Decided to make this a separate thread so things are a little organized.

Well, overall I loaded some maps/firmware from Arga and it ran so well in the garage I had to see it under load. So I bungee cord in the box and off I go!

First off, I want to say how shocked I was, it worked very very well. I was concerned since my AFR's were really lean everywhere, but there was no stumble, everywhere it felt like it just MIGHT be getting lean, it got a full point richer before my brain could sound any alarms.

The real issue, something I might need help with even getting started - at higher revs, maybe 5k, the motor gets super stumbly bad wrong. Like, I might not be able to "power through it", but I don't want to.

I'm running open waste gate, and staying out of boost thanks to having the MS's boost cut working. That's definitely not what I'm hitting, I'm in decent vacuum, the map's pretty flat out there, and all the other cuts have a different feel to them.

When I was using them stim, I remember seeing the ignition lights slowly getting brighter, till a narrow band of RPM's where they would both flash on and off together, totally different from before.

I'm hoping to get that out of the logs? The trouble is it's like a 30 minute log, is there some way to cut out sections?

Let me know what to post, and I'll do it. Thanks everyone!
-Abe.

 

Ok, here's a post - it looks like the MAP is falling for some reason? I don't know if it's a cause or an effect, but this is a slow free-rev in my garage.



My maps are pretty flat - I'm not sure how to save them? Some one tell me and I'll post whatever you ask for. :-)

The issue is the flat spot at ~4300 rpm, the motor just plain stops, and only heavy throttle can make it pass there. On the road I don't know if I even CAN do it.

I've heard of there being issues with where you take your boost signal on the manifold, that you can get funny readings at certain resonances.... Maybe that's what I should look at, but it could be something else entirely, something in software. If it was losing sync, how would I know?


<edit> Zoomed Version of Data:
http://abefm.smugmug.com/photos/236275309-O.png

 

After much log staring, all I can think is that it's really rich.

Log from last night's street run, see the RPM get jumpy? It's more obvious there

http://abefm.smugmug.com/photos/236290874-O.png

MAP goes down, pulse width and RPM up

 

zip and upload the entire map.

 

When you say "map", you mean the MSQ file?

Here it is. The MSQ and this morning's data log

I put three markers in it.
1) before I decided to run through a few gears, hitting this mysterious cut each time, I think the second cut was a overboost cut.
2) At the end of that little set of pulls
3) a few moments later, one more and I tried to stay in it.


I've got another thought, perhaps it's the VICS actuator, moving a flapper in the manifold. I know I had it set to switch around 4-5k. Two things, 1) I think I put a few hundred RPM hysterises on it. 2) I could swear it does it at lower RPM freespining.

It definately seems like an ignition thing, and like I may have mentioned, on the STIM, running other types of signals (the stim won't do 99+ wheels), the ignition does weird weird things for a narrow band in higher RPMs. There's no inductors in my circuit, so I can't imagine any resonances there, but... Well, look at it, tell me what you all thing, I'm lost.
-Abe.

 

yeah. that works!

 

I took the MS out of the car, set it to CAS 4/1, and get the SAME ISSUE between 4350 and 4390 RPM. The two LEDS blink on and off, slowly (5 hz?).

This tells me it's unlikely to be noise from anything in the car, as it's in another room.

There might be cross talk from very close circuits on the proto area. It kinda "fades in" coming and going at the ends of each spectrum. It's no multiple of any rev limit or anything.

I'll keep digging, but right now I'm very lost!!

 

Ok, more info. I'm still working on a table of all this, but I've noticed something, all else being held constant, the effect depends on a combination of CLT and RPM

i.e. if I raise temp, the RPM's it happens at goes up. And, it's not like just warm up or something, 100* will be, for instance, 2265 rpm, and 200* will be like 2270. If I go back and forth to any of the many temps on my table, the RPM where I get the coils cutting out is repeatable.

It certainly seems like software to me now. My best guess is something in the interupt handler is taking a long time to clear.

Is all the relevant version information in the MSQ I posted earlier?

There's two cuts I'm noticing, so far. One is only a few RPM wide, both cylenders cut. Then, at high revs (6k?) one or the other cylender will drop entirely. Which one seems random.

 

Oh, god... This is horrible. I am pretty sure it's a software issue, and being an MS-II you'll have nothing to say about it, but you can read my thread and stare at my pretty pictures and be happy it's not you.

http://www.msextra.com/viewtopic.php?p=176434

The last two posts have data logs and links to here:

<scratch that, you probably can't get stuff from there anyway>. I'll repost:

Ok, here's the grail for this problem: I'm pretty convinced this is all in software. I looked at the signals coming into the ECU daughter board, and they look fine, even as I cross the borders where things get weird.

A few of these pics are obviously spark cut, cutting 3 out of every 7.

But sometimes, the voltages are just going high or low and staying there for hundreds of milliseconds, even seconds. It seems pretty totally random.

It's a big mess. I think I'm running a v.15 that a friend gave me, perhaps I should go back to v.16.

Anyway, this was all taken with the stim. I'm going to put pictures of the scope traces online. There's two files, each mark in the files correlate to a picture (they should).

I don't know what "audit logs" are good for, but I'm attaching that too.
All the scope pics are here:
http://abefm.smugmug.com/gallery/4067828
And each one (will have) has comments on them.

It's very obvious the outputs are doing BAD WRONG things - what I don't know is how to get the attention of someone on this board who knows something this technical. It's not crossed wires, a wrench in the block, oil in the coolant or something else. This is relatively scientifically traced out over several careful experiments - so I would appreciate comments in a similar vein. I can certainly make dumb mistakes, but what I would most like is someone to actually suggest something worth checking or give me some ideas.

Thanks. I don't mean to cop an attitude, I'm just getting a little frustrated here. It seemed obviously software and now I'm even more sure of it. I'd like someone who knows the hardware and the code to lend me a hand or ask me a question.

Thanks!!
-Abe.

Example, pulling sets of sparks off each channel:

 

Wish I could help with some of this. I really hope you dual posting this on MSEFI or an equivalent forum where people actualy understand MSII and have insight.

 

Trying it. So far, I got several posts of people argueing how I'd be overboosting if I unhooked the wastegate arm, because boost pressure will blow it closed.

Another post was just that boost would fluctuate violently.

Telling them the logs I posted never show pressure toping 80 kpa had little effect.

It's just a slow board.... 2 posts a day isa busy day. Thanks though.

 

That's lame. Where's Arga when you need him?

 

Damn, I'm trying. We've been PM'ing. Usually, though, by the time I read his posts or e-mails he's solved it. The guys brain runs 4x the speed of mine.

I'm not sure what's causing this 5k problem. We're running the same code and I don't see this. I'm not running idle, boost control, fan control, etc but none of that could be coming into play.

It's midnight now. I'll look at the logs tomorrow.

 

Heh - no, you've been great. A lot of the simple stuff I'm able to see that I probably didn't need to ask for help in the first place. I sure wish this was something dumb I overlooked like facing my AIT sensor out instead of in.... :-)

The only thing that's really come to me, I've heard there's issues with different versions of msq's and firmware versions. My MSQ I think I brought over when I flashed - I was running a newer version of the firmware, and when I went to Arga's code, I had to flash the ecu. I *think*, but don't really remember, that I brought my old msq over to try to keep all the ignition and temp settings, etc.

Testing details:

The JimStim is set up for CAS 4/1 [01000000](sometimes I run it in 4g63 [01100000] mode).

My inputs look something like this:


My data collection looks something like this:


The stuff I got last night, showing:
Channel 1: Spark
Channel 2: Spark
Channel 3: Crank pulse at daughtboard
Channel 4: Crank Pulse @ JimStim

A representative picture from last night:


You can see the inputs never miss a beat, but the outputs miss pulses, exchange widths, etc.

Attached is latest MSQ and Audit files. Perhaps it's my MSQ from the newer version of the code not playing nice with the older version of the firmware? I hope that's something recorded in the files.

In the mean time, since the MSQ is saved, can't I load on the newer firmware and try it out, and then go back if it doesn't fix anything?

 

Instead of the 10k resistor on the pull up, should that be a cap?

 

IT'S ALL JOE'S FAULT! HE MADE ME DO IT!!!

'Course, I agree with him. :-) Caps can't really pull up anything, perhaps an open line like that, it might help, but the idea is that the OEM sensor can sync a lot of current (comparitavely) but doens't source any. So a large resistor will bring the voltage of the input to the transistor up when the sensor isn't "engaged", but when it is, you have a path to ground and the...

and the.....

Hey, Joe! What's up with that? 10k and 10k means the input to the transistors are seeing 2.5V as "low"??

Maybe the 10k coming in should be switched with something a lot lower! Like, 10k pull-ups, and 500 ohm current limiting?

It sure doesn't seem like it would hurt to let "low" be low, no? And, where does the current going down that line come from, cause if it doesn't source from the transistor, then the "upper" 10k resistor to Vcc should be limiting it.

Er... Anyone?

 

Did we have 10k and 10k written on my wall? That seems really wrong to me, but of course I can't check it at the moment. Actually, I don't think we had any resistor at all between base and sensor. 10k @ 5v = only 0.5ma in the first place.

The current going down that line (to the sensor) is indeed sourced through the 10k pullup from Vcc to base. Thus, no resistor is needed (or desired) on the base to sensor line. No cap either, that'd make things worse.

 

That's about what I was thinking. I think I'll short out those input resistors, see if things get better. Things have been working well on the 'scope, but the tale of the car is a different one.

Well, yeah - actually, the inputs to the CPU look perfect well into where the output gets crappy. But, why tempt fate?

Guess that's more soldering for me. Too bad, I was going to buy soda and deoderant today.

 

Abe's Rule of Thumb #49:

If you need an air tools to get the connector to fit back on, there's a chance it got "too hot" somewhere along the line.

Yep, the coil melted onto the connector, and filled with bubbly plastic. A couple snapped razorblades and nearly-bent screwdrivers, and I decided to break out the big guns.

 

Well at least you'll have some adapters for your cop harness. Even when not melted the harness connection at the coil is a PITA. I started lubing mine a little just to make the connection come off easier.