No those are 3 different logs. (took the screen shots at different times)

But what about the negative spikes?

 

the average is the average for that cell.. when looking for knock use maximum value... as well as look at logs to see if its erratic or consistent. i have no clue about you rpm dimp i assumed u may have sunken the clutch because of something then presume with your log. the dip off in the load seems very wierd is is consistent?

 

Nope the RPM was consistant through the pull. Even if i put the clutch in a little and let off theres no way it would look like that. Thats one logged line that just has an arbitrary value. That is the major problem with 2byte load and rpm etc. From what Im reading at least.

The errors sometimes appear and other times do not. Wich is consistant with what ive read about 2 byte load. So why not just use the 1byte stuff? Is 2 byte that much more accurate?

 

Interpolate.

Scorke

 

Interpolate the spikes? Yeah I can do that, but my question is how does it affect the map tracer? Those spikes are still values and if all the tracer is doing is taking averages it does not "interpolate" these spikes.

 

Map tracer is unfortunately not the best way to represent the data, it's just finding the cells that correspond best to the data it's seeing,load/rpm, not always the ACTUAL value that is being read.

The only way you can be sure the data you are recording in ecuflash is accurate is to record it with something else with a higher capture rate, or capture less data through the OBD2 port so its at least fast enough capture rate that you can eliminate jumpy readings.

DO NOT WORRY about a knocksum of one unless it is is always present, 1 count of knock through the stock ecu at 2-3k rpm will not kill a motor.

Scorke

 

Thanks for the info. It seemed like thats how the tracer was working. Its too bad there isnt a map "plotter" to just plot the raw data on the cells of the graph. I guess it would be the same deal though you have to average each load for multiple plots. Now that I think about it Im assuming when you run the cursor over the traced cell and it brings up a little spread sheet those are maybe the data that used for the averaging?

 

lol, this is the funniest thing I've read all day. Everyone should be using maptracer to tune. EvoScan MapTracer shows you the TRUE Actual ECU Target AFRMAP in the Maptracer. and shows you the TRUE actual AFR wideband reading. And the Interpolation in the mouse over data grid, shows you the ACTUAL interpolated target AFR. And in the bottom grid you can increase or decrease the map as needed on the actual correct cells.

If i read anything like this again on this forum, I'm gonna kill myself.. lol.

 

What I was trying to explain is that Map tracer can highlight cells that you are not actually putting the engine through during a pull.

You might flash into a load cell for a split second during or after a pull that is not of any concern when your making adjustment after, where as you can follow the trace and logically see what areas it's hitting need to be adjusted if you consider what sort of path it should be taking.

All cars go from top to bottom in terms of RPM scale on the left throughout a pull, but it's rare for you to always have a map trace that shows a linear progression through every adjacent cell, it moves and jumps around some, all i was saying.

Scorke

 

Yes correct, even a knock count up to 10 is fine, and 99% likely just engine noise. Also watch for crank walk clicking the crank shaft sensor setting off the knock sensor too. True knock will normally set off a knock sum in excess of 40 or 50. set your knock sum alarms in EvoScan or EvoScanGPS to 8.

 

wait, you trying to explain that it can highlight cells you are not putting the engine through, then the next sentence you say it did go into that cell for a split second. so which is it? did it go in that cell or not? my guess is 100% yes it went to that cell if the maptracer says it did. I think what you are trying to say, is that you don't know for how long its in that cell, or what corner of that cell (interpolation) its closest to. Both can be seen clearly from the little popup window of each cell.

So when tuning I look at each cell on its own merits.. To be honest, at the end of the day it just comes down to the wideband AFR graph across the RPM range. if your wideband readings are too low (rich), then in the map grid underneath take out some Fuel (do this by increasing AFR values=leaner) around that RPM/Load, keep the values around that cell smooth by looking at the other cells just past that too. (getting the correct cell from the maptracer is key, hence the highlighting, the average/max/min or the value displayed I don't look at) once that AFR curve looks good. do more runs, and move from low boost through to your high boost setting, and it will slowly hit all the load cells, and then do a final check over, as some cells you changed will affect AFR for other boost setting levels. (haha if you're like me and always turning boost up or down depending on warm/cold weather, racing/street/highway use.)

Then once all the AFRs are happy. then start looking for timing tweaks to get some smaller performance increases. Most Mitsubishi stock timing maps are actually quite good as a base to start tuning from. Look at as many stock timing maps you can. and look at some tuned maps.

Haha you're right this thread has turned to "tuning 101"

 

Max knock count is 36. I also would not be so quick to tell people a knock sum of 10 is most likely engine noise, not true in most cases.

 

I agree, although I have seen knock sums above 40 on a few evo8s. When I use the expensive knock sensing headset, I have not heard knock sums less than 10 being real knock.

Another good measure if you have some knock is to start datalogging Knock voltage, its the raw level of noise (coming from the knock sensor) in the engine.

 

Interesting stuff here. What your saying is that 10 counts is no biggie? When I was knocking pretty bad due to various reasons including mechanical and tune wise I would see 10-12 counts sometimes and that was pulling quite a few degrees of timing and then dumping fuel in dipping to super low 10.0AFR's. Super low timing and sub 11 AFR does not make for a peppy evo from what Ive felt with my car. So how can this be ok? Also your saying 40 counts is more like real engine knock? I feel like if I saw 40 counts on my log I would be more inclined to thing that it WASN'T real knock and there was a vibration coming from somewhere or it some sort of "phantom knock"

Also you said that you can change the values in the ignition table and fuel tables in the maptracer? I wasn't aware that you could. I tried but couldn't see a way to do this. It would be easier then referring to ecuflash. Mostly to lower the error of selecting the wrong cell. Those colors all blend together after a couple hours of tuning know what i mean?

 

Also what RPM and LOAD values are most of you guys logging for use with tuning?

Ive got the 2Byte load/rpm/airflow mod setup in my Mutt. But my 2byte Rpm gets spikes so sometimes I log 1byte to speed things and and get less spikes. Although my 1byte rpm well get fluctuations in the lower on %60 of my pulls. Kind of annoying.

My 2byte load also will generally have one spike or dip in it. I can "interpolate" this though so its no biggie just annoying.

I have the tephra v5.10 setup with a 1.2 2byte to 1byte load factor and same factor in my evoscan eval. Does anybody use this? My map tracer wouldn't work with this value logged instead.