Wideband data for 3.2 getting 10 mpg city - what do you think?

[Discseven]

Engine’s an ‘85 3.2 in an ‘80 911. City mpg is at 10. Highway 27. These results have been double checked. City used to be 18. Drop to 10 mpg occurred at some point during engine's rebuild. Don’t think there’s any need for engine or rebuild details given the wideband data clearly speaks for itself. The issue in hand here will be immediately apparent---O2 delta between 1-2-3 and 4-5-6. Wideband = Bosch 4.9 sensors, Innovate controllers, LogWorks software. To rule out hardware error, sensors were swapped side-to-side as were controllers. Test data remained consistent.

My prior thinking as well as that of others was this being a fuel injector issue. On seeing the wideband data here, and given the injector testing as was done, the finger (in my mind) is now pointing to either the ECU or the injector harness cable. Am interested in alternate views—please see what you think.

I’ll put rebuild notes at the end as I know this matters to some.

Warm engine at idle tests:
Green trace = 1-2-3 side of engine
Blue = 4-5-6
Vertical axis = Lambda
Horizontal = Time in seconds




Idle mixture screw out 5 turns
Lucas injectors
W6DPO Bosch spark plugs





Idle mixture screw out 5 turns
Bosch injectors in 1-2-3 / these are original to engine #0-280-150-158
Lucas injectors remain in 4-5-6
NKG Iridium spark plugs installed





Same config as prior setup only with…
Idle mixture screw now turned out total of 7 ¾ turns


Here's interesting part… The remaining Lucas injectors in 4-5-6 are removed and replaced with Bosch injectors. Engine now has all Bosch injectors… idle mixture remains at 7 3/4 turns out...




Expectation at this point was to see both O2 trace lines nearly overlapping, that being on the potential that the Lucas injectors failed to some degree—and—there being fresh Bosch injectors with even flow now installed. Instead, the wideband pattern of the all-Bosch injectors is uneven side-to-side just as with the all-Lucas injectors. Bosch are indicating richer than Lucas.


More injector detail to consider… Bosch injectors were removed from the engine years ago and shelved. While the wideband system was being installed here, the Bosch injectors were serviced by Mr. Bill. His report…




The Lucas injectors that are now removed from the engine were installed and in service for a number of years prior to the engine’s rebuild. They did deliver 18 city mpg. This fact together with O2 wideband profiles for the all-Lucas and all-Bosch injector config looking similar, and the combo wideband of them being "unusual" in this case… a smoking gun as I see this. IOW.. Injectors don’t seem to be at fault—something else is afoot.

Spark plugs cannot be the cause of the delta anomaly seen but am noting them changed as a matter of record. The switch from W6DPO to Iridiums was advocated by Sal and Ivan---who I credit knowing far more than I about this stuff.

While switching plugs, made sense to test compression. Was done with roughly a 2 second crank and the fuel injection harness disconnected. I suspect cranking longer would result in higher psi. In any case, purpose was to rule out the presence of any cylinder failures. Was good to confirm...TY Sal.





W6DPO plugs that just came out. These were pulled, cleaned, and reinstalled about 500 miles ago when I started into this 10 mpg puzzle. In reality, the color here is more brownish than is seen on my screen.



Engine profile for the interested…

Mods from stock during rebuild:

964 Cams
1.45 matching overlap
SSI Exhaust
2-in-1-out Dansk muffler
Clewette ignition wires
Blaster coil
W6DPO spark plugs

Stock:

Heads rebuilt to spec by Xtreme - new valves & guides, seats faced
Pistons & cylinders were well in spec so all were recycled with new Goetze rings - break in with fossil then replaced with Motul syn 20-50 and new filter.
Polished journals
New journal bearings
Rod large ends rounded and finished to spec with new bearings
New bushings on rod small ends bored to spec
New squirters
New timing chains

Injectors are not stock—are Lucas low impedance—but were installed years ago and so were with the engine doing 18 mpg city. They were “serviced” during rebuild under oddball circumstances. In short, injectors are under suspicion here. ECU is also questionable as the one installed now is NOT the one I used to have pre-rebuild. Won’t get into the saga of these parts as it’s just noise compared to solving the MPG puzzle in hand. Suffice it to say the ECU is also on the suspicious list. Will see about these parts as puzzle-solving unfolds.

After the rebuild and once on the road, power to redline was there through all gears. Were it not for fuel & mileage gauges staring at me, I’d not be writing this. Immediately, and with little Motronic experience I admit, I molested fuel-things hoping for a quick fix. AFM disk, ECU fuel setting screw, mixture and idle screw being included in my initial forray. Varying degrees of performance resulted. Obviously no quick fix was had. As the engine now stands, everything that was molested has been returned to original settings with the exception of the idle speed and idle mixture screws. Warm idle is a steady 850 rpm. Mixture screw is 5 turns out—now changed to 7 ¾.


Did city and highway wideband recordings. That collection is largely redundant so am not positing that stuff. Does exist as good baseline fodder should it be needed.

WOT to redline may be of interest...



This is with Bosch-Lucas injector combo. Did not get to redline in 4th and no 5th at all.

For you who suggested my getting into wideband... you were absolutely right. I regret my hesitation as it now stands. Thank you for egging me on-and-on!
.

[scarceller]

The W6DPO plugs are not correct for the 3.2, you need standard plugs with normal simple ground strap like these NGK 6637 BPR6EIX Single Iridium Spark Plug

[scarceller]

If the stock injectors are causing different reading between banks you likely have a bad flowing injector. I suggest you swap 1-2-3 injectors with 4-5-6 and check AFR again. If the AFR readings swap banks you know it's the injectors.

Be very careful with used injectors, even if they have been cleaned and bench tested they can easily still not work properly in car because of age.

[scarceller]

Your idle mixture target should be 0.95 lambda NOT 1.0
The idle mixture is always set between 13.8 to 14.4 AFR

[scarceller]

One more point: if you look at the intake you will notice that the throttle body is skewed toward the driver's side, this favors air flow slightly to to 1-2-3 side. Meaning that you get slightly more air flow toward that side, the air is not 100% even between 1-2-3 and 4-5-6 but it's not much different. And if the air favors the 1-2-3 side that would result in slightly leaner reading on that side but what you have shown above is that 1-2-3 runs richer than 4-5-6.

Again, my suggestion is swap stock injectors between banks to be sure it's not injector related.

Then, within the DME it has only one single transistor that fires all 6 injectors at once, I doubt the issue is DME internal. But it could be injector harness related, be sure the big harness plug near the drivers shock tower is clean and tight. Also with key in RUN engine off check the 12volt pin of each injector harness that plugs into the injector. Be sure they all have exact same voltage reading. You can also check that voltage level with engine at idle for each injector, you will need to back probe the injector plugs to do that.

[Discseven]

Sal... Iridiums are installed. They went in with the compression tests.

.95 lambda at idle. That was hit with the Bosch injectors in for 1-2-3, Lucas in for 4-5-6 and the mixture screw at 5 turns out. Leaving that mixed injector combo in the engine danced in my head... Couldn't do it, even to just test mpg.

Just spoke with Bill (Mr. Injector) about the injector wideband readings. He leaned toward the situation NOT being injectors but did not take a bold stance on that. Mentioned I'd be sending him the Lucas to see where they are at. He said send the Bosch too and he'd recheck them under warranty. Your idea for swapping injectors is another fine idea. Is on the menu with a wrinkle...

I just sent the ECU to ProgRama in Boca to check it. When it comes back, either it is good as is or there was an issue with it... and Joseph fixed it. If there is a fix, I'll install it and of course wideband it immediately. If ECU checks out good without needing a fix, am swapping the injectors as you suggest.

Injector harness... running 12v through it tomorrow.

By the way, wideband... You played a large role in that happening here. Thank you for the repetition of advice in that regard. More, you're a supernova of all this air-fuel intel. So more thanks your way for raining it down on the arena I'm dancing in here.
.

[Joesmallwood]

Could there be an exhaust leak? Intake leak?

[mysocal911]

Originally Posted by Discseven

Sal... Iridiums are installed. They went in with the compression tests.

.95 lambda at idle. That was hit with the Bosch injectors in for 1-2-3, Lucas in for 4-5-6 and the mixture screw at 5 turns out. Leaving that mixed injector combo in the engine danced in my head... Couldn't do it, even to just test mpg.

Just spoke with Bill (Mr. Injector) about the injector wideband readings. He leaned toward the situation NOT being injectors but did not take a bold stance on that. Mentioned I'd be sending him the Lucas to see where they are at. He said send the Bosch too and he'd recheck them under warranty. Your idea for swapping injectors is another fine idea. Is on the menu with a wrinkle...

I just sent the ECU to ProgRama in Boca to check it. When it comes back, either it is good as is or there was an issue with it... and Joseph fixed it. If there is a fix, I'll install it and of course wideband it immediately. If ECU checks out good without needing a fix, am swapping the injectors as you suggest.

Injector harness... running 12v through it tomorrow.

By the way, wideband... You played a large role in that happening here. Thank you for the repetition of advice in that regard. More, you're a supernova of all this air-fuel intel. So more thanks your way for raining it down on the arena I'm dancing in here.
.

The last place to send a questionable DME ECM for evaluation! Since scarceller has been very supportive in this thread, he should've been the 1st choice.
From what I've seen of the ProgRama "repairs" over 20 years, it shouldn't have been even considered.

Good Luck!

[scarceller]

Also, nothing you showed so far would result in 10MPG city. If anything you should be getting decent MPG running lean in 4-5-6

It's still a mystery, what's the cause of 10MPG?

When you noticed the 10MPG was the stock O2 sensor installed and connected? And now it no longer is in use?

[scarceller]

Just noticed you are using a non stock ignition coil?
Blaster coil?

Do not do that, please install a stock coil. The DME is fine tuned to properly charge the stock coil. See my video on what it takes to understand coil charge times and every coil is different.

https://www.youtube.com/watch?v=a7dFdIdsZeU

[mysocal911]

Originally Posted by scarceller

Just noticed you are using a non stock ignition coil?
Blaster coil?

Do not do that, please install a stock coil. The DME is fine tuned to properly charge the stock coil. See my video on what it takes to understand coil charge times and every coil is different.

https://www.youtube.com/watch?v=a7dFdIdsZeU

Not really! The DME ECM just waits a constant time period (~ 1.0 - 1.5ms, spark dwell) before switching-on the output TO-3 Darlington transistor, irrespective of RPM,
to develop the next spark energy. It operates basically the same as a points ignition system, not like the later ignition systems where the time delay for spark energy
development varies based on RPM. Oscilloscope waveforms of the DME ECM pin 1 verify this.


When Blaster coils are used in a 911 3.2, they provide a spark energy basically the same as the OEM Bosch coil, based on their tested parameters as provided here;

Typical Ignition Coil Values: https://www.systemsc.com/diagnostic.htm

As indicated by the coil parameters of the two Blaster coils and the OEM Bosch coil (911 TCI), the electrical parameters are with 10% of one another.
Furthermore, since the coil series resistance is a little higher on the Blaster coils, less heat will occur to the DME ECM output transistor coil driver
when coil current saturation occurs.

[scarceller]

Originally Posted by mysocal911

Not really! The DME ECM just waits a constant time period (~ 1.0 - 1.5ms, spark dwell) before switching-on the output TO-3 Darlington transistor, irrespective of RPM,
to develop the next spark energy. It operates basically the same as a points ignition system, not like the later ignition systems where the time delay for spark energy
development varies based on RPM. Oscilloscope waveforms of the DME ECM pin 1 verify this.


When Blaster coils are used in a 911 3.2, they provide a spark energy basically the same as the OEM Bosch coil, based on their tested parameters as provided here;

Typical Ignition Coil Values: https://www.systemsc.com/diagnostic.htm

As indicated by the coil parameters of the two Blaster coils and the OEM Bosch coil (911 TCI), the electrical parameters are with 10% of one another.
Furthermore, since the coil series resistance is a little higher on the Blaster coils, less heat will occur to the DME ECM output transistor coil driver
when coil current saturation occurs.

Not correct, the DME has a 3D spark dwell table and varies spark duration per crank rev based on system voltage and RPM. the table looks like this:

I have tuned this table for different coil types in the past.

Is the blaster coil similar to the stock coil specs, that I don't know but I take your word for it. Or one could bench test it to find out the exact time it takes to fully charge and saturate the coil and various voltages.

[Discseven]

Originally Posted by Joesmallwood

Could there be an exhaust leak? Intake leak?

Joe... could be but doubt it. Intake nuts were changed a few months ago to narrow nuts and torqued. Exhaust I checked a few weeks ago but not all of them. On your note, will check them all to confirm one way or another.

Originally Posted by mysocal911

The last place to send a questionable DME ECM for evaluation! Since scarceller has been very supportive in this thread, he should've been the 1st choice.
From what I've seen of the ProgRama "repairs" over 20 years, it shouldn't have been even considered.

Good Luck!

Thanks Dave. At this point, what's done is done. Will see how they handle my case. Sal's been beyond supportive, more than anyone knows from this and the prior "disk turn" threads. You too have helped a great deal---albeit you have a heavy hand at times! Had I known Sal worked on ECUs, would have been no-brainer to put it in his hands. Now I know. Had conversation with him about this yesterday after seeing your post. (Login was glitching for me yesterday.)

Originally Posted by scarceller

Also, nothing you showed so far would result in 10MPG city. If anything you should be getting decent MPG running lean in 4-5-6

It's still a mystery, what's the cause of 10MPG?

When you noticed the 10MPG was the stock O2 sensor installed and connected? And now it no longer is in use?

Sal, when 10 mpg was discovered, that was immediately after rebuild. Stock narrowband was installed at that time. In the prior "disk turn" thread dealing with this matter, input to me was that 3.2s don't need the O2 sensor connected. I doubted this and so disconnected my sensor and checked milage. No difference. It has been disconnected since. Now, wideband 4.9s are installed. I have not run a formal mpg test but can tell from the fuel gauge and odo that the same 10 mpg city continues.

Whether the imbalance side-to-side is related to 10 mpg or is a separate puzzle is also unknown. Am prioritizing solving the imbalance. Perhaps the 10 mpg will solve with it / perhaps not. I hope the ECU is at fault for both issues!

Originally Posted by scarceller

Just noticed you are using a non stock ignition coil?
Blaster coil?

Do not do that, please install a stock coil. The DME is fine tuned to properly charge the stock coil. See my video on what it takes to understand coil charge times and every coil is different.

https://www.youtube.com/watch?v=a7dFdIdsZeU

Yes on using Blaster coil. I tried to get a "black" Bosch coil during rebuild as the news was to avoid the silver-cased coils. Supposedly the "silvers" were made in SA and inferior. Black Bosch coils were NA at the time. Checking Blaster specs to the stock Bosch, specs matched so I didn't hesitate to go with the Blaster.

Setting mpg and O2 imbalanced engine sides aside... engine idles well and pulls with power to redline through the gears. I say this as reference while not knowing what difference would be evident between running a Bosch or Blaster coil.

Originally Posted by mysocal911

Not really! The DME ECM just waits a constant time period (~ 1.0 - 1.5ms, spark dwell) before switching-on the output TO-3 Darlington transistor, irrespective of RPM,
to develop the next spark energy. It operates basically the same as a points ignition system, not like the later ignition systems where the time delay for spark energy
development varies based on RPM. Oscilloscope waveforms of the DME ECM pin 1 verify this.


When Blaster coils are used in a 911 3.2, they provide a spark energy basically the same as the OEM Bosch coil, based on their tested parameters as provided here;

Typical Ignition Coil Values: https://www.systemsc.com/diagnostic.htm

As indicated by the coil parameters of the two Blaster coils and the OEM Bosch coil (911 TCI), the electrical parameters are with 10% of one another.
Furthermore, since the coil series resistance is a little higher on the Blaster coils, less heat will occur to the DME ECM output transistor coil driver
when coil current saturation occurs.

Am on the sideline interested to see where this coil discussion goes.

If anyone has a source for the original black Bosch coil, please post it.

Should be hearing from Joseph at ProgRama today or tomorrow.
.

[wazzz]

Karl, just curious to know if you are now running open loop or closed loop. If closed loop, since you now have a setup with two wideband sensors, which one of the O2 controllers did you plug back to the Motronic ECM? And have you tried to swap them? It is probably a stupid question, but since the Motronic only has one input for the return signal, just asking anyway.

[Discseven]

Tested injector harness for integrity yesterday. (This should have been done before sending ECU out so am guilty---cart before the horse here.)





Pin 14 serves 1-2-3 side of engine. Pin 15 serves 4-5-6. When looking into connectors to injectors, and with connector's quick-release bar facing down, full continuity runs to right side female slot.

Harness is good --- 12.64v from ECU plug to all 6 injector connectors.
.

[Discseven]

Originally Posted by wazzz

Karl, just curious to know if you are now running open loop or closed loop. If closed loop, since you now have a setup with two wideband sensors, which one of the O2 controllers did you plug back to the Motronic ECM? And have you tried to swap them? It is probably a stupid question, but since the Motronic only has one input for the return signal, just asking anyway.

Gilles... Way I set dual wideband up here is to see & record O2 data using LogWorks. No controller is connected to ECU so engine is now constantly open loop in regard to O2. Early on in this expedition---in the "which way to turn..." thread that I recall you were a part of---there was input that a 3.2 does not need the closed loop provided by a connected O2 sensor for the engine to run correctly. I checked this proposal by disconnecting my O2 sensor and running a mileage test. No difference in mpg or performance with or without the O2 sensor connected. Given my 10 mpg city, perhaps this discounts the credibility of my engine's results in this regard. That said, given the credentials of those who offered this input, I doubted only because I'm an empirical monger.

I did swap the O2 sensors side-to-side in the exhaust after seeing the first trace lines... (green is 1-2-3 side / blue is 4-5-6, scale is lambda.)



Had to question if imbalance seen was a sensor or controller fault or truly the exhaust O2 levels. I also swapped the Innovate controllers. Readings did not change for each side of the engine. I concluded hardware was working correctly.
.

[Joesmallwood]

Originally Posted by Discseven

Gilles... Way I set dual wideband up here is to see & record O2 data using LogWorks. No controller is connected to ECU so engine is now constantly open loop in regard to O2. Early on in this expedition---in the "which way to turn..." thread that I recall you were a part of---there was input that a 3.2 does not need the closed loop provided by a connected O2 sensor for the engine to run correctly. I checked this proposal by disconnecting my O2 sensor and running a mileage test. No difference in mpg or performance with or without the O2 sensor connected. Given my 10 mpg city, perhaps this discounts the credibility of my engine's results in this regard. That said, given the credentials of those who offered this input, I doubted only because I'm an empirical monger.

I did swap the O2 sensors side-to-side in the exhaust after seeing the first trace lines... (green is 1-2-3 side / blue is 4-5-6, scale is lambda.)



Had to question if imbalance seen was a sensor or controller fault or truly the exhaust O2 levels. I also swapped the Innovate controllers. Readings did not change for each side of the engine. I concluded hardware was working correctly.
.

Karl, one more thought after seeing this chart. The lean side has some spikes in it as well as being higher. Could this be a leak somewhere on the intake side? A crack, gasket, hose, or something else?

I may be overlooking this answer, but could the DME be reading the lean side to over correct fuel dose and driving down the mpg? You probably wouldn't feel this in the power of the engine, but it would definitely show up in fuel use.

[scarceller]

These engine do not like to idle above 1.0 lambda they will lean misfire at idle if lean. That's likely the cause of the spikes in lambda on the 4-5-6 side. The green line for 1-2-3 looks like decent lambda around 0.95 and that's a good setting for mixture.

The mystery still is why are the 2 banks so different?
My suggestion is swap injectors from RHS bank to LHS and see if the issue moves to other side. If it does it's an injector issue, if it does not you likely have a air leak on the 4-5-6 side. I really suggest swapping injectors and using only the stock injectors.

The other test worth while is a cyl balance test unplugging one injector at a time and observing what that does to AFR, if one injector effect AFR more than another sister injector on that same side you know that cyl is having an issue. Also spark plug color is worth looking at. The WBO2 data is ONLY one data point, you can't put all your faith in just one data point. For example: maybe you have a air leak at the head gasket in cyl 4? that would make you think the entire bank 4-5-6 is lean but it's not, you may have 5 and 6 running on target and only 4 is lean because of a cracked plastic spacer or gasket at the intake port. Remember, the WBO 2 sees the average O2 in the exhaust stream from all 3 cyls

[scarceller]

One more thought, double check firing order of the dizzy and be sure the spark cables are going to correct cyl. Firing order is 1-6-3-4-2-5 (clockwise rotor rotation). No disrespect here, but I've seen stupid things cause issues in the past

Also the 964 coils are exact same part number as the stock 3.2 and you can find plenty of used good ones on ebay and you get 2 coils for a decent price. They are the good German black ones, just search for "porsche 964 coils" I see a set of 2 right now for $49

Enjoy!

[mysocal911]

Originally Posted by scarceller

Not correct, the DME has a 3D spark dwell table and varies spark duration per crank rev based on system voltage and RPM. the table looks like this:

I have tuned this table for different coil types in the past.

Is the blaster coil similar to the stock coil specs, that I don't know but I take your word for it. Or one could bench test it to find out the exact time it takes to fully charge and saturate the coil and various voltages.

Actually, the spark duration is basically a function of the spark energy stored in the coil during the time the Darlington transistor output pin1 is initially not grounded, typically about 1ms on the 911 3.2.
So at 6K RPM with a spark duration of 1ms, this leaves about 2.3 ms to charge the coil for the necessary spark energy.
Once one knows the key parameters of an ignition coil, no actual in circuit testing is needed.