doh.... that's what I get for not having the software installed on my work laptop...

 

I'll take a log on my drive in to work tomorrow, but, did anyone confirm if the vehicle speed cell works on USDM EVOs? I tried it on the drive home tonight and it was way off. With 2x scaling, it was too high for mph, but didn't correlate to kph either.

 

I think someone else said it was off on the USDM. The ECU code is quite different with huge chunks of code missing from JDM - I think a lot of the OBD II stuff isn't there as it only has one cat and one lambda sensor. Doesn't complain if you unplug the wastegate solenoids. Has lean spool. Lots of minor differences. JDM one is good to modify though, like a 90s ECU in terms of what it lets you do to it without complaining.

 

Okay, here's a log from this morning, using your new Data.xlm file. I got some knock (9) on the pull, but I think I may not have let my car fully warm up. Plus I got some horrendous knock (38!) when I accidentally hit the rev limiter.

Anyway, the way it looks to me (I'm a novice, remember), it seems like the Load lines up really well for Timing. I don't recall how you said to decode the AFR, so I'm not sure if that one is working for Load (128=14.7? I can't remember).

Anyway, let us all know if you think this is working. If we've finally figured out an accurate way to determine Load, that would be awesome.


** EDIT: I removed the file after I realized it was my baseline file (a Vendor map). If you want it for Load Calc analysis, PM me. **

 

Looks good to me, goes a bit wacky on over-run, but that is not where we are really struggling to determine load. The 2550 is empirical from my own data to make it line up with the timing. Are you able to post your timing map in the areas where it didn't knock?

On another note, looks a bit rich, but advanced? What boost does it run? Obviously your MAP sensor isn't the same as mine, maybe it is a 1 bar MAP sensor if present at all?

What gear was this as the speed did seem to follow RPM, you could help work out if the speed reading is correct on the US model...

 

Sorry, I guess that was rather stupid of me to think that you could evaluate accuracy of the calculation w/out the maps! I'll post the relevant sections shortly.

 

jcsbanks,

I was thinking about this a little more today.

I don't think you need to include any lag times in the formula. The IPW already has any deadtimes, etc, inherently calculated. The IPW is the resulting pulsewidth that the injector needs to be open to compensate for these delays.

So, the formula should just be:

C*IPW/AFRMAP
where C is some constant (so far, 2550 based on your data)

The constant can be fine tuned and exactly found through more logging on several different cars.


Eric

 

It is because IPW includes the lag time that I am subtracting it. I don't want it in there, because I have fuel, and I have AFR, so I'm trying to get air.

Agree?

Maps would be good Rob W.

 

I am going to have some 'free time' tonight to do some logging. Is the updated .xml file for EVOScan located anywhere in this thread?

 

Yes post #56, you need to log AFRMAP and IPW to calculate load. Ignore MAP, will be interesting to see how vehicle speed compares to your speedo.

 

jcsbanks,

Yes, I see where you are going. OK, makes sense to me.

I was just trying to think of a formula that doesn't need to take the deadtimes and latencies into affect. Because with this formula, the forumla would need to change with battery voltage, for example.

I agree with you though, we should leave it in. It should give a more exact load value.

I'll look at some old DSMLink logs and see how well the IPW correlates with g/rev. I would imagine it would diverge moreso in lower RPM ranges, where the latency is a bigger part of the injector pulsewidth. For example, at idle the latency may be 10% of the IPW, whereas at WOT it would be less than 1%.


Eric

 

See attached file (same file as before, but with maps added).

Boost is set at 20psi with an MBC.

 

Oh, also, regarding vehicle speed. I tried to log some numbers this morning, but it was just too hard to compare speedo to laptop with traffic and all. I'll try again tonight.

But, my impression is that it's actually closer than I originally thought. (maybe offset by 10mph over the whole scale?) I'll try to figure it out tonight.

 

Rob W.,

Great data. Looks like the load is following pretty nicely.

I also checked a couple of points pretty quickly where you had knock, and the Evo ECU seems to be acting like the DSM ECU so far (about 3 counts of knock = 1 degree of timing pulled). I'll have to look at a lot more data to verify this, though.


Eric

 

Will be interesting Eric.

Rob, looks to fit indeed, as would 0.35 deg retard/knock count approx?