30 Aug

Well that was exciting!

Took the car out to do some early morning runs on the freeway when traffic’s light. Did 12 runs without a problem and headed home, happy that the rev-limiter problem seems to be solved. On the way home I figured, what’s one more run, lucky #13… and at a rather high speed, had a rather spectacular “poof”…

Midlana suddenly filled with smoke and oil mist. At first, I thought that I’d blown the engine, because in the mirror I could see that I must look like an F1 car that just blew the engine. However, there weren’t any terrible noises and the engine was idling smoothly as I pulled off the freeway. There was oil everywhere: in the engine compartment, in the passenger compartment, on the inside of the windscreen, all over the rear view mirror, it’s going to take a while to clean up.

I get out and looked back down the freeway, happy to not see any familiar looking engine parts or big oil hemorrhage, just a thin line. Okay… start looking around the engine compartment… no holes in the block, all the oil lines looked intact, so where’d it come from? Since the dry sump has an 8-qt tank, so there was sufficient oil so the engine could be started to see where the leak was… and there it was. See the big brass-looking bolt on top of the remote oil filter block (fifth picture from left)? See how it’s backed off about 1/8”? That’s not good because there’s an O-ring under the head of the bolt. As purchased, the bolt is held in solely by friction; there’s nothing preventing it from slowly backing out, and indeed it had. I instantly remembered someone pointing at that block about 6 months ago, saying that it had failed this same way for him. I’d put it on the mental list of things to do but never got around to it. Well I am now! The bolt was drilled so that safety wire can positively retain it.

The last picture shows the $0.25 part that failed, the O-ring… well it was allowed to fail. As the bolt backed off, oil pressure stretching it until it popped out of the seat. On the side of the road I tried tightening the bolt, but the damage was done; the O-ring had been stretched so much that it wouldn’t go back into its groove, and all tightening the bolt did was kink the O-ring and not fix the leak. New O-rings (a replacement and a spare) are on the way. This failure could have been so much worse in about a hundred ways. However, listening to some of the 10th-anniversary stories of what Hurricane Katrina survivors went through, I have nothing to complain about.

With this issue diagnosed, and the rev-limiter puzzle sort-of solved (there are a few aspects of the “fix” that don’t seem to quite fit, but I’m tired of dealing with it), I’m going to switch back to the weaker (8 psi) wastegate spring and see if the problem reoccurs. I don’t expect it to since switching to the 15 psi spring didn’t solve the problem the first time. After that, it’s on to setting up the knock sensor and other wonderful features.

29 Aug

While it’s been quiet, things have been happening, and because it’s too hot and humid to be driving the car, you get an update.

To make a long story short, the premature rev-limiting may be solved. The problem is/was that under full-throttle acceleration, sometimes the car would reach redline and sometimes MAP would drop off about 300-500 short of redline. It turns out that the problem is intermittent, which really messes up testing if you don’t realize that! You do a run and see a problem, so you make a change and do another drive. The problem is gone, so you figure you found the trouble and that it’s fixed, only it isn’t. Change the wastegate, it’s fixed, no wait, it isn’t. Change the boost control valve, it’s fixed, no wait, it isn’t. Re-gap the plugs and swap in new ignition coils. Okay, cool, it’s finally fixed… no wait, it isn’t. The only way to deal with this is to make one change at a time and do multiple runs to determine if it’s really fixed or not.

After changing all the obvious things – and somewhat grasping at straws at this point – attention turned to checking less-obvious things such as the rev-limiter function. As an experiment, the rev-limit was increased from 8000 rpm to 8500 rpm. Changing the rev-limiter is the only thing that’s ever had any direct effect on the problem, and beforehand, has never worked right even twice in a row. As of today, three successful runs have been performed without a problem, the engine reaching 8000 rpm without issue. However, more testing is required before drawing any hard conclusions.

I asked the ECU manufacturer if there were any known issues with the rev-limit code, and the somewhat arrogant response was “No, there is not a rev limiter bug*.” That’s a pretty lofty statement; per that logic, anyone reporting a rev-limit bug doesn’t warrant attention because there is no bug – that’s called circular reasoning. To be fair though, maybe the problem is something else, so I can’t be too sure of my own claim until more runs are performed without failure.

* I got a similar reaction when reporting that their manual contained an error regarding how to connect the flex-fuel sensor. Their response: “No one has ever had a problem with this.” The long pause after they realized that their manual was wrong was very satisfying.

16 Aug

I finally figured out datalogging and boost control, so now it’s down to test drives, datalog reviews, sw changes, and more drives, only this isn’t the best time to be doing that – or just about any other outdoor activity. At around 90 degrees F, the airflow around a open-top car stops making it feel cooler and starts making it feel warmer; at a humid 100+ deg, I’m a slow-roasting turkey in a hot-air oven. Midlana handled it fine, which is a good data point for the cooling system, but it wasn’t any fun for me. It’s frustrating not being able to use my day off to further tuning, but it’s just too miserable to be out driving, especially since it’s not our usual “dry heat.”

The data logs are for fine tuning the ECU control loop for holding commanded boost. It turned out that the parameters I had were too small, restricting the ECU’s ability to maintain a boost setpoint. So that’s getting dialed in, in addition to how MAP dropping off above 7500 rpm, which means finding space to do full throttle pulls in third or forth without scaring people. Typically I run out of room short of that and it’s one reason why I didn’t realize I had the MAP problem in the first place. After finally getting the datalogger sorted, all I needed is one good run to see if it’s solved, which was late in the day, and the anomaly is still present.

Click on the thumbnail and welcome to my world. The more you study the plot the more information will become apparent. The vertical blue line mark where MAP drops off unexpectedly. I’ve looked at just about every engine parameter and haven’t figured out what’s happening first to cause it. The problem is that MAP is a component in just about every engine parameter table, so when MAP drops off, it’s hard to tell if it’s driving all the changes, or one of the changes is driving MAP and it’s just an innocent bystander. This may take a while to figure out. Given that the problem only occurs above 7500, it’s not an issue in 99% of normal driving. It’s when I get out on track that it becomes a bigger deal. If I can’t figure it out on my own I’ll either have to give up and just accept a 7500 rpm limit, or take in to a shop that really knows this ECU.

In a related matter, I got lectured by the ECU manufacturer about their unit being in my street car. Because aftermarket ECUs don’t have government emissions approval, they’re subject to government scrutiny about their product being used on public roads. No problem, I thought, Midlana is 100% road legal and smog-exempt thanks to the SB100 specially-constructed car exemption. Yes it is and no they don’t care; they don’t want to risk appearing to be supporting someone doing something illegal with their product (even if I’m not). So in an odd twist, I’m doing them a favor by not giving their ECU any publicity, so there won’t be any more mention of the manufacturer.

9 Aug

Did a bunch of test drives to configure the new single wastegate, and the MAP problem is still there! Boooo; yup, boost drops off exactly the same as before at the same 7500 rpm. The thing is, with the 4-port boost control valve, the ECU’s closed loop boost controller should adjust the duty cycle to compensate, as the 4-port controller allows it to gate MAP pressure to assist the spring, holding the wastegate shut. I haven’t found the logger variable that reports the duty cycle of the control valve… that’s the first step. The second step is figuring out why it’s (presumably) not changing in response to the MAP drop. I sent off some plots and the tune to the ECU manufacturer so we’ll see what they say.

2 Aug

A couple more shots of the new wastegate and heat shields. Also shown are the newly-added intercooler fans; adding them now while everything is apart saves time later. Last shown is the growing electrical circuits aft of the firewall.

It’s easy and fun installing the big parts, but another thing to really finish it 100%. That means fully welding everything, trimming hoses to the proper lengths, having the correct ends, torquing all hoses and fasteners, tying off all hoses and wires, and adding sleeving and tie-wraps as needed. While this modification has dragged out for a month, other than waiting for the proper fan connectors to arrive, it’s completely finished.

This is mentioned because a builder on the Locost forum just “completed” his car build, but it wasn’t done. What that means is part’s aren’t mounted in their final position, stiffening brackets aren’t yet added, wires are still hanging loose, and the ECU tune is terrible. With it in this condition, he drive 30 miles to get the car inspected, only to have it break down on the way. Separately, the engine destroy itself on the dyno once there, no doubt due to driving with a half-baked tune. A successful build can’t be rushed; things have to really be complete else it risks reliability and safety, potentially to the driver and those around him. My comment in his build thread that perhaps he should slow down went unanswered, but I think my mechanical engineering buddy said it best, “No one knows anything; the results speak for themselves.” Take your time and make it right.