. I have to run Midlana there at least once, but it’s the furthest away, typically the most expensive, and has a really low noise limit. We’ll see.

18 Oct

Car-related activities took a backseat for a week due to a work-related out-of-town gas turbine class. What I was most impressed by weren’t the electronic controls, but rather the mechanical backup systems, which were the control system before electronics. Even though it’s all electronically controlled today, all the mechanical systems remain in place, standing by in case the electronics fail. I’m in awe of those systems, using springs, rocker arms, linkages, bellows, spool valves, magnets, and very few electrical parts. It’s easy to think that the systems are complicated, until realizing that they are in fact beautiful and elegant mechanical solutions. My hat is off – in a big way – to the mechanical engineers who designed the systems, many of whom likely used nothing but slide rules and a lot of brain power to come up with solutions.

The class answered a question I always had; there’s a scene in the movie “Cast Away” where Tom Hanks swims away from the floating wreckage of his cargo jet as one engine kept running on a broken wing. I thought that was ridiculous, but surprisingly, Hollywood actually got that right. Once lit, all a jet engine needs to run is fuel, and since it’s got an engine-driven mechanical fuel pump, an engine can indeed operate all on its own with no power at all, a tribute to its elegance and robustness.

Somewhat related, when I got out of class on Thursday, out to the west were some seriously huge and mean looking clouds. It turned out those storm clouds were dumping tons of rain on trapped motorists on the roads north of Los Angeles. We got a little rain but avoided the worst of it; I read that some of the side roads will require months to clear.

The only car-related activity I did was to swap in the 16-lb wastegate spring and switch to a purely mechanical-activated wastegate, but haven’t driven the car yet.

Yes, I will admit I gave a bit of thought about getting hold of a used gas turbine auxiliary power unit (APU). The trick is the transmission and how to throttle it. I have some ideas there but it’s just pie-in-the-sky. I’m fine with where I am for now.

10 Oct

The dash oil temperature now reads right, but MAP continues to read low (only) at idle, with it reading 15-20 kpa, while the ECU, which is reading the very same sensor, sees 22-24 kpa. That’s nice.

Anyhow, Midlana went back onto the dyno to see whether the cause of the MAP drop could be found. At first, the tuner felt that the clutch might be slipping, because if it slips even a little, MAP will immediately drop. He said it doesn’t take much slip to cause MAP to drop. Later, he got suspicious of the closed-loop boost control, so it was disabled and the boost run at a fixed duty cycle. Like magic, all MAP variations disappeared, indicating that the PID control loop needs more work. With that “solved”, boost was increased and then the MAP-drop started happening, so he finally saw what I was experiencing.

At the end of the day we had no smoking gun, but do have some things to try. First is to swap in a 16-lb spring and disable boost control completely, running a hose directly from the intake tract to the wastegate. It’ll be run that way for a month or so with logs always running, to see if the MAP-drop still happens or disappears, then go from there. Depending how that turns out, the next thing to try is going back to a 3-port controller. He felt that the combination of a weak (5-psi) spring, 4-port controller, and poorly-tuned PID may have been causing the problem (falling out of boost and being unable to recover).

The tuner told me about one customer’s car with a twin-scroll turbo setup that was impossible to tune because that boost was all over the place, very sporadic and unstable. He said that it wasn’t until the owner switched to a regular T4 turbine housing did the problem go away. When I told him that I’m running a twin-scroll setup as well, he just said, “Hmmm.” I’d love to try some different turbine housings, a 1.0 A/R housing and a non-twin scroll unit, but at $550 each, it’s just too expensive for something that would potentially be used only once.

4 Oct

As a previous post explained, oil temperature is fed into an analog input on the dash, and that the value recently started reading “51 deg C” when cold. The dash allows setting up sensors as either resistance-based or voltage-based. It turns out that the sensor was set as resistance-based, but because the ECU already has a pull-up resistor on the sensor, that’s a voltage, so it has to be voltage-based. How on Earth it worked fine for five months and only now started acting up is beyond me. I don’t touch the dash configuration unless necessary, but today the sensor type was switched to voltage and presto, it started working… again. Very odd.

The other issue which has been going on even longer is that dash MAP reads low. The ECU always sees the correct value yet the dash reads low (only) at idle. As posted earlier, I thought the dash used the MAP value in the CAN packet but I forgot that I had to give up on it because the dash scales MAP as a percent and I can’t change it – percent what!?). I spoke to AiM and they said to check that the dash has the latest code – it does. They also said to connect the sensor ground of the dash to the sensor ground of the sensor, so I connected the two. Got everything buttoned up and started the car, and it idled like crap. I let it warm up and it just kept running worse and worse, so an earlier tune that I knew worked fine was loaded – it ran like crap as well. Then, the light bulb went on, and I remembered having this same problem when I connected the flex sensor ground to sensor ground (imagine that) and it also ran like crap. Sure enough, disconnected the ground and the engine idled perfectly smooth. I tried grounding the dash input signal ground straight to the chassis and it didn’t make any difference to the displayed value. On the other hand, the displayed value was displaying fine, so not sure what’s going on there.

Drove the car some yesterday and if it wasn’t for the MAP-drop thing, it’s running great! Just have to get that behind me!

3 Oct

Well it’s back, the MAP-drop anomaly, which was noticed soon after updating the ECU firmware. I’ve done a lot of runs, logged more variables, but it’s still a mystery and worse than ever. Now it happens nearly every time at a lower rpm, around 5500-6000. The only good thing is that it lowers the speed I have to be going to invoke it (so instead going to jail for a year, it’ll just be a ticket…) Actually, I’m fed up with it so the car’s going back to the dyno next Friday. With it happening now so easily and repeatedly, it should be easy to replicate. The problem will be, once it is replicated… “now what?” I’m just too close to the problem and need a second (and third) set of eyes on the problem. It was fun as a science experiment, until it wasn’t, and now it’s annoying.

I know what you’re thinking, “why not just go back to the previous version of firmware?” Well I could, but since no one else has this same problem, it seems unlikely that it’s root cause. I suspect there’s something specific with my engine that’s causing it. Also, I rather keep the ECU upgrade because they made a lot of improvements; I don’t want to give them up because some are fairly important. Also, because the ECU is fairly new, there’ll be a lot more updates, and I don’t want to be left with an old version because of this. May as well hit this thing face-on and figure it out now.