tephra

iTrader: (6)

yeah I was going to...

I figured it would change the WDGC to quickly otherwise...

mrfred

iTrader: (50)

This might go back to how you thinking of doing averaging. Were you thinking of averaging the MAP sensor reading from each cycle through the ADC routine, or were you perhaps thinking of measuring the MAP sensor several times each time the CPU traverses the ADC routine? Maybe keep the original ADC read, and then add another ADC read in your add-on code. Then for each time through the code, you'd have two reads of the MAP sensor.

tephra

iTrader: (6)

yeah well my averaging routine has the last 3 ADC readings stored in RAM.. not hard to expand that..

of course a lot of this will be trial and error - but my feeling is that the WG needs probably 1/10 sec to react to wgdc changes..

jcsbanks

It depends on whether you think the "noise" is electronic/ADC or whether it is the pulsation due to the airflow from the inlet manifold through each inlet valve.

tephra

iTrader: (6)

bit of both...

if you just so happen to sample the MAP when the pressure drops then you dont get a true representation of whats happening.

so if we avg it out a bit then that will smooth things a little..

jcsbanks

You'll certainly need a slower response time and can get away with much more filtering if you use it to control boost rather than for fuelling/timing for transients.

Imagine a standard car at peak power running say 1500Hz on the MAF sensor. With an exponential averaging over 10 ticks then you still have a rate of 150Hz which is quicker than one engine revolution even at the RPM limiter.

For a boost controller I would suggest you want to update it every time the solenoid duty cycle is complete, otherwise you're wasting calculations. On most ECUs I've come across this is about 15Hz, so you do have ample filtering opportunities running it off the slow main loop.

tephra

iTrader: (6)

mmm well once I start in ernest I will experiment with different loop timing...

tephra

iTrader: (6)

right back on this topic. I hope to include this in my next patch, which will be out soon and have a this, 2byte->1byte load mod, as well as some fixes and cleanups.

so I want to finalise the averaging algorithm so I can release it.

without posting graphs, weighted 5pt, weighted 3pt, and normal 3pt (SUM/3) basically give the same result within a few pt's of a psi.

so in keeping with KISS principle (yes I know MrFred you said it at the beginning lol!)

I think I will just go with a regular 3pt average, ie =(LAST, NOW, NEXT)/3

So we lose one "future" map reading, but at the rate we run the avg routine and get new ADC data I don't think it will matter to the boost routines.

anyone have an objections?

ps the 2nd graph I posted on pg1 was "laggy" because I screwed up the plotting points :P hehe

tephra

iTrader: (6)

dan l

iTrader: (3)

I believe it is intake manifold harmonics and normal pulsations as well. The intake manifold is pretty much a musical instrument with respect to vibrations.

I for one would like to see the normal jittery map signal and log this. I don't want my boost spike that I'm trying to tune out being lessened by the math. It would also be nice to have the option to turn the map averaging on/off. I'm still unsure if it will help or hurt when tuning the boost but I'd like to have the option to turn it on/off as with the rest of the tephra mods.

Thanks for the great features guys, so far your doing a great job of implementing things people need and with no bugs as far as I can tell.

tephra

iTrader: (6)

yes option to enable/disable is in.

problem is that with the boost routines if you happen to run them on the spike (up or down) then the error correction goes haywire and starts changing WGDC needlessly..

also you will still be able to log both real and avg map signal, its just the avg will be used by the boost routines... (if enabled)

tephra

iTrader: (6)

ok I implemented a simple averaging algorithm ((last2 + last1 + last)/3) and this is the result:



Much smoother.

ps - Ignore my boost curve, its not tuned properly...