95 lt1 injector offset questions

[spfautsch]

Slope flow matching data: Average flow rate at 43.5psi = 428cc/min. Your set is Flow matched within 1.9%
Injector offset matching data: Average offset at 3bar & 13.5V 962uS. You set is matched within 1.4%

In my notes Fuel pressure at Idle with Vac hose off is 47psi (which is what the engine should see under WOT)

**** So we need to use the 47psi injector kPa value ===
1 psi = 6.89476 kPa
47 psi = 324.05372 kPa

The fact that they gave you a specific offset and static slope is promising.

Unless you have an adjustable FPR I think I'd regenerate your data using 43.5 / 300kpa unless you have an overwhelming amount of faith in the accuracy of the gauge that 47psi was measured with. If it's a typical shop grade gauge that registers 0-100, a 3-5 psi indicator error along with a 3-5% linearity error is not unheard of. Here in the states we have the "miracle" of Harbor Freight which is a huge importer for extremely low grade instruments such as the one I purchased. Mine would read 46 psi on my 4 bar vehicles. Most troubleshooting guides will tell you a 4 bar system won't even start at 46 psi. I replaced the original junk gauge with a $20 liquid filled gauge from the hardware store and it reads in the relative neighborhood. But I still wouldn't use it as a calibration reference.

I was more interested in the method you used for interpolation than the data. I'd really like to develop something that works in libreoffice. But that would probably be info better suited for a different thread.

Good luck...

Edit: Another simple check I would suggest is measuring your injector impedances from the PCM connector while having a helper jostle the harness around. This would tell you if all the injector leads in your harness are intact.

[Terminal_Crazy]

Hiya

Quick update. I still have some weird shit going on...

OK I've spent the afternoon recalculating the Offsets
I recalculated IFC at 44.1 (47psi) which should be very lean from 38.51 (about 14%))

I see what you save above, so can still pull it back to 42 later.

I used the data of the website to recalc the Voltage Offsets.
At 13.6V the difference is about 11.5% lower
Reset the Low PW adder table

I Flashed with Kur4o's 13_10_2018 EEHack 4.90
I Have NOT altered the in car laptop to view the InjectorPW so didn't see the same issue as befor until I got the log up on my desktop.

First start, by 86C had idle BLM's at 138 154 - not too shabby
as it got hotter BLM's moved.
141 154
143 154
144 154
146 147
148 148 OK
went for short drive RH BLM 16 topped out at 160
Trimalyzed data which really increased the low map areas up & went out again.

Car ran OK and was generally happy
Surge was still present but car did idle at 7mph quite happily.
Surge came on as soon as increased throttle 1.2% and went off after about 15mph (8%tps)


I might pick up some O2 sensors and replace them as well as checking all the grounds and cables etc.

I still don't KNOW why the InjPW would be different in Open Loop ?



Interpolation of data:
I don't use MS Excel and my use of LibreOfficeCalc is limited
I don't think interpolation is available as std. There seemed to be some extension available for Excel.
The maths to fit curved data was well outside my field.
I started to read an glazed over.
So my program just uses linear interpolation.
For day to day use I have ARM processor based computers as well as PC's and Linux boxes.
My main computer runs RISCOS, written by Acorn from the original BBC micro back in the 80's who then went on to develop the ARM processor.
For quick calculations I either use BBC Basic an easy & fast programming language or write programs in C/C++

Anyways. for the interpolation, I take a data table such as
0 5
10 100
decide how many points I want (say 5 across the columns and 5 between the rows )
Precalculate the size of grid needed to fit the data in so i end up with
0 x x x x 5
x x x x x x
x x x x x x
x x x x x x
x x x x x x
10 x x x x 100

Then move across the cells. Dividing the difference up as a fraction of the difference (5-0 =5. 5/5==1 ) so we get 0+1x1, 0+2x1, 0+3x1, 0+4x1, until we hit the next row
I then calculate the next data row 10 to 100 so that is (100-10=90. 90/5 == 18 ) so we get 10+1x18, 10+2x18, 10+3x18, 10+4x18
this gives me
0.000000 1.000000 2.000000 3.000000 4.000000 5.000000
x x x x x x
x x x x x x
x x x x x x
x x x x x x
10.000000 28.000000 46.000000 64.000000 82.000000 100.00000

Then traverse the grid for each column which gives us
0.000000 1.000000 2.000000 3.000000 4.000000 5.000000
2.000000 6.400000 10.800000 15.200000 19.600000 24.000000
4.000000 11.800000 19.600000 27.400000 35.200000 43.000000
6.000000 17.200000 28.400000 39.600000 50.800000 62.000000
8.000000 22.600000 37.200000 51.800000 66.400000 81.000000
10.000000 28.000000 46.000000 64.000000 82.000000 100.000000

IF you can explain how to calculate this any better to fit curved tables i'd be interested.
As it is, each grid virtex (4 points) are plotted linearly onto a plane.

Mitch

[kur4o]

I see you are out of ideas, so why don`t you jump in and give it a try of the ls injector patch. Anyway you already have the needed cal data.

When I put the ls injectors I figured that is not possible to interpolate the tables good enough, and since ls injector data is superior to anything gm have done previously I just made the patch to utilize the improved ls inj data.

The PCM calculates how much air enters the engine in grams/second. Than slash that to the commanded AFR and find the needed fuel in grams/second.

The grams of fuel needed are converted to inj pulse width with the help of engine rpm and fuel flow constant.

So you need a dead spot fuel flow constant. It should be 43.5psi /3 bar at WOT, if it is other than that check your regulator or gauge.

Than the PCM adjust the voltage offset. At lower voltage the injector don`t open fast enough and the pcm compensate for the less fuel flow by adding some ms to already calculated inj pulse width.

Unfortunatelly the fuel flow of the injectors is not linear from zero. It gets linear after about 3-4ms range. Below that range they flow less in most cases and the PCM compensate by adding more ms to inj pulse width. The curve here is what matters most.

If any of this components are not accurate you will get all kind of weird conditions running lean and rich at the same time under different engine loads that require the same pulse width.

Tuning at this point is not possible.

Tuning air tables through injector constants is really bad idea that should be avoided on all costs.

Ve table should be smooth and linear accross the table. The ve value is interpolated almost anytime between 4 adjacent cells and you can jump from lean to rich really easy if the transition is not smooth enough. Maf is much easier to tune so you can give it a try. You should always tune VE and MAF at open loop until you match the commanded and the actual afr.

At this point I can only suggest to start from scratch and do it all over again the right way. Tune everything in open loop and than switch to closed loop.



On the right side I saw 02s drop to zero on several occasions with no apparant reason, suggesting bad injector.
Don`t focus that much on inj pulse width, 02s give better idea what`s happening.

I hope you can duplicate that inj difference at start up. In open loop the two sides must have almost identical readings. Some fluctuations are possible due to the data stream updates at different times accross banks.

[Terminal_Crazy]

On the right side I saw 02s drop to zero on several occasions with no apparant reason, suggesting bad injector.
Don`t focus that much on inj pulse width, 02s give better idea what`s happening.

I hope you can duplicate that inj difference at start up. In open loop the two sides must have almost identical readings. Some fluctuations are possible due to the data stream updates at different times accross banks.

I need to sort out this weird issue first.
I hadn't seen the InjPW go to zero but when they match up they are much less stable than the lhs in the offset display.
This seems to have started about 12 DEC and when I started playing with the EOIT so i'll try an earlier bin again that was ok.

It's in about 7 of the last 14 logs. I need to adjust EEhack graph so it displays the PW taller

Strange thing is, the car sounds and runs ok.

Thanks for the input.
Mitch

[spfautsch]

OK I've spent the afternoon recalculating the Offsets
I recalculated IFC at 44.1 (47psi) which should be very lean from 38.51 (about 14%))

Once again, why are you using 47 psi to calculate flow? Do you have an adjustable fuel pressure regulator?

[Terminal_Crazy]

Once again, why are you using 47 psi to calculate flow? Do you have an adjustable fuel pressure regulator?

That was measured fuel pressure.
I recalculated the voltage offsets with the online generic data rather than off the supplied sheet.
Change from previous data was about 10-14% lower. IFR raised back to rated 42# an increase of about 14%
I then recalculated offsets back at the 300kpa pressure and there were minor differences.

Reflashed with EEHack 4.7
Initial fire up was a bit rich, blm’s sat about 151-121 so I resloped the VE Tables.
This change has brought the low corner of the VE table up quite a bit and doesn’t drop off like a ski slope now.
Blm sat at 126-127.
Needs a run out now.

New pcm and o2’s are on their way.
Issue withe InjPW seems to have gone so maybe just a wierd flash issue.

Mitch

[kur4o]

I got a little confused what you are trying to do.

First you need that fuel flow figured. If you have the fuel flow at 42 pounds at 4 bar, you need to convert it to 3 bar flow.

The formula is x*0.86602540378
x= fuel flow at zero vacuum at 4 bar fuel pressure, [zero vacuum value should be the lowest value in the ls data].

Than you have to convert the voltage offset. You should have a ls data table called injector offset vs vacuum vs volts . Take again the row at zero vacuum and interpolate to lt1 voltage . It will take some time but it can be done by hand easily.

[spfautsch]

That was measured fuel pressure.

With a factory pressure regulator?

[sherlock9c1]

For what it's worth, on every older LT1 I've measured, the fuel pressure has always been on the high side, usually 3-4psi higher than spec.

[spfautsch]

For what it's worth, I doubt you (or, more specifically Terminal_Crazy) was / were measuring with a recently calibrated, laboratory grade instrument. One can easily spend $300-$400 on such a device and it may still be off 1-3 psi "out of the box".

My point is that I'm not sure I would skew all my injector data based on measurements taken with a $20-$50 instrument. And bear in mind a fuel injector "test kit" that sells for $50 likely includes a gauge that's worth about $5 and cost about $2. Just my $0.02, for what it's worth (considerably less than $0.02).