Relative newbie with a bunch of new mods

[guciodog11]

Hi everyone,

After dropping in a new longblock I threw on some new parts such as ID1000's and Perrin parallel rails. Otherwise the car has been tuned for a TD06-20G, FMIC, Helix 880's and a K&N SRI. The car is still unregistered as of now, and only plan on driving it around in CL of course.

To date I just scaled the injectors, set latency that I found online, threw on and scaled an AEM 3.5 bar MAP and disabled some CEL's. The car runs well on closed loop, but cold startup is kinda crap, with AFR's bouncing all over before the O2 kicks in.


I was wondering what the correct procedure is to begin retuning. I read a ton of guides that say to replace with stock injectors or an intake, but since I do not have access to either, I'm not sure on how to proceed.

Since the injectors are scaled already, can I skip over that and go directly to scaling the MAP? I don't know nearly enough to mess with fueling tables or timing, but I'm set behind the theory of tuning them.

I of course plan on getting a proper dyno tune, is there anything I can do to make their job easier? I know MAP scaling takes a long time with logging and the like so I feel like that's the only thing I'm semi-capable of doing as of right now.

Thanks!

* Registered users of the site do not see these ads.

[benflynn]

Set the CL to OL tables to zero, scale the MAF, adjust boost, and repeat. There are some good spreadsheets on romraider and a tuning guide here somewhere or there somewhere

[86Dreams]

MAF takes a long time too. I think I spent north of 7 hours on my MAF curve.

Anyway. Do you have actual Injector Dynamix 1000CC Injectors? If so the flow rate should actually be 995. What latencies did you set?

Did you scale the pulse width tables for cranking?

[guciodog11]

Thanks guys. I'm afraid to do full WOT runs just in case I do something dumb. I've been reading guides for the past few weeks so I know what I'm doing on paper, but I don't trust myself that much to beat on my motor. MAF from what I read is just a bunch of hours with RR logger and casual driving, I can def do that easily.

I was aiming just to stay in CL and out of boost to get the MAF down right, much less risk of blowing anything up that way ( i hope!)

Yup the injectors were BNIB right from ID, set the flow rate to 1k for now with a 05 WRX FPR.

I think I used:
8.00 10.00 12.00 14.00 16.00 (Volts)
2.60 1.69 1.25 0.98 0.79 (latency)

Only thing I modified with the cranking PW was reduced low temp (80C and below) values to that of warm values, and set idle 950 across the board to reduce that annoying high idle that's used for cat warmup)

Starts up within half a turn of the motor regardless of temps.

Only problem I have is the O2 heater circuit takes forever (2-3 mins) to activate and get the o2 up to speed. Initial OL idle is wonky, with AFRs jumping from super rich to super lean, right up until the o2 kicks in and goes to CL, then it's a rock solid 14.6. I'm assuming the crappy AFR's are due to an improperly scaled maf? Need to do a boost leak test, but I went over all manifold vac ports and it all looks solid, i hope my FMIC isn't cracked or something.

[86Dreams]

You cannot tune MAF in CL. you have to force open Loop. and you have to account for tip in enrichment. You have to account for the effect of any compensation table as well.

The idling issue could be due to a number of reasons. I had a similar issue with ID 1000s. I gained a lot of help in this thread on RR https://www.romraider.com/forum/view...p?f=15&t=13832
try the flow rate and latencies in that thread and see if it helps. they are in post 3 by jon7009

[guciodog11]

Originally Posted by 86Dreams

You cannot tune MAF in CL. you have to force open Loop. and you have to account for tip in enrichment. You have to account for the effect of any compensation table as well.

The idling issue could be due to a number of reasons. I had a similar issue with ID 1000s. I gained a lot of help in this thread on RR https://www.romraider.com/forum/view...p?f=15&t=13832
try the flow rate and latencies in that thread and see if it helps. they are in post 3 by jon7009

I was going off this guide, I thought it was possible to do CL MAF tuning.

https://www.romraider.com/forum/view...hp?f=33&t=5404

Thanks for the link, I appreciate it!

[86Dreams]

Originally Posted by guciodog11

I was going off this guide, I thought it was possible to do CL MAF tuning.

https://www.romraider.com/forum/view...hp?f=33&t=5404

Thanks for the link, I appreciate it!

I see, yeah thats a different procedure then. I never used the RR MAF tool. I did mine by hand by forcing open loop logging AFR & MAF and then rescaling. it takes a few iterations to get it dialed in that way. I dont have any opinion about how RR does it, but I guess follow the instructions.

[guciodog11]

Originally Posted by 86Dreams

I see, yeah thats a different procedure then. I never used the RR MAF tool. I did mine by hand by forcing open loop logging AFR & MAF and then rescaling. it takes a few iterations to get it dialed in that way. I dont have any opinion about how RR does it, but I guess follow the instructions.

I'll give it a try then thanks!

Looks like the guy in the romraider forums had the same problems as me, I'll mess with the scaling and see if I can improve things.

[JSR84]

Originally Posted by 86Dreams

You cannot tune MAF in CL. you have to force open Loop. and you have to account for tip in enrichment. You have to account for the effect of any compensation table as well.

The idling issue could be due to a number of reasons. I had a similar issue with ID 1000s. I gained a lot of help in this thread on RR https://www.romraider.com/forum/view...p?f=15&t=13832
try the flow rate and latencies in that thread and see if it helps. they are in post 3 by jon7009

That is incorrect. There is absolutely no issue with scaling a MAF in Closed loop. Tip in isn’t an issue if your process and data collection is correct. When you start reaching high MAF voltages that require you to go into open loop then you switch your process. But there is no reason idle, cursing and moderate load can’t be done in closed loop, it’s actually easier. The ECU does the math for you.


Anyways, trying to scale a MAF and dial in injectors is incredibly difficult. You need to be able to trust one set of parameters. If you can’t, you’re just guessing really. ID does provide good data for their injectors, but it’s not perfect and the error can cause issues.

Your cold start idle issues is most likely the initial cranking PW. That table exists to provide addition fuel to help wet the port wall when the engine initially starts. If you simply reduce that tables by the difference between your old injector scale and you new injector scale you’ve probably taken too much out. Try adding some back in, then observe the change. Make addition revisions as required. Just adjust the entire table together, and adjust A,B,C,D by the same multiplier.

The high idle you speak of is to warm the engine faster, not necessary the Cat. What they do is retard the ignition timing pretty substantially on a cold start to help warm the cat. This makes the exhaust note quite loud and can also cause oil dilution and cylinder wall wash. But they care more about emissions than those other things, hence why they do it.

Also, what year and model?

[guciodog11]

Originally Posted by JSR84

That is incorrect. There is absolutely no issue with scaling a MAF in Closed loop. Tip in isn't an issue if your process and data collection is correct. When you start reaching high MAF voltages that require you to go into open loop then you switch your process. But there is no reason idle, cursing and moderate load can't be done in closed loop, it's actually easier. The ECU does the math for you.


Anyways, trying to scale a MAF and dial in injectors is incredibly difficult. You need to be able to trust one set of parameters. If you can't, you're just guessing really. ID does provide good data for their injectors, but it's not perfect and the error can cause issues.

Your cold start idle issues is most likely the initial cranking PW. That table exists to provide addition fuel to help wet the port wall when the engine initially starts. If you simply reduce that tables by the difference between your old injector scale and you new injector scale you've probably taken too much out. Try adding some back in, then observe the change. Make addition revisions as required. Just adjust the entire table together, and adjust A,B,C,D by the same multiplier.

The high idle you speak of is to warm the engine faster, not necessary the Cat. What they do is retard the ignition timing pretty substantially on a cold start to help warm the cat. This makes the exhaust note quite loud and can also cause oil dilution and cylinder wall wash. But they care more about emissions than those other things, hence why they do it.

Also, what year and model?

Thanks for the info and help!

It's a '95 L with an 04 STi ECU and a built 257.

How long do the inital cranking PW table affect the PW? Maybe I'm misunderstanding the functionality of it, but my thought is that it is only relevant/applicable while cranking. Initial startup for me is perfect, one spin of the motor and it fires right up. Never had to crank more than a 1/5th of a second. The wonky Idle starts about 15-20 seconds after startup, and persists until the o2 heats up, ~ 2-3 minutes in. From what I've read this is far too long, and I've been down this rabbit hole for weeks now. No CELS's wiring and sensor is all good.

I'm also thinking its the injectors, that's why I'm hesitant to do any logging since I can't tune for both MAF and latency.

EDIT:

Looks like the scaling values I used started at 8 volts, while my table starts at 6!

Gonna use these values and see what happens: https://www.clubsub.org.nz/forum/ind...times-on-a-g4/

[86Dreams]

Originally Posted by JSR84

That is incorrect. There is absolutely no issue with scaling a MAF in Closed loop. Tip in isn't an issue if your process and data collection is correct. When you start reaching high MAF voltages that require you to go into open loop then you switch your process. But there is no reason idle, cursing and moderate load can't be done in closed loop, it's actually easier. The ECU does the math for you.

I completely disagre. Its easier to eliminate those variables all together, then open the throttle to steady state and observe commanded vs actual AFR then adjust. What are you doing subtracting AF Learning and AF Correction under CL operation? Give me a break, those are always going to be time delayed and averaged from some event or from multiple load cells, there is more inaccuracy from doing it that way and it will take you longer.

[benflynn]

Is it not how all the self tuning ecus work?

[JSR84]

Originally Posted by guciodog11

Thanks for the info and help!

It's a '95 L with an 04 STi ECU and a built 257.

How long do the inital cranking PW table affect the PW? Maybe I'm misunderstanding the functionality of it, but my thought is that it is only relevant/applicable while cranking. Initial startup for me is perfect, one spin of the motor and it fires right up. Never had to crank more than a 1/5th of a second. The wonky Idle starts about 15-20 seconds after startup, and persists until the o2 heats up, ~ 2-3 minutes in. From what I've read this is far too long, and I've been down this rabbit hole for weeks now. No CELS's wiring and sensor is all good.

I'm also thinking its the injectors, that's why I'm hesitant to do any logging since I can't tune for both MAF and latency.

EDIT:

Looks like the scaling values I used started at 8 volts, while my table starts at 6!

Gonna use these values and see what happens: https://www.clubsub.org.nz/forum/ind...times-on-a-g4/

I'll preface this by saying I'm not too familiar with the in's and out's of your gen of ECU.

The Initial Cranking PW tables can effect fueling right up until Closed Loop takes over. If those tables aren't high enough the initial fuel puddle isn't robust enough and will scavenge fuel from the proceeding injections. Since this extra fuel requirement isn't being accounted for, the fuel that actually enters the cylinder isn't of the appropriate mass and causes a lean condition.


Definitely ensure you have the correct data for your injectors, I would even go as far as confirming with Injector Dynamics that you're using the correct info, they're customer support is awesome. Outside of that it's hard to say exactly what's going on without a log, but it sounds like the Initial Cranking IPW to me.

Originally Posted by 86Dreams

I completely disagre. Its easier to eliminate those variables all together, then open the throttle to steady state and observe commanded vs actual AFR then adjust. What are you doing subtracting AF Learning and AF Correction under CL operation? Give me a break, those are always going to be time delayed and averaged from some event or from multiple load cells, there is more inaccuracy from doing it that way and it will take you longer.

That's fine, but you're disagreeing based off ignorance.
There are different ways to skin a cat. However, it's clear you don't understand what is going on in Closed Loop operation and how to use it to your advantage when calibrating.
The only thing you need to avoid are transients (Which you also need to avoid when using Open Loop), and if you know how to use excel, Mega Log viewer or any other data based program then you don't even need to worry about that. You create filters that help remove that data.
2 delays exist in the system. First, the time the ECU calculates the fuel mass to when that results hits the Oxygen Sensor, and guess what? That's present in both Closed Loop and Open Loop. The other "delay" would be the time it takes the ECU to take the input from the Oxygen Sensor and compare it to the target and apply the correction to the fuel mass calculation. When using the factory Oxygen Sensor, these delays are a non-issue. They happen so fast, the result won't adversely affect the process.
Calibrating using Open Loop does work, and it's necessary when trying to hit voltage ranges that are outside of what is possible in Closed Loop, but your statement about not being able to calibrate a MAF Sensor in Closed Loop is FALSE, plain and simple. Don't want to take my word for it? Fine, there are loads of examples and tools all over the place.

Please research the Dunning Kruger Effect and place yourself on the scale. I'm slowly coming out of the valley and even get knocked down a bit from time to time. I think you're on the "Peak" still.

[86Dreams]

Originally Posted by JSR84

I'll preface this by saying I'm not too familiar with the in's and out's of your gen of ECU.

The Initial Cranking PW tables can effect fueling right up until Closed Loop takes over. If those tables aren't high enough the initial fuel puddle isn't robust enough and will scavenge fuel from the proceeding injections. Since this extra fuel requirement isn't being accounted for, the fuel that actually enters the cylinder isn't of the appropriate mass and causes a lean condition.


Definitely ensure you have the correct data for your injectors, I would even go as far as confirming with Injector Dynamics that you're using the correct info, they're customer support is awesome. Outside of that it's hard to say exactly what's going on without a log, but it sounds like the Initial Cranking IPW to me.



That's fine, but you're disagreeing based off ignorance.
There are different ways to skin a cat. However, it's clear you don't understand what is going on in Closed Loop operation and how to use it to your advantage when calibrating.
The only thing you need to avoid are transients (Which you also need to avoid when using Open Loop), and if you know how to use excel, Mega Log viewer or any other data based program then you don't even need to worry about that. You create filters that help remove that data.
2 delays exist in the system. First, the time the ECU calculates the fuel mass to when that results hits the Oxygen Sensor, and guess what? That's present in both Closed Loop and Open Loop. The other "delay" would be the time it takes the ECU to take the input from the Oxygen Sensor and compare it to the target and apply the correction to the fuel mass calculation. When using the factory Oxygen Sensor, these delays are a non-issue. They happen so fast, the result won't adversely affect the process.
Calibrating using Open Loop does work, and it's necessary when trying to hit voltage ranges that are outside of what is possible in Closed Loop, but your statement about not being able to calibrate a MAF Sensor in Closed Loop is FALSE, plain and simple.

Please research the Dunning Kruger Effect and place yourself on the scale. I'm slowly coming out of the valley and even get knocked down a bit from time to time. I think you're on the "Peak" still.

Having tried both I think its much easier to do it in Open loop even if youre using software. This is because the engine operating state is easier to observe while youre logging. And the supplied AFR is going to be more consistent. Seeing the AFR stabilize to a value in concurrence with stabilization of the the load state is much more obvious than if you have 3 feedback controls affecting it.

I went at great lengths to apply compensations to see the true AFR at any time by accounting for the Fuel Trims, just like youre suggesting. In the end I found it to be much more straight forward to do it in full open loop.

When you do it manually with an external gauge there are fewer columns of information to consider and its very easy to determine how far back you have to offset the column of information from the wideband.

Even then when you apply software techniques to the problem you end up with data that is less obscured, and which leads to much simpler algorithms.

[guciodog11]

Thanks for the additional details, I'll definitely start playing with the cranking PW's now.

ID was super quick to respond with custom values for latency for everything I needed, here's what they sent me if anyone else is looking:

6.000 9.000 12.000 14.000 16.000
4.108 2.145 1.245 0.977 0.790

These values are for ID1000's, running 43.5 psi (stock) values, measured at idle with vac hose blocked off.

Hopefully I'll get an updated map uploaded today and see if it makes a difference!

[JSR84]

Originally Posted by 86Dreams

Having tried both I think its much easier to do it in Open loop even if youre using software. This is because the engine operating state is easier to observe while youre logging. And the supplied AFR is going to be more consistent. Seeing the AFR stabilize to a value in concurrence with stabilization of the the load state is much more obvious than if you have 3 feedback controls affecting it.

I went at great lengths to apply compensations to see the true AFR at any time by accounting for the Fuel Trims, just like youre suggesting. In the end I found it to be much more straight forward to do it in full open loop.

When you do it manually with an external gauge there are fewer columns of information to consider and its very easy to determine how far back you have to offset the column of information from the wideband.

Even then when you apply software techniques to the problem you end up with data that is less obscured, and which leads to much simpler algorithms.

Fair enough dude, the only thing I took issue with was your use of the word "cannot". Everyone has their own ways, I've never had an issue using closed loop. As I said, there's more than 1 way to skin a cat.

Cheers

Originally Posted by guciodog11

Thanks for the additional details, I'll definitely start playing with the cranking PW's now.

ID was super quick to respond with custom values for latency for everything I needed, here's what they sent me if anyone else is looking:

6.000 9.000 12.000 14.000 16.000
4.108 2.145 1.245 0.977 0.790

These values are for ID1000's, running 43.5 psi (stock) values, measured at idle with vac hose blocked off.

Hopefully I'll get an updated map uploaded today and see if it makes a difference!

Let us know how it goes.

[86Dreams]

Originally Posted by JSR84

Fair enough dude, the only thing I took issue with was your use of the word "cannot". Everyone has their own ways, I've never had an issue using closed loop. As I said, there's more than 1 way to skin a cat.

Youre not wrong, I should not have been so rigid about my opinion. You can use either approach. my personal experience was that I felt like i got more consistent results from full Open loop once i got comfortable with the process.

Originally Posted by JSR84

Cheers




Let us know how it goes.

Seconded. I can think of some other things to check if youre still having issues.

[guciodog11]

Went ahead and flashed with this calibration.
https://imgur.com/a/WvsgNTe

Pretty much the same issue, leans out just as hard, but maybe a few seconds shorter than before. RPM's drop by 250ish just like before during lean condition.

Here's a shot if basic parameters, if it helps.
https://i.imgur.com/UNf0fME.jpg

Did basic diag earlier:

-checked all o2 sensor wiring, sensor and heater loop for continuity (blew a main relay a year ago, just fired up again this year)
-brand new OEM FPR
-secured all vac hoses, def no leaks
-new AEM 3.5 bar-problem existed before and after install.
-new Denso OEM front o2

All I can think of is putting in another MAF, but don't want to spend money on it if I don't need to.

Thanks for the help guys!

[86Dreams]

So this is still to do with the idle right?

if so what does your MAF Compensation (IAT) table look like?
what does your Engine Load Compensation (MP) table look like.
What injector latencies did you end up using?


Also which 04 sti ecu? ADM JDM EDM or USDM?

[guciodog11]

Yup still rough OL idle, before CL kicks in after 2 mins cold start. After that it's solid 14.6 AFR.

for latency:

6.00 9.00 12.00 14.00 16.00
4.18 2.15 1.25 .99 .79

I'll have to pull up the MAF and Engine load comp tables, haven't touched them since my last tune 10 years ago or so.

It's a USDM ECU.

[86Dreams]

Did you try these?
6.5 9.0 11.5 14.0 16.5
3.58 2.092 1.332 0.98 0.752
do you have a log of the idle with the wideband?
did you change the warm up enrichment tables at all?

[benflynn]

Post ur rom

[guciodog11]

I'm working on getting wideband output into the ecu for proper logging but I'll get a video of startup today, along with the ROM. Damn work preventing me from diagnosing it. I'll post up both tonight!

Didn't change warmup enrichment from stock at all.

Thanks a ton for the help everyone, I really appreciate it!

[guciodog11]

Reflashed with:
6.5 9.0 11.5 14.0 16.5
3.58 2.092 1.332 0.98 0.752

Same result.

I did not have time to datalog, I def will tomorrow.

Below is my ROM along with definitions since i'm running Merp.
https://drive.google.com/file/d/11Is...ew?usp=sharing

"subaru standard" folder is for ecuflash, ecu_defs_merpmod.xml is for Romraider

Here's a screenshot of the MAF comp and engine load comp tables:
https://imgur.com/PZXgqfE


Thanks!

[86Dreams]

I knew one guy that sorted this out by adjusting the intake temperature compensation table.

Your observing a hunting idle right?

I would get this on my 02 after tgv deletes on a usdm ecu. then when I had a jdm motor i had a jdm ecu and i would get a hunting idle on the usdm ecu but not the jdm.

I never quite figured it out but always suspected that part of the problem is hard coded into the way the feedback control with AF correction works. essentially what I think happens is at cold start the AF correction routine destabilizes itself and so it over corrects one way then back the other way then back the other way and so on until its warmed up. for me tapping the throttle to break it out of the warm up phase would put it into a regular idle.

This is part of why I have the opinion that I do about scaling the MAF in OL.


But a log of more than anything might be revealing.
personally id like to see RPM, MAF, AF Sensor 1, AF Sensor 1 Current, AF Sensor 1 Resistance, CL/OL Switch, Final Fueling target, coolant temp, intake temp, MAP, injector pulse width

this thread has some interesting discussion
https://www.romraider.com/forum/view...p?f=15&t=13088