I'll post on msextra when I have alpha code ready. I guess I can post here too.

Ken

 

While waiting for the updated code, I went ahead and added a second air solenoid wired to turn on with AC signal (again, the AC compressor is on a 1 second delay relay, effectively giving me a feedforward mechanism of sorts).

Now the extra air is more pronounced. From idle of 1050, when AC switch is turned on, the 2 solenoids raise the idle to ~1290, AC clutch engages. RPMs drop to ~900-1000 with just 1 big dip. Minimal overshoot, settles at ~1050-1100 idle. Fairly solid. And I haven't tuned the PID settings yet!



This picture shows the difference with the 2 extra air solenoids vs the single solenoid (ghost lines). Not the stability of the RPMs and the stark difference in idle valve duty!


The idle valve duty only responds to the initial big dip, then is pegged at minimum duty, so it's actually now fighting to bring the idle down, which is much better than the other way! I think my hysteresis settings should be reduced, that's probably another factor. The idle variation almost equals my hysteresis settings...

I think I can safely reduce the idle target to 900 now...turning on the headlights will still upset the balance, but I expect the improved code to take care of that, and perhaps even reduce the initial droop even further...

 

now you need to time the solenoid opening better, it's happening too early. You want to make the leak start as the a/c opens up so they cancel each other out.


the second ghosted log makes me wonder about your AFRs, the spark angle and idle valve are locked solid, but its hunting around like a bitch still...makes me think the fuel needs fine tuning.

 

That's a 1 second delay. Will try reducing it. You're running 500ms, right?

 

something around that

 

The good news is, I got the changes made in ms3 and ms2.

More good news is I tested them in ms3 and they work.

The bad news is I left my USB->Serial converter that I use with the MS2 at work all weekend, and I'm not going back to the office until tomorrow. So I should have the ms2 changes tested tomorrow evening.

Ken

 

Ken, email me the ms3 firmware and I'll play with it this afternoon.

 

I tested my own firmware with said changes on the stim, seems to work well. Will test on the car later.

 

I'll be emailing it to you and y8s later today when I'm done with work.

Ken

 

The AFRs are pretty much dialed in. This pic is a comparison of the 2. The only thing that stands out- intake air temps are higher in the one with some oscillations. Also ~ 1 degree more retard (temp based?)



Attached are the relevant logs for your perusal. And the MSQ. Lemme know if there's anything to tweak. Quick before Ken posts the updated MS3 code and you get all distracted on me

 

Fantastic! When I do get to try it, I'll disconnect the 2nd solenoid, so we can have an apple to apples comparison and see the difference the code changes have!

Thanks for the quick work on the fix!

 

Overlay AFR, PW, RPM, MAP.

I can't do it here.

 

Battery V is stable, no differences there.

You may be on to something! PW is fluctuating with the higher IAT...?

What setting connects the 2?

 

pull up the GVE and see what the VE% value it was fueling off. maybe the fuel map just needs work. check for any gammae enrichments at the time as well. Gair or gbaro or something.

 

Hmm. Small bumps in gVE corresponding to the oscillations. The stable idle had a flat GVE. But again, I didn't touch the VE tables between the 2. The only difference was IAT ~90 stable, ~120 oscillations...



Oh I get it! I have a step in my VE table:

Here's the stable one:


Here's the one with oscillations. Note the 4 cells used to calculate VE compard to the previous one using 2 cells.


So do I make the lower row richer, or the upper row leaner, or average them out? Any other areas in the VE table standing out? This is VEtable 3, which is used when AC is on (hardware trigger along with spartktable 3). EDIT: I think this shows I need to richen the lower rows.



Hmm..so my wanting finer resolution in the idle zones actually hurt me in this case! I should pay more attention to the transitions between idle cells, since there is less room for the software to interpolate the values between cells.

 

whats odd is it's showing it's fueling at 49-50% VE, but your maps in that area doesn't ask for that amount.


but i realize you're oscillations, now looking at the logs, aren't bad at all.

looks like we are talking +/-50RPM, min the and max RPM is really exaggerating the oscillation. Had that been on a normal log I wouldn't even had seen it.

 

I'm table switching both spark and VE, triggered by AC. Look at the VE table 3.

The reason I decided to switch VE tables was to get the correct fueling at idle for the spark tables. Previously, switching just the spark tables, I was a bit rich with the AC off and a bit lean with the AC on...

Oh definitely! Still, I think it could still be smoothed out some more. When logging the idle, I purposely don't rev it up, so these small oscillations come out

 

(zack morris's phone)

No way.

 

Greg: you have code waiting in your inbox. Turns out on ms2, only idle was messed up. I had EGO and boost right.

While I was there I boosted the D term a bit for both idle and EGO.

Ken

 

Thanks Ken! But I have a DIYPNP! Could you send me a microsquirt s19? Thanks again!