Mystery CEL that I can't shake

[Dhan]

The problem
CEL at around 3k rpms.

What isn't stock
Balance shaft eliminated
Supra SMIC w/hard pipes
Small 16g (set to stock boost)
Open downpipe
ECMlink V3
LC-1 WB

Extra info
No boost leaks
Doesn't burn coolant or oil
Healthy engine (don't remember numbers, but it passes a compression test)
Timing is good to go
No a/c or radio

Background
So for a while now I've had this issue where city and highway driving, at around 3k rpms, the CEL will illuminate. It'll stay on if I keep the revs around that 3k mark. The car's performance doesn't seem to change.
The car came with a hacked up MAF. I replaced it with a stock one, but that didn't change anything. So I hooked up ECMlink but the ECU won't save the CEL code(s).
I added a wideband recently in an effort to narrow things down, but it didn't really help me.

Where I'm at now
Apart from what's stated above, I've put the car back to its stock configuration. Turbo is set to OE level, OE 450cc injectors are cleaned and reinstalled, MAF has been switched with a working one (from another of my DSMs), and vac lines are good to go. Plus, I put everything back to its defaults in ECMlink, so the car is running factory rich (I'm assuming).

So, I decided to make a new log in hopes someone here could see why there's these CELs all the time.

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

Thanks for reading

 

[1davho]

You probably have dsmlink set up to illuminate the check engine light with so many degrees of knock retard. I can't look at your log because I don't have my laptop with me, but log knock retard and see how many degrees you have around the time that it comes on. If you are getting a lot of knock, back off timing in that area a degree or two and see if that helps.

 

[Dhan]

Ah, okay! I see it now. There's several spots where there is more than 5* under the KnockRet tab, so you must be correct. I didn't realize that was a feature of Link.

Thank you so much for your help sir!

 

[1davho]

No prob, its a good feature to leave on. Knock is detonation and if u get too much knock and don't tune it out, that is how you melt pistons and valves. Good luck. Just read all the forums and play around with it, u will catch on.

 

[iaintmaci]

Do yourself a favor before you start changing a bunch of timing problems. Pull out your knock sensor and take a look at it. Make sure that all of the goop inside is not gone. If it is, either replace it, or try to put some rtv silicone in it to see if it helps. I pulled so much timing trying to get rid of the knock that my wife's Neon would leave me sitting at a light. Pulled my knock sensor one day and almost all of the filling had come out. A dab of silicone and it works great again. I was able to run the stock timing maps without any knock at all.

 

[Dhan]

^^ Good point! I was pretty sure the PO replaced it but I checked anyway.

Looks all right to me.....

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

I've been going into the LIVE ECMlink Settings tab and tweaking the Timing Adjustments Table, one notch/cell at a time, then pressing "Save all to ECU". Hard to discern if it's doing anything though.
What's considered too extreme of an adjustment? I don't want to adjust the timing to the point that I damage something. Here's where I left it before the laptop died on the highway....

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

Like I stated above, CEL (5*+ of KnockRet) comes on at about 3k rpms +/- 250. So that's where I was adjusting things. Also trying to keep things smooth between the cells.
This is the only feature I'm changing right now. The idle is about perfect and in line with the videos and wiki info on ECMlink's website. Haven't tried any WOT driving.

PS. I am keeping up with the boost leaks (and fixing them).

 

[iaintmaci]

I never mess with the sliders. I always look at the timing maps. I track the datalog to see where I am knocking and then start to adjust from there. To me, it seems that it is a cleaner, smoother, and more accurate control that way.

 

[c2ypt1c]

I've been going into the LIVE ECMlink Settings tab and tweaking the Timing Adjustments Table, one notch/cell at a time, then pressing "Save all to ECU". Hard to discern if it's doing anything though.
What's considered too extreme of an adjustment? I don't want to adjust the timing to the point that I damage something. Here's where I left it before the laptop died on the highway....

If you're trying to tune out knock, you want to lessen the timing advance. You're making the probelm worse by advancing the timing (raising the sliders). Watch this video:
http://www.ecmtuning.com/demos/basictune.html

The timing sliders have a global effect on the timing map, whereas the direct access table allows for pinpoint accuracy for load vs rpm.

The 1g timing map is notorious for being too advanced, while the 2g map is the perfect starting point. If you're getting knock outside of WOT, try using the 2g map:
v3configs [ECMTuning - wiki]

and here's a direct link to the 2g map:
http://www.ecmtuning.com/images/forums/v3configs/DSM-2G-stock.eda

^ This should get you where you need to be

Also, having the knock sensor active at 0% tps might be giving you some phantom knock. Try raising it up to about 10%

There are some other issues I see in your log that you should be aware of:
- you're not going into closed loop because your wideband is wired into the front o2 pin and you're not simulating a narrowband signal. This is making you run super lean during cruise and idle. Not to mention your airflow (mafcomp) isn't dialed in.

To enable narrowband simulation, go to the ECUinputs page. From the "Wideband(WB)" drop-down menu, select FrontO2. Copy to ECU.
Then go to the NBO2sim page and check "enable narrowband simulation."

- your tps is a little off. tpsvolts should be around .63v at 0% (use the tps adjust tool)

For more details read my guide:
http://www.dsmtuners.com/forums/articles-tuning-ecu/448002-tuning-ecmlink-v3.html

 

[Dhan]

I never mess with the sliders. I always look at the timing maps. I track the datalog to see where I am knocking and then start to adjust from there. To me, it seems that it is a cleaner, smoother, and more accurate control that way.

If you're trying to tune out knock, you want to lessen the timing advance. You're making the probelm worse by advancing the timing (raising the sliders). Watch this video:
http://www.ecmtuning.com/demos/basictune.html

The timing sliders have a global effect on the timing map, whereas the direct access table allows for pinpoint accuracy for load vs rpm.

I see what you guys are saying now with the difference between the tables versus the sliders. Much more accurate!

The 1g timing map is notorious for being too advanced, while the 2g map is the perfect starting point. If you're getting knock outside of WOT, try using the 2g map:
v3configs [ECMTuning - wiki]

I copy/pasted the 2g map and I could tell the difference. Still needs some fine tuning (which I expected), but you're right, a big step in the right direction.

You mentioned a few other things.

• Adjust the knock sensor activity from 0% to 10%, I did that.
• O2 not being simulated through the wideband. I could've sworn I already did this when I setup the wideband! Apparently not. Took care of it and the O2 is cycling like it should between .20V-.80V +/- a little.
• You mentioned dialing in the MAFcomp...I'm trying to read into this a little more and I watched the video. It says to start by doing a cruise log, applying the template, then make adjustments from there. So that's what I've been doing.
• Messed with the TPS quite a bit. I get it dialed when the car's off, but when the car's idling, either the throttle position will be off or the TPS voltage will be off. I can't quite get it to 0% ThrottlePos and .63V, it'll always be +/- .1V or .2V. I'm wondering if the TPS itself needs some attention?

http://www.dsmtuners.com/forums/articles-tuning-ecu/448002-tuning-ecmlink-v3.html

Read through this as well and followed a lot of the links. Very helpful! Things are coming together now and I'm slowly getting a handle on things in ECMlink.

So after applying the 2g maps, I reset the sliders to zero and did another cruise. The 2g timing lowered the counts of KnockRet, but there was still some present. So I began using the Track Datalog in the TimingMaxOct table to tweak the trouble spots; usually only one or two cells at a time.
Like I said above, I'm also using the MAF Comp sliders to dial in the airflow properly. Which seems to be the other major component to getting a "good tune".

I had a couple questions though....
1. When I start to tune WOT logs, do I use the OpenLoopMax and Min tables?
2. Why are the timing sliders there if we can adjust everything with the tables? What does "global" mean in relation to what we're doing?

Also wanted to say thanks for all the help so far guys. I can handle your everyday mechanical issues just fine, but the ECU aspect is foreign territory for me Thanks again! Really appreciate the assistance

 

[c2ypt1c]

Don't get too caught up with knock outside of WOT. Many people just ignore it all together and activate the knock sensor at 50% and/or higher rpms. If you're running a 2g map you should be ok. Like mentioned previously, it could be phantom knock caused by low rpm vibrations, etc. An easy way to tell...run e85

I had a couple questions though....
1. When I start to tune WOT logs, do I use the OpenLoopMax and Min tables?

The min tables are pretty much useless (active with lots of knock for a long period of time), unless you are running dual maps for meth or something. The max tables are what you want.

2. Why are the timing sliders there if we can adjust everything with the tables? What does "global" mean in relation to what we're doing?

Sliders are helpful to make quick adjustments when tuning. With DA, you have to turn the car off, reconnect, save the changes, etc.

 

[1davho]

Don't run e85 with stock injectors. Bad idea