Da Groj

Loren wrote:Smell for burndedness

I will do that.

Doug - I hear you loud and clear. If Megasquirt didn't have the solid reputation that it does, I'd be rethinking this project... But - it's a hobby - this is "fun." Truly, I am learning.

Pulled the MS out. Opened it up. Not smelly. Visually - nothing burned, leaking, broken, wet, dirty. All pretty clean and nothing obviously wrong.

Hmph.

The fact that nothing inside the MS appears to be fried is a good thing. That means it's either a less visible internal failure of the MS (not likely, but not impossible), or it's a wiring problem.

You've checked wiring from the injectors to the end of injector harness. So, next would be to check from there to the ECU. Sometimes extra hands are required for that kind of thing.

Loren wrote:Sometimes extra hands are required for that kind of thing.

That's what I was thinking.

Upcoming tech day at Steve's...

So the tech day was this evening. Loren came by and applied his electronics knowledge.
We re-traced the wiring continuity from the injectors all the way back to the ECU. Everything checked out. So we opened the megasquirt box and Loren found some corrosion that we cleaned. Then tested the continuity of the injector circuits in the ECU. It all checked out. So, we're hoping maybe the corrosion was shorting traces on the board and we put it back together and started the car. No dice, same dead cylinders.

Then Loren had the idea of putting the stock ECU back in the car. We did. It ran on all 4 cylinders. Took it for a drive.

So, my Megasquirt lasted about 2 years.
It's out of warranty. DIYAutotune will repair it. $75/hr. Cheaper than a new one... at least up to 11 hours...

I just PM'd you this thought, but... be sure it's not something in your tune. Step back to a known good tune file. You could have flipped a setting accidentally, or have a corrupt file.

Also... my apologies for not suggesting swapping the ECU sooner. I didn't realize that you had a TRUE plug-and-play installation. I usually do things the hard way, and switching back to the stock ECU is never quite that simple. In your case, it took 3 minutes! We could have done that a lot earlier in the troubleshooting process.

Damn, 2 years out of an ECU...the ECU was still communicating with your laptop and getting sensor output readings despite not cranking?

Loren wrote:I just PM'd you this thought, but... be sure it's not something in your tune. Step back to a known good tune file. You could have flipped a setting accidentally, or have a corrupt file.

Also... my apologies for not suggesting swapping the ECU sooner. I didn't realize that you had a TRUE plug-and-play installation. I usually do things the hard way, and switching back to the stock ECU is never quite that simple. In your case, it took 3 minutes! We could have done that a lot earlier in the troubleshooting process.

Installing an old tune was one of the first things I tried.
And no worries about the ECU suggestion - I should'a thought of that - I know what I did to install it (mostly) and I was so focused on the car itself it never occurred to me.

aw614 wrote:Damn, 2 years out of an ECU...the ECU was still communicating with your laptop and getting sensor output readings despite not cranking?

Oh, no, it cranked. And the engine ran. On 2 cylinders. One of the things Loren diagnosed was that the injector circuits in the ECU are set up to run cylinders 1 and 3, and cylinders 2 and 4. For whatever reason, the ECU was not sending signal to cylinders 2 and 4. And that's what DIYAutotune is gonna get paid to fix.

So it pains me to have to post this, but after some email exchanges with a guy named Matt at DIYAutotune, it turns out I managed to flip a setting in my software. Here're the simple directions I was given after Matt reviewed my tune file:

"Please go to Basic / Load Settings -> Engine and Sequential Settings, and change the "Injector drivers" setting from "Standard drivers" to "Additional drivers". Let me know if that does not fix the problem."

So it seems that when set to "standard drivers" only 2 banks of injectors fire. "Additional drivers" is the correct setting, and, lo and behold, all 4 injectors work now. I have no idea how I changed that setting, and when I loaded an old tune way back when this first started, I obviously did it wrong (panic will do that to ya), as the old tune is correctly set and should have worked.

I'm going to go crawl under a rock now.

I think that's awesome. Sucks in a way... And yet... It's good that the MS is actually reliable like it's supposed to be.

Told you, those guys know their shit!

Native wrote:I'm going to go crawl under a rock now.

Don't feel bad I spent an hour spraying weeds with killer only to learn a couple days later I forgot to put the concentrate in 2 months ago. So basically I watered only the weeds. They are dead now tho' probably a surprise to them? Hopefully your car stays alive now that you stopped flipping it off and using it for a table.

Native wrote:So it pains me to have to post this, but after some email exchanges with a guy named Matt at DIYAutotune, it turns out I managed to flip a setting in my software.

You probably emailed Matt Crammer, lead engineer and author of Performance Fuel Injection Systems. Like asking Moses.

^^^
You are correct, sir - that's the "Matt." Quite pleasant in his emails as well. Good to know they've got quality, there.

So in continuing to chase down that throttle bog I've been bitching about, I've found that intermittently, but repeatably, the needed tpsdot measurement that triggers an acceleration enrichment is "missing." A datalog shows the throttle input, but no rate of change. Zero. I've googled and found nothing. So I shot off an email to my new best friend Matt at DIYAutotune. Perhaps he can help.

Also, baby's got a new battery. It's technically a motorcycle battery. Same 200 CCA as the Odyssey PC625 I had, and it weighs 9.5 pounds, which is about 5 pounds less than the PC625.

Matt answered:
"It's the TPSdot uses # WOT setting? You have the ECU considering the throttle to be "fully open" at 11% even at high RPM. Movement beyond the "WOT line" is considered not to affect the airflow and ignored for TPSdot calculations."

I get it. I guess I'll make some adjustments next time I get a chance.

Makes sense.

So I've been continuing to mess with the enrichments, and the %WOT settings, and I've gotten pretty close to proper throttle response. So close, in fact, that I brought the car up to FAST Monday this week and shared the keys with a few folks. I know exactly where to find the "imperfections" and needed to confirm that I wasn't just driving around them. For the most part, the tune got good reviews. Philip found a hiccup around 3000 rpm, which is certainly possible, and Chris had an interesting theory about motor mounts that I'll need to look into.

In the meantime, I'm going to keep working on it. It's pretty good, but I want it better. The worst of it now is at really low rpm, like 11-1200 ish, say, idling through a parking lot, with a big enough throttle stab it'll still bog (and rock the drivetrain, which is what Chris was thinking about - maybe I've torn a motor mount). And the higher rpm ranges after a full throttle release then hard back on it, as we sometimes do at autocross, it'll still bog.

I also noticed that, say approaching a traffic light. Clutch in and off throttle to do a lazy rev-match downshift. As the revs are dropping, there's a range where a throttle blip does nothing - the enrichment is too late. I have that datalogged (haven't looked at it yet) from my drive last night so I hope to fix that, too.

I'm actually catching myself thinking about the next step in the project, but it's been so long I've been messing with this, I'm not even sure what that is. I think I'll need to install the intake air temp sensor and get things working with that, but I have to go back to my notes and check. Heh.

{bad French accent] Five months later... [/bad French accent}

I've been messing with the tune, and got it to the point that the only bog was when you were fully off-throttle, then back on it - say in an autocross, and not a left-foot braker, you get off the gas, on the brake for a turn, hit the apex then back on the throttle - that's when I get a full-lean bog. Can't get it to go away.

So, today, I started a complete new tune. Adjusted my AFR table, adjusted my timing table (a tad less aggressive), and re-autotuned the fuel table. The good news is it took me about 4 hours to get darn close to where I was with the prior tune, which took uncounted hours. The bad news is I still have the off-on throttle lean-out bog. Shit.

I've so been avoiding the "enhanced acceleration enrichment" programming. I really don't wanna mess with "sucked from walls" and "adhered to walls" and so forth, but it looks like I may have to. Or hand it over to someone with a dyno and pony-up ( $$ ) . Nah.

And, I've got to check the motor mounts like Chris told me to. Not sure how that could cause a lean condition on a data log, but who am I to question?

Native wrote:And, I've got to check the motor mounts like Chris told me to. Not sure how that could cause a lean condition on a data log, but who am I to question?

He is trying to correct "the lean" condition.

So here's an interesting tidbit:
I've known that the AFR on the gauge is about .3 richer than what the megasquirt logs show. So, I've put priority on the gauge, and just have tuned by adding .3 to any AFR settings I make (if I want 14.7, I program for 15). Well, I got curious about making them match, and did a little reading. What i came across was that the gauge itself may not be properly displaying signal from the wideband controller, which is really "the authority" one wants to tune by. To check, hook up the controller to the ol' laptop and monitor it directly.

What I found is that my gauge shows .1 -.2 richer than what the controller says. And that means that my AFR settings in TunerStudio are "off" by the same .1 - .2. So if I decide to fiddle with the wideband voltage settings in megasquirt, all I'm looking to do is make it read .1 - .2 richer. I'll probably just drop all my AFR targets by .1 and call it good...we'll see.