Hilton

Ok.. update

I switched injectors to some Bosch Design III 19lb ones I had (the 4 hole ones with EV6 body). AFR's are still different, but the injectors are much less noisy, and the engine does seem smoother.

I just logged a minute of idle with the WBO2 on each side, with my non-cat coding plug so that there was no O2 sensor fiddling the results. Stupidly I forgot to adjust the CO pot first - so its a little rich overall, but at least I have some measurements which are reliable. I'm happy now that all is running smoothly, so its time to just sort out the maps for good AFR's and chuck some cats on it for the test.

WBO2 measuring 1-4 side: 14.3 AFR
WBO2 measuring 5-8 side: 14.1 AFR

(these are average over 600 samples (1 minute of idle) as reported by sharkplotter).

So tomorrow the cats go back on, and I'll do some sharktuning in non-cat mode to sort out the AFR's, then copy that to the cat map and do some logs using NBsim.

I'll take the same measurements on my '89 when I get the chance, to compare, and to verify the result that 1-4 runs leaner than 5-8.

Also semi-interestingly, I cut the rubber boot off the incorrectly reading spark plug boot. The plastic isn't cracked, but depending on which compass point I push the probe against the end where the wire screws in, I get readings from 1.6 to 3.6 kOhm.

dprantl

You probably already know this, but a '7' on NGK's scale is not the same as a '7' on Bosch's scale. BP7ES is two ranges colder than stock.
Dan
'91 928GT S/C 475hp/460lb.ft

69gaugeman

It looks like you only fail at idle when coming to a stop. I am just throwing this out there, but maybe an ISV issue?

jcorenman

Hilton,

That's an interesting test, and much more comprehensive than most. The low NOx and high CO in the most-recent test does sound like too-rich in general. Was that with stock LH maps? Or had you remapped the LH based on the old MAF?

What is confusing is that the LH should have been running in closed-loop mode throughout the test cycle, i.e. with NBO2 sensor regulating AFR. That should result in AFR's cycling a few percent above and below stoich, and keeping the cat's happy. Where that can fall apart is a goofy NBO2 sensor, or O2-adjust hitting the limits at +/- 20%-- which shouldn't happen unless the fuel maps are way off, or the MAF cal is off, or the Temp-II sensor is wrong.

Are you able to log with Sharktuner during the test cycle? Do you have any logs from the previous test?

Your plan sounds good-- stock plugs and stock cats with NBO2 (or NBsim) connected to LH, coding plug for cat mode. Make sure the LH stays in closed-loop mode over the normal driving cycle, i.e. O2-adjust should never go zero except on decel (assuming no WOT or hard acceleration). Ideally O2-adjust should stay in the +/- 10% range over the whole range, if not then take a look at tweaking the fuel map.

When is the test?

Hilton

Yep - I considered using the cooler plugs after the fail before this one, when I had slightly high NOx but no working hall sender, as a means to combat NOx. I thought 7 was only one range cooler than stock? i.e. BPR6ES is equiv. to WR7DC? However I decided to stick with stock rather than more unknowns, until I've passed the darn test and can install an x-pipe etc. on this car.

The problem appears to be just off-idle, when pulling away. I did suspect idle switch or ISV, but both check out fine, and the ISV% in sharktuner is consistent.

It was with stock non-cat setup and no O2 sensor; my old O2 sensor was faulty (must've been dropped - it had bits of ceramic loose in it) and a replacement didn't arrive in time for the test. The replacement is on hand this time.

The IM240 test is used in Phoenix AZ judging by other posts on this forum, albeit with gr/mile targets. The Aussie pass values are proscribed in an Australian Standard, so not sure whether the pass values are the same in AZ as here.

I'd like to log the test - the only question there is whether I can get NBsim working reliably to run closed loop for the test. I've noticed that sometimes O2 adjust goes way lean on startup with the NBsim, so will try to figure out today under which circumstances that happens (I have a techedge 3A2 wideband controller and 4.9 sensor). If I can't be sure of NBsim, I'll just use the stock narrowband sensor.

Tomorrow morning - so this evening (its 8am here now) is swap the factory cats back on, and some logging/tuning runs on the non-cat maps. I'll spend an hour adjusting the non-cat map, and then see how it copes in closed loop.

I have a 40km drive for the O2 adaptation on the way there tomorrow too.

dprantl

I've always used this as reference to the Bosch/NGK heat range scales (it's pretty good reference for many other spark plug characteristics too):

http://vw.zenseeker.net/Ignition-SparkPlugs.htm

Dan
'91 928GT S/C 475hp/460lb.ft

jon928se

Following this with interest and sympathy for Hilton, as my 88SE went through the same test when I imported 6 years ago.

First test was with no Cats and it nearly passed - was running a bit rich so CO too high.

2nd test was with Cats (I think the same set that Hilton is using now) but no O2 sensor - still running in non cat mode.

I was lucky that the tester remembered me and was receptive to my request not to drive the car using the highest gear possible (equates to about 2000rpm in 5th gear at the fastest part of the test) and on coast down means that fuel cut off doesn't work because revs are too low. Using only 1-3 gears the SE passed on it's 2nd attempt.

jcorenman

I think it is not a good idea to run cats without a NBO2 sensor-- there is something about the alternating rich/lean cycle that is needed for the cat's to be able to burn HC's and reduce NOx. I don't recall the details, but I think running the NBO2 sensor (or NBsim)-- as you have planned-- is the best option.


That sounds like the warmup issue for a WBO2 sensor. Starting from cold there is no problem, the WBO2 needs 30-40 seconds to warm up and start working, while the engine needs minutes to get up to temp before the LH starts using the NBO2 sensor signal.

But on a warm restart, the engine is already warm and the LH only waits maybe 20 seconds (or less, if the ignition was only off for a few seconds).

During the warmup period some WBO2 controllers sit at zero volts (which indicates a lean mixture), which would drive O2-adjust positive (and the mixture rich)-- so this sounds backwards.
Other controllers go open-circuit during warmup, which causes the LH to ignore the NBO2 sensor-- O2-adjust then sits at zero until it gets a valid signal. I think the 3A2 calls into the the zero-volts category.

While it would be interesting to log AFR data, in the interest of passing the test I think using the real NBO2 sensor would be your best bet.

Logging with ST2, even without AFR, would still be interesting-- just to confirm that the LH stays in closed-loop mode throughout the test. (Be sure to include O2-adjust in the log data).

Good luck!!

69gaugeman

And?

jon928se

Just spoken to Hilton on the phone.

It passed Woohoo.

Ironically, however he is now stuck in the Testing centre car park with what appears to be either a failed LH ECU or the PEMS have come loose - so his wife is on a rescue mission.

jcorenman

Yippee, Congratulations!!

Oh crap, I wonder what he said that angered the car-gods??

Hilton

LOL yep, the good news is I passed.

0.32g/km THC
0.30g/km NOx
2.8g/km CO


The bad news is I'm still there LOL.

Just walked to a local hardware store and bought a cheap set of allen keys so I could get at the ECU's - the PEM is seated securely, but when I turn the ignition on, I get the dreaded clicking injectors noise.

Its possible the image on the PEM is corrupt - I know from experience that a bad ROM image can cause the same symptoms, when swapping in chips I've burned on my eeprom burner. Or its possible that the ECU has died - although I suspect its the ROM image, as the ECU is a 3 year old JDS rebuild.

So in the meantime, my lovely wife has picked up a spare LH and is heading down here.

Either way, I have the important bits of paper I needed, on the 6th go at the test. I also managed to log the second run with the Sharktuner, so I have some handy data on rpm/load for anyone who needs to tune for the IM240 test (I couldn't log AFR's for it however - had NB sensor in the factory cat for the test).

Now if there's some way to force-flash an image onto a PEM, Id love to hear about it

Hilton

To clarify - I'm getting the message "No ECU with Sharktuner Programmable EPROM module was found" when I turn on the ignition.

I've tried File > New in the ST software but get the same message. It loads the EZK PEM just fine.

Unseating and reseating the PEM made no difference. I can get it to say "ECU connection lost" if I try loading a binary immediately after a battery disconnect - so it appears it can handshake.

jcorenman

Just to clarify... No start, clicking injectors with key on, correct? (Pull fuel-pump fuse to avoid flooding the car).

Telling ST to load-from-PEM's gives the "No ECU with PEM..." message, correct? EZK won't load?

But you can open an EZK file (or load a new EZK file) and sent it to the EZK's PEM, is that right? So the ST box is communicating OK at least with the EZK.

It sounds like a problem with the K-Bus but it would be unusual for it to be working fine, and then fail. This is the '87?? The wiring is slightly suspect because the factory never used the diagnostic bus.

Check the 12-pin diagnostic connector, is it properly plugged in and snug? I've seen one where the female contacts were worn and the plug fit loosely, making intermittent contact.

It could also be a badly-corrupted or broken LH PEM, but it has to be pretty dead for the ST to not see it at all, for loading. There is no way to force the load, if loading from ST doesn't work.

And those are great test results, you nailed it!!
So what happened between the test and the no-start? Did you do anything to the car?

Hilton

I can load EZK from PEM, I can't load a file into the LH, or read the LH PEM.

My lovely wife drove down with a spare LH, I plugged it in and drove home, then took her out for dinner

I just had a few minutes to swap a stock chip into the suspect LH - it works fine, so the LH is ok. I'll have more time next weekend to investigate whether the issue is with the wiring, connector, or the PEM itself (handily I have another 928 with the later diagnostic port - so I can test in isolation).

When I got to the test, car drove great down there with nothing plugged into it. Got to the test center, turned on the laptop and plugged in the ST, and after restarting it to begin the test, had trouble with the O2 adjust going off to one extreme (-20%) and the car running really rough.

I reloaded the binary into the PEM, and had the same behaviour, until I turned off O2 adaption and then turned it back on, at which point the car ran fine for two runs through the test (whew!). I turned the car off, printed out the results and had a chat with the guys at the gov't test place about 928s, then tried to drive out and had clicking injectors.