Idle, Air Con and AFR wierdness

[arran]

Hi all,

MSL and MSQ attached. Since getting the car running about 6 months ago with this ECU I have had success with getting my closed loop idle to work. The engine is a 13B Mazda rotary turbo. A couple of months ago I had the AC gassed. I've had problems getting the idle to behave which has been perplexing because it shouldn't make that much difference. I think however I am seeing the problem in data logs. If I free rev to 3k and lift off the throttle the engine will very likely stall.

The reason for this I think is because of the AFR, it goes full lean on decel with the AC on, but behaves on the same decel with the AC off. At t=793 secs I turn on the AC.

So....what's the story, why am I getting such a big lean spike just because the AC is activated?

Any help greatly appreciated!

ArranCurrentTune.msq

2014-01-25_12.00.27.msl

[Matt Cramer]

The VE table at idle looks a bit rough - I think it's just moving into a different VE table cell that's tuned differently.

[arran]

Yes, that makes sense! Thanks Matt, I should have realised.



Using MLV tuning console I was watching the way the trace jumps around the bins and the afr wasn't making sense. That would be because at 1300rpm it stops looking at the main fuel table and goes to idle ve table.

Looking closley at the idle ve setup screen, i have idle ve activating below 1300rpm but i have the idle ve table x axis stopping at 1100, so there is a gap of 200rpm. What is the recommendation on how to maximise resolution? What happens below 800, does it go back to the main table? What happens between 1300 and 1100?

And

With setting "leave valve closed above: (rpm) = 1800, should the valve be left closed above 1800rpm? Following setting the minimum duty to 68.6% it does not seem to be doing that. I'd want it to bottom out at 68% when in or near CL idle, but when nowhere near CL idle shouldn't it close?

Thanks again,

Arran

[arran]

Update on the stalling with AC on - made some changes to idle VE table this morning and it is definitely improved. It still stalled a couple of times but i can feel it is tending less. So idle VE is definitely the place to be looking which is a vast improvement over the continual adjustment of the PID settings that didn't seem to be making any difference which was quite frustrating.

[arran]

Over on Rx7 club i sort of hijacked a thread on BAC stuff
http://www.rx7club.com/megasquirt-forum ... st11671435

Thanks to Aaron Cake it appears that i need a flyback on the BAC.
I added one this morning and it changed the duty cycle behavior substantially.

I have adjusted it all again to start OK (still some tweaking needed but it is OK), warm up OK, and respond to a throttle blip and setting in to idle rapidly. Having done all that I went for a drive this afternoon and stalled the hell out of it EVERY TIME i lifted off with the AC activated.

Needless to say this is causing me some anxiety.

So I disabled Idle VE and autotuned the hell out of the AC stalls using the main fuel map only. After about half hour of tooling around the burbs I felt I had it sorted, no stalls, everything seemed perfect I had not had a stall for 10 minutes and a couple of dozen clutch in from 3k or 4k engine deceleration tests. When I arrived home and pulled into the drive way just to drive the point home the car stalls again as I am coasting up to the bump where the gutter is. #$%#$^$&$%#%#^&*(%# was my reaction.

In the MSL at t=1785 there is that stall, I can see the injector msecs are increasing but the afr continues to go skyward. In that attached MSQ I have enabled enhanced acceleration enrichment following the drive and have not tested that yet.

ArranCurrentTune1.msq

2014-02-02_19_modified.msl

[arran]

I'm still wrestling with idling issues with the AC enabled.

I have an input to Megasquirt connected to the AC system so MS knows when the compressor is engaged but I have not connected up a MS output to control the compressor. I'm planning on doing this.

Am I wasting time with Firmware 1.2.4 trying to make this work?
In the AC idle up settings, what is "Idleup duty%" supposed to do? There is a setting to increase the idle target, which would seem to be the same thing as idleup duty. What i'd like it to do is to just instantly add duty as the compressor is engaged as it is an instantaneous load. But it does not do that. As the compressor engages there is invariably a substantial drop in rpm before CL idle lifts the idle to set point after maybe 1 sec.

I notice reading the Firmware 1.3 change log http://www.msextra.com/forums/viewtopic ... 25&t=48995 that there are a couple of changes to Closed loop idle.

This one:
Fix bug elaw found in idle where in certain circumstances, the the ramp to target won't ramp.
Re-arrange/re-name idle control valve/algorithm settings.

What seems to happen is that I can repeat the same test 10 times and it will be perfect, no undershoot, nice ramp down to target, but on the 11th time it will want to stall. It seems that the dash pot adder just does not engage sometimes. I can't see why.

I am making improvements to work around the dashpot adder. I have added fuel to the main fuel table around the idle area which made the idle very very rich, so I enabled VE idle again and specified a 1 sec delay so on the decel the active point will dance through the main map idle area and have fuel added to try to counteract the lean out, then idle ve activates with leaner afr. I have also smoothed out the main fuel table to remove a number of weird hills and troughs

I also noticed that at night, lights on, AC on, fan on, the battery voltage will drop to around 11.5. This seems very low so i am going to change the alternator. I have increased the pwm idle voltage compensation numbers for voltage below 12.3

The car idles and falls into idle absolutely perfectly without AC.

Any advice greatly appreciated. current MSQ attached

[arran]

I continue to have trouble with the car stalling when returning to idle with the AC on.

I think I understand what is going on. It is related to the Closed loop initial value table. When I drop into idle the value needed for AC off compared to AC on is very different and I can't see how to set it up to cover both AC on and AC off.

Say the engine is cold the CL idle vs temp adds to the target rpm, that is fine. As the engine drops into CL idle the initial value is selected and the idle settles from there. When the AC is turned on, say on a warm engine now, the CL target is increased to somewhere near where it was when cold no AC, but the AC load is large and the initial value needs to be higher because the (negative) rpm dot is much higher.

In the AC idle up settings, does the "Idle up duty" add to the CL Initial value table values when AC is on?
How does "Idle up duty" differ from "Idle up target increase" ????????

Appreciate any advice. Thanks

[muythaibxr]

Idle up duty is how much is added to the duty when tha AC button is on. This is added to the initial value every time you return to idle. Are you using MAT or CLT for the initial value table? Idle up target increase is how much to add to the target idle speed when the AC button is pressed.

The dashpot adder not being added is on my list of things to investigate.

[arran]

Hi, thanks for have a look at the dashpot adder.

I am using CLT rather than MAT for the initial value table.

Over the last couple of days I have changed:
1) The Voltage calibration, the MS was reading low. I have calibrated it to the voltage as measured from the battery terminals.
2) I have changed the alternator just to be sure because I had a spare
3) I have spent a lot of time tweaking the Initial value table. I went in to idle valve test mode with AC on and AC off and data logged how the engine responded to % changes to idle valve duty as the water temp increased. I used that info to populate the initial value table, and did further tweaking today on the road.

It still stalled a couple of times when I was stationary in the garage when I got home, the water temp was quite high at the time, around 100 degrees Celsius (the ambient here today is currently 37!) so it may have been fuel vaporisation, though it still didn't stall with the AC off!

I'll look more closely at what effect the "idle up duty" is having.

Thanks again

[muythaibxr]

I would recommend using MAT for the initial value table. That way you can account for MAT increases causing a need for more airflow to get the same RPM. This method automatically accounts for CLT because lower CLT should have higher targets, and the target RPMs are used for that axis's lookup.

My guess for the dashpot adder not being added is that you are using the dashpot factor setting. That should be set to 0 for most people.

[arran]

Sorry, I'm not sure what a Dashpot factor is.

I have "dashpot decay factor" of 3 secs, and Dashpot adder is 3.9%
The MSPro manual calls it "dashpot decal factor" (not my typo!)

I'll try my luck with MAT lookup.

Thanks, Arran

[muythaibxr]

Sorry, the decay factor is what I was referring to, and it is not set in seconds. It is a multiplier that affects how quickly the dashpot goes away with respect to rpmDOT. If rpm is dropping fast, the dashpot stays on, if not, it doesn't. i would suggest setting that to 0 and using MAT for the initial values table.

[arran]

Recommendations have been applied- changed to MAT initial value, and changed dashpot decay factor to 0

I'm not sure how the MAT table should look, I've rescaled the MAT axis to be between 0 and 80 celcius, and left the numbers that were in there from the CLT setting alone for now.

Thanks, i'll see ho wit goes on a road test.

[muythaibxr]

Just look where closed loop held the duty for several MATs. The scatter plot feature of the log viewer would probably be useful there.

[arran]

Went for a half hour drive after making the couple of changes suggested by Ken.

I had one stall with AC on before I got out onto the main road but I pulled over and tweaked the Closed Loop initial value table and increased the "AC idle up duty cycle" to 5%. I had not understood what this setting was doing until this thread, thanks very much for that.

I did the rest of the drive without another stall! I'll see how it behaves through a cold start but fingers are crossed that the changes to Dashpot Decay factor, MAT Initial Value Table and the AC idle up duty have nailed it.

Another observation:
Sitting at idle when the AC is engaged, ie you push the button, there seems to be a delay before the idle duty is added. The RPM dips down a couple of hundred RPM for a few secs. It isn't a major thing because the car does not stall, but the revs tend to dip before CL compensates. The idle valve should immediately do something like add 5% duty, just when when the AC is activated.

Also, I do not currently have the engagement of the compressor controlled by the MS. I have a spare digital Out in the engine bay ready to hook up if I want to. Aside form the obvious disengagement of the compressor over TPS=X and RPM=y, with Firmware 1.2.4, are there any other circumstances when MS will briefly turn off the compressor?

[jsmcortina]

If you use the MS3 control of your compressor your should eliminate the RPM dip. The MS3 raises the idle and then (after a selectable delay) turns on the compressor.

James

[arran]

It's been a few days and I have had a chance to take the car for a trip and look further at the issue.

Car is still stalling with AC on. I have not yet connected up a MS output to control the Compressor

The short story is I know what stops the stalling. If I disable the Air Con idle up the car idles beautifully whether the AC is on or off. I'll let the data logs tell the story, attached are two data logs both taken an hour or so ago, one with AC idle up enabled and one with AC idle up disabled.

With Idle up enabled, particularly on a cold engine it seems, the car still frequently wants to nearly stall, or stall, as I pull into neutral and let the revs fall from 3000. At wits end on this problem, I simply disable the AC idle up. You'd think that without the assistance of the idle up that the stall tendency would increase, but it didn't. The RPM obviously dropped a 100 rpm or so when the AC is turned on, but the perfect idle I've been mentioning I have with the AC off was retained with the AC on.

Why that is I have no idea. In the file AC_idleup_on at t=251 there is a stall. The ECU seems to be doing what it should, as the stall progresses injector pulse width goes up, as does Idle PWM, but why why why does the AFR head skyward? In MegaLogViewer with Tuning Console as the stall progresses the fueling in the main table is going from 37, 38 and then 46! but the AFR is still going leaner, how can this be?

Same thing happens at t=276 though the engine does not quite stall here, it recovers, but the VE table in tuning console heads into what should be rich condition, but the AFR goes full lean. Same at t=290. Something is lying to me, I don't think it is the AFR because the actual engine behavior matches the lean condition.

In the MSL file ac_idleup_disabled, at t=333 the idle returns beautifully, as it does at t=363, and t=377, and so on. You can see when the port k variable is low when the AC compressor is active. So either AC on or off the engine returns to idle perfectly.

Is there something the MS is doing that it shouldn't when AC idle up is enabled?

[Jobro]

The AFR is not necessarily going lean when the WBO2 system starts reporting lean.

The conceptual problem many people have with WBO2 systems is "they measure the air fuel ratio", they do not measure air fuel ratio. WBO2 systems measure the amount of excess oxygen in the exhaust, ie Lambda. Furthermore they read lambda assuming the air fuel mixture was burnt. I suspect you have having a misfire event. I have been having similar problems with a similar engine but a different engine management system. If you have a misfire during the commencement of this stall condition, the WBO2 system is going to determine there is X amount of unburnt air in the exhaust stream, lets call it 2% immediately before the beginning of the stall condition, then something happens and you begin to stall. Think about each individual cycle, if a cycle fails to fire for some as yet undetermined reason (no one knows exactly why as yet) then the WBO2 system might measure 3% 4% 5% remaining O2 in the exhaust stream. This is different from 3,4,5% when the mixture was successfully lit off. What I am getting at is your stall is probably misfires and your misfires are failed combustion events which are going to be putting abnormal amounts of HC (unburnt fuel), CO incomplete combustion, and O2 (unburnt oxygen but should have been) out the exhaust pipe. Your mixture might be rich or lean or even 'correct, no body knows but your WBO2 system will always report lean because there is unburnt O2 in your exhaust stream because you probably have a misfire to be stalling.

[arran]

Thanks, whatever is occurring it is triggered by the AC Idle up setting, which should not happen.

To all, I note Firmware 1.3 has just gone Release Candidate 1, is there any possibility that the behavior will be changed in that Firmware?

[arran]

Well I've gone and done what I said I'd never do when running a rotary engine - installed beta firmware! Well sort of, I just installed 1.3 RC1. I am not able to work out the problem with random occasional stalling because despite what the log says, the dashpot/initial value for the idle valve just does not actually do it's job.

The new version complained about MAF I don't have, and closed loop idle became disabled.

I immediately noticed that the dashpot behaved differently, with the same settings as in 1.2.4 as the engine settled into idle it did so much much slower from about 1200 down.

I'll update again after i drive it