I took the 85 in for the two year smog test. This was the first test of the new 396 Super Ram motor. It failed miserably for high CO. I took it home and did some data logging to see what was going on. The O2 sensor was stuck in the upper 800's. So I replaced the O2 and got the same results. This is where it gets wierd. The problem occurs only when the transmission is in Park or Neutral. It just so happens the smog test is done in Park at idle and 2500 RPM. When in gear the O2 readings fluctuate normally.

Any ideas are welcome. I'm baffled.

 

I learned an awsome trick about 15years ago (from Professor Overdrive, Summit Racing) and I've never failed since. Fill your tank with high-test and add a quart of rubbing alcohol to the tank then drive a little, I bet you pass.

 

Thanks Tim. I will do that before testing again. But with HC reading 500ppm and CO at 4%, I don't think alcohol will be enough. There is most definitely a problem.

Two years ago this car passed with an HC reading of 7 and CO at 0%. At that time I had the Super Ram, AFR heads, a mild cam, 360 CID and the same exhaust.

 

new hot engine may have weaken the cats, did you run the car/cats real hard and hot when it was smogged?

 

If that passes you, it'll buy you some time to dial it in. I'm sure you know the routine, TPS, Air/Idle, MAF and yada yada. Clean everything up and dial it in, see if your runnin the right plugs and so on.

 

This is true, could be a cat thing as well

 

Possible pulse width limitation issue at very light load.

When you get below the min pulse width, two things happen:
The pulse is limited
The INT is reset

As result the ecm won't go any leaner (even in closed loop).

This can be an issue with larger injectors and double fire mode. Single fire mode helps in this regard, since the injector offsets are only added once per 2 revolutions. You can run leaner in single fire mode than in double fire mode as a result.

Look for a single fire/double fire mode switch. I'm not sure this is supported in the $1F mask for the 870 ecm. Might be an issue.

 

I was thinking pulse width may be an issue. At zero load I have the pulse width set to zero. I tried backing off the pulse width at low load but it didn't have any effect. Maybe I should try pulling some more out.

 

I checked the bin file for a single/double fire mode switch. There was none. I found a constant table that may be relevant. It's called MINIMUM ASYNC PULSE WIDTH.

It is currently set for 1.69 msec. In my data logs, the areas where I have problems have a pulse width of .8 to .9 ish. Is it possible the ECM is over riding what I see in the log?

 

Min Async pulse refers to Accel enrichment. You want the min (synchronous) pulse width. Sorry I don't have a detailed $1F hac or access to $1F source to provide an address.

In $32B and $6E codes, an 0.8-0.9 ms displayed pulse width in double fire mode would likely run up against the standard bpw limit value of 1.6 ms and suffer from this effect.

Are you using 42lb injectors or larger? If so, I'd say this is your problem.

You may be able to reduce the min pulse width value if you can find it, but you risk injector mis-fire if you go much below 1.6 ms in double fire mode.

With single fire mode, you can raise the min pulse width to over 2 ms and still be lean enough for idle and coast down decel, even with 63 lb. injectors and still avoid injector mis-fire under most conditions.

If you can't fix it in the tune, I can think of a couple of work arounds.

Smaller injectors/and or reduced fuel pressure.

Increased alcohol content of the fuel to lower the stoich ratio and raise the minimum requred bpw to achieve stoich will lean things out. A blend of E85 and pump gas with appropriate tuning should pass.

The ethanol smell may be hard to disguise, but it may not matter.

Good luck.

 

just wondering, why is idle or off idle rich? dial back on the fuel tables, what does the wide band read?

don't go on a maybe,

go here:

http://www.amazon.com/CRC-05063-Guar...owViewpoints=1

or heet


http://honda-tech.com/showthread.php?t=2403686

 

I suspect you have identified the problem. The injectors are 42#. There is no table for minimum pulse width that I can find. I'll try reducing fuel pressure to see if it makes any difference. If so, I'm thinking some 32# injectors may be the least expensive way out of this.

 

I'm learning something here, obviously you are datalogging and on top your tune and aware of every driveability nuance,as I am with mine.

why would you knowingly go to a smog station with a rich condition? why are those 42's still in the car when you have administered multitude tuning attempts prior to your needing to smog?

Not only that, did you feel the tune was spot on to allow you drive your car in that state?

 

In addition to the things we talked about last nite, I wonder what would happen if you intentionally miscalibrated your MAF at those two narrow RPM test points to force the motor to run leaner? I'm wondering if you got the BLMs to show say 140-145ish in those areas it might help........? Maybe tequilaboy can chime in on this idea?

 

The following suggestion is of a highly speculative nature, so proceed at your own risk:

Here are the relevent parameters in $32B for example. Notice they are both equal to the same raw hex value of 0x006F and are in sequence.

;----------------------------
; FUEL OUTPUT PARAMS.
;----------------------------
LC3A4 $006F ; 1.693 msec MIN BASE PW, (msec * 65.536)
LC3A6: $006F ; 1.693 msec Default Pulse Width if calculated PW is <= to
LC3A4

Looking into the $1F bins that I have on hand with a hex editor, there are 3 values = 0x006F which are in sequence, beginning at address: 0x385. Coincidence? Maybe...Maybe not. Why there are 3 equivalent (0x006F) values in $1F is not explained.

One of them may be the 1.69 ms MINIMUM ASYNC PULSE WIDTH already discussed. This would seem logical. Check the address to confirm.

Assuming that the logic is the same, you could attempt to reduce these values by small amounts and see what happens.

0x0069 would be 1.6 ms
0x0062 would be 1.5 ms
0x005C would be 1.4 ms

After changing, be sure to save with your editor to update the checksum.

If you go too low, the injectors may misfire at the smaller pulse widths and cause some rough running. Just go low enough to maintain some INT activity at your lightest expected load condition.

 

Thanks for the comments.

This is a bit more complex than you want to make of it. I did not go to a smog station knowing I had a rich condition. That would make me an idiot.

I fine tuned the MAF tables to achieve optimum BLMs. I had the BLMs nailed to 128 +/- 3. That is well within spec. What was not obvious was the fact that the rich condition only occurs at no load. That appeared during the smog test with a high CO. Maybe you can explain to me why BLMs and INTs are normal but the burn is rich.

The only way I can see the rich condition is through the O2 voltage which is high (.8 to .95 mV) and not transitioning. My ECM does not allow me to data log for LV8. That makes tuning a lot more difficult.

 

Interesting. I haven't attempted anything in hex yet. Before I go to the dark side I'm going to try a couple other easier changes. First I'm going to try playing with timing in the problem areas. Second, I can try lowering fuel pressure.

Thanks much. I think you are definitely pointing me in the rightdirection.

 

Forgot I had a TunerCat $1F definition file.

The 0x006F at 0x389 is the Minimum Async Pulse Width as suspected. One out of 3 confirmed.

So the 0x006F values at 0x385 and 0x387 could be what we want and fit the pattern.

0x385: 00 6F MIN BASE PW, (msec * 65.536) (Presumed)
0x387: 00 6F Default Pulse Width if calculated PW is <= to 0x385 (Presumed)
0x389: 00 6F Minimum Async Pulse Width (Confirmed)

 

If you are using TunerCat, I can always edit the 1F definition to add these Constants, and send you a copy if you wish to test them.

This way you won't need to bother with a hex editor.

Shoot me a pm with your email address if you want.

 

Not trying to simplify a complex issue, but want to comment simply.

I don't know what your xdf looks like, and the tables/scalars or flags that you may have or not have, I suppose since you have an 85 year ecm it is different than my 89; which is the only experience I have.

Trying to read between the lines here and I feel something is missing,

you don't have a waveband gauge installed is that correct?

if this is so I see your problem, because the issue you have would be "in your face" apparent if you saw the gauge gyrate to rich, on no load situations and everything in between.

your rudimentary no datalog for LV8 needs "supplementation", you know when the DCFO kicks in, because it is in an extreme low load scenario, i call it negative load, cruise withh a fixed speed via cruise control maybe be another no load", a different no load nonetheless, dependent on the grading of the road, that can fluctuate,

If you have an automatic the load during idle in park or drive is depend on the stall speed and the quality of the torque converter, too loose and the load is lower, tighter the load maybe higher

To me load and power application varies with throttle pressure no matter how teenie...IMO

I have heard, and don't hold me to it, that the obd1 technologies transmit data extremely slow and because of this narrow band....well you get it..... you are sharp guy. you know when you are chasing your tail when you don't have to.

I am just thinking out loud and not offering any unsolicited advice, in which I am surely not qualify to give, if it were solicited.