ECMlink Help with tuning SD. What next?

[enriquez2000]

I was trying to read through to see what your base idle is. It looks like iys idling almost at 1100 rpm based on your picture.

You may want to try adjusting fcoffset by 200rpm to compensate and see how that does for your catching of the idle.

If youre not targetting 1100rpm then you have other issues

 

[justacanuck]

I was trying to read through to see what your base idle is. It looks like iys idling almost at 1100 rpm based on your picture.

We have tried a variety of idle rpm settings. It seems to like 850-900 best.

You may want to try adjusting fcoffset by 200rpm to compensate and see how that does for your catching of the idle.

I've also tried various settings for this, typically between 100 and 200. I haven't found that it provides any significant improvement or really any difference at all. I would love to learn more about how this feature is supposed to work.

If youre not targetting 1100rpm then you have other issues

Yes, we are definitely having many other issues

Given the parameters in TmngMaxOct table, why does the ECU vary the timing so much in idle conditions (i.e from 5-20 degrees?)

What causes it to go from idling fairly smoothly to surging during the same drive cycle? (better example log attached)

What else might cause the system not to want to run at 5 degrees BTDC but still seems to run quite well in other conditions?

I'm reading some of the documentation again... I had recently tried setting the idle a little bit lower. Maybe I need to set my idle back higher, and or try settings the FCoffset parameter back to zero, or at least closer to it.

 

[We're on Boost]

I would love to learn more about how this feature is supposed to work.

Basically, your ecu cuts the fuel to 0 (InjOn) whenever you take your foot off the gas (IdleSw=1) and your rpm is greater than your target idle + FCoffset. (There is actually a delay of about 0.7 seconds to do this cut).
Then as your rpm drops (foot still off the gas), the fuel is supposed to come back on when rpm drops below Target idle + FCoffset. This happens with no delay, very fast.

I like to keep this switch 500 rpm above my target idle so there is plenty of time for things to get with it before the rpms get too low. So my Coasting FC Offset is set to 500. Actually English Racing set it to that right at the start when they had the car, and I've just kept it that way.

On my car then, target idle is 1100, FC offset is 500, so when I'm coasting down with foot off the gas my fuel comes back on at 1600 rpm. If I'm driving along at 1500 rpm and take my foot off the gas, there is no fuel cut at all.

The timing is also jumping around way more than I would expect.

My timing is very jittery like a sawtooth shape whenever IdleSw is 1 and I'm below my FC rpm (1600). But it's smooth when IdleSw is 0. And it's smooth when IdleSw is 1 but I'm above 1600 rpm.
I don't know why the idle timing is so jittery. But I'm very glad that timing is smooth when at high power or even just at higher rpm. The kind of "smooth" I'm talking about is like straight horizontal lines without the sawtooth rubbish.

I haven't looked at your latest logs enough yet - probably tomorrow night. After I get back from my drive to get some E85 which is 43 miles away!

We should have at least one log from you that logs the whole run from cold start to the end. Log should start before cranking and the ISCposition should be 120 there. Then at the end, keep the log running for 5 seconds or so after key off. ISCposition should go to 90 there, after key off. It's ok if it's a long log. You can add words to the file name that call attention to things at XXX seconds in the log, or whatever.

 

[Justin DuBois]

Closed loop has very active timing adjustments to try and improve emissions and smooth out the idle, but the more modified the setup, the more random this appears - and SD mode is a hard mode to get to idle smooth anyway. MAF idle has always been better for me. One thing I did look at is the timing cells in the DA table targets for the idle range of operation. I found the (edit) 2G tables transition away from 5 deg right away, and this was not helping the idle hunting. By reverting those cells (edit) to the EVO 8 defaults it, did help. More of the cells were set to target 5deg in the idle rpm range, and this took some lumpy behavior out of the idle area. If you are running non-stock cams, this gets even more sensitive.

The other thing is the injectors. Some larger injectors just don't like the low end. Stock sized injectors also idle better than anything else. I'm actually happy that I was able to get 1000cc RC low-z injectors to idle well enough, as I hear these have been challenging to work with historically.

 

[justacanuck]

Basically, your ecu cuts the fuel to 0 (InjOn) whenever you take your foot off the gas (IdleSw=1) and your rpm is greater than your target idle + FCoffset. (There is actually a delay of about 0.7 seconds to do this cut).
Then as your rpm drops (foot still off the gas), the fuel is supposed to come back on when rpm drops below Target idle + FCoffset. This happens with no delay, very fast.

I like to keep this switch 500 rpm above my target idle so there is plenty of time for things to get with it before the rpms get too low. So my Coasting FC Offset is set to 500. Actually English Racing set it to that right at the start when they had the car, and I've just kept it that way.

This is a very useful explanation. I just changed mine to 470. It does seem to be helping. Log attached.

I haven't looked at your latest logs enough yet - probably tomorrow night. After I get back from my drive to get some E85 which is 43 miles away!

We should have at least one log from you that logs the whole run from cold start to the end. Log should start before cranking and the ISCposition should be 120 there. Then at the end, keep the log running for 5 seconds or so after key off. ISCposition should go to 90 there, after key off. It's ok if it's a long log. You can add words to the file name that call attention to things at XXX seconds in the log, or whatever.

Unfortunately, I didn't read your entire post before I started logging today, so I didn't capture the ISC position before startup. However, I did capture the keyoff, and we can verify that the ISC goes to 90 after the keyoff.

I have attached the long log from this morning's warm-up cycle. After it got to normal operating temperature, I tweaked SD (bringing up VE around the idle area: was ~52, now set closer to 58), and I reduced injector deadtime to bring back fuel trims at idle (deadtime was 115, now down to 50). I also adjusted the idle rpm (increased closer to 1000rpm), and tweaked BISS so ISC settled at 30, but you can see the recurrence of the problem I was dealing with several times. (log 17-01.elg)

The smaller log 17-02.elg contains the settings when I finally quit. It seems to work much better now, possibly almost "fixed".

I do believe that setting the coasting FCoffset far beyond the recommended settings helped to improve the performance in this situation. Thank you so much for sharing your advice and experience.

 

[justacanuck]

Closed loop has very active timing adjustments to try and improve emissions and smooth out the idle, but the more modified the setup, the more random this appears - and SD mode is a hard mode to get to idle smooth anyway. MAF idle has always been better for me. One thing I did look at is the timing cells in the DA table targets for the idle range of operation. I found the EVO 8 tables transition away from 5 deg right away, and this was not helping the idle hunting. By reverting those cells back to the 2G defaults it did help. More of the cells were set to target 5deg in the idle rpm range, and this took some lumpy behavior out of the idle area. If you are running non-stock cams, this gets even more sensitive.

I had increased the timing on the top left corner of the table to mostly 12s, ie. in between Evo8 and 2G base settings.

Next time I work on this, I do believe I will try increasing the base timing around the idle area a few degrees more i.e. closer to 15 degrees like the 2G timing at 1000rpm. This car does not seem to like only 5 degrees ign advance at idle, at all.

 

[Justin DuBois]

Woah - ok I said it backwards - and I edited the post - EVO 8 timing tables were better for me with more 5's-7s in the idle cells:

You must be logged in to view this image or video.

The 2G timing table was causing more idle hunting, as the transition from 800 to 1000 was a big jump from 5 to 10-12 deg

You must be logged in to view this image or video.

I'm suggesting a smaller transition in timing will help with non stock idle vac and higher rpm's.

Justin

 

[justacanuck]

Woah - ok I said it backwards - and I edited the post - EVO 8 timing tables were better for me with more 5's-7s in the idle cells:

You must be logged in to view this image or video.

The 2G timing table was causing more idle hunting, as the transition from 800 to 1000 was a big jump from 5 to 10-12 deg

You must be logged in to view this image or video.

I'm suggesting a smaller transition in timing will help with non stock idle vac and higher rpm's.

Justin

I appreciate the clarification. We have taken steps to minimize the hunting by keeping the differences in that area smaller.

This car doesn't seem to like the timing so low at idle, so I'm thinking we might have to try moving it a little higher.

There is also a reasonable possibility that this problem got worse for us because we moved the intake cam advance from 6 degrees advanced, back to 3 (i.e. closer to stock). Maybe doing that contributed to making this issue worse in our specific application. It's also possible that the cams in this car are not stock, and the apparent 6-degree advance that was set on the intake cam wasn't actually advancing the cam, but potentially the 6 degree setting was just degreeing the cams properly. (i.e. I understand that cams, particularly aftermarket cams, may or may not be degreed very accurately.) They guy we bought it from did say it had been rebuilt previously, so it could very well have aftermarket cams.

I still really don't understand why this car doesn't seem to want to idle with only 5 degrees of advance. Whatever is causing this must be related to the idle quality challenges on this car.

I'm anxious to get to the bottom of it because the next car I will be working on is very likely be even more challenging with HKS272 cams, bigger injectors, bigger turbo, stroker etc.

 

[enriquez2000]

I appreciate the clarification. We have taken steps to minimize the hunting by keeping the differences in that area smaller.

This car doesn't seem to like the timing so low at idle, so I'm thinking we might have to try moving it a little higher.

There is also a reasonable possibility that this problem got worse for us because we moved the intake cam advance from 6 degrees advanced, back to 3 (i.e. closer to stock). Maybe doing that contributed to making this issue worse in our specific application. It's also possible that the cams in this car are not stock, and the apparent 6-degree advance that was set on the intake cam wasn't actually advancing the cam, but potentially the 6 degree setting was just degreeing the cams properly. (i.e. I understand that cams, particularly aftermarket cams, may or may not be degreed very accurately.) They guy we bought it from did say it had been rebuilt previously, so it could very well have aftermarket cams.

I still really don't understand why this car doesn't seem to want to idle with only 5 degrees of advance. Whatever is causing this must be related to the idle quality challenges on this car.

I'm anxious to get to the bottom of it because the next car I will be working on is very likely be even more challenging with HKS272 cams, bigger injectors, bigger turbo, stroker etc.

Honestly if youre not sure whats going on with the cams you are going to be fighting this forever. Guessing will only keep adding to your frustration.

If it were me Id redo all timing and make sure my cams were degreed properly.

Cant fix mechanical issues with tuning. You can hide some but not fix

 

[Justin DuBois]

I've got a set of Web's I'd love to try - but the last time I tried with stock timing gears, it idled like a big block motor boat, loaping along. had to pull them out. Now with timing gears I could go and try again.

I wonder - is there a lobe on the cam that would point up and peek at TDC? That would be too easy.

 

[We're on Boost]

The smaller log 17-02.elg contains the settings when I finally quit. It seems to work much better now, possibly almost "fixed".

The 17-02 log looks pretty decent.
I've been looking at the part after the throttle blip where the rpm gets down into the 700's at 8.4 seconds.
At the same place the wideband gets maxed out full lean (even more than maxed out - it hits 4.14 raw volts )
Maybe you could say that it's all weak there because it is still recovering from the quarter of a second that the InjOn was 0, back a second earlier (from 7.282sec to 7.521sec).
And 750 rpm isn't so bad. It's practically stock idle speed.
Like you've mentioned, the ISC doesn't respond to it at all. I'm not so sure that the ISC is supposed to respond to things that are that short of a duration though. It doesn't seem like it from most of what I've seen in other logs of mine and other people, but I haven't quite figured it out.

The throttle blip itself got you a TPS Delta of 2. Wow! Congratulations! It actually did something too. InjOn time about doubled a few ms later, and that stayed for a couple tenths of a second.

 

[We're on Boost]

There must be another idle control timing table built into the ECU that we simply cannot see.

I just noticed that the ? help on the RPM/TPS page says that timing at idle is varied to help control idle speed. Not just the ISC. So maybe this is where the unexpected timing numbers are coming from, and why they don't match the chart we can see. I think it would be interesting if there was an option to turn that bit of "Smart" off and make it just obey the table that we can see and adjust our selves.

You must be logged in to view this image or video.

 

[enriquez2000]

I just noticed that the ? help on the RPM/TPS page says that timing at idle is varied to help control idle speed. Not just the ISC. So maybe this is where the unexpected timing numbers are coming from, and why they don't match the chart we can see. I think it would be interesting if there was an option to turn that bit of "Smart" off and make it just obey the table that we can see and adjust our selves.

You must be logged in to view this image or video.

This would be a question/request for [email protected]

Its been a while since Ive seen anything new added but who knows

 

[Justin DuBois]

This would be a question/request for [email protected]

Its been a while since Ive seen anything new added but who knows

Light bulb - the DA timing table is used for Openloop, and I think it must also affect the closed loop targets.

 

[99dsmer4g63]

After it got to normal operating temperature, I tweaked SD (bringing up VE around the idle area: was ~52, now set closer to 58)

What’s the airflowperrev now? I’m assuming you inflated your actual numbers.

-Daniel

 

[justacanuck]

What’s the airflowperrev now? I’m assuming you inflated your actual numbers.

-Daniel

I worked to keep airflow/rev the same, 0.25g/s

 

[justacanuck]

Light bulb - the DA timing table is used for Openloop, and I think it must also affect the closed loop targets.

Definately DA timing affects closed loop.

But something else is also acting on it.

 

[justacanuck]

UPDATE:

Recap: We had retarded the intake cam, from 6 degrees advance, closer toward "normal", and set it to 3 degrees advance.

I had to increase VE several % to get airflow back to 0.25g/s. The vacuum at idle with this new cam timing was slightly more robust, and after changing the settings a bunch, e.g. coasting FC offset, I did manage to get it running a little better, but the problem kept coming back.

I was convinced that this new cam timing wouldn't solve the idle problems and that it was possibly worse. Unfortunately, we didn't do a WOT pull with this new cam timing, but I'm not sure that would have been a good idea. It seemed to want to knock more, making us a little nervous, so we changed direction and advanced the intake cam 3 degrees again (i.e., back where it was set before @ 6 degrees intake cam advance).

Obviously, this changed the base timing, so we used the "ground timing connector to ECU" feature. It idled great at 10BTDC. I had to use the throttle a little to finish adjusting the cam/ign trigger to set base timing back to 5 BTDC because it would hardly run at 5 degrees. As soon as we uncheck the "ground timing connector.." the timing jumps from 5 degrees to closer to the DA timing table. However, it still fluctuates way too much, in my opinion, fluctuating from ~9 degrees to ~20 when that part of my table is fixed around 14 degrees. I managed to capture a good log of the behaviour while we set the timing. (attached below)

My son says it is running pretty well again, but I haven't logged much since we did this, so I'm not sure I even fixed the VE around idle yet. But i do believe setting the coasting FC offset at ~500 makes significant and noticable improvement. But I also still have to re-check for boost leaks.

As you can see in the log, it still runs extra rich around idle right now. It started snowing the day after we reset the cam timing and I haven't gotten back to it yet.

I'm really starting to wonder if I should try adjusting the DA "load factor." The help says, "This directly affects the calculation of LoadFactor used, among other things, as an index into the main timing and fuel tables. LoadFactor can be considered the ECU's idea of cylinder pressure (engine load)."

Recall I had to set my open loop thresholds way down to avoid staying in closed loop while also in boost. The engine goes from vacuum to boost at a load factor of 0.7. Perhaps having a larger/faster turbo and other modifications would justify changing this setting?

To me, it almost seems that the ECU is behaving like there are another couple of rows of data below the 0.3 load factor in the DA timing table that we don't see in the ECMLink.

I have already ordered a degree wheel because we are going to have to use it while building the other car (a stroker with 272 cams)

Is it even possible to check/verify cam timing while the engine is still in the car?

 

[We're on Boost]

As soon as we uncheck the "ground timing connector.." the timing jumps from 5 degrees to closer to the DA timing table. However, it still fluctuates way too much, in my opinion, fluctuating from ~9 degrees to ~20 when that part of my table is fixed around 14 degrees.

Well that part of the log, from 173 sec to the end, looks good to me. I can even see there that when rpm starts to drop a little, that is when timing jumps up to a high advance. Which would help pick up the rpm. So it's like it says in the help, timing at idle is varied to help control idle speed. It's not all up to the ISC, which probably can't really change that fast anyway.

As you can see in the log, it still runs extra rich around idle right now.

It isn't really running rich though because the -10% combined fuel trim takes care of it. 10% + or - is well within the range of what closed loop can do. My car idles with combined fuel trim of about -14% after the car is fully warm, like after 20 minutes of running, and it seems fine.

 

[Justin DuBois]

Well that part of the log, from 173 sec to the end, looks good to me. I can even see there that when rpm starts to drop a little, that is when timing jumps up to a high advance. Which would help pick up the rpm. So it's like it says in the help, timing at idle is varied to help control idle speed. It's not all up to the ISC, which probably can't really change that fast anyway.


It isn't really running rich though because the -10% combined fuel trim takes care of it. 10% + or - is well within the range of what closed loop can do. My car idles with combined fuel trim of about -14% after the car is fully warm, like after 20 minutes of running, and it seems fine.

Have you tried adjusting your global dead time to dial in the warm closed loop fuel trims for idle closer to +/- 2% This should have little to no effect on the WOT tune. Then you can use the Combined FT ve table adjust to set the SD table. In theory, this will give you better open loop values in the ve table for cold start.

 

[justacanuck]

It isn't really running rich though because the -10% combined fuel trim takes care of it. 10% + or - is well within the range of what closed loop can do. My car idles with combined fuel trim of about -14% after the car is fully warm, like after 20 minutes of running, and it seems fine.

Like Justin said, my understanding is thay we should tune to get combined fuel trim closer to zero at idle, when fully warmed up.

 

[1990TSIAWDTALON]

Within +-5% on STFT and LTFT and it should idle great at stoch, 14.7 AFRs.
Also, on 91 octane, you will be lucky to get more than 5-7* of timing before knock comes in. How much boost/compression ratio will also determine if it wants to start pinging or not.

 

[We're on Boost]

I'm having trouble figuring out where we can purchase replacement knock sensors in 2024. I've read some older threads, and read a few mentions of alternate Mitsubishi part numbers that could work.

If you end up trying one of the non-OEM knock sensors that came up in this thread, or even trying to verify your old one for good or bad, we might be able to evaluate how close it is to a known good OEM sensor by comparing its behavior to logs like this where the knock sensor voltage has been logged along with the ecu computed "Knock Sum" which is the same number we see in ECMlink as RawKnock. I have several logs like this. In this particular one I had a lot of knock. Most of the other logs were more like everything was normal.
So keep that in mind for something that could happen! You would have to be able to log knock sensor voltage. On a 1g ECMlink I think the best input for that would be the IAT - so you'd have to temporarily quit using it for Speed Density.

Degrees of retard = Knock Sum x 0.352 (about).

You must be logged in to view this image or video.

 

[1990TSIAWDTALON]

Put a toggle switch on the IAT input and switch between the two.

 

[We're on Boost]

On a 1g ECMlink I think the best input for that would be the IAT

I'm thinking the IAT input might not work at all for this, because maybe the knock sensor voltage is ac rather than dc. So then an ac voltmeter might be more like what you'd need (Vac on a multimeter). I can try to check this out sometime in the next week or 2 here on my car. I still have the ecu+ that made my old logs, and I'm going to run the knock sensor wire to it again pretty soon. When that's going I can see what the Vac and Vdc scales on my multimeter read on it - if either one of them agrees with the ecu+ which for sure has a proper input for this voltage measurement.