Strange AFR Behavior @ Idle

[ErnieJones]

Hi Folks,

Hoping that I'm missing something simple. Microsquirt Module on Ms Extra on the latest firmware release. This is a new module and along with it I installed a wideband so I'm going through a bit of a learning curve as the narrow band seemed to be much more reliable and easy (if not accurate).

What I'm seeing is this, at hot idle the AFR will read rich, call it 12:1 and this is with a VE setting of 38. I then lean it out by dropping the 38 down to something like 32. The AFR in Tuner Studio then reads close to stoich which makes sense - it's been leaned out. I then hit the apply button and the ARF swings way lean, 17:1 or so

I then richen the AFR back up by changing the VE back to 38 - it then reads roughly stoich. I then hit the apply button and it immediately drops down to 12:1.

Get the picture? When I adjust the cells relating to the idle AFR and get it right, and then hit apply, it goes way lean or rich and I end up back in a cycle between a VE of 38 or 32.

No ASE, no WUE fully warm. No advance changes either.

Tell me I've missed something really simple in the settings or that this is a known issue caused by novice user mis-configuration I can post up a config and log but I was thinking that this may be something common. Let me know if it's not and I will

[kjones6039]

Come on Ernie...... Show us you msq and log! I for one, would like to see that afr doing it's thing!

Ken

[ErnieJones]

Yeah, yeah, yeah - I know it sounds stupid

I just did a logging session. These are the general paramters;

stoich (+/-) is 32 in the VE table
rich (+/-12:1) is 38 in the VE table

It doesn't always do it but it does more often then not. I managed to log some of the weirdness. You should see that 32 is stoich at some point in the log and at the end of the log, 38 is stoich (!!).

Anyway, I'm at a loss but I hope it's going to be a 'bonehead' moment because otherwise this could be really hard to track down. In simple terms, it seems that stoich is changing all the time from between 32 to 40... Seemingly, with no other changes and the engine at full hot.

[Matt Cramer]

This sounds like the "burn stumble" effect (where the ECU hiccups when you click Burn) is putting you in a different part of the map where it's running a little different.

[ErnieJones]

This sounds like the "burn stumble" effect (where the ECU hiccups when you click Burn) is putting you in a different part of the map where it's running a little different.

Burn Stumble is an interesting term. In my case, almost every time I hit burn, the engine will stall. This behavior was the same with my first ECU. I chalked it up to it being a single cylinder engine but even if I rev the engine up to 8,000 RPM and then hit burn, it will, almost every time, stall. I don't know if that's 'normal' but 1.0 did it and 2.0 does it too.

I just saw something interesting that might shed light on the issue....

Engine running at stoich with VE set to 34

I shut off the engine/ECU (key off) turned it back on and the 02 sensor read very lean, about 18:1. I let it run (barely) and waited. The 02 still stayed at extreme lean. So I then tried hitting the throttle and it was so lean it immediately died. So it's not that the 02 is displaying incorrectly, it really is that lean.

Just to keep it simple, with it idling at stoich the VE setting was 34. Turn off engine/ECU and start it and it went to about 18:1 Turn off engine/ECU and start it again with same VE and idle and it's back at stoich.
I just did this several times with NO setting changes and most of the time the engine would not start.

I then went in and set the VE to 40 and then the bike will start - it was too lean again even though the previous setting of 34 was stoich.....

Now I've got key on, not running, indicated RPM is zero, and the VE table cursor is bouncing between the two lowest points even though the engine is off and RPM is zero.....

Is it possible I have a goofed Microsquirt board?

[ErnieJones]

Once again with the engine running I see that it's too lean. So I richen up the same sections of the VE table until it reads stoich. I let it sit at stoich long enough to ensure it's settled and accurate. I then hit apply and the AFR goes rich to 12.x:1

[grom_e30]

even my 1000cc 4 cylinder bike engine stalls when i hit burn unless i have it over 5000 rpm then it usually catches before the engine stalls. where is your o2 sensor? i had mine stuffed down the tail pipe today and it took the sensor about 45 seconds from reading full lean in the air 22afr to reading 14 or so afr in the exhaust. what wide band are you using as well?

[piledriver]

what are your afr targets set to?

are you using idle ve?

~2 second lambda delay at idle is not unreasonable near idle.

[ErnieJones]

even my 1000cc 4 cylinder bike engine stalls when i hit burn unless i have it over 5000 rpm then it usually catches before the engine stalls. where is your o2 sensor? i had mine stuffed down the tail pipe today and it took the sensor about 45 seconds from reading full lean in the air 22afr to reading 14 or so afr in the exhaust. what wide band are you using as well?

OK, so don't read anything into the stalling when burning - thank you for confirming that is 'normal' behavior
02 Sensor is new Spartan Wideband and it's the original version (cannot detach sensor). It is located 2' from the exhaust port which, according to the docs, is where it's supposed to be. I also calibrated it according to the docs on Alan To's site and it was almost exactly perfect out of the box.

what are your afr targets set to?
are you using idle ve?
~2 second lambda delay at idle is not unreasonable.

AFR targets for idle is set to 14.7
Idle VE? Not sure what that means but if you mean am I in the idle section of the VE table, yes. The engine idle is 1450 RPM and I'm always working in the same column once warm.
Thank you for confirming that slow 02 response is normal. I noticed that immediately after going WB from NB and was wondering why such an exotic sensor was so slow (so to speak).
I do wait until the AFR stabilizes before making a change.

But this is the thing, stoich is moving and it's either 32 or 38 on a regular basis. Fully hot, no ASE/WUE or other stuff. Just plain old hot idle with 20 degrees timing.

"Once again with the engine running I see that it's too lean. So I richen up the same sections of the VE table until it reads stoich. I let it sit at stoich long enough to ensure it's settled and accurate. I then hit apply and the AFR goes rich to 12:1"

I can't make heads or tails of it. It is as if the settings are getting corrupted somehow. I think I will try re-flashing the little bastard to see if that does anything. But I do believe there is some kind of corruption issue going on here.

One thing does come to mind, I will check the IAC to ensure it's not moving for some reason. That's the only thing I can think of that could make the idle rich or lean the same amount every time. If that thing was stroking back & forth the same amount, at will, then this would make it rich/lean....

[grom_e30]

i think you going in to it to much, split the difference 38-32=6 so divide that by two add that to the lowest number and use 35 ve, i target for 14afr at idle and mine does wander a tiny bit depending on temp ect but i happy with it between 13-16 afr at idle

[ErnieJones]

i think you going in to it to much, split the difference 38-32=6 so divide that by two add that to the lowest number and use 35 ve, i target for 14afr at idle and mine does wander a tiny bit depending on temp ect but i happy with it between 13-16 afr at idle

Na, there's an issue and it's a serious one The engine won't even start at times because the AFR goes too lean. Just to make this a bit clearer, let's say the bike is too rich at idle, locked at 12:1. I go in to the VE table, change the idle numbers in the cells and the AFR locks in at 14.7 - it stays there all day idling perfectly until I hit the apply button. As soon as I do that, the AFR goes way too lean (17:1) and stays there.

The reverse is also true. Now that the idle VE has been adjusted to what was perfect stoich at 14.7 (until I hit the apply button) and it's now become way too lean, I go back in and change the VE cells to richen it up. Once again, I get a perfect 14.7, and it will lock in on that until I hit the apply button and commit those changes to the ECU's VE table. As soon as I do that it will go way too rich (12:1) and I'm back to where I began. I change the VE cells to lean it out to a perfect 14.7 - it locks there and then I hit the apply button and it instantly goes way to lean (17:1) and locks there.

It's a cycle that is repeating itself all day long. I did pull the IAC connector at full hot, so the IAC is locked at the proper setting, and the issue is still present so it's not the IAC inadvertently creeping around that's causing changes in the idle AF ratio.

I'll have to come back to it later with a clearer head. Right now I'm way too frustrated. The smart move will be to let it sit for a bit until a light bulb comes on at some point because it's not driveable in it's current state and it's behavior is consistently moving between locking in at 12:1 or 17:1 with nothing in between. That part is consistent. I'm always moving the idle VE from 32 to 38 back to 32 and back to 38 in an effort to straighten it out. Very peculiar, I suspect it will be interesting when I do finally case it.

I'm also curious as to why the cursor in the VE table will still move rapidly between the two upper left cells when the engine is not running and the RPM is zero. I wonder if that may be symptomatic of this issue too?

[vw_chuck]

What injectors are you running on it? Flowrate and brand?
What does the pulsewidth do when you hit apply?

[vw_chuck]

What injectors are you running on it? Flowrate and brand?
What does the pulsewidth do when you hit apply?

[grom_e30]

One thing I did notice from the log is you nearly always only half sync up when the engine is running.

[ErnieJones]

On the half-sync, that's probably just the logging session and it had numerous hard starts/failed starts in that log. When running it's fully synced (I have the little green lights on the bottom of the dash set to show sync/half sync).

On the injectors, they are Keihin (Japanese) and I have two of them. I was running on an 'old' one yesterday and I changed it out for an identical cleaned and testede one last night. Same thing. The flow rate and dead time calcs etc are as follows;

Flow Rate = 139.5cc
496 msec offset @ 13.2 v

Settings in TS;

Injector Dead Time = 0.496
Battery Voltage Correction (ms/v) = 0.066
Req Fuel = 6.0
AFR Target @ idle = 14.7

The pulse width appears to be doing what it should. ie: if it's 2ms and I enrichen the mixture it will increase to something like 2.3 ms and vice versa.

Here's the thing though. I just went out to start it to check these things and it was way too lean even though it had been shut off at stoich. I had to richen up the AFR table by the same amount (again) just to get it to start. I'm really starting to wonder if the microsquirt module itself may have an issue. If I shut it off at stoich, and then start it up 15 minutes later and it's too lean to start, then something is up. Something is changing.

[grom_e30]

The ve table should not affect starting its not used until the engine goes past the cranking rpm setting.

[ErnieJones]

The ve table should not affect starting its not used until the engine goes past the cranking rpm setting.

That makes sense but it's not what I am observing. When it fails to start it appears to be symptomatic of it deciding to go to 17:1 If I set the VE back to 38, it will start. If the ECu decided to go to the 12:1 side then there's no starting issue other then it's too rich. Which makes sense, too rich you can still start and too lean, not going to happen. Probably the reason why I'm not seeing your logical theory is because it's past the cranking point and firing (at the time when the VE would be in use) but it's not completing the start because it's too lean.

Just to try and bring this back to basics for minute (I was outside walking around thinking about it).

There is no issue getting a perfect idle at stoich. It will lock there and sit there all day idling perfectly at an indicated 14.6:1 - literally, for hours.

I achieve this perfect idle by adjusting the VE cells and NOT writing them to the flash.

The minute I write the new VE table changes to the flash memory, that's when it screws up and my perfect idle will either go 12:1 or 17:1. It will always be one or the other and it's consistent as can be. The same setting that got me a nice stoich idle 2 seconds earlier will either go way rich or way lean when committing that VE setting to the module.

The new VE settings do appear in the VE table after flashing. So they appear to be committed.

At times, I'll be able to achieve a perfect stoich idle and with the settings committed to the flash memory. This is rare but if I adjust it enough it can happen. If I power cycle the ECU it will either;\

start perfectly with the stoich idle or
it will fail to start because it's too lean or
it will start and it will be at 12:1.

There's a pattern here of either 12:1 or 17:1 or stoich. It's like throwing spaghetti against the wall, I never know what will stick. This is kind of what makes me wonder if the ECU may be pooched in some weird way. Even if we just isolate this one sentence, "If I adjust the VE table to get a nice stoich idle, and then write those changes to flash - it will then either stay at stoich or immediately go 12:1 or immediately go 17:1" This behavior is like clockwork. It will do one of those three things and it will also do one of those three things if I power cycle the ECU - despite no changes to the VE table.

This is why I'm starting to wonder about the Microsquirt Module itself. Or, perhaps, a configuration setting that could cause this weirdness? Maybe I've stumbled upon a problem with the code? I don't know. I can't see any reason for this behavior at all. I'll look into it more closely later because I'm frazzled now

[grom_e30]

have you tried making a change to a non running engine, say if it is rich at 38 shut engine down then burn a lower number is say 34 then start the engine see how that works for you??

[ErnieJones]

No dice

The issue isn't one of finding the correct VE value for stoich - I know it's 30-32 (depending on the injector settings). It's that no matter how you set the VE cells for stoich one of three things will happen when I hit apply and burn it to firmware;

- It will remain at stoich
- It will go straight to 12:1 and stay there
- It will go straight to 17:1 and stay there (for as long as the engine will run when it's that lean).

Same thing happens on startup. If I power cycle the ECU I have a 1 in 3 chance it will be correct. If it's rich, it will be 12:1, if it's lean it will be 17:1 and if I hit the jackpot it will be 14.6:1 All this with no changes to the ECU.

I'm sure we're all getting frustrated with this thing 'cause it's a hard one to fathom. I sent an email to Bruce to see if he could shed some light on it. I have no idea what could cause the ECU to randomly alter the fueling when it's power cycled to either 12:1, 14.6:1 or 17:1 with no other apparent changes. Seemingly at will and randomly.

[piledriver]

I sent an email to Bruce to see if he could shed some light on it.

You are using B&G firmware?

If not try the idle VE and idle timing (and perhaps cold timing) features and see if it helps.