RomRaider

Thanks.

Yes. I noticed that you were using Kelvin. I'll probably use Rankine since most of us in the US are accustomed to Fahrenheit.

Haven't implemented it yet. My logs don't have IAT so I need to add a box to enter alternate data.

I'm liking this tool to look at partial throttle timing!

Here is a log that puzzles me a bit. According to the spreadsheet, I show no knock events. Yet, when I scroll through the log, I see that I have negative feedback at times 477703 and 646578. Shouldn't I be correcting for these events?

This spreadsheet was meant to look for knock due to base timing which is a function of rpm and load.

I excluded knock events due to throttle changes. Those have to be addressed but I think they should be addressed somewhere else.

For example, let's say your tip-in is messed up and you get a knock when you hit rapid changes in throttle. That throttle change can occur anywhere on your base map. If you start pulling timing from each area that you see knock, it still wouldn't solve the problem.

Anyway, that was my logic. Not sure if it is valid.

Okay, that sounds logical.

mickey, NICE work!
way better than I imagined it could be!

The other neat feature that people should be aware of is that any value on Mickey's Summary tables on the Knock spreadsheet can be clicked to see the individual datapoints that went into that specific summary cell. I use this to decide whether or not to believe the knock. one row of -5 is probably BS, whereas 20 consistent rows of -1.4 is probably really Knock.

I am goig to try this guy out to fine tune my knock and AVCS.

I'll let ay know how good it works.

Gabe

I added IAT to the spreadsheet so that it can calculate theoretical air density based upon IAT. There is a box where you can enter in a default value for IAT if you didn't log it.

This spreadsheet is great! Thanks Mickey. I am also using your road dyno which is great too (I added it to Airboy's sheet and posted it up). Now all I need is a better tool to dial in intakes and MAF scaling. Especially for the learning ranges. Don't you have one of those made that can be used to tune LTFT using STFT within say 20-30 minutes?

Also, does anyone know what the Feedback correction persistence value in MS is for an 07 STI?

Thanks,

Gabe

It actually requires either 20-30 minutes using careful driving or 1 to 3 hours of random commute driving. It was incorporated into the injector scaling spredsheet that robertob is still using.

I deleted it because it turned out to be more complicated than it had to be plus it gave different latency results for 16 bit or 32 bit ecus. As it turns out, the defs are different for latency for 16 bit or 32 bit. I'm hoping to write something simpler soon. I sent my DW 650s for dynamic plus analysis at DW so hopefully, I'll have a solid reference point.

BTW, I think kascade is incorporating this into the next logger so it may be moot.



You should be able to see this in your logs. You don't get knock?

The way I dial in the MAF calibration is to log MAF voltage, LT and ST trims. I keep the throttle as steady as possible and log as much data as possible at different speeds.

When done, I take all of the data, add the LT and ST trims together, sort by MAF voltage and then average all of the overall corrections (LT+ST) for each MAF voltage together. Then you can look at all of those corrections across the MAF range and figure out what to modify.

If lower MAF voltages are off a great percentage on average than higher MAF voltages, then you have a latency issue. If the curve is off by the same percentage across the entire MAF voltage range, you need to adjust fuel injector flow scaling. I usually use a 3rd order poly fit to get the equation for the MAF values and then work off of that to get rid of 'noise' in different intakes.

I think you could do this in a formulaic manner by normalizing the corrections so that they are all positive or all negative by shifting the whole curve up or down. Then use latency to adjust the shape of the curve. Next would be injector scale to adjust the offset of the curve and then finally, I would datalog the same variables (ST, LT, MAF) and make small corrections to the MAF curve itself.

This assumes the MAF is calibrated properly, but I've been adjusting latency, then flow, then MAF recently and I found it works a bit better. I wish I was better with Excel and I'd try to do it.

Yeah I see it and it looks to be 600-750 MS. Just wanted to confirm with anyone.

I think I can write some VB to do this. Not so much the injector scaling, that is easy with a WBO2. My concern is the learning ranges...

Gabe

Here is the spreadsheet that robertob is still using.

http://www.romraider.com/forum/viewtopic.php? ... r&start=32

Click on the tab, Correction, for the graph.

I've noticed the most variation in the low MAF range. The high MAF range seems to be smooth on most intakes.

Bump, I copied version 12 that was posted on nasioc to this website.

I'm going to change this to make use of merchgod's learning viewer tool. That way, instead of resetting the ecu, the user just needs to copy the "initial" FLKC table from merchgod's tool.

Also, the column and row headings need to be changed to look more like merchgod's.

I also want to take advantage of the FLKC index offset parameter in the new defs. Not sure how to best use that right now.

Now THAT will TOTALLY rock. I always have to try to figure out "well, it thinks it knocked once here, but is that really just the first time it's seen the existing FLKC value?"

BTW, how does the FLKC table offset parameter work?