1997 F-Body ECM

[Tom H]

Getting back to my work to understand this PCM now that the majority of the reprogramming stuff is out of the way...

I have not yet found a way into the code space of the TPU if this is even possible. Second best is to understand the register set and how the code makes use of it. The TPU is mapped between $1400 and $16FF in the ESide address space. The CPU uses CSI01 chip select for access.

The first eight word locations set the base injector pulse width. To gain an understanding of this parameter I used a number of my past projects. First, I make the PCM think it is running an engine using the Optisim I wrote about earlier. This tool basically allows me to drive the PCM with Optispark, MAF and VATs signals. I also needed to default some inputs like TPS, ECT, IAT, MAP. I then changed the ESide programming to prevent it writing to the locations of interest. The reprogram also includes a custom 8192 baud interface into the ESide that I use to control the location of interest.

OK, so with that setup I found the following:

• $1400 through $140E are word wide locations
• They are clocked with a derivative of the CPU clock -> EClock -> TPUClock: CPU clock /4 = EClock, EClock / 48 = TPUClock
• Programming the pulse width longer than the time it takes for two revolutions of the crank results in an overflow condition. A steady ON condition with a one cycle pulse OFF.
• The code programs these eight locations in the TPU interrupt service. Other code accesses to these locations mostly just turn them off where the engine isn't running

The minimum pulse width is 184uS when programmed to$0000. I believe this is the time it takes for the injector to react to the input signal.That is the lag between turning the injector on and fuel starting to flow. In any case, here is the measured and calculated results of the testing. The error is most likely my army surplus scope that was last calibrated who knows when.


InjectorPW.JPG

It will be important for me to find where the minimum pulses is set. Folks who want to change injectors might find that modern injectors respond quicker or that larger injectors are slower. My best guess is that this will be based on the TPU Clock and should have a setting around $0C --> 12 decimal.

-Tom

[Tom H]

There is another group of eight word locations addressed at $1580 thru $158E. These eight locations are read and clear only. What I mean by that is writing any data to the location clears it. The value you read is a reflection of the previous data written to $1400 - $140E words. The code uses this to implement the decay of prime and other adders to the fuel. Reading the locations also cause them to clear.

Still looking for the minimum pulse.
-Tom

[Tom H]

Think I will try to change the way I am posting this info. This all is getting quite complicated, probably I should write up a pdf with all the explanations in one place.

For now, here are some details of the end of injection. Locations $141C, $141E, $143C, $143E set up the end of injection timing. The first pair $141C, $141E are programmed with a number from 1 to 6. This number represents the number of 90 degree delays inserted. The second pair $143C, $143E further delay the injection by 90/16 or about 5.625 degrees. For the second pair programming ranges from 00 to F0, the lower four bits are not used.

I will go on now to separate the pairs. I believe it likely that they apply to bank 0, bank 1, but I will find out. I also plan to find out if the lower nybble of the second pair is active in the TPU or not.

Last up is location $14D6. When this location is written with bit 14 set, it updates the TPU to use the parameter pairs described above. You will see it's use in the code like this

		LDAA	$14D6			; READ UPDATE REGISTER
		LDD	$14D6			;
		LDAA	$40			; ADJUST EOIT COMMAND
		STD	$14D6			;

Gradually I am discovering more and more of this part. Process is long and painstaking though.
-Tom

[kur4o]

regarding the eoit target, we found some data from earlier version of tpu, the target is 720 - (x * 5.625 ) and is applied as degrees BTDC,
This is where is stops spraying so start of injection is much complicated math, using commanded bpw engine speed and is applied per cylinder, due to sequential system.

I think there were some 720 degree counters that are also sent to tpu, but need to look at some old diss.

Max time is also capped,and that is the injector DC%, you can do some math based on low res pulses time * 8 that equals max BPW, if above that EOIT will overflow, since it will overlap on next cycle.