2 Attachment(s)
TBI 7747 Idle Fueling Issue while Cold - Too lean, AFRs not responding to bin changes
Been lurking here for a while as I've been doing TBI tuning on/off over the years. Searched a bunch on this forum, but haven't seen other treads with the same or similar issue.
Basic problem: Not running as rich as I'd like at idle when cold. Results in a stall if driver or reverse are engaged shortly after startup. Also the crank time is extended, but one problem at a time. Changes to the target AFR in the tune don't seem to have the expected effect. Idles fine warm, at the desired AFR (13:1). Cold it also tries to idle at 13.5-14:1 and does so poorly. I'd like to get that down into the 11:1 range when warming up, but changing the Open Loop AFR vs Temp table doesn't impact Actual AFR. I've tried lowering that table by 2 across the board and AFR doesn't appreciably change. This is after the choke tables should be out of the picture (idling for more than a couple minutes). Has this issue until about 70C coolant temp where the engine is warm enough it'll idle okay at 13:1.
Engine info: 355 w/ vortec heads, 9.5:1 CR, performer RPM intake manifold, carb to tbi adaper, headers, dual exhaust, 212/222 hydraulic flat tappet cam. GM 80 lb/hr injectors, stock fuel pressure.
I'm using a stand alone WB02 (LM-1) and watching it as I'm connected with an Ostrich 2.0 and ALDL using TunerProRT. I've used the WB02 recently on a carbed vehicle and it seems accurate (I free air calibrate when I move it between vehicles, engine behavior matches expectations when AFR is too rich/lean).
Latest bin and log file from cool idle (14C) where it is running leaner than desired attached. I'm thinking there must be something overriding the target AFR tables, but haven't found anything obvious yet. If I heavily skew the main VE map in the idle region, it will richen up some and idle better. However, that produces an overly rich idle when warm. Running exclusively in open loop (255 deg C enable value for closed loop).
Any suggestions would be appreciated!
2 Attachment(s)
Progress - learnings and a rant
A few things that have seemed to have improved cold idle and general performance:
1. Didn't find an exhaust leak, so I ruled that out.
2. Did remember that the gas in the tank was at least 6 months old and likely bought during the winter...not much volatility left. Tank was low, so I put nearly a full tank of fresh in and that really helped. Can hit 10:1 on command now.
3. AE needed major increases. Thanks to this site, found that 5000+usec (5+ms) values are often needed on aftermarket intakes. Huge help with cold and warm tip-in and upshift at steady throttle performance. Was always going lean temporarily, especially on upshifts.
4. Went back to a stock non-async bin and applied changes from there (again, this forum was a big help finding that). Saw Async briefly on one of my earlier bins, then a major lean spike and nearly stalled during coastdown. Do. Not. Want.
5. I'm in Open loop all the time (WB02 tuning), so I set the idle target AFR to the same as non-idle open loop AFR target (14.1:1) to prevent step changes in pulse width when dropping into "open loop idle" mode.
6. Adjusted VE table significantly in a few places. (high map, low RPM)
7. Several other changes (PE target AFR, AE CTS multiplier, choke AFR at low temp, idle air tolerance, etc)
Slightly different problem now. Cold start seem better (at -3C), but I do get a stall if I drop it in gear (drive or reverse) without letting it idle for a couple minutes. Made some changes in the latest cal that I haven't tested on a cold start, so we'll see if they help (AE CTS multiplier, MAP AE PW).
Now for a rant. Context - I work on automotive embedded controls as my day job. This includes writing software and doing calibration work. When you write the software and know exactly what the calibrations you added do, there's no guessing involved (well, not much anyway). If you need an extra variable to log, you add it in the software and log it (whether internal RAM or over CAN). Trying to do the same type of work but on a 7747 ECU is maddening. I give major props you those of you who have more patience than I on the guess and check work needed to make these ECUs work on modified engines, especially without a WB02. Between the lack of variables (what all parts of the equation are doing what to determine the total PW?, what threshold am at I for AE?), the slow rate of data (1 second/sample when the ECU is calculating at ~12.5ms per loop), not knowing how all of the different calibrations are actually used or what they even are, and on and on, this work many of you are doing is much more difficult than doing it at an OEM or wherever the code and cal are being developed. Ostrich and EEPROM chips help, but still, ugh.
FWIW, this problem isn't limited to OBD1 GM stuff...EFILive/HPTuners and the like all suffer from many of the same issues. Lack of full control over spark and fueling, limited data logging, can't actually change the software, etc. There's just a bunch of stuff running in the background, not well understood or even known, that is impacting engine behavior. These tools are almost always working from a limited set of parameters, many with limited descriptions/understanding of their actual function.
Thankfully the next vehicle I'm doing aftermarket EFI work on has a megasquirt. Open source code, fast data logging, built in WB02 closed loop control, all calibrations defined, etc. Sure, host of other issues when you start from scratch, but at least there won't be a bunch of stuff lurking in the background, changing things without me knowing about it. It'll be a nice change from the ole 7747.
bin and latest log attached if anyone made it this far :)