Hey,

I've been trying to correct a problem with a Megasquirt install for the past couple of days. I suspect that I know what the problem is but would like to confirm.

The relevant setup is as follows:

-13B, street port, big turbo
-550CC primaries, 1000 CC secondaries
-MS I, V 2.2 board, firmware 029t
-Ignition run by dizzy, locked around 10 degrees

Car runs like a champ except for a major stutter around 5200 RPM that occurs under boost. Slightly noticeable at 6 PSI, major problem at 10 PSI.

After examining a datalog, it appears there is a lean spot just before the stutter. I have attached an image of an example to this post.

Looking at the settings in Megatune, the injector stage scaling factor seems to be the cause. It's set to 255, with a stage on point of 4000 RPM and a delta of 1.

As you can imagine, the VE table looks like two hills with a valley in between since the 1000CC secondaries are coming on full force. The bottom of this valley is (drum roll please) just before 5000 RPM.

So it looks like the injector stage percent needs to be brought down to a more sane level, and then the second half of the VE table scaled accordingly...

Thoughts?

 

You need to set that scale factor to suit the injectors in that application. The 255 factor assumes identical primary and secondary injectors. Also The scale factor is not a '%', but a number that is divided by 512, and then multiplied by the base pulse width.

This is documented at the msnsextra site:
http://megasquirt.sourceforge.net/extra/stagedinj.html

In your case, primary flow = 550, secondary = 1000. Using the equation above:

Scale factor = 550/(550+1000)*512 = 182.

There's no WB data in that log snapshot, but the engine should go way rich when the secondaries come on line at 4k unless you've pulled the VE values way back to compensate.

Some other thoughts: If that turbo provides high levels of boost below 4k, you may run out of primary injector before hitting the staging point which could be catastrophic. Try TPS-based staging, or at least use the second transition trigger set to 100kPa or so. Also, set that delta to a higher number to add some hysteresis to the system. That'll prevent staging on/off jittering if operating steady-state at the set point.

 

you can also use the 2 parameter staging in case you get to 4000 rpms off boost or something... Just set it to the "or" setting, and set the first parameter to RPM, and the second to MAP... then you can have the rpm parameter stay at 4000, and the MAP parameter at 100-110 kPa.

 

Good to know my suspicion is correct. I was hoping I was wrong because it means the car needs to be retuned. Hence the post to confirm...

One thing I have to say though is that to be honest, I've always been totally confused about how the injector scaling works. The documentation is a little vague and worded strangely. Every other standalone referrs to injector staging as a percent of the primaries, or percent of total flow. The Megasquirt uses what is basically an arbitrary number...well, not arbitrary at all, but it sure looks like it after a long day of tuning...

Hence the VE table for this install has two hills with a valley in the middle. The valley is at 4K. The car was originally tuned with 4 x 550CC, and I guess the scaling was never changed when the secondaries were upgraded.

Yep, that's likely how I'll set it next time I have access to the car.

 

The docs on the original msnsextra site I linked to are sparse, but contain the fundamentals needed to configure the feature properly. Unfortunately, in the new msnsextra manual, the descriptions for the staging factor have been paraphrased incorrectly. I'll send them a note to update.

As I'm sure you've seen, documentation for the base MS is second to none. The extra code and other variants, on the other hand, are a moving target from a documentation standpoint. Most programmers would rather hack in a new feature than document the last one. No complaints here, though, as the results are amazing for those willing to spend the time to sort out the details. Such is the nature of open source...

In any case, to clear up your confusion:

Staged Pulse Width = Original Pulse Width * Scale Factor / 512

With a scale factor of 255, staged pulse width is ~1/2 the original pulse width. Note the actual calcs subtract opening time prior to scaling, and then add it back in after.

The seemingly strange scale factor number (0-255) is due to data type and binary math issues. I suspect it could be massaged to display as a '%' value in the msnsextra.ini file for someone with that inclination, although with clearer documentation I doubt it's needed.

I'd recommend spending some time going through each and every setting to verify proper configuration prior to tuning any further. There may be other even more critical parameters incorrectly set. Pay particular note to the rev limiter and overboost settings.

Is this project affiliated with the CP folks?

 

I've been trying to figure out how to contribute to the project but have been stumped since it seems everything has already been done. Now, I think I know what I can do: documentation! The documentation of the 'Squirt is great, but some of it reads like it was written with the assumption that the reader know the source code.

Those were the first things I checked.

 

Your help (and anyone else's help) doing documentation would be MUCH appreciated!

James and I are so busy coding for ms2/extra that we haven't had time to spend on documentation for that or for msns-extra. Phil R has done a pretty good job with the new msns-extra documentation, but if you can think of ways to make it easier to understand, by all means do it. Send it to me, James, or directly to Phil R (on msextra.com).

If you really want to write docs, when we get ms2/extra's interface kinks worked out (problems with megatune really) your help there would be much appreciated too.

Ken

 

During the winter I should have spare time to help out with the documentation. I'm pretty used to writing clear and understandable documentation so it shouldn't be too big of a chore.

 

I can do it when I have to, it's mainly an issue of me having time for docs or code but not both... and I'd much rather code

 

I can't stand coding. Well, that's not necessarily true. When I'm in the right mood I can spend many hours staring at a screen but any other time it makes me crazy. Never get into assembly though. Just when I would have started working seriously with microcontrollers, I was getting out of electronics. I think the last time I did anything in assembly was in grade 10 computer science.

 

haha, the last time I did anything with assembly was Wed.

It's not so much that I like coding, it's that I like that I can write code, and then see the results immediately... as in I have a car running on ms2 now that wouldn't run on ms2 previously.

It's the same thing for the staging and trailing code that I'll be starting on, or the x-tau code I'm working on now... the x-tau code will hopefully make the corolla with 20v 4age and ITBs more responsive, and make the AFRs more controlled, and help drivability. As an added benefit, it should be easier to tune than the current ms2 x-tau code.

I don't need to go into the benefits of the rotary trailing code and staging code, those are obivous.

But I like being able to immediately and physically see and feel the results of my code.

 

I think that's a big plus. If I had started some electronics projects with payouts similar to working on engines, then I probably would have stuck with the hobby. However now with the Megasquirts my interest is increasing...

Regarding the car, changing the staging settings to their proper values cleaned things right up. Still some tuning needed up top, but that car is a rocket. Made nearly 300HP at 6 PSI last time it was dyno'd.