Looking at the magnified logs, I noticed that RPMdot has a lag to it. The graph below shows that even if RPM was going up, the RPMdot still read negative! It took 5 ticks from the bottom of the droop until RPMdot became positive.



I changed the RPM lag factor from the default 50 to 87 (which is what I saw Ken using at MSextra). Now it seems to have helped. It only took 3 ticks till RPMdot became positive.



I may need to check my lockout settings, as I have sacrificed some smoothing for faster response. But even before I checked the logs, it felt better. Maybe placebo effect. :P

I also disabled the close valve when revving-- it solved the droop! I edited the ini to allow a 1 second PID delay (previous minimum was 2). Full load, no droop coming down from revs. Now small blips can lead to a weird idle bump which doesn't come down, I need to fine tune the lockout settings.



I also tried switching from 3.1.1 mariob v17 to the official 3.1.2beta- I still couldn't start the car! Just smelled of fuel. I even loaded an old msq from 3.0.3u, no go. Wonder why...

Next target: Find out what the F is causing that weird idle duty shape :P

 

OK, I'm done with the idle AC droop! Including droops on deceleration from revving.

The final piece: raised minimum duty to 36, and maximum to 85.
P90 I80 D5500 min RPM 900 max RPM 1500.



With this setting, the code struggles to bring the RPM down to the 975 target at no load, but settles in right on the money when AC is on. I could probably reduce the dip lower than the 40- 60 it is now (by decreasing max rpm), but that would make the steady state RPM too jittery.
To be fair, I went back to the start and tried tuning I then P with minimal D. It was terrible. So yeah, it's extreme D for my car with the large throttled volume. It's not for everybody- but at least the option is there.

I'm MOVING ON with my life LOL. Next issue: open loop deceleration settings, closed loop entry/lockout, overrun settings.

Before I start, can anyone tell me exactly WHEN I am supposed to enter closed loop? In the graph above, I enter closed loop just after the RPM goes down to target. 1 second delay (edited ini).

Regarding overrun fuel cut: what's the theory behind it. I saw Brain's settings were cut fuel at 1700 rpm, 2 sec delay return fuel 1600 rpm. Is that still valid?

Attached are my msq and logs. Someone please point me in the right direction for correct deceleration/ closed loop entry.

 

Not bad. Here is mine with my improved firmware that has full A/C control. The A/C duty added depends on the coolant temp, so it works extremely well across the board. I have also modified the firmware to allow a higher closed valve duty than the minimum PID duty (like a minimum dashpot adder).



There's an immediate bump from the non-A/C idle (850rpm) to the A/C idle (1000rpm) the moment the A/C switch is pressed (user0 changes from 1 to 3, bottom graph).

I is happy. Finally the A/C works exactly like stock.

 

Awww man! My solution sucks

 

But again, you also have that silly blow through throttle.

 

Man up and get a turbo.

 

Ok enough about the supercharger!

Can I get some handholding on the deceleration/overrun transition from open loop to closed loop?

 

The die is cast. Changing to RX8 injectors and Maruha cams (mild cams- 252in 256ex 8.5mmlift).

Adjusted req fuel from 11.7 (stock 1.6 injectors- 230cc) to 6.3 (rx8 inectors-425cc). Cranking PW is in %, so I shouldn't need to mess with it. Adjusted injector open time from 1.2 (1.6 injectors) to 0.8 (RX8 injectors).

Has anyone determined the exact injector open time for the RX8 injectors yet? And battery correction factor?

Anyway...fingers crossed that the swap will be painless...

 

One of the MS coders tested an RX8 yellow injector and posted a spreadsheet of the results

http://www.msextra.com/forums/viewto...41357&start=20

 

Awesome! 800ms@ 12.1v = 733ms@13.2?

Doing the math...batt correction factor = 73?

 

Where can you get this custom firmware?

G

 

By buying one of Reverants MegaSquirt units from him.

 

^ this. My only source of income is the Megasquirts I build. So support the cause and consider buying one of my ECUs.

 

Stellar response.

 

PM me or post the latest news on the MS3!

 

Still waiting for the developers to open source the code. I have (literally) a dozen MS3 daughterboards on my shelves that I don't want to sell without adding my mods to them.

 

Dammit.. With the cam/injector change, it's idling like crap. Crapcrapcrap.

I seem to have some sort of misfire. Some sort of poor combustion, feels like the first time I installed the MS.

Car has occasional hiccups at idle, worse with load. RPMdot for some reason is way up, making the idle code go nuts! It was worse yesterday, I suppose there's some sort of break-in period for all the parts to get to know each other?

 

OK, it settled down a bit. I was considering a vacuum leak at the injectors, and kicking myself for installing 2 things at one time (never learn). So I reinstalled stock injectors.

Still has that stumble. MAP is a bit inconsistent, but it may be due to the intermittent stumbles the engine gets. I reduced PID effect by increasing max rpm to 2000 so the valve duty isn't driven as hard.

This is a graph of no load idle (AC on at the far right). MY MAP has gone up from 30 with the old cams to 45! Is that right? AC load went from 40 to 50! Is that just more air being let in by the cams?



This is a shot of the stumbles.


Hope this is familiar to someone.

 

Too much overlap?

 

^ Possibly. Too little knowledge to reply for sure, am frantically catching up on my reading

Posted a separate thread about it:

https://www.miataturbo.net/showthrea...279#post772279