1991 Eagle Talon Tsi AWD

[justacanuck]

This is the Build Thread for 1991 Eagle Talon Tsi AWD.

We dragged the car out of the back alley, covered in snow and it had not run for several years. When we got home, we put some fresh fuel and a battery. Cleaned up some connections and to our surprise, it started right up. I ran like crap and eventually, I smelled something burning. Turns out the ISC that I plugged in was completely seized up and plugging it in had fried parts in the ECU. I wasn't too concerned, because I wanted to upgrade the ECU anyway, so we proceeded to send it away for an upgrade by ECMTuning.

While we were waiting for the ECU to get repaired over Christmas, we cleaned up the throttle body, blocked off the FAIV circuit, and replaced the ISC controller. Once we got the ECM back, we found the TPS sensor was also faulty and would not return to zero in extreme cold so we replaced that as well.

Despite the weather being -10C (17F) we managed to get the car started. Brought it up to operating temps, and tweaked the BISS settings and increased the idle speed to get it idling a bit smoother with ISC in the 30 range while idling. Once I felt that we had to most basic settings configured, we took it for a drive around the block. It seemed to be running OK, it was getting late, and the weather was getting worse so gave up for the night.

The next day, the temperature had dropped significantly (i.e. -23C or -9F) and we could NOT get the car to start again. I found out that this car has one of those old MAFTranslators. I was set to 5, F, E, B. I figured it would be better to disable any adjustments the MAFTranslator and set them all to ZERO. I thought it best to work this direction, at least until we can remove the translator entirely and switch to native MAF settings in ECMLink, but so far this has not improved the start-ability in this very cold weather.

I've been reading for a few days, tweaking the odd setting on the car, and trying many times to get it started again but still not having any luck. So far we have:
- Adjusted the lowest of the MAFComp sliders (e.g. Minus ~18) to try and remove what appears to be some excessive air in the 0-100Hz range (AirflowPerRev is just over 1gm/rev on cranking.
- Adjusting the CoolantTempFuelAdj at -2F to be closer to the setting in the 19F column (ie. ~32 instead of ~47)
- Adjusted the CrankingFuelAdj similar to above (-2F set to ~106)

The changes above do seem to have created the intended effect of removing some excessive AIR from the logs, and I was guessing we were probably getting some excess fueling (injOn time is now closer to 8ms during cranking instead of +10ms).

Unfortunately, I'm still not having any luck, so any advise regarding settings for VERY COLD START environment would be very welcome.

 

[justacanuck]

I really don't know what I should reasonably expect for log data on a very cold start. Maybe I just have a bad front O2 sensor?

 

[pauleyman]

I really don't know what I should reasonably expect for log data on a very cold start. Maybe I just have a bad front O2 sensor?

cant be o2. ECU doesn't look at o2 when the engine is cold.

 

[justacanuck]

cant be o2. ECU doesn't look at o2 when the engine is cold.

Thanks for that. That is what I thought.

I just realized I'm not seeing CrankingFuelAdj-ustments in the logs either. I also can't find any "Key Start" signal in the logs. I was definitely cranking.WTF I don't understand why that would happen.

 

[justacanuck]

I think we may have found the problem, or at least narrowing in on it.

This thread shows a picture of the potential harness/location for the problematic circuit, or at least his similar problems were solved. I'll certainly check the connections around that area.

https://www.dsmtuners.com/threads/my-crankingfueladj-seems-to-not-be-working-at-all.540792/

It sounds like that wire/location is supposed to supplying 12V to Pin 108 on the ECU.

According to some older documentation, pin 108 from ECU goes to a device which may actually be called the "Inhibitor switch". I hope this problem isn't tied into other features that may be disabled, like the neutral safety switch, clutch pedal switch etc .

 

[justacanuck]

Back probing the connections from the start relay switch indicates power from relay to Pin 108 on the ECU but no "KeyStart" signal. We swapped in another ECU and surprisingly, we got KEY START when cranking. It still doesn't start but that is obviously another issue. I'm guessing it is too rich, now that we have extra fuel from CrankFuelAdj with "keystart" signal. I'm going to turn down the crank fuel adjust and see if that makes it better. I may try a factory ECU instead of the ECMLink modified ECU, just to see if it will start.

 

[justacanuck]

Tried the factory ECU and got it to start. It didn't start easily, but it started.

I also found the manual for the MAF-Translator so we will tweak those settings and seen if the car starts easier and runs better.

We will run the car like this, at least for a little while until the weather warms up. Eventually, I will remove the MAF translator from the equation, and install a new MAP sensor, IAT and LC2 wideband controller.

I plan on sending back the broken ECU for further diagnostics/repair.

 

[justacanuck]

Adjusted the MAF Translator to better match our injectors, and the Factory ECU started the car perfectly today at -14C.

So, I'm still not sure what to think about our problems with ECMLink.

I am loving the software though! It's really easy to understand and use.

 

[justacanuck]

Started the car today and checked vacuum (~14) and tweaked the BISS settings. MAFT adjustments, as per manual, sounded like it might be too lean. Now that it is significantly warmer outside, and we had the engine warmed up a bit, we swapped the ECMLink ECU into the car and it started right up.

We tweaked the BISS screw a little bit, and adjusted the MAFComp table after a few trips around the block, to get our STFTs closer to zero.

If anyone cares to provide feedback on our data log below, I would be most grateful.

Thank you,
Josh

 

[justacanuck]

Reading this thread, I came to believe there are 2 things should happen next:
- fix the global fuel settings (I believe the last log was too rich)
- connect our Autometer wideband to the ECU so we can log it and make some better decisions moving forward.

I connected the blue wire from the autometer gauge, to the EGR input on the ECU (Pin 15) and the black wire to ECU ground (PIN24) and made those configuration changes in ECMlink.

This wideband gauge operates from 0-4V with 0v=10:1AFR and 4V=17:1AFR. I know our Direct Access Max Oct table has some settings as low as 9.3 but I'm kind of expecting those might end up being updated to something closer to 11 anyway. Of course I still have the factory front narrowband O2 to help us out, so I'm not relying on the new gauge, at least until we can confirm it works properly.

The car started up fine and seemed to be running OK. We brought it to operating temperature and did some (smaller?) adjustments to the lower frequencies of the MAFComp table.

We took it for a drive around the block, keeping the throttle inputs fairly small. The log seems to indicate everything is working as expected but it seemed to want to die a few times when dropping the throttle input to zero while pushing in the clutch.. Tomorrow I'll take it for a longer cruise and get a few decent pulls, with slightly higher throttle.

Logs attached.

 

[justacanuck]

So, with the WB logging properly, we were driving it around, and the more I was looking at fuel trims and comparing the logged AFR, the more something didn't seem right. I decided to back up and check for vacuum/boost leaks properly.

We found several leaks. One potential source of vacuum leak between GM MAF and the intake of the turbo. Another small pinhole on an interconnect boot, on the cold side of the intercooler piping. We also found another HUGE leak at the connection between the intake and the throttle body. Turns out my son had put the throttle body gasket on the wrong way (upside down?). We fixed all those leaks and tested again. It almost holds pressure now, but we still have a relatively small leak around the throttle plate hinge itself. We had put some fresh o-rings in place of the deteriorated factory seals, but I guess we are going to have to purchase the proper seals.

I tested vacuum again and we are still only getting ~14". I was kind of expecting to see something closer to 20".

The remaining leaks seemed relatively small, so I figured I would start again with our STFT/MAF calibration. The engine seems to be running significantly smoother and the problem encountered previously, dying out on downshifts, seems to have gone away. Even the MAF sliders seemed to require less adjustment than before.

We are still using relatively small throttle inputs and keeping the boost very low but still get some pretty big knock retard, >4 degrees, at ~3000rpm, in open loop, as soon the boost starts to hit.

I'm pretty sure I even saw some knock during open loop operation. It doesn't make sense to see so much knock, particularly at such low throttle input, low RPMs, and very mild boost levels. I tried swapping the "Direct Access" MaxOctTiming table with the factory EVO3 table and it still started knocking far too easily in my mind, at around the same spot. I guess I'm going to have to go back to the beginning, again, and check the base timing. This doesn't seem right. I'm going to do some testing on our FPR as well. Maybe it isn't increasing fuel pressure like it should when the boost hits?

 

[justacanuck]

I should have mentioned that we shipped one of the ECUs back to ECMLink because we didn't see "key start" signal. Turns out they found an intermittent solder joint and have repaired it free of charge.

In the meantime, right after we had finished wiring in the wideband sensor to our EGR input circuit, we lost the ability to see, and log, SPEED at the ECU. The newly repaired ECU should help us confirm a problem with the car rather than another problem with the ECU. I have already ordered some new reed switches, so we will take the dash apart, test and repair those wires and/or replace the reed switch itself.

This project is getting to be a little more challenging than I expected.

 

[We're on Boost]

We are still using relatively small throttle inputs and keeping the boost very low but still get some pretty big knock retard, >4 degrees, at ~3000rpm, in open loop, as soon the boost starts to hit.

I'm pretty sure I even saw some knock during open loop operation. It doesn't make sense to see so much knock, particularly at such low throttle input, low RPMs, and very mild boost levels. I tried swapping the "Direct Access" MaxOctTiming table with the factory EVO3 table and it still started knocking far too easily in my mind, at around the same spot.

Was it really 4 deg of knock retard or was it a RawKnock of 4 raw?
In you log dated Jan 27 I see 1.4 deg of knock retard and the RawKnock there is 4 raw.
That is at 1375 seconds. Bummer that Speed isn't working because that's how I usually guess at the load, rather than just looking at the LoadFactor item.
Anyway that 1.4 deg of retard at 1375 seconds, like you say, there's nothing going on there to cause it.
Except - that area around 3,000 rpm seems to be a noisy rpm region in these engines. Mine is, a little noisy there, and again right around 4,000 rpm sometimes. And I've seen the same with other people's engines.
So if I see 1.4 degrees of retard there I don't pay much attention to it, except that I'll keep an eye on it hoping it won't get worse.

If it's really 4 degrees of retard and it's at light throttle, then I don't know, I guess I'd try reducing the timing in that rpm area and a little before, just to see if it even helps.

There's one other thing besides just a noisy engine or actual knock that I can think of, and that is that when the hydraulic lash adjusters go bad, or start to go bad, they get noisy and it shows as knock. If you want to see a real extreme example of that, check out this thread, which went for a long time. In the end it was bad lash adjusters. Lifters, HLAs, whatever you'd want to call them. Looking ... Oh yeah, it was called "crazy knock under part throttle .....
And there's some discussion of "false knock" in there.

https://www.dsmtuners.com/threads/r...nder-part-throttle-low-load-open-loop.543706/

Going back to that 1375 sec in Jan 27, I guess I would say that 27 degrees is plenty of timing for 2500 - 2600 rpm and load factor of .65 - .70.
It's a "normal" amount though. It "should" be ok. That's what mine was (as tuned by English Racing) until I reduced it by quite a bit on both the min and max octane timing maps in just that small area. It didn't make a heck of a lot of difference in the amount knock. So I wouldn't be surprised if it doesn't make much difference on your car too, but it might if your retard is 4 degrees. My typical retard was a little less than that.

 

[dwb]

it seemed to want to die a few times when dropping the throttle input to zero while pushing in the clutch..

Your throttle cable might be too tight or something, both the throttlepos and idlesw are oscillating at idle. I also see you are simulating the idle switch with the TPS, and you are using TPS adjustments. If possible, adjust the TPS manually instead of using TPS Offset/Scale. Is there a reason you are simulating the idle switch from TPS since it seems your idle switch is working? If you can get it to recognize when it should be idling, you will have less stalling and all around better idle.

Another thing that helped me with near-stalling and cold starts was to enable factory setting for the FPS (assuming you still have the FPS inline with the FPR boost source).

 

[Spleen8urLSX]

Also not likely directly related to your problem but something to make life a bit easier, stock 1g maf’s are cheap, take the GM maf and maf translator out of the equation for now now that you have ecmlink at your disposal. It sounds as if someone a while back was using it for its piggyback capabilities, no need now with ecmlink so to simplify your process take it out of the equation as there’s no other gain to be had using a GM maf on the current set up.

 

[justacanuck]

Your throttle cable might be too tight or something, both the throttlepos and idlesw are oscillating at idle. I also see you are simulating the idle switch with the TPS, and you are using TPS adjustments. If possible, adjust the TPS manually instead of using TPS Offset/Scale. Is there a reason you are simulating the idle switch from TPS since it seems your idle switch is working? If you can get it to recognize when it should be idling, you will have less stalling and all around better idle.

Another thing that helped me with near-stalling and cold starts was to enable factory setting for the FPS (assuming you still have the FPS inline with the FPR boost source).

View attachment 720236
View attachment 720238
View attachment 720239

Yes, we did have a tight throttle cable. I have since resolved that issue.

I was simulating the idle switch because we were having reliability issues with it, and we were planning to use that ECU input for something else, perhaps antilag trigger.

I will definately try enabling the factory FPS settings. Maybe that is part of our issues. The regulator does what it should (i.e. increase pressure when I simulate boost) but perhaps those missing factory FPS settings are messing with the actual fuel delivery while driving.

Thank you very much for taking a look at our logs and providing your advice. We really appreciate the extra eyes and brains

 

[justacanuck]

Was it really 4 deg of knock retard or was it a RawKnock of 4 raw?
In you log dated Jan 27 I see 1.4 deg of knock retard and the RawKnock there is 4 raw.
That is at 1375 seconds. Bummer that Speed isn't working because that's how I usually guess at the load, rather than just looking at the LoadFactor item.
Anyway that 1.4 deg of retard at 1375 seconds, like you say, there's nothing going on there to cause it.
Except - that area around 3,000 rpm seems to be a noisy rpm region in these engines. Mine is, a little noisy there, and again right around 4,000 rpm sometimes. And I've seen the same with other people's engines.
So if I see 1.4 degrees of retard there I don't pay much attention to it, except that I'll keep an eye on it hoping it won't get worse.

If it's really 4 degrees of retard and it's at light throttle, then I don't know, I guess I'd try reducing the timing in that rpm area and a little before, just to see if it even helps.

There's one other thing besides just a noisy engine or actual knock that I can think of, and that is that when the hydraulic lash adjusters go bad, or start to go bad, they get noisy and it shows as knock. If you want to see a real extreme example of that, check out this thread, which went for a long time. In the end it was bad lash adjusters. Lifters, HLAs, whatever you'd want to call them. Looking ... Oh yeah, it was called "crazy knock under part throttle .....
And there's some discussion of "false knock" in there.

https://www.dsmtuners.com/threads/r...nder-part-throttle-low-load-open-loop.543706/

Going back to that 1375 sec in Jan 27, I guess I would say that 27 degrees is plenty of timing for 2500 - 2600 rpm and load factor of .65 - .70.
It's a "normal" amount though. It "should" be ok. That's what mine was (as tuned by English Racing) until I reduced it by quite a bit on both the min and max octane timing maps in just that small area. It didn't make a heck of a lot of difference in the amount knock. So I wouldn't be surprised if it doesn't make much difference on your car too, but it might if your retard is 4 degrees. My typical retard was a little less than that.

We checked base timing and it was almost TDC so we adjusted to ~5 BTDC. I thought that might have been the end of our troubles, but we still have other issues (more on this below, in response to other feedback)

Thank you for your feedback and advice. I really appreciate someone helping us out.

 

[justacanuck]

Also not likely directly related to your problem but something to make life a bit easier, stock 1g maf’s are cheap, take the GM maf and maf translator out of the equation for now now that you have ecmlink at your disposal. It sounds as if someone a while back was using it for its piggyback capabilities, no need now with ecmlink so to simplify your process take it out of the equation as there’s no other gain to be had using a GM maf on the current set up.

Yeah, there was an SAFC2 installed previously, to go along with the MAFT. Before today, we had set the MAFT base to setting 4 which is supposed to match the stock injectors.

We were thinking along the same lines about getting rid of unnecessary stuff, so tonight we tried to bypass the MAFT and wire the GM MAF directly into ECMLink. First try, we applied 5V power instead of 12V. Got check engine light/DTC for air flow sensor malfunction (Code 0012?) and of course it ran like garbage. We corrected this issue, wired the GM MAF to 12V but the problems didn't seem to go away.

I'm starting to think this isn't a GM original MAF and the MAFT was covering up the inappropriate behavior (ECMLink documentation seems to suggest that non-OEM MAF sensors may not work with ECMLink set to 3" GM MAF)

Unfortunately, I don't think we have the factory style MAS sensor at all. We might just move direct to SD since we have IAT and MAP sensors already, we just need to wire them in. This should be very easy, as soon as our ECMLink SD cable kit arrives.

In the meantime, we might try an alternate GM-MAF and enable the FPS factory settings as suggested. *edit: Tried an alternate MAF, enabled FPS factory settings. No change.

Thanks for your feedback. We appreciate your input.

 

[justacanuck]

Just for the sake of completeness, we still have a bit of a leak around the throttle plate seals. I have ordered new replacements.

I have also ordered some reed switches to restore the lack of speed signal to ECU (I hope)

 

[Diamond Star Werks]

To help it start in cold weather you need to increase the fuel supply, not decrease. Fuel vapor is what burns and there is significantly less fuel evaporation at cold temps. This is why they have a FPS solenoid which blocks vacuum to the regulator to increase fuel pressure thus fuel supply on cold starts. After the car is warmed up the FPS is turned off and vacuum/boost is allowed to be applied to the Fuel pressure regulator. The FPS is not used again until the next cold start.

On the 02 sensor, if you log closed loop you will be able to see when the ecu is using the o2 sensor and when it is not. The o2 sensor is only ever used in closed loop(normally 15-30 seconds after startup), so it 0 effect on start ups.

I see in the last log posted you've returned the cranking fuel adjustment back to stock which is good. In the initial log you decreased your fueling, this coupled with the MAFT and the leaks you've found and fixed is likely why it wouldn't start on the ecmlink ecu. The other issue in your first log is that it shows your cranking fuel adjustment as zero as it is not sensing your key in the start position. So it isn't adding the extra cranking fuel.

I am for moving to SD tuning, I like it.

In the second log your car runs for 30 mins and still is under 170*. I'd say your thermostat is missing or stuck open.

As for the knock I would check your sensor to make sure its not starting to ooze. Also fix the temp issue as well. Engine will react differently with proper operating temps.

 

[Spleen8urLSX]

Yeah, there was an SAFC2 installed previously, to go along with the MAFT. Before today, we had set the MAFT base to setting 4 which is supposed to match the stock injectors.We were thinking along the same lines about getting rid of unnecessary stuff, so tonight we tried to bypass the MAFT and wire the GM MAF directly into ECMLink. First try, we applied 5V power instead of 12V. Got check engine light/DTC for air flow sensor malfunction (Code 0012?) and of course it ran like garbage. We corrected this issue, wired the GM MAF to 12V but the problems didn't seem to go away.I'm starting to think this isn't a GM original MAF and the MAFT was covering up the inappropriate behavior (ECMLink documentation seems to suggest that non-OEM MAF sensors may not work with ECMLink set to 3" GM MAF)Unfortunately, I don't think we have the factory style MAS sensor at all. We might just move direct to SD since we have IAT and MAP sensors already, we just need to wire them in. This should be very easy, as soon as our ECMLink SD cable kit arrives.In the meantime, we might try an alternate GM-MAF and enable the FPS factory settings as suggested. *edit: Tried an alternate MAF, enabled FPS factory settings. No change.Thanks for your feedback. We appreciate your input.

You can’t simply hard wire the GM maf even with ECMlink. You either need their plug and play cable that’s specialty built by them (and no longer offered) for the GM maf that they explain on their wiki why it’s necessary, or the maf translator set to the base settings for stock injectors while making your other corrections in ECMlink. Being that the previous owner installed Denso 560cc injectors I’m assuming the car is on a 14b or 16g which would never come close to flowing enough to NEED a GM maf so again, your best bet is to pick up a plug and play stock 1g maf to get it going and then go from there once you get the bugs worked out whether that be speed density like you mentioned above, 2g maf, Evo 8 maf, etc. I certainly hope you didn’t cut the maf pigtail off the cars engine harness and try to hard wire the GM maf that way.

 

[justacanuck]

To help it start in cold weather you need to increase the fuel supply, not decrease. Fuel vapor is what burns and there is significantly less fuel evaporation at cold temps. This is why they have a FPS solenoid which blocks vacuum to the regulator to increase fuel pressure thus fuel supply on cold starts. After the car is warmed up the FPS is turned off and vacuum/boost is allowed to be applied to the Fuel pressure regulator. The FPS is not used again until the next cold start.

I see in the last log posted you've returned the cranking fuel adjustment back to stock which is good. In the initial log you decreased your fueling, this coupled with the MAFT and the leaks you've found and fixed is likely why it wouldn't start on the ecmlink ecu. The other issue in your first log is that it shows your cranking fuel adjustment as zero as it is not sensing your key in the start position. So it isn't adding the extra cranking fuel.

I am for moving to SD tuning, I like it.

In the second log your car runs for 30 mins and still is under 170*. I'd say your thermostat is missing or stuck open.

As for the knock I would check your sensor to make sure its not starting to ooze. Also fix the temp issue as well. Engine will react differently with proper operating temps.

Yes, we found a problem with the Key Start signal in the ECU. Sent it away for repair. We have another identical ECU that we have been using in the meantime. The keystart signal solved our no extra crankFuelAdj and it started good.

The temperature issue is another story. It is possible the thermostat is stuck open but when we started our testing, the outdoor temps were VERY cold (below -20Celcius or ~-5F). The engine would get nice and warm while stationary, but the engine temps would drop very quickly while driving. We put cardboard in front of 1/2 the radiator, just to keep the temperature in normal operating range. That worked well to maintain decent operating temperatures. Today, the outdoor temp was closer to +10Celcius (+50F) so we removed the cardboard and the engine temp seems pretty normal and well regulated.

I'll take a closer look at the knock sensor... but the issue seems more like a phantom knock since it spikes at fairly low RPM, then slowly declines, rather than getting worse with higher RPM as one would expect.

 

[justacanuck]

You can’t simply hard wire the GM maf even with ECMlink. You either need their plug and play cable that’s specialty built by them (and no longer offered) for the GM maf that they explain on their wiki why it’s necessary, or the maf translator set to the base settings for stock injectors while making your other corrections in ECMlink. Being that the previous owner installed Denso 560cc injectors I’m assuming the car is on a 14b or 16g which would never come close to flowing enough to NEED a GM maf so again, your best bet is to pick up a plug and play stock 1g maf to get it going and then go from there once you get the bugs worked out whether that be speed density like you mentioned above, 2g maf, Evo 8 maf, etc. I certainly hope you didn’t cut the maf pigtail off the cars engine harness and try to hard wire the GM maf that way.

Yes, I had read that article but it didn't click for me. When they said ECMLink supports GM-MAF in ECMLink, I assumed that meant native support, i.e. not requiring the special cable, that they don't sell anymore. I guess that explains why we couldn't get it to work properly.

No, we didn't cut the main wiring harness, just the wires coming from MAFT so it would be easy to put it back together. (The wiring is pretty hacked up already so I wasn't worried to much about making it worse )

I guess we will go dig through the box of old parts to see if the original 1G MAF is still here somewhere.

 

[dwb]

Yes, we did have a tight throttle cable. I have since resolved that issue.

I was simulating the idle switch because we were having reliability issues with it, and we were planning to use that ECU input for something else, perhaps antilag trigger.

I will definately try enabling the factory FPS settings. Maybe that is part of our issues. The regulator does what it should (i.e. increase pressure when I simulate boost) but perhaps those missing factory FPS settings are messing with the actual fuel delivery while driving.

Thank you very much for taking a look at our logs and providing your advice. We really appreciate the extra eyes and brains

Sounds good. Keep in mind the FPS is only active (solenoid closed) under very limited circumstances at idle/closed throttle (and probably some other variables like cold start) when vacuum is actually reducing the base fuel pressure. So it helps bring back up to 37psi (1g turbo base fuel pressure). It would be very bad if the FPS were active under boost since it would see no manifold pressure.

 

[justacanuck]

Sounds good. Keep in mind the FPS is only active (solenoid closed) under very limited circumstances at idle/closed throttle (and probably some other variables like cold start) when vacuum is actually reducing the base fuel pressure. So it helps bring back up to 37psi (1g turbo base fuel pressure). It would be very bad if the FPS were active under boost since it would see no manifold pressure.

Thanks for that. Based on this, I don't expect the FPS settings would fix any of our problems. I suspect we have a weak fuel pump and/or a restricted fuel filter. We have replaced the fuel filter and I have ordered a new fuel pump to eliminate those potential problems.

Since that GM MAF cable isn't available, and we can't connect the GM MAF directly to the ECU, I think we are just going to go direct to SD.