Hey guys, first of all Merry Christmas!!!

I'm wondering if anyone can shed some light on how the PRP duration and gain and offset work. I can't seem to get my AFR's to stop swinging so much in closed loop. It is really apparent at low TPS. I can hear the engine go lean, then rich, then lean, then rich.
It bugs the hell outta me, and I've tried to lower them all without results.
I'm running 90 pph inj's at 15 PSI, Flash EBL. VE is dialed in, AE is good, BLM's aren't wandering, but the AFR swings too much. You can feel it.
Can anyone help explain this stuff to me? I have searched for good threads on PRP stuff but haven't been able to come up with good info.

Any help is greatly appreciated!!!!

Mike

 

I think I found some stuff that may fix it.
THanks Rbob lol

 

FWIW

lowered airflow and gain multiplier tables a lot.
Integrator is happier now

works mo betta


-edit
Can you enter negative values in the tables???

 

Not in the proportional gains (PRP) tables.

RBob.

 

Table addresses ??
Interested to how you corrected it.

 

I don't know crap about hex code, so I can't help you there.
Rbob probably knows

 

That would be a right click on the table description, then a left click on Edit Itexm XDF Info which will bring up a new window with that particular XDF items information. There in the third box down you will see the address. Now as far as setting up the table itself I become pretty clueless but I do know how to add constants and flags, and even tables if I'm just copying from one XDF to another.

 

cool. yah. got it

hey oldred-

let's say I wanted to add IAT to the 7747. how would I do that
I know there's a lot involved, just wondering what the MAJOR steps would be
one has got to be making a new table and flag, right

 

Mine isn't functional yet but I think its just a flag and a table, maybe two.

 

For $8D the proportional gain tables are at: $84CF & $84DE

The math for the two tables is:

gain = (L84CF[gms/sec] * L84DE[O2error]) / 256

The gain is then added or subtracted from the injector PW. If the O2 is currently reporting rich the gain is subtracted. And is the O2 is lean the gain is added.

The purpose of the proportional gain is to force the AFR to oscillate.

When going with larger injectors the gain becomes too high. This is becuase the gain calculation doesn't know anything about injector flow. The fix is to reduce the gain values. I tend to use the gain vs airflow table for minor adjustments. And the gain vs O2 for major. In that case it is an even across the table change, all locations equally.

RBob.

 

RBob, you're a frickin' animal.
Is the PROM tuning guidebook the best place to start learning how to edit and write code? probably...just curious. THanks, man

 

Does that apply to big injectors only? What about taking little 46# injectors and running them at 40-50 psi, how does that effect the prop gains?

 

IIRC the Patch Method Demo thread (should be in the tuning guide book sticky) has several posts on how the instruction set works. One of the math routines (a 8 x 16 or 16 x 16 multiply?) is broken down instruction by instruction. Along with an example of it doing a multiply with actual values.

To add IAT code to the '7747 is tricky. Can only tap into the minor loop fueling code, which is at least a semi-decent location for it. As the IAT affects the fueling the most. Will need to read the ADC input for the IAT, then do something with it. Most important is to watch how much time is taken for the conversion.

If you look at the stock '7747 code in the minor loops. The ADC conversion is started then the code continues to do something else. And then comes back to the ADC and reads the value. This saves ececution cycles and is good coding practice.

RBob.
----------
It is any time the injector flow rate is increased over stock. OTOH, an increase in displacement over stock has the opposite affect. Trouble is, usually when the displacement is increased the injector flow rate has to be increased in a greater proportion. Something about folks bolting on better heads, more cam, and better exhaust.

Best bet is to take the ratio of injector flow increase and reduce the proportional gains by the same ratio. Can then roll any displacement increase back into the gains by increasing them. At the least this should get it close.

RBob.

 

I thought it was: gain = (L84CF[gms/sec] * L84DE[O2error]) / (256*256)

The only problem with using just the Airflow table is that is maxes out at 64 gm/sec in the $8D code. The prop duration tables can be messed with too.

 

If you divide a 1 byte by 1 byte multiplication by 65536 (256 * 256) the result will always be less then 0. Don't see how that would work out (??).

RBob.

 

You are right. The two 8_bit mult. gives a 16_bit value. The upper byte is retained so it is a divide by 256 and the remainder is discarded.

 

I brought my prop gains table down by 20% and it seems to have stabilized things pretty well. I'm getting areas of rich, dead on, and lean but thats because I'm messing with my timing but before I would have cells right next to each other 20% different then each other. Seems to be helping.

 

I agree!

Look in Chapter 6-7, especially "Source code for dummies....Ask questions here! "
That's where I learned to do the dis-assy and reassembly.
Would never have done it on my own.
Cobra289 also made a summary document with links available on Moates server under source code/hacs.

 

cool. thanks JP

 

I am pulling my hair out with this PRP stuff. Can't seem to get it right.
I datalogged last night and despite what I thought the problem is still present.

Here the INT and BLM seem to be close to 128, but the AFR is lean.


here's another weird one


here, the BLM is lean but the AFR overshoots rich


Here, the BLM is lean, but the AFR never gets low.


Here are my O2 settings. I'm using the ZT-2 simulating NB O2

 

Where are your BLM/INT coming from ?
Are you running a separate NB sensor ?

 

no. ZT2 for wideband and NB input to ECM

 

But you're running C/L. The BLMs are ok and that's what the ECM is working towards. However, I didn't catch the mods on the motor but you may need to modify the O2 R/M/L voltages.

 

I've raised the mean rich/lean voltage as well as the lean and rich voltages to match. lowered closed loop target AFR.
stats on setup are in first post.
Thanks guys, keep it coming!!!
Maybe I'll bump the voltages another 50 mv higher and see what happens. Also gonna try to shorten the window for fast R/L down from 130 mv to 25 mv
Mike

 

I am thinking that the ZT2 simulated NB output isn't correct. Look at the last analysis screen you posted. The 2nd & 3rd lines from the end. The WB is showing 14.8 and 14.6, which brackets stoich. The NB entries are 323 mV and 991mV. Having a NB of 991 mV and the WB at 14.6 doesn't seem correct.

Then when the WB is reporting lean (see the blue highlighted line) the NB reading is too high. At 17.3 the NB should be in the low double or single digits. It is reporting 208 mV. Then a little further down in the same screen the WB AFR is 14.6 - 14.4 and the NB is over 1000 mV.

May be worth pulling the WB sensor and using a NB O2. Just to see what happens. May want to do some VE Learns and see if the VE table changes much.

RBob.

 

I have been told ( I think it was HaulnA$$ ) that the WideBand is not as acurate in NB simulation mode,
within such a small range, that a NB uses for input to the PCM.

Anyway, was running a LC-1 to simulated NB input to the PCM, had some of the same problems
put in a good AC Delco NB and all was good again.

How Only using the WideBand to log.

 

I'll try the NB tomorrow and see what happens. Have a new AC delco 3 wire NB I'll slap in
Thanks guys

 

I was interested to see how the changed O2 values looked compared to a stock $8D.
If the routine runs the same way in your setup then your values could be causing some trouble.
I did use a 53 mV window for both sets of values rather than the 103 that was indicated. That opened the window Waaay up.
Take a look.
If this confuses the issue I'll remove it.
Jp

 

well, the stock stuff looks right and mine looks AFU.

maybe I'll throw the $8D values in it and see what happens

 

THanks JP, that graph really cleared things up. I haven't test drove yet, but I will. Will definitely post back results.

 

Well, tried the mv settings in the graph above and no real change.
It is overcorrecting too much with the BLM. say BLM is 120, it will pull too much fuel causing the AFR to shoot up to 17/18. yah, 18. I tried dropping all the prp stuff to 10% of original, no change. Tried stock EBL PRP stuff, no change.
In open loop, it runs great. AFRs are steady so I'm eliminating the WB 02 at a culprit.
One thing I changed a long time ago was the BLM update rate- from 450 millisec to 200 millisec. Could the computer be tracking the AFR too fast?

It really only acts up at low TPS. like when I'm stopped and give it some gas to go it will accelerate, then the AFR's lean out, truck slows down, then AFR's come back into the normal area, truck accelerates like normal. After tip in, the AFR's really lean out and it lean bogs.

I think this might be a combination of too much AE and possibly the BLMs tracking on the extra fuel from AE, pushing AFR's wayyy up through overcorrection. THat's why I brought up the 200msec BLM update rate thing.
Thoughts?

 

BLM update rate- from 450 millisec to 200 millisec

That maybe a very good reason for the overshooting AFR.

 

The $8D idle constants (rich, lean, mean) and non-idle tables (rich, lean, mean) have a bias (stock AUJP is 101mV) that is subtracted when in DIVERT to atmosphere and AIR to cat. conv.

Under normal operation (AIR not to exh. ports) the ECM will subtract the 101mV from the constant/tables to create the window for the FAST O2 and SLOW O2. The O2 value read from the sensor is compared to this value.

With that, the constants/tables may not work the way you want when using non-$8D code.

I do not trust any of the simulated NB outputs. The switch point is very steep and doesn't allow a good range of AFR targets with the O2 constant/tables. A better way is to convert the WBO2 to a less steep NBO2 value in the code. It gives a more linear curve and more accuracy so that the O2 constant/table values can be acheived. In doing so, you can target a 16.xx AFR for low KPA, 14.xx AFR for mid KPA, and 12.xx AFR for high KPA.
Overall, you need a more linear curve than a stock NBO2 curve to do this with with reliable and accurate results.

Does the code you are using already have this WBO2 to NBO2 function built in with a curve that can be changed with software?

I have found that the PROP GAINS are usually less of a problem compared to the inaccurcies of the steep portion of the NBO2 input. Making the NBO2 curve more linear is the way to go when using a WBO2 as the closed loop sensor.

The other nice thing is that at low KPA you barely feel the difference between DFCO and the 16.xx AFR when not in DFCO. You can also make the transition to PE MODE almost seamless if you choose.

 

yah, it isn't the prop gains at all. Good point, junk.
I've smoothed out the O2 settings and I'm gonna put in an AFS 74 this morning and see what happens. THanks guys. Curious to see what happens with this AFS 74.

 

No change with new 3 wire heated O2, AFS 74. The AFR's according to wideband hover around 15.3 when in closed loop, oscillating around that.
Same problem as before. I've tried stock PRP tables, all PRP stuff zeroed out, really high Rich lean mv values for the INT, along with raising the mean R/L accordingly. Changed BLM update rates. Changed VE. CHanged AE. The computer pulls too much fuel when rich and doesn't add enough when lean. Same light throttle big sine wave AFR osciallation. 13.0-18.0. It is ridiculous. Wideband is reading correct. The engine runs great when around 14.7, like it should, but for some reason the computer is pulling way to much fuel and isn't adding enough. Really starting to **** me off.
O yah...in open loop, the truck runs like a raped ape. It isn't mechanical. It's in the tune somewhere, and I've run out of places to look.
This overcorrection with BLM is what's doing it. But then, why does my AFR hover around 15.3 when BLM is 128 and INT millivolt settings should have 128 BLM reading around 600 MV, which is a hell of a lot more rich than 15.3.?

 

Jusr to be sure that we are still on the same page. From your opening post:

I can't seem to get my AFR's to stop swinging so much in closed loop. It is really apparent at low TPS. I can hear the engine go lean, then rich, then lean, then rich.
It bugs the hell outta me, and I've tried to lower them all without results.
I'm running 90 pph inj's at 15 PSI, Flash EBL. VE is dialed in, AE is good, BLM's aren't wandering, but the AFR swings too much. You can feel it.

Running 90 #'ers on a 350. One thing is that all proportional gain tweaks need to be from a steady state condition. Just an even no throttle change drive down the road.

Leave the 3-wire heated NB in for now. Use that for the O2 feedback. We need a starting solid starting point, and that will be better then a simulated NB.

Open both you current BIN and the original EBL BIN in TP. Copy the orignal BIN parameters over to your current BIN. This is the proportional gain stuff, duration, BLM update rate, INT delays, etc.

With 90# in place of a 65# injector the prop gains will be about 33% too high. Reduce the Prop gain by 33% in the 'PRP - Gain Multiplier vs Airflow' table. All entries, can use the toolbox & multiply by .67

As we are starting out fresh, lets slow down the INT update rate. This entry "INT - lean O2 INT delay multiplier", set to 80%. This will reduce down the INT update rate when lean & going leaner from the 50% value it is at.

Then the "INT - Delay vs Airflow" table. I'd say double the values across the table. This will slow down the rate that the INT updates.

This is important as the INT can take over when the prop gains are off. This can be seen as the INT rising and lowering during steady state driving. And it usually overshoots which will cause large fluctuations in the AFR (this may be what you are feeling/hearing).


Flash that into a bank and next time you drive it see how it is. It should be a lot tamer in AFR changes. When looking at the data log note if the INT is oscillating. If it is then slow it down some more. If when driving there is an area that swings too much, tap the space bar.

Then look at that area of the log file (highlighted in green), and get the gms/sec airflow rate. That defines the area of the "PRP - Gain Multiplier vs Airflow" that needs to be changed.

Once changes are made give it some drive time before deciding whather it is OK or not. It is easy to make too many changes in too short of a time frame.

In closing, make absolutely sure that there are NO exhaust leaks. Any leaks make closed loop worthless.

RBob.

 

GOT IT GOT IT GOT IT!!!!!!!!!


SWEETNESS!!!!!!!!!!!!!!!


finally figured it out, thank to everyone here.
Started with stock PRP settings, multiplied everything by how much bigger my injectors are now (~100lb/hr) vs. stock(~61lb/hr), so I multiplied all PRP stuff by .65.
I disabled closed loop, ran open loop for a while. Monitored the narrow band's voltages, and when I was right around 14.7 on wideband, the narrow band was reading 850 mv. so, for ***** and giggles, I made 850 mv my mean R/L for the INTegrator for all Kpa values, gave it a -50 range for lean and a +50 range for rich. Guess what. SOLID 14.4-14.9. Like butta. PRP gains are functioning correctly. I realize that I basically tuned my NB to the WB, but based on the response from the engine when I raised INT voltages it is correct.
Truck runs great, no oscillating AFR's in closed loop. I bet I can even get it to idle closed loop finally.
I thought around 450mv was 14.7? I guess for my truck with this engine and this narrowband sensor it is different. The wideband and Narrowband finally agree, and the engine runs like it should.
I know what you're thinking...850 mv should NOT be the mean R/L point, but nothing else is working. THere are no exhaust leaks, as I have replaced all gaskets on both banks as well as collector gaskets.
The new AFS 74 NB O2 will definitely stay in, as well as the WB for confirmation. THANK YOU THANK YOU THANK YOU thirdgen !!!! RBOB- thanks for the advice, as well as all you guys that helped me out. Can't tell you how happy I am to be finally through this mess.
My next step will be to examine the spark plugs in a few days, as well as some more open loop VE learn sessions targeting 14.5 to nail this puppy down even more. I will adjust the voltages to get it **** on 14.5 for closed loop, since all fuel is pretty much 10% ethanol or whatever they put in fuel.


GREAT SUCCESS!!

 

yep. idling closed loop now, 1.1 msec Sync pulse width. just as it should be.
Had to raise the idle O2 settings also, but whatever
AE is off a bit, but holy hell- i'll take it. LOL.

 

You certain there isn't an air leak in the exhaust somewhere? I ran my car with more radical cam and heads. 80# injectors on a 350 running at 20psi. Definitely dropped PRP particulalrly at lower airflow. I actually ran lower O2mV than stock especially at idle.

 

you're saying my pw is too high at idle- I agree
some of that is 650 usec of offset

man, I'm sure there are no leaks. My idle mv settings are pretty high, and I'm sure they can come down a bit to lean it up but I'm just stoked I'm idling in closed loop. definitely gonna tweak it some more.

 

Been fine tuning my O2 and VE all day, here's what I've come up for INT/O2 settings. yes, I know, very weird. BUT- WB shows 14.2-14.7 in closed loop so I'm not complaining. Not to mention how much better the engine runs. I've needed to get VERY specific on the INT for this- notice the 4.34 mv constant. The dang NB O2 won't swing a lot, so I need to make the values pretty close to eachother

 

If you are using those tables with a stock unheated/heated NBO2 then you have gone too far. Setting the tables that far away from the NBO2 stoich point means that you have to set the min. O2 error real small like you have (4.34mv) for an INT update to occur. You are pretty much running open loop at that point or sloppy closed loop. A little noise on the wire will give a change of 4.34mv (1 bit). The A/D will have that kind of error just due to system voltage and noise internally.

You would gain a lot using the simulated WBO2 to NBO2 output if you need to have the tables that far off. The WBO2 should be reading around .45v if the WBO2 says stoich. You could use a normal looking table and gain more variablility in tuning.

I would tune it closed loop using the WBO2 and just look at what the NBO2 is doing to give an idea if they agree.

Just my advice above, but if you like it that way then leave it that way.

 

If I run closed loop with anywhere near stock NB O2 settings with the NB O2 OR WB simulated NB O2 I run lean. 14.7 on wide band = 1052 mv on wideband's NB simulated output. that ain't right

14.7 on WB according to actual NB's output- 845 mv to 858 mv.

You want me to send you the log?

 

That is very strange. The simulated NBO2 should output .45v when the WBO2 is at stoich (14.7 for gas). That is a definition. The simulated NBO2 output should be .45v if the AFR is at stoich by design. It sounds like the simulated output is programmed wrong. Post the curve that is programmed into the WBO2 to create the simulated NBO2.

I am OK with the WBO2 reading stoich and the stock NBO2 reading rich. The NBO2 isn't that accurate so that is fine.

EIDT: take a look at the www.14point7.com JAW software. It displays the AFR, Lambda, WBO2 voltage, NOB2 voltage. It gives the curves I am talking about for the JAW WBO2 controller. This is just an example to show how the WBO2 and simulated NBO2 volts should be. Well kind of, the JAW curve is a little wrong around stoich.

 

sorry if I sounded pissed, it's just that this thing baffles me. I don't know why the computer likes the NB to be so high, or maybe the other way- the NB reads stoich so high. Or, for that matter, why the WB's simulated output reads stoich so high which giving a WB output of 14.7. I'm using the Zeitronix ZT-2, not real sure on what the curve looks like. I'll head over to the website and see what I can come up with
----------
check that pic out. could explain a lot


http://www.zeitronix.com/questions/O2questions.htm
----------
well, junk, looks like it was pilot error. my bad from the start

 

that still leaves the questions as to why the NB still puts out 14.7 at the really high mv value for 14.7. Trust me- If I leave the NB running at 500 mv as the mean R/L it runs lean. no power, WB at 17, plugs white.

 

Exactly, you see how the WBO2 stoich should read .45V.

As for the NBO2 reading, they aren't that accurate. Slightly rich (14.6 for gas) could make the stock NBO2 read .8 volts. Keep in mind the WBO2 has a +/- .1 AFR error as speced by the designers.

If you use the simulated WBO2 you should get stoich at around .45 or .5 volts. The stock NBO2 could read anything there. Tuning it like you are doing for best driveability is the right way to do it.

 

This is what I run in the $8D code in closed loop using 10% ethanoll gas. It runs 16.5 AFR for 20-30KPA, 15.5 AFR for 40KPA, 15.0 AFR for 50KPA, 14.6 AFR for 60-70KPA, and 14.1 AFR at 80-100KPA.

The $8D subtracts 101mv (AIR DIVERTED) in normal operating conditions. So subtract 101mv from the values I posted for idle and non-idle tables. Do not use these values they were tuned for just this engine.

 

I hear ya, junk. there is more going on here, though. your tables make sense for your app. I see you like to run lean under low load areas and more rich under the high MAP areas. makes sense-after all, your F-body weighs what...3600 pounds with you in it? yah, 450mv is stoich on the ZT-2, and according to the graph the simulated NB from WB make sense.
I keep using the actual WB's AFR as a baseline because it TOTALLY corresponds with engine behavior and power output. That ZT-2 is dead on for WB output. That's why I'm fudging my INT so much to get the NB in line with the WB. Here's the kicker- had no issues with the stock one-wire O2 immediately after I installed the engine a year ago. BLM was 127/8 at idle in closed loop. Something about needing to change settings when you stray from stock that I sometimes overlook

 

I also hear ya on the simulated NB output from the WB being .450 when WB says 14.7. Yah, it should be like that, but it just isn't. frickin messed up. looking at my logs I get simulated NB output from WB values ranging from 1052 to 252 for 14.7. if that makes sense