If you think about it in reverse:

why dont we just have 2 rows? 1000rpm and 10000rpm... then let the ECU interpolate linearly over the revrange..

:wink:

 

I have successfully ran my car in the drivability range on a 4x4 section of a standard map just to see if it would work. I ignore all knock below 140kpa so it might not have been the cleanest but it did work. With the big map I will be able to run the map out to 405-420kpa in some nice steps of 15 or so, and 500 kpa for the racecar in steps of 20. The RPM is good to 10,500-11k as pointed out for the race car but I can have very precise control where I need it, hold timing to a certain value, set up dual control map switching by setting one portion of the big map to 92 octane timing numbers and let say Q16 above a certain Kpa value, then repeat on the map switching if need be.

Now thats just load, when I look at the timing numbers and some of the places my car has developed knock zones since the stock turbo to the 35R it becomes even easier. Before it was spool and then the 6k range. I had 5750, 6k, 6250, 6500, 6750 etc. Now I worry about 4750, 5, and 5250 and then again about 7750 or so. There are alot times I can run 9* at 6k on a stock turbo and 11 or 12* at 6500 but I HAVE TO RUN 9* at 6250 still.

Dave I owe you a semi load of beer for this latest greatest invention

 

Ah, very interesting. So the ECU isn't really interpolating, but just switching states mid way between cells?

What about when you step in more then 1 degree. Say you have timing increase 4 degrees in a 500 RPM gradient. Does it add a degree every 125 RPM? or does it add all 4 degrees right in the middle?

 

It adds a degree every 125 RPM in your example.

It is doing interpolation. The example Tephra gave is just showing that the timing maps only have a 1* resolution though, so the interpolation changes the value half way through.

 

will the big maps only be on the alt. maps or will they be on both the std. and alt.?

 

both

 

The nice thing about more cells is you get to manipulate the interpolation curve. Right now,with only 2 cells, the curve is linear. Adding cells between them allows you to ramp up timing sooner (or later) if you have the need/desire to do so. My car was always touchy between 5.5-6k RPM and there was a 5* jump. The knock happened around 5750 RPM so it forced me to lower the timing at either 5.5k or 6k to be sure I didn't have too much timing in that area. By adding a row at 5750 RPM with the same timing as I had at 5.5k I could hold timing steady from 5.5k to 5750 RPM before it ramped up, allowing me to get the most timing possible at 5.5k and 6k without 5750 RPM being a bastard,lol.

 

I'm a bit late to this party, but I'm on board with finer adjustment resolution. I have seen a few 'fancy' standalones that can do .1*
I.e. I have a spot right at 4750 where advancing a whole degree causes 3-5 counts of knock. I just hold out the advance, but it takes away from power above that rpm point.

i think the OP was trying to see if we could rescale the routine for timing by lowering the max value allowed and increasing the step increments. Not that I know how it works, but it should take up the same space.

I seriously need to start delving further in assembly...:/

 

Its been pointed out that some stock ECUs do this already (Subaru being one). I would imagine the Evo is, but we are limited in both the datalogger not dealing in less than whole numbers and ECUflash not being 100%. Look at the fuel maps in the raw values vs the AFR target and you'll understand what I mean.