richyvrlimited

iTrader: (1)

Arrrrghh, sorry for the confusion everyone

Thanks for the correction Reverant

Greg G

LOL!

Arg. An hour of my life wasted trying to figure out why the car suddenly runs like crap. 2 plugs went bad. Replaced all 4, again purring like a kitten. Which makes me wonder-- does all this tuning lead to shorter plug life?

Braineack

iTrader: (62)

if you run crazy rich it can.


Dunno why you guys still want so much D, after the code change I reduced it significantly. IIRC I'm around 30 | 20 | 10

Any more than 10 and it significantly slows the ramp to target speed and takes forever to finally get to my target.

Greg G

With the code change, I think Ken included some amplification of the D response along with the sign correction. Unfortunately, the beta he sent me doesn't start the car for some reason. So I'm using the Mario v14, which has the sign correction but no amplification.

miatauser884

iTrader: (11)

So I was playing with idle today after the mario v14 upgrade. I'll seemed golden, then out of nowhere when the car normally settles to its rpm, it shoots to ~470 rpm and idles at 13.3 afr. It won't recover it just stays there on the verge of conking out, eventually it does die.

I noticed some anti stall settings in tunerstudio. Did you have to change anything substantial greg when retuning. It seems odd. I'm wondering if somehow my dashpot adder gets weird when I increase the rpms to flash. I don't know why the rpm would target so low.

Any ideas?

I ended on P20 I105 and D130, it seemed as though I could have gotten away with numbers as low as P20 I20 D10, but as I kept adding I, it required more D every time.

Greg G

All I had to do was adjust the WUE, then it was acting like a better version of v5! Zero stall events for me during a long city run, with lots of stop and go. I have not yet played with the new pid tolerance settings and the antistall code.

Is your issue AC related? Because without AC (and with a working vTPS), I have had almost zero issues since 3.0.3. All my idle issues are AC load related, aggravated by the large throttled volume of the supercharger. Make sure you have the VE tables nicely dialed in.

miatauser884

iTrader: (11)

It was idling perfect for half an hour, then all of a sudden it would shoot to ~500 rpm and stay locked there. Very weird. Before this happened it would stay at 750 rpm +/- 20rpm. Very nice. The craxy part is how long it would idle at under 500 rpm with 750cc injectors. I was amazed. It might just be that I'm trying to idle too lean and the car just doesn't like it. It wants to idle at 13.5. I want it to idle at 14.7. i'll play with it some more tomorrow. What do your PID values look like?

No AC on mine. When I turned the headlights on at 475 rpm, the car didn't die. Kept on idling.

Greg what do your idle cells look like? Do you take the approach of leaving a block of cells the same, or do you let VE analyze do its thing?

Greg G

I don't let VEA touch my idle cells. The spark table is flat as a pancake in the idle cells, 10 degrees. The VE table is almost as flat. At least that's what I found worked for me. I wanted gVE to be steady as a rock during idle.

I just switch both tables when AC is on, to maps with more advance and more fuel (but just as flat).

miatauser884

iTrader: (11)

What are your map values and rpm values around idle? What do you consider the "idle area"? Are you table switching for normal idle? Idle was great yesterday and went to **** today. Did you follow the MS3 instructions for tuning, or do you have some other techniques? This is the final issue with my car that it keeping it from being really enjoyable.

Greg G

Been busy IRL for the past few weeks, but I finally have a rough understanding of how it all fits together, using the 3.1.1 mariob v14 modified firmware. The antistall and duty tolerance, while they may be considered bandaids, are really useful!

It's hard to put in words, so let me put it in a song...LOL

Note the RPM range is 700-1500, so it looks lumpier than it actually is.



I use the antistall just once, to catch the big droop from the AC compressor, and again when the lights turn on.

PID settings for this are: P 30, I12, D80. Not yet ideal...the corrected sign for the D term resulted in better action (settles down oscillations faster), but still not strong enough to kill the initial droop. The antistall function does this nicely...but it comes in too late, when the RPMs have already dropped significantly. I may ask mario to add RPMdot as a second parameter to trigger the antistall. But the antistall has to be armed only below 900 rpm.

JasonC SBB

Crank up the D. D works off of RPM dot, ya know.

Greg G

Will modify the ini to allow crazy D and see what happens. I like the immediate response of the antistall though. It's the shape we wanted from the D term. It kind of lets you run milder PID settings (so its more sedate around target) and yet have the instant response to droops. Just need to find a way to trigger it earlier

Oh- if you make the antistall threshold too high, you get into a crazy oscillation with antistall

Greg G

Hmmm. Looking at steady state idle, I think the mild~30 rpm variations may be traced to small fluctuations in PW. Looking deeper into PW calculations, I now think that

1. I should focus on a flat PW rather than a flat gVE. gVE is just one part of PW calculation.

2. The increased resolution in the idle VE area is actually hurting me, since the values I can input are only up to single digits (meaning I can't input 44.5, only 44 or 45.) So I actually limited the ability of the MS to interpolate! It was good when I thought I wanted a flat gVE, but not for PW. >.<

The locked cells are the idle area- this is the VE3 map, used with AC. Yes, it needs a lot of work outside of the idle area >.<



I need to reduce the MAP cells in the idle area (y axis). So where should I put them? Arg. Why do I need a 16x16 VE table again?

miatauser884

iTrader: (11)

I've done it in the past, but I'm not really sure why keeping all the VE cells in a large area is a good idea? As MAP and RPM change, the required fuel for a given AFR is also going to change. Shouldn't the VE table reflect this?

I too feel that i suffer because I can't specify the VE in .1 increments. I think .1 increments in VE would be beneficial. Otherwise you have to adjust the rpm values to interpolate to a desired value, which can also work against you.

Braineack

iTrader: (62)

yeah, that's taking it a little too extreme, your fueling demands at 45kPa are not teh same as 25kPa

Greg G

The thought process I had at the time was just keep everything flat and let it settle down. I think it works for the spark table, but not the VE table. Ah well, at least someone may learn something from all this

I'll remove the 30, 40 and 50 rows from the fuel load to allow the VE table to interpolate better in that area. Where should I put the 3 rows? Is there any area that needs more resolution?

Greg G

OK I completely revamped my thinking regarding the fuel tables. I got rid of the 30 and 40 rows, allowing for smoother interpolation. Then I tapered the VE tables.

Results?



Whoa! Smooth!

Handled AC well too! Note antistall is not even triggered. Minimal action from the idle valve.


I was happy enough UNTIL AIT WENT UP! Bigger droop, triggering antistall. Minor oscillations returned with higher AIT. Idle valve is much more active now! Why is that?


Gair looks screwed up, and is screwing up PW. How do I kill it? I already copied Brains "anti ideal gas law table"...

Braineack

iTrader: (62)

you have a lot of AIT corrections. if you copied my table and you're getting that, it's not correct. it should zero out gair so it's always at 100%, iirc my ait correction value is set to 50%, maybe yours is set at 100?

you'll probably have to retune you fuel maps a little if there was always correction on the tabe.

Greg G

Yup 100%! I set it to 50 now :P

Weird though how Gair spikes like that. I'd have thought it would act more like WUE.

Greg G

Upgraded to mariob v17 firmware-- but I edited the ini file to allow crazy D values (greater than 200).

I turned off all the "bandaids"- target rpm hysteresis 0, tolerance 0, antistall at max value of 255 rpm below target (target at 975, so antistall threshold= 720 rpm)

Results are interesting! The jump in spark advance shows the table switching triggered by the AC switch; extra air solenoids bump up the rpm before the compressor starts (on a ~1.5 sec time delay). The red markers represent the headlights turning on then off...

p25 i 12 d1200


p25 i 12 d 1600


The scaling is fixed, so the can be compared easily. RPM scaling is 700-1500 to exaggerate idle oscillations.




If you compare it to previous pictures I posted, the idle duty ramps up really fast as soon as the dip starts. Previously it would start ramping up near the bottom of the dip. Even when I used the antistall, it would open the valve up at 800 RPM. Seen here, it opens up earlier at ~1250 rpm, just after the peak of the idle bump! I also noticed that because it ramps up early, it doesn't need to open the valve as much.

The 1200 kills the dip well. The 1600 D does this as well, but I think it doesn't have enough P to deal with the slower changes in RPM, clearly seen when I add the headlight load to the AC load.

The idle without AC is set to the bottom of the valve range, so it has no way to react to the extra air solenoid bumping the RPM up. I might try closing the throttle opening to make it idle with the valve more open, so it can kill the initial rise in RPM from the extra solenoid.

The corrected D really works! I think I need to add some P to kill the shallower oscillations.