Printable View
In the fiero community, there is CONSTANT argument and debate about multec injectors and their effect on idle. I'd like to help shed light on what makes multecs behavior so much different that a level of mysticism has developed around the name (not in a good way either).
What does "fixing factor for Multec injectors" KF94MAP do in $88?
Funny... I've never noticed that one. But it makes me laugh because I'm betting the name derives from empirical testing which revealed a difference in behavior and a resultant brow wiping related to trying to rewrite code to account for several different types of injector. I'm imagining something like "Buck it! We'll add this value for Multecs and be done with it."
I'll have to take a look at the code to see how it actually works.
should have been some notes in the item(present in all MPFI 60V6 masks that I've made XDFs for)... if not:
WHEN DIVIDED BY ADMAPF+28, THIS FACTOR PRODUCES A SUBTRACTED CHANGE TO BPW IN MSEC. ITS UNITS ARE THEREFORE 'KPA*MSEC'. TO PRODUCE THE COMPUTER UNITS FOR THIS CALIBRATION FROM THE MAP AND DELTA BPW VALUES, USE THIS EQUATION: N = (65.536*DELTA BPW)*(2.711* MAP) WHERE DELTA BPW IS IN MSEC AND MAP IS IN KPA, OR, N = 2.711*(DELTA BPW)*(MAP) WHERE DELTA BPW IS IN COUNTS AND MAP IS IN KPA.
kind of(really) cryptic, I know, but that's the official word from one of the GM code engineers back in the day.
using A1(since I've spent FAR more time documenting and modifying it than 88) since they're going to be really similar for this stuff. I don't have this completely documented in my disassembly, so some of this is going to appear unnecessary.
the subroutine at C1DC deals with the offset stuff. if above the max value in the F94 table, all of this is skipped.
the F94(injector offset vs low BPW) table lookup is done, then added to the final BPW value(and then stored to the stack). MAP is then grabbed(filtered and has an offset added), B cleared and then all placed into X. A receives the KF94MAP calibration value(B stays clear) and an integer divide occurs(effectively, D/X with the quotient ending up in X and any remainder in D). so, we've just divided a MAP value by the KF94MAP value and I'm actually just going to write out an example to better understand this myself.
using a ported BCFA calibration, assuming a target pulsewidth of .73mSec, the low PW adder will add .17mSec for a total of .90mSec, stored to stack.
MAP(let's say a raw 48 a/d counts, should be 28.07kPa MAP) value used here isn't raw MAP, it's filtered and rounded up when necessary, so let's still say 48 counts. it has d28 added to it(so a total of 76 counts in the accumulator), then pushed into the stack with a cleared B and pulled into X.
B stays clear and A receives the calibration value(d50 for BCFA), then the integer divide occurs. should compute as 3200/4C00 (or 12,800/19,456 for humans, or since these are 8-bit values being used with a 16-bit instruction, 50/76), which is .658 real-world. the microcontroller should have 0 in X and 50 in D.
X is saved to the stack and pulled into D(ignoring the remainder of 50) and stored as byte FE. another value is pulled out of the stack into D, this would be the final BPW with the low BPW adder accounted for. now, byte FE(which has a value of 0) is subtracted from BPW, taking it from .90mSec to the same .90mSec....
and that's the end of the multec correction adder stuff. after this, the voltage offset is added to these values and the BPW is pretty much done being modified.
doesn't seem too significant to me, at least in that scenario. i'll do an abbreviated version of another calibration's values using the same scenario: ASHH(92 w-body 3.1/3 speed). at .73mSec, there is an adder of .12 for a total of .85mSec. has a KF94MAP value of d162. so, 162/76, which is 2.13. so, 2 counts of (1000/65536) are subtracted from the BPW, that's .03mSec, final PW of .82mSec....
still seems really insignificant. let's get nuts. going off of how the code works, increasing either the KF94MAP value or operating at lower MAP ranges will increase the amount of change caused by this code. running significantly larger injectors or a smaller cylinder volume should help decrease the total injection time as well.
let's max out the calibration value to 255 and assume that injectors that flow 50% more fuel per time-period are installed(but otherwise follow the 3.1 offset curve). now we're looking at .49mSec, which has an offset of .18mSec(.67mSec total). KF94MAP of 255. let's say around 15kPa in a strong overrun situation. 13 a/d counts would be 15.15kPa, offset that by 28 counts to end up with 41, so our critical math step ends up at 255/41, which is 6.22. that's .09mSec removed from .67 for a total of .58, making it 86.5% of what would have happened otherwise. that's kind of a significant change proportionally.
now let's operate this exact engine in a more common situation: 50kPa at 2.2mSec, which is probably reasonable for a 55MPH cruise. no offset at 2.2mSec. KF94MAP of 255, 108 a/d counts is 50.2kPa, offset by 28 is 136, math ends up at 255/136, which is 1.875, so only subtracting .015mSec from the pulsewidth....
so, much like the rather cryptic description given by GM, it appears to remove some measurable injector on-time when in low MAP situations...... and that's about it. setting up a "correct" KF94MAP value that isn't well within a margin of error using an O2 sensor is probably not going to happen. that's a lab-level measurement, I would just leave it as-is or lower it to a trivial level.
if that leaves you confused(a lot of people follow the math but start getting tripped up once assembly instructions are brought up), the last paragraph is all that's really necessary for a summary. even I learned a thing or two today and I'd like to think I'm among a very small handful of people who understand the MPFI 60V6 code(though this situation does show up in the 93-95 3100/3.4RWD SFI code as well) at this level.
I'm not running multecs and it's set at 219. So if i set it to 0, will it 0-out the calc or should it be set to 1?
Basically,.the higher the number and the lower the kpa, the larger the subtraction from bpw? That's how i understand it. VERY informative write up!
correct. if you want to get into proportional changes, then the lower the BPW is to start with, the more significant of an effect it will have as well.Quote:
Originally Posted by f85gtron
0, 1, anything up to probably 50 or so isn't going to have much if any effect on the BPW in most situations. even with Multecs, I would be curious to see how it effected the drivability of just running 0(or patching past the code entirely).
I'm having a hard time trying to figure out why the code engineers justified making such a weird system that only reduces pulsewidths at really low pulsewidths to begin with and why the MAP component is involved with it... it's just all kinds of weird to me.
Thank you, Robert, for this explanation.
It seems like the opposite of a low PW adder. I'm only guessing but it hints that Multec injectors may to take longer to stop spraying fuel when fuel pressure is low. In order to tighten emissions and reduce the need for INT / BLM to try and correct for this issue the engineers built a function that tried to account for the error and close the injector early. If true then the low MAP qualification is used in place of actual fuel pressure measurements so increasing / decreasing pressure via AFPR would alter the needed corrections.
Curiously when early TPI V8 engines were switched to Multec injectors by a tech following a GM bulletin there was no new calibration supplied and no correction was included. It would be interesting to have early Bosh and Multec equipped engines side by side to compare data.
The issue with the 2.8 in the fiero when changed to Multec injectors is a hunting, unstable idle condition. That calc routine must be the repair? Some of the 88 year cars started to see the multecs. Couple those with 160 baud and you've got serious hunting, like 3-400 rpm each way swings.
C3 vs P4... processor speed is same in both. I believe it's 2MHz. 160 baud is communication rate through ALDL. Bob R. of Dynamic EFI uses his own "Lockers" board to access data without going through ALDL and it's as fast as P4 if not faster.Quote:
Couple those with 160 baud
I did not know that. I switched over to the 7730 and i swear it manages the motor MUCH better. Feels quicker response, even on a bad tune. Must be the placebo effect?
Well... there are definitely differences between the ecm families. Newer = better performance in most cases but I don't have enough info as to what's different. More or faster a/d conversions? Faster peripheral hardware? Off hand I really don't know. But I do know the processor clock is the same for both C3 and P4.
I'm assuming it has to be somewhere along those lines as well, but I would think the fuel pressure regulator would be able to keep the pressure difference between the manifold's vacuum and the fuel rail's pressure equal(keeping flowrate equal).... unless getting close to a full vacuum causes the FPR to become non-linear(specifically, rail pressure being too high, hence the need for trimming pulsewidth)?Quote:
Originally Posted by 1project2many
seems like it would have been less code intensive to just setup a small 2D table or a threshold/multiplier combo to account for that though.