M-9353-BB302 Injector Voltage Offset work for sharing/ feedback

I've wanted to address fueling on my primary LT1 car for a while. The dyno tune from 2009 was good on power and driveability but pretty rich all over. I started checking the tune after learning a lot more about it over the last several years and found some things that needed doing. Primarily, O2s weren't relocated, the VE tables weren't touched. I stumbled across Injector offset voltage and realized I could have some noticeable fuel waste and rich condition. Those that played with it and got it close with Bosch-style Ford 30# reported better all over the place.

Searching for the answer from someone else's hard work (hey, I admit it) I came up with no clear answers. I have attempted to do this with fairly simple math and Excel's scatter chart functionality to get what I needed, pretty close. Unless I just completely don't get it.

Ford gives us 8 data points of offset vs voltage. GM uses 32. GM is missing 3 values that Ford gives (11V, 13V, 15V). Ford gives us fuel pressures closest to GM's 43.5 with 39.15psi and 44.95psi. So I divided that into chunks to make up some increments in between to get close to 43.5psi. Then figured out the corresponding multiplier for 43.79psi (close enough, and way better than leaving values meant for the much faster response Rochester 24#ers). That estimated multiplier is 1.05512.

Next, I laid out 35 values, knowing I'd not use 3 of them (11V, 13V, 15V) but graphing the voltage vs offset curve would help me to find the in-between values I needed to generate and paste into TunerPro.

Here's the shape with 8 data points, as Ford hands it to us for the M-9353-BB302:
Attachment 14122

Now with 35, but with 27 unknowns as zero:
Attachment 14125

So I left the chart open while I played with the values in the column. I threw some meaningless fudged numbers in below 6V so that end of the chart is not relevant. Hey, at 6V, it's not even running, right?
Attachment 14126

Next, I multiplied these values by 1.05512 to get offsets for those voltages at 43.79psi: EDIT: ***I changed this to 1.0538 when I resumed work on this at post #3***
Attachment 14127

I've attached the spreadsheet I created, tried to make notes on it as to what I was doing. 2 sheet workbook. Feedback welcome. Column F- red numbers are multipled, blue values are filled to make the graph smooth between REAL DATA points.

I changed offset voltages, dropped the low pulse adder by an arbitrary 11% and added the O2 relocation values. The car noticeably rips better and harder than before. Throttle is more responsive, revs quicker. Definitely less fat at idle and part throttle. I left PE mode alone as it was already really good. I also dropped the idle from 875 to 800 and it's stable, though I may bump it up a hair. I do plan to datalog it as soon as it stops raining again, to make sure all is good, not leaned out where it shouldn't be. Car is as follows:

94 M6 Formula 76k mi.
ported 374 casting heads with 1.94/ 1.56 valves + appropriate springs, 7/16 studs, raised plates
Comp Ultra Pro Mag 1.6
short travel 875-16 Comp lifters
Comp 503 cam
approx. 10.9+:1 ish SCR
properly ported stock intake
TPIS 58mm TB
Stock bottom end + resize big end + ARP
Ford Racing 30# M-9353-BB302
better than stock pump but not overkill 255lph, lol
Stock regulator
44psi @KOEO
Pacesetter long tube headers
mandrel bent true dual exhaust/ X pipe with glasspacks tucked/ stacked into the old muffler location
SLP CAI
TFS ceramic elbow
retains unmodified (clean) MAF
Stock 3-channel ABS 10-bolt with 3.42s
LCAs/ relo brackets

Datasheet for the M-9353-BB302
https://performanceparts.ford.com/pa...9593-bb302.pdf

I've been agonizing on how to alter the "Injector Offset" parameter table in TunerPro (using EEx 4.1 xdf). I revisited the work above and started doing pen and paper exercises with the Ford datasheet. Staring at a pdf really doesn't help me work the numbers into my brain in a useful way. The top left table of the Ford Datasheet is a set of constants, to be multiplied by the appropriate multiplier in respective tables.

ALOSL (lbs/sec) multiples against its corresponding multiplier for each given pressure in the FNPW_LSCOMP table
AHISL (lbs/sec) multiples against its corresponding multiplier for each given pressure in the FNPW_HSCOMP table
FUEL_BKPT (lbs/) multiples against its corresponding multiplier for each given pressure in the FNPW_BKCOMP table
MINPW (ms) of .776 is the minimum consistent pulsewidth these injectors deliver at Ford factory pressure of 39.15psi
FNPW_OFFSET multiplies its whole table by its corresponding multiplier for each given pressure in the FNPW_OFFCOMP table

Now, since we lack a multiplier for 43.5psi in each multiplier table, we need to find that, or get as close as we can so we can use a reasonable value for each table. Again, I used Excel to graph this and stretch the x axis to get the values as close to correct as I could manage.

1) For minimum pulsewidth, I fudged by just calculating the proportion of 39.15 to .776 and got 43.5 to .851. I also tried using the Bernoulli equation, which we normally use to alter the flow vs pressure change, to find a relative minimum pulsewidth at 43.5psi. That method returned .776ms at 39.15 vs .8179ms at 43.5. Converted to usec, those are 776 and 817.9usec. Math geeks, please correct me on which of these 2 methods, or neither!! This may be off!! Time or pressure, or a blended value of both, not sure which is better... or even accurate.
So in TunerPro the STOCK "Injector Offset" table shows:

(Again, this is the stock table)
BPW | Adder
usec | usec
488 | 381.5
549 | 351.0
610 | 305.2
671 | 274.7
732 | 244.1
793 | 213.6
854 | 198.4
915 | 167.8
976 | 152.6
1037 | 137.3
1098 | 122.1
1159 | 106.8
1220 | 106.8
1281 | 91.6
1342 | 76.3
1403 | 76.3
1464 | 61
1525 | 61
1586 | 61
1647 | 45.8
1708 | 45.8
1769 | 45.8
1830 | 45.8
1891 | 30.5
1952 | 30.5

Continuing on from here, I'm going to use .818ms as the Minimum Pulsewidth. With the calculated offset at 14V, That minimum pulsewidth would take a total of 1.6863ms for these injectors at 43.5psi. Low pulsewidth adder falls at and below a point where flow becomes linear, somewhere between 122 and 124usec. The table shows a parameter data point of 854usec. **This pulsewidth intersects with 14.0V and 14.4V offsets in the Injector Voltage Offset table.** So, I think BPW below 854usec is where the low pulsewidth adder is particularly important for this injector. Some have had to experiment with different injectors to see what they do vs. what the engine wants with each specific injector/ manufacturer to find where to increase and where to decrease values. Some have zeroed out the table completely and dialed it in by feel/ AFR at low RPM/ part throttle. I guess the point is that at/ below the 793-854usec BPW is where most of the "adding" would be, at the "knee" of the low slope and somewhat less so above, ending in a zero value at or somewhere before 1952usec (max BPW in the low pulsewidth adder table). I'm thinking this can give a much closer place to start with dialing this in.

2) My 'close enough' value for Low Slope multiplier at 43.5psi is 1.076. Multiply this by the Low Slope Constant and you get .010784748 lbs/sec in the low slope. Multiply this by the length of the pulsewidth in sec/msec/usec at any given RPM vs MAP vs Hz/ g\sec and you get the delivered fuel mass for that pulse in the low slope, if the resulting mass is < .00001336 lbs.

3) My 'close enough' value for High Slope multiplier at 43.5psi is 1.0584. Multiply this by the High Slope Constant and you get .009041 lbs/sec in the high slope. Multiply this by the length of the pulsewidth in sec/msec/usec at any given RPM vs MAP vs Hz/ g\sec and you get the delivered fuel mass for that pulse in the high slope, if the resulting mass is > .00001336 lbs.

4) My 'close enough' value for Fuel Mass Breakpoint multiplier at 43.5psi is 1.0091. Multiply this by the Breakpoint Constant and you get a breakpoint fuel mass of .000013360484 lbs. as the switch point from low slope to high slope.

5) These are my slightly tweaked Injector Voltage Offset values with these injectors at 43.5psi:

Volts usec
0 | 213
1.6 | 213
3.2 | 6011
4.8 | 6011
5.2 | 5004
5.6 | 4211
6 | 3540
6.4 | 3112
6.8 | 2777
7.2 | 2532
7.6 | 2319
8 | 2166
8.4 | 2014
8.8 | 1861
9.2 | 1739
9.6 | 1617
10 | 1525
10.4 | 1434
10.8 | 1342

11.2 | 1281
11.6 | 1190
12 | 1129
12.4 | 1068
12.8 | 1007

13.2 | 976
13.6 | 915
14 | 854
14.4 | 823
14.8 | 793

16.4 | 396
18 | 122
19.6 | 0

Let's back up a bit... To understand and apply correct fueling, we need to understand the density of the fuel. Ford datasheets use lb/sec for low and high slopes and the breakpoint (where flow becomes pretty much linear). Injector Dynamics graphs their injectors for pw vs cc/min flow. The also have a good explanation of the Ford datasheets and have a good article on how that looks in a graph as well.

Water has a specific gravity of 1.00, weighs 8.325lb/gal, and has 453.59237cc per lb. Gasoline varies. Some race fuels have a SG as low as .715 to .728, but pump gas varies from about .710 to .770SG. For this exercise, I'm going to assume a SG of .740 to represent E5-E10 93 octane pump gas. This gas would weigh 6.1605lb/gal. It's helpful to convert flow from weight to volume, from lb/sec to cc/min. To do that, I'll multiply 453.59237cc/lb of water by 1.26 to get how many cc in a lb of this assumed fuel ( = 571.5263862cc/lb).

Low slope = 0.010023 x 1.076 (43.5psi multiplier) = 0.010784748lb/sec
x 60 = 0.64708488lb/min x 60 = 38.8250928lb/hr
38.8250928lb x 571.5263862cc = 22,189.56498186363936cc/hr;
/60 = 369.82608303106056cc/min; /60 = 6.1637680505176776cc/sec
/1000 = 0.0061637680505176776cc/ms; /1000 again = 0.0000061637680505176776cc/usec


High slope = 0.008542 x 1.0584 (43.5psi multiplier) = 0.0090408528lb/sec
x 60 = 0.542451168lb/min x 60 = 32.54707008lb/hr
32.54707008lb x 571.5263862cc = 18,601.509344220544896cc/hr;
/60 = 310.0251557370090816cc/min; /60 = 5.16708592895015136cc/sec
/1000 = 0.00516708592895015136cc/ms; /1000 again = 0.00000516708592895015136cc/usec


Using a calculated offset of .8683ms at 14V, we can now get a very close idea of fuel per pulse. Remember, offset is the time it takes for the injector to start to open. The offset time it takes the injector to start to open is accounted for in every pulsewidth commanded by the PCM, adding whatever offset is in the inj. offset table at the batt. voltage detected at the time the pulse is commanded, plus calculating the pulsewidth of actual fuel needed. It's amazing that a 25-26 year old computer can do this 4x per revolution up to 7,000RPM. That's 386-387 times per second at 5800RPM.

Now, for the breakpoint, the amount of fuel at which flow becomes 'linear' for the injector.
That's 0.0000132400lb x 1.0091 (approx. multiplier at 43.5psi) = .000013360484lb.
x 571.5263862 = .0076358691384029208cc.

Here's where I start to lose how to do this...

Calculated against the low slope, this mass of fuel occurs at a net pulsewidth, excluding offset, between 1.220ms and 1.239ms 'on' time.
Calculated against the high slope, this mass of fuel occurs at a net pulsewidth, excluding offset, between 1.464ms and 1.478ms 'on' time.

If I start calculating delivered fuel using high slope just above 1.22ms - 1.239ms it's of course lower than the delivered fuel AT the breakpoint, by about 73cc/min! What I'd hoped to see is the point at which to stop adding low pulse adder, in terms of ms / usec. In practical terms, I can again fudge and say low pulse adder needs to happen between the lowest data point in the table (.488ms / 488usec) up to 1.24 - 1.47ms (somewhere between 378 - 458cc/min), with minPW somewhere in between at about .818ms/ approx. 305cc/min. Though at least (I think) I can safely say that no adder is needed at all in the cells 1525usec and above in the Injector Offset Adder table in TunerPro. But that could be as low as the 1281usec and above cells.

Edited out, still working.

The low pulse offset adder is indeed needed for non linearity of delivered fuel with very low open times. At ls1 pcm data the table spans till 4ms at lt1 pcm it is set to 2ms[Which can be easily expanded to 4ms with some simple patch].

I guess the ford`s high low scope is where there is non linear operation. Some drawing will explain it much better.
Usually the adder is used to flatten the curve of non linearity.

The hard part here is to convert draw from lb/s to pw in ms. Some complex math equation will be needed for sure.

What was driving me nuts was when I graphed the low and high slopes, they reached zero at the same place because of the offset. So, again I fudged (I hate that). I simply subtracted 0.2383312 from the offset for high slope to make the crossover happen at the right value. Including the offset time, that would be between 2.088ms and 2.108ms. So that gave me the area I was interested in for compensating for offset and adding fuel to get from the low slope line to the high slope line. Shown as the difference in value between blue and red lines at each data point. I added 5 values, down to zero at the bottom of both slopes for graphing purposes. The minimum value for the LT1 offset table is 488usec / .488ms. Area of interest highlighted in yellow:

Attachment 16348

In this screenshot from TunerPro RT, the stock values are in red, my previous blind guess values from about 2 years ago in green and the current values from this exercise in black. Note where this injector hits minimum pulsewidth and goes linear. I'm curious and will have to find an answer, but I wonder if the Bosch Ford injectors are slower to open (offset) but reach fully open quicker than the stock Rochester 24#? As small as these resulting values are, I can see why some had good results zeroing out the whole table. Value increments are limited to about 15.267 (15.3)

Attachment 16349

The actual values I ended up with I got by zooming in on the graph and bending up the low slope against the offset difference to the high slope (column C) to get that commanded fuel value to match using low slope flow rate. I figured I could get very close. So that would be:
0.000ms | .201
.0305ms | .195
.061ms | .191
.122ms | .182
.244ms | .163
.488ms | .122 Attachment 16350 Attachment 16351
.549ms | .111
.610ms | .101
.671ms | .092
.732ms | .083
.793ms | .074
.854ms | .064
.915ms | .054
.976ms | .045
1.037ms | .035
1.098ms | .025
1.159ms | .016
1.220ms | .007
1.239ms | .001
1.281ms | .000


And for reference, here are my 2 offset tables with values derived from this exercise:

Attachment 16352

Quote:

---

Originally Posted by kur4o

The hard part here is to convert draw from lb/s to pw in ms. Some complex math equation will be needed for sure.

---

I feel pretty confident I've got that. (post #4)

Quote:

---

Originally Posted by kur4o

The low pulse offset adder is indeed needed for non linearity of delivered fuel with very low open times. At ls1 pcm data the table spans till 4ms at lt1 pcm it is set to 2ms[Which can be easily expanded to 4ms with some simple patch

---

That is good news. I can definitely see there being (big) injectors that don't hit linear flow until a point after 1.952ms. I'd be interested to know more about this simple patch, though I wouldn't likely use it for my Firebird. It would be good for other projects, mine or someone else's. Having crawled through this over a couple of weeks, I think I could do it with another Ford injector a lot quicker. I may take a couple more, like a 36# and a 80# just to see how they look in this rough math modeling method...

Also, I noticed you did a lot of work on the EEX xdf project. Thanks are definitely in order!

I loaded these settings into my tune. Stone cold start, immediately fired, no issues with idle as it warmed up. The only other change I made in fueling was to resize the injectors from 33.01 to 31.61. I also made a correction to something I found my tuner from 2009 missed. The boundary for Main to Extended spark advance at 4,000RPM wasn't matched. Very minor.

I can tell it's not as rich. It runs warmer but not hot. Before it ran cooler from too much fuel basically everywhere. It now thinks the injectors are smaller than they were so it's adding a smidge of fuel in places where needed to meet target AFRs, I'm sure. I really need to get a wideband setup on this car. My TBI truck runs an LC-2 and has instrumentation out the wazoo but my Firebird has nothing but stock gauges. I have a spare driver's side A-pillar I was intending to poke holes in for a triple pod, may be able to get to that by spring.

I can say seat of the pants tells me there is quite a difference. Throttle response is dramatically better and it revs quicker. Next step is datalogging it, been a while since I looked into that. TTS Datamaster used to have like a 20 free recording limit, I'm pretty sure EEHack also has built-in datalogging. I can at least get some idea of AFRs with the narrowbands...

I'll do a similar analysis of a couple more Ford injectors like I said above. This all is probably the right idea and probably close, but I am not suggesting anyone use what I've done here without some input from people who know a lot more about this than I do.

Sub'd - Thanks for sharing your work!

I have been playing with the data for these injectors as well, and found the below links to be very useful for defining the differences between the GM and Ford data:
http://injectordynamics.com/articles...racterization/
http://injectordynamics.com/articles...racterization/

I am still working through the numbers based on the above, with a goal of having these injectors be indistinguishable from the stockers in my 95 9C1.
I still have plenty of work ahead to install, tune, tweak and validate.

Congrats on the real world driveability improvement!

Jim

Quote:

---

Originally Posted by JimCT_9C1

Sub'd - Thanks for sharing your work!

I have been playing with the data for these injectors as well, and found the below links to be very useful for defining the differences between the GM and Ford data:
http://injectordynamics.com/articles...racterization/
http://injectordynamics.com/articles...racterization/

I am still working through the numbers based on the above, with a goal of having these injectors be indistinguishable from the stockers in my 95 9C1.
I still have plenty of work ahead to install, tune, tweak and validate.

Congrats on the real world driveability improvement!

Jim

---

Thanks for the interest! I'm almost done with the BB-9593-LU47 sheet, then doing the BB-9593-LU80. I'm working toward an Excel workbook that you can just populate with the Ford tables from their datasheets and each worksheet in the workbook does the rest. Except for the voltage offsets and the multipliers for 43.5psi since they don't exist, but those have graphs that let you match/ smooth to the static values for fuel pressure and VBAT increments that are common to all Ford datasheets. I'd also like to do a duplicate worksheet to deal with 4 or more bar pressure for later GM systems, might be able to set that across the board with a filtered cell on sheet 1 of the workbook. Kur4o also mentioned that HP Tuners has larger VBAT offset tables, across 4ms instead of just 2 with the $EE / EEx definitions. I'd like to be able to address other tuning systems' characterizations for Ford Injectors if I can.

You can look at the injector patch thread here.

http://www.gearhead-efi.com/Fuel-Inj...njector-tables

There is an xdf that contains the patch. You can use whole patch or only low pulse extend patch and than populate the table with your values.

I do have a better updated version of the patch but it is not set up in a xdf yet. It reads voltage from injectors at a spare pcm pin and some improvements at stratup. If you are willing to try it, I will do some xdf for it. I run it for 2 years now with zero issues and much better overall drivability.

There is also some screenshots with table size and format for ls1 pcm. If you want to add support for that size tables too.

Quote:

---

Originally Posted by kur4o

You can look at the injector patch thread here.

http://www.gearhead-efi.com/Fuel-Inj...njector-tables

There is an xdf that contains the patch. You can use whole patch or only low pulse extend patch and than populate the table with your values.

I do have a better updated version of the patch but it is not set up in a xdf yet. It reads voltage from injectors at a spare pcm pin and some improvements at stratup. If you are willing to try it, I will do some xdf for it. I run it for 2 years now with zero issues and much better overall drivability.

There is also some screenshots with table size and format for ls1 pcm. If you want to add support for that size tables too.

---

Sure, I'd very much appreciate a chance to try the patch you already have. I'd like to know more about the version in development and how it uses the spare pin. Sounds very interesting. For 0411 PCMs in HP Tuners or any other platform's definition table I'd just need to see the range of values in the table and understand any rounding or incrementation they do, or if they take any decimal value entered exactly. I've never tuned anything OBD2, but have walked a friend through tuning his H/C/I 2000 Tahoe, so I have a copy of his current tune and VCM editor on a computer. I see these parameters and I'll start studying them in more detail to get my head around it:

Flow rate vs kPA (flow rate)
Flow rate mult vs volts (flow rate)
Min injector pulse (limits)
Default injector pulse (limits)
Offset vs volts vs vac (offset)
Short pulse limit (pulse correction)
Short pulse adder (pulse correction)
Boundary (injection timing)

There is unsused pin that reads 0-25v. Possibly left for fuel pump voltage but unused in code.

I repurpose it to read injector voltage for better precision. It needs a wire from pcm to one of the injector feeds[close to engine] I was looking at logs of that voltage and you can see how voltage drops at wot or rapid changes. I had to do it since the ign feed for pcm is very inaccurate going all around the car dropping the voltage the pc sees. At some cars can be good but at least on mine there is some voltage drop that cant be fixed, even with better wiring. You also needs to run fixed fuel pressure all the time.[Disconnect the vacuum reference tube].


The patch uses 3 tables from ls1 calibration.
low pulse offset,
injector offset [vacuum vs voltage]
fuel flow vs vacuum.

Since the tables don`t match in size, some interpolation is needed between lt1<->ls1 tables.