dimented24x7

Quote: Originally Posted by CalEditor If you wanted to run a 350 TPI you could use the 454 segment (no crank sensor). That is interesting... Which years did not use the crank sensor? How did they work, then, like the older CPI systems with a standard HEI dist.?

This ad is not displayed to registered members. Register your free account today and become a member on ThirdGen!

CalEditor

My mistake I should have said Big Block. They still used them in 1998 on Big Blocks with EFI

dimented24x7

Its been a little over a year now, and I think Ive gone about as far as Im going to go with this thing. I've covered all the major tuning food groups: spark, fuel, closed loop fueling, EGR, idle speed/IAC PID, and misc. constants and routines (no e-trans coding, running a stick). Theres still alot of unknowns as Im basically trying to wrap my mind around something that was designed by an entire engineering division at GM/Delphi, but I think theres probably enough there to tune with. Now that Ive gone through it, I can comment on the code overall.

One of the first things noticable about the code is that its for a 32-bit MCU, and is much more complex than the previous 8-bit HC11 based ECMs/PCMs. For one, it has the OBD-II communications in it, which adds a decent ammount of overhead and routines to the code overall. As the PCM also has SFI, the fueling and other routines are driven by the CKP pulses, which arrive 45 degrees ahead of each cylinder firing. This means that the fueling is calculated in REAL time for each cylinder and outputted to each fuel injector. Additionally the fueling is calculated by either the MAF (stable engine operation) or by SD with varying corrections to the calculated VE from the MAF sensor when operation is unstable. The fueling routines are also the biggest change with these PCMs. Everything is done in terms of cylinder airmass, including dynamic fueling for transient engine operation and crank fueling. No more injector pulsewidth or duty cycle when the MAP or TPS changes.

The idle routines are basically an evolution of the routines in the $0D computers. The idle speed is calculated based on various inputs (cool temp, selected gear, A/C, alternator/torque converter drag, etc) and passed off to the PID. The PID itself is linearized in terms of percentage of total IAC airflow, just like the $0D, and primarily uses a proportional/integral control approach.

The spark routines are much the same, with many enhancements. One of the biggest changes is the addition of the CKP sensor. Now the timing can be calculated exactly for each cylinder as the CKP pulses always arrive at the midpoint between each cylinder. This also means that the location of the distributer isn't too critical. The PCM assumes that the distributer is off, and calculates its location in degrees relative to TDC from the time intervals between the crank and cam pulses, and compensates as needed. Pretty neat, eh?

There is one big dissapointment, though. The closed loop fueling routines. These are basically the same as the old routines. The BLMs make a second appearance as the long term fuel trim and the INT is the short term fuel trim. The routine is basically split down the middle for each bank of cylinders. The code looks like it had potential as it had the beginnings of improved logic and traditional PID control, but it was hastily slapped together. Some routines are riddled with errors and logic flaws.

Never the less, overall its not too bad, and it should serve its purpose quite nicely. In the next few days, Ill post the updated code.

scuzz

That's some really in depth, cool stuff, dimented.
Thanks for taking the time to help all of us.

93V8S10

Dimented, just wanted to say Thanks for all of the work you have done with the $0D systems and now with OBD II. Your ability's are truly amazing!

dimented24x7

Thanks. I cant wait to get it in the car and try it out.

Hopefully it will be usefull in trying to figure out how the later PCMs work. One thing that has been obvious in working with the PCM is that this computer is pretty much the outer limits of what I can accomplish due to its complexity. You'll get a good idea of this when you see the code when I post it. Its amazing how much these computers have evolved over time since the first CC carb/TBI ECMs.

Zalfrin

not related to the PCM hacking, but check out http://www.uvscanning.com for a customizable scanner that works with AVT units. You might also be interested in http://www.powrtuner.com The powrtuner is defunct now and was mainly for W-bodies, but there is a lot of good info on the board. Luckily, I have a w-body and the software isn't locked down like HPT, so I can hack the bin as much as I like and reflash with no problems.

dimented24x7

Have you had a chance to try the UVScan? That looks cool. I think I might buy an AVT scanner and try it out.

The only possible problem could be doing crank sensor relearns and similar procedures. Not sure if they support that like some of the commercial ones do. On the other hand, I found most of the code that handles the relearn, so it may be possible to do it by hand thru hyperterminal in windows.

Zalfrin

Yep, I use it all the time for scanning. It's a much better scanner than the one that is built in to Powrtuner software. There are a ton of extended PIDs mapped, and you can enter your own if you find something that is not already mapped.

I don't think CKP relearn is in YET, however the bus traffic associated with it has been captured and will be getting added in to UVScan at some point. Right now I think the only real-time controls are turn on/off CEL and clear fuel trims. UVScan 2.0 is in Beta Testing with lots of enhancements. Gauges, support for LC1 and PLX wideband logging (using a seperate serial interface from the AVT841), etc.

The really nice thing is you actually have support from the developer, he's around a lot and take's suggestions. It is a spare time project, but he has been pretty good about getting out updates.

dimented24x7

Zalfrin,

Is there a particular AVT scanner thats the best to have? It seems like the 841 is the one thats most referred to. Also, it looks like it has an A/D input. I assume that this can easily be incorporated into the datalog? It would at least save me some time by not having to use the linear EGR input to read in the wideband.

HaulnA$$

See post #54 in this thread. The 841 (rs-232) or the 842 (USB) are the only ones worth messing with. JMHO

dimented24x7

Thanks. Forgot you had mentioned them earlier.

Zalfrin

843 is also a possibility (ethernet on the input side). Not sure how it works with uvscan since it looks for a com port, maybe theres a virtual com port set up? I know one of the guys (a moderator, actually) on uvscan forums is using an 843.

841 is the most common as that is what the powrtuner software (where this project is coming from) was based on. Yes, they all have 3 A/D ports and you can log those via uvscan, as well as set up fake PIDs (so for example you can log mg/cyl directly rather than using a spreadsheet to calculate, or similar).

dimented24x7

I posted the latest iteration of the 98 vortec code on moates.net. Its posted on the source code and hacks section of the file server. Its not complete, but most of the major elements of the engine calibration are commented. Its also not of much use yet as there are no freeware flash utilities that support these PCMs, and the major tuners encrypt their bins. If tunercats hadn't sold the OBD-II tuner to Junk Performance, which promply made it ****ing useless, I could also have written a freeware TDF. Instead, I'll just be using tunerpro and the old trusty standby, the PP-II. I'll post up an XDF as well when Im done with it.

dimented24x7

I finally got around to starting up the car with the SFI.

Specs:
350 with trickflow 195 heads
Edelbrock victor EFI with 36 PPH injectors
Isky mega hydrolic, 221/221 .465/.465 with a 108 LSA
98 blackbox

I had lots of fun breaking in the cam when I first started it. The first three or four starts, it did nothing but chug a couple of times and crap out. After switching out the timing tables and some other things, I did get it to start, but forgot to disable the transmission abuse logic, so it would cut the fuel out above 2000 RPMs, which made it interesting to try to break it in. Did get it running OK, though with some major changes to the timing and idle calibration. I initially had the stock settings, which basically blend the MAF in when the operation is steady, but since I have a cam with a decent ammount of overlap, the car was pretty much running in SD. It idled OK despite running pig rich, but due to all the lag introduced by the fueling logic, as soon as I tried to touch the gas, it would want to stall and pop out the intake. Conversely, when I closed the throttle, it would head all the way out to the rich limit, and want to stall again. I can see why the LS1 guys just lock these in SD. The blending matrix is a mess on motors with cams cams. The matrix is so complicated that it looks like it would be nearly impossible to street tune. My solution was to disable all the convoluted fueling logic and just lock it in mass airflow. Not surpisingly, it runs much better. No lag, and the engine idles within a few percent of my desired lambda/AFR. The only change I had to make was to add a little transient TPS airmass to compensate for the brief lean tip-in due to the fact that I have a 1000 CFM tb on the car.

I also cant use closed loop. At idle the LTFTs and STFTs peg out and flood the motor with gas. The stock closed loop code in these sucks. There are errors, and there is very little control over when and how much corrections can be applied. I used to just run closed loop at part throttle before with my old PCM, but it looks like Ill be full open loop for now.

Heres some vids of the car after the break-in. You can see/hear that in order to rev the engine, you have to gradually build it up, or it just craps out. Also, I had to hold the throttle open half-way just to get the motor to fire up. Otherwise, it would just chug and stall.

http://www.youtube.com/watch?v=VvYUXjcA8uQ

http://www.youtube.com/watch?v=HMKfoI1Pbus

Now running on just the MAF with no blending or filtering. I can snap the throttle open with ease, and the motor runs much leaner (cant see it in the vid, but its around 13.7-14.1). It also starts more normally as well. The engine idles with around 12 inHg of vacuum, which put it way out on the VE tables before. Runs much better with just the MAF. Sort of makes you wonder why GM even bothered with the mixed mode fuel logic to begin with Might lower emmissions a bit, but it brings a lot of excess baggage with it.

http://www.youtube.com/watch?v=t6cXmHkANEM

dimented24x7

Heres a snapshot of the blending matrix. This probably isn't defined in such packages as HP Tuners, but it should be. The table only represents a portion of the logic as there are also other routines to handle applying the MAF and so fourth. This probably works well on stock engines, but if you have something thats a long way off, its probably best just to go straight SD or MAF and skip the headaches. I dont even know how you would tune this on the street or a dyno. It not only handles how the current fuel mas is applied, but what percentage of the fueling is the steady state airmass, MAF airmass, transient MAF/MAP/TPS airmass, and dynamic airmass. The numbers on teh side represent which quadrant the MAP and RPMs are in. The letters are the multipliers. In addition, you also have to be careful that the multipliers add up to unity, or the long term fueling will gradually head to full rich or lean.

Attached Images

BlendingMatrix.GIF (38.0 KB, 21 views)

afgun

What are you using to edit?

dimented24x7

Tunerpro. I wrote XDFs for the spark, idle, fuel, system, and emmissions. Once they're done, I will release them (still working on some of the emmissions stuff).

afgun

NM; re-reading your other thread...

__________________
--andrew