High gear coast down does seem to fit well with 2.73 gears: https://datazap.me/u/tequilaboy/mmal...7392-7431-7408

Must be some slip present during acceleration in 2nd gear which lead me to believe it was a 3.07.

 

.

Sounds like a plan. My 383 with a super hogged out, oversized, siamese TPI still reacts similarly to the PE values of a stock 350. By that, I mean the lower RPM ranges and the mid rpm ranges (where you can't get much relief from knock) required PE fueling. I only made it through 1-2 pulls on a dyno before it started knocking -- meaning that was MY "ragged edge". I tried lower values and observed HP loss. Cooling was/is the main secret for keeping mine up close to 30-deg timing. I'm talking about the ranges where you're down around 24-deg. I tend to agree with you that higher gearing -- along with pushing air through a proverbial straw raise "virtual" load. If you're running 6" rods vs my 5.7", slight lighter weight pistons might also be more prone to knock? Whatever the case, I wouldn't be able to judge the compression you're running w/o more info and that really wouldn't help. It is what it is. (TB seems to be saying just that in a different way?)

I think you'll probably have a whole lot easier time getting good power of that engine using a short-runner setup...probably an HSR, MR, or even the new Edelbrock hi/low ram? I think we see its not a good match to a stock TPI.

(That said, I remember reading some large engine 454? actually was setup with a TPI years ago...in a truck. Probably has a LOT less compression...and probably more retard in the cam? At one point, I tried researching the cams used in 89 (and 88). While I couldn't conclusively find the orientation, I found one company (Oregon cams maybe?) that made matching cams. From that I got the impression our stock 350 engines ran some retard in the cam. It would certainly help build HP in higher rpms. Comparitively, the cam you have now might be advanced...as many street engines are. So, you might have several things preventing the TPI from being an acceptable intake for that engine

I wouldn't plan on keeping it long. So, fight with it accordingly.

 

Still some knock retard even at 23 degrees advance: https://datazap.me/u/tequilaboy/mmal...7097-7926-4565

It may benefit by going a bit richer. You could try a PE target AFR of something like 11.5-11.7:1 (11.49:1 = +5.47%) (11.71:1 = +3.12%) under 4,000 rpm, but you're already flirting with the line of death. 12:1 = 0% was initially chosen for simplicity (and line of death considerations) and is generally rich enough for most naturally aspirated engines at peak torque.

I assume that you're running a fuel with some ethanol content (E10). Setting the stoichiometric afr ratio to about 14.12:1 (from 14.73:1) will provide general enrichment and also enrich the PE target proportionally. This will also shift the BLM down a bit.

Cells 10 (and 11) shown here: https://datazap.me/u/tequilaboy/mmal....07&tmax=11.00

Note: Cell 10 (typical cruise cell) is of particular interest since the BLM value from this cell is stored in SAM cell B and is used to initialize cells 1-15 on the next ignition cycle. Cell 0 is stored in SAM cell A and will initialize Cell 0 only.

Cell 0 can't seem to make up its mind: https://datazap.me/u/tequilaboy/mmal...0.00&tmax=0.41

Side note: if the engine really is that knock sensitive, you may want to consider using a more knock resistant fuel. ~E33 is simple to mix up if you have access to E85 and should be around 96 octane. Mix 2 gallons of 93 for every 1 gallon of E85 and set your stoichiometric target AFR to about 13:1.

 

Proposed spark table (w-125% Load range shown for visualization purposes)

With the virtual load approaching 319 now , you have a huge range of load (or virtual load) variation (from 208 to 319) covered by a single column in the main spark table. Re-scaling the load by 125% and revising the column (labeling) will provide additional tuning resolution (4 columns) for actual high load operation.

With the re-scaled load (255 = 1.0 gm/cylinder as opposed to 255 = 0.8 gm/cylinder), the internal load signal will behave just like the virtual load 1.25 signal shown here. The virtual load will be unchanged (255 = 0.8 gm/cylinder and 319 = 1.0 gm/cylinder). Hope this is clear.

https://datazap.me/u/tequilaboy/mmal...2857-1696-1699

Might need to tone things down a bit above 3200 rpm in the 200, 220,240 columns. The above table was not intended to the perfect be-all, end-all spark advance table, but to provide additional flexibility over a wide range of load variation.

 

I have to admit being curious how [to combat knock] you went DOWN from 20-deg advance in lower rpms + added 12% fueling in that same rpm range THEN...
BACK to prior values of 20-deg timing + no add'l PE fuel AND ended up back where you were -- with NO knock?

It SEEMED like you said those things helped -- then they weren't even needed?
I'm sure I missed something!

 

Gregg, for your curiousness on the spark table, I did do some experiment runs with the very low advanced [down to 9] on some of the lower RPMs, but I never had knock until 2500 RPM from the data logging, thus went back to the 20 and original fueling until after 2500 RPM. I think tequilaboy has figured out some of the limitations I was running into to find tune spark at these higher loads by improving my fidelity to tune in those areas. Loaded up his suggestions here and will be reporting back with logs and observations.

 

Went for a run with the suggestions from tequilaboy and the engine feels like it has opened up a new level of power under full throttle - the change to AFR ratio to 14.12 seems to have made a ton of difference along with the spark suggestions - thanks for the help, this is a drastic improvement over anything I have run so far on this engine.

Looking at the datalog, I see the effects of the scaled load variable, it now has granularity where it was flatlined before. On the run, the load variable maxed out at 224 at 4600 RPM under full throttle and never got to the peak at 255. This is definitely the answer, which also reduced the knock and knock retard. Seems I only had one knock, right at 3200 RPM and 214 load when I was running 26 spark advance - so maybe I change that one cell down to 24 -or live with the very rare knock.

 

Example data from latest log: https://datazap.me/u/tequilaboy/mmal...2131-3270-3290

Note the behavior of the load signals, afr desired, inj bpw, and O2 volts. It is now able to run richer than was possible before. Some knock retard is still present here, but that is not the main focus of this example.

We have covered some advanced topics here, so it is understandable that there may be some confusion.

Line of death (math limitations) -- Fixes were applied to the latest bin and are included in the test data above.
Maf limitations (briefly)
Load limitations
Virtual signals
Load scaling
AFR compensation for modern fuel
PE Fuel
PE spark
Main spark
PE entry vs tps
BPW method

 

Fuel trims are now trending a bit rich but with a much tighter range than before (108-129): https://datazap.me/u/tequilaboy/mmal...5708-6426-4939

Cell 2 is the only real outlier and not a big concern.

Cells 10 and 11 look pretty good: https://datazap.me/u/tequilaboy/mmal....39&tmax=11.00

As does Cell 0: https://datazap.me/u/tequilaboy/mmal...0.00&tmax=0.49

 

New load signals added: Load var 1.25 (cur) and Load var 1.25 (old) which track with the original virtual load signal as expected.

Data filtered for Load var 1.25 > 250 (and Load var itself > 200):

https://datazap.me/u/tequilaboy/integer-virtual-load-variables?log=0&data=4-8-9-19-20&solo=4&mark=127-134-222&trim=19&tmin=249.95&tmax=307.00

You can see 3 knock events in this log (and two apparent data glitches which can be disregarded).

 

Virtual Signals and Load Scaling: Are these two categories possible with $6E tuning, or is this unique to earlier/different bins? (I'm surprised you can change the load values versus [just] the spark values in the cells!

 

Here is the $6E load scalar address (from arap_hac.src disassembly):

LC5B0 FCB 80 ; 1.25, Ld VAR FOR SCALING LV8 SCALE FACTORS
; cal = arg x 64 (64 is correct, however when using this value as a factor (80 * 60 = 4800) works better for calculating virtual load directly from the maf and rpm signals). The internal math must be slightly different but arrives at a similar result as can be seen)

Maf gm/sec * 60 sec/min = Maf gm/min. (Maf gm/min)/rpm = Maf gm/rev. So thinking the internal math is really like this: Maf gm/sec * 60/rpm * 80/64 * 64.

The virtual signals shown in these examples were created from existing data signals using google sheets after the data was first exported to csv using DataCat from the appropriate DataCat dcf file. The csv file was also edited and uploaded to DataZap for display (and sharing) purposes.

In TunerPro, it is possible to create virtual signals in the adx file for direct display in TunerPro or export.

 

I'd have to refresh my memory on assembler (from 1985ish). But, maybe not necessary if you convert to hex and search for the "string" of load constants with FIND/REPLACE? Or do you write assembler these days? Now that I think about it, the load variables can't come out of thin air, so they (too) should be able to be displayed/edited with TunerPro, right? Maybe you're seeing if I can figure that out from the disassembly instructions? And your recent help editing to change MAF table display values!!!