Why is my AE always "late"?

[seth900]

I think I've had this problem with MS-II for as long as I've had it, over a year at least. With several different code versions, basic MS-II as well as Extra. I'm running in fuel only mode with stock Saab Direct Ignition system.

It's just that for some reason I'm always getting a very lean dip before the enrichment kicks in, causing bucking and jerky behavior so that I've actually learned to drive the car with very smooth throttle movements to avoid this. Needless to say, I'm getting so tired of this that I'm actually thinking of switching back to the OEM system.

When looking at log output, it shows accel enrichment triggering and PW going up. However, this is always followed by lean readings from the WB for a few cycles, then extremely rich readings for some more cycles before it stabilizes.

I've played around with different AE settings a lot, from small to large values to everything in between. Increasing the AE PW just causes an even richer spike after the lean spike. That's what makes it look like it's always late. I'm using 100% MAP based AE, but I've tried with different blends with TPS as well. The only difference is that TPS is harder to trigger since I've got a pretty big throttle which makes it quite insensitive.

Lag factor for MAP is 100, and 90 for RPM. Changing these will obviously just make things worse. I've installed three small fuel filters in the MAP signal line to smoothen the signal, but behaviour was exactly the same as without them except there's naturally less signal noise.

Currently I'm running 315cc/min (at 2.5 bar fuel pressure) injectors, in an 2.0l inline-4. I used 580cc/min injectors earlier but I was limited to 2 squirts/alternating with them so I initially thought that this was the problem
But now with the smaller injectors I'm able to run 4squirts/alternating or 2squirts/simultaneous and it really isn't much better.. I know the OEM system uses 2 squirts/simultaneous and accel works perfectly even with no TPS (it's a LH-Jetronic with hot wire MAF).

So in summary, AE triggers fine but it just seems never to be there early enough? Probably this is a tuning issue, but what could it be?

[muythaibxr]

That is a very loaded question.

There are tons of things that can affect throttle response:

1) What are you triggering accel enrichment from (MAP? TPS? Both?)?
2) What algorithm are you using (EAE? X-tau? Normal AE?)?
3) How many squirts/cycle are you using?
4) If you're using normal Accel Enrichment, what is your threshold set to?
5) Did you ever *just* tune VE with no AE?
6) how did you wire your injectors if you're using alternating?

All of these things (and some stuff I left out) can affect throttle response, can make it very difficult to get good response.

With answers to the above questions, we should be able to do a better job of helping you get better response though.

Ken

[seth900]

I'll happily provide all necessary details.

1) Just MAP but I've tried a 50/50 blend and pure TPS as well
2) Normal AE
3) 2 squirts/simultaneous, but I've also tried 4 squirts/alternating
4) 60Kpa/s
5) Yes, and steady state AFR is pretty fine
6) 1+4 and 2+3, but right now I'm using simultaneous

[muythaibxr]

I'll happily provide all necessary details.

1) Just MAP but I've tried a 50/50 blend and pure TPS as well
2) Normal AE
3) 2 squirts/simultaneous, but I've also tried 4 squirts/alternating
4) 60Kpa/s
5) Yes, and steady state AFR is pretty fine
6) 1+4 and 2+3, but right now I'm using simultaneous

Most of those settings should be alright.

However, MAPdot is always going to be a bit more laggy than TPSdot once you get TPSdot tuned right.

other things can affect response too, like VE-table lookup lagging behind actual conditions, etc... I've found that you're always going to have a slight lean-spike when you do a hard-stomp as well.

My recommendation would be to switch to TPSdot with a very low threshold... if you don't have much TPS noise, set TPS lag factor to 100, and the threshold to 2-5%/sec.

If you're getting a lean spike followed by a really rich spike, you probably also want to shorten the accel shot, shorten the adder, or both. The lean spike is caused by AE/VE lookup not reacting fast enough, where the rich spike is caused by AE over-reacting after it activates.

Also, you could try EAE as well... Just enable it without lag compensation first, and tune it so that normal driving is OK, then turn on lag compensation and try some quick blips. You will end up with a slight lean spike, but response will still be good, and you shouldn't go super-rich after.

Ken

[seth900]

Thanks for the response.

I've been considering TPS, but the main problem is that it is pretty noisy especially at higher lag factor settings. So noisy that I have to set the threshold above 5%. It's a new sensor but a cheap non-OEM one though.
Any advice in getting rid of the noise? Also because of the large throttle bore (Saab original, 55mm in 2.0l engine), TPS reading is fairly non-linear. With a quick stomp, the fastest I can get is maybe 100%/s, and TPS shows only 1-2% at cruise.

As for the accel shot duration, I've already shortened it to 0.1ms with 0.1ms taper to zero PW.

So it seems my choices are either getting TPS to be more sensitive or learning to tune EAE.

I actually tried enabling EAE once but it started oscillating so bad that it was not drivable at all. But maybe I'll have to delve into it now.

[muythaibxr]

Yeah, on some engines, EAE's default settings work fine, on others, they're completely wrong.

In any case TPS noise is usually because of a bad TPS, or bad grounds on the TPS.

Ken

[seth900]

I did some testing, and judging from the logs it seems that when applying throttle at idle mapdot is lagging behind tpsdot some 0.2ms! No wonder I'm stumbling no matter how much AE I'm throwing at it.

I seem to remember that I didn't have these problems with MS-1, though it could be that I used tpsdot back then. But maybe mapdot calculation is slower in MS-2, could this be possible?

I never really got TPS accel working, but that's probably since I'm getting ~15 TPS noise which can correlate to over 200Kpa/s mapdot values. I'll first try swapping in a genuine Bosch sensor, the grounds and cable should be pretty good quality since they're original OEM stuff.

[Keithg]

What is your map lag set to? With the later versions of the beta code,
there is a filtering mechanism which allows a much less laggy map
signal to be used.

KeithG

On 7/19/07, seth900 <jcarlson@iki.fi> wrote:
[quote]

I did some testing, and judging from the logs it seems that when applying
throttle at idle mapdot is lagging behind tpsdot some 0.2ms! No wonder I'm
stumbling no matter how much AE I'm throwing at it.

I seem to remember that I didn't have these problems with MS-1, though it
could be that I used tpsdot back then. But maybe mapdot calculation is
slower in MS-2, could this be possible?

I never really got TPS accel working, but that's probably since I'm getting
~15 TPS noise which can correlate to over 200Kpa/s mapdot values. I'll first
try swapping in a genuine Bosch sensor, the grounds and cable should be
pretty good quality since they're original OEM stuff.

[muythaibxr]

I actually saw that MAPdot lagged behind tpsdot in ms1/extra as well... I don't think this is specific to ms2/extra.

MAPdot definitely lags behind TPSdot on a quick stomp though due to things like the manifold filling effect. It physically takes longer for the actual pressure in the intake manifold to go from the old value to the new value.

TPS actually changes a bit ahead of the manifold filling, so by the time the engine needs the extra fuel, it gets it.

This is the basis for some code I'm going to try to work on soon that will use the TPS to determine when the manifold filling effect will occur, and to shoot a bit of extra fuel in ahead of what EAE and the VE lookup would do in order to get EAE started, and prevent the lean-spike on a hard stomp.

Ken

[mtlcafan79]

Is that going to be an MS2/extra feature, or still just part of MS3 initially?

[seth900]

What is your map lag set to? With the later versions of the beta code,
there is a filtering mechanism which allows a much less laggy map
signal to be used.

KeithG

MAP lag factor is set to 100, so that shouldn't be an issue. I've smoothed out the signal with small carb fuel filters instead.

You're using the same B202L Saab engine, right? Would be interesting to see your accel enrichment settings, if you're not suffering from lean spikes. I'm using a T7 head though, but it shouldn't make that big of a difference.

[muythaibxr]

Is that going to be an MS2/extra feature, or still just part of MS3 initially?

I'll try to get it in for ms2/extra 3.0, one of the later 2.x releases... It will most likely not be in 2.0

Ken

[muythaibxr]

What is your map lag set to? With the later versions of the beta code,
there is a filtering mechanism which allows a much less laggy map
signal to be used.

KeithG

MAP lag factor is set to 100, so that shouldn't be an issue. I've smoothed out the signal with small carb fuel filters instead.

using carb fuel filters will significantly slow down the response. You may want to try removing them and using a slightly lower lag factor.

Ken

[PSIG]

Is your issue only around idle? 55mm is considered a fairly small TB for a 2.0L and you should have no issues above idle with your MAP signal. Can you post your msq and a log showing the problem?

David

[rb26dett]

i have a 70mm throttle on my 2l and didnt even try to tune it for ae. i had no issues at all with it. i doubt its the throttle thats the issue. does the map line come off the plenum or one of the runners? that could be the cause of some wobbles in the map reading.

[rb26dett]

having thought about this for another few seconds, i also have a huge plenum. a smaller one would "fill" faster, so perhaps its actually a plenum volume to throttle bore ratio that matters? just a random thought for you. i saw a pic of that engine in a search and it might be about 1 litre or so internal volume. i have at a guess over 3l of volume there.

fred.

[thechuckster]

fwiw: i have an undersized plenum for my 18RTE Toyota and i get the same wild oscilating accel enrichment at low rpms.

I have great response on boost, but off boost, the car has similar behaviour (- bucking/stalling when trying to accel away from idle.

I have throttle about same size (60mm from memory) so after the throttle cracks open, there's very little change in MAP (and thus in AE) with further pedal effort until RPMs lift. This makes smooth driving in peak hour traffic difficult and i use a lot of fuel.

To get any reasonable off-boost accel response results in an AE configuration that's way too twitchy and prone to false triggering. It does suggest that its very difficult to tune 'around' design flaws in an induction system.

I think there are three ways address the problem - larger plenum (lots of engineering work), a smaller TB, or changing the actuating arm on the TB so the butterfly doesn't open linearly with pedal effort.

(fwiw 2: My tuning helper reckons i should put a better engine in that has a proper plenum)

[rb26dett]

possible reason is that oem usually uses maf or vaf and thus can afford to go smaller for response and still know whats going on accurately at all times, not just under steady state conditions.

having read that, i'm glad i made it so big. and of course the point was for minimal pressure change on each induction pulse, and thus maximum air injested per stroke (at high rpm and boost).

fred.

[muythaibxr]

(fwiw 2: My tuning helper reckons i should put a better engine in that has a proper plenum)

If you figure out a good EAE tune, that'll get rid of most of your problems... Then the only one left to solve will be quick blip response.

Ken

[seth900]

I'm running the MAP line from a vacuum port in the plenum of course, the same line that goes to the FPR, ignition system, boost gauge etc.

I read on the MS main forum that using small fuel filters instead of lowering lag factor would give me better response. So now I'm a little confused as to which actually is better. I suppose the idea behind the filters is to use them as small accumulators, or what is it?

But as far as the main problem goes, I did some experimenting this weekend and wired the injectors differently. Now all the injectors are on just one line just like the OEM system. This surprisingly cured 90% of my problems and I get pretty good response now. No bucking at all anymore, just a slight lean dip on somewhat faster stomps that only shows on the AFR gauge.

I'm at a loss as to why this made a difference, since I was running 2 squirts/simultaneous already. Shouldn't this run the injectors identically as a completely batch fire system? The injectors used to be wired as 1+4 to inj1 and 2+3 to inj2 earlier.