I already wrote this up here: (i think the pics look better there)
http://www.buelletinboard.com/forums/showthread.php?21477-Using-Datalogs-for-Diagnostics
on to business...
I promised a few people in a few places Id post the methods for advanced use of datalogs and interpreting data.
To get started, you need to be able to understand the functions of MegaLogViewer's (MLV) scatter plots and their corresponding axis (3 of them). Im not sure if the unregistered version is capable of graphing the plots, so if not, youre outta luck. Its a powerful tool for the minimal price, relative to other diagnostic/logging software that is in the industry. Not to mention, it has the field naming of inputs already ported to Buell ECMs.
For this example, ECMSpy was used to record the logs, and MLV to view them. Just because Ive dealt with this pairing of software for years now. Perhaps I will get into Spy for Mono and do some stuff with that, but since the current setup works for all practical purposes, I'll stick with that.
Most of you know that software setup is good for UP TO 2007 Buells. This particular example is on a 2007 XB12S, with an LC1...logging/tuning was done with the full WB output, NOT a simulated NB.
Ok, so the start of the problem was dead feeling under hard accel and high RPM. Naturally a Log and Tune session was called upon. This is where the issue first started to show up in terms of data analysis. A log file was generated, then analyzed with the corresponding maps. MLV then spit out a new fuel map. I noticed there were quite a few cells that were at and above the max 255 8bit value. This prompted me to install the actual AFR gauge so that I could watch the realtime AFR independently of what I was seeing in the logs. There was a couple reasons for this; I wanted to see if the data fed to the log via ECM was accurate (having MLV suggest values above that of the max injector pulsewidth made me question the data stream), and also to see what the AFR was doing under certain circumstances with enrichments, etc enabled/disabled.
Now thats out of the way.....
I changed any values the VE analyzer returned that were above 255, and made them 255. If You do not do this, Spy interprets the values above 255 and then attempts to manipulate the adjacent cells and it gets all jumbled up. It will show places in the maps that are "0" and move other values around. By maintaining a MAX of 255, it prevents this from happening.
Then those maps were burned, and the result was logged again. Once the datastream accuracy and signalwas verified, I was able to start plotting suspect paramters in MLV's scatter plots.
It was evident shortly thereafter what the problem was.
Here are the graphics of the plots....(excuse the horrible image compression).
For the first plot, I set up the axis as follows:
X= rear cyl pulsewidth (in ms....the units doesnt matter, we are concerned with its duty cycle,..)
Y= AFR as measured by LC1 WB
Z= TPS 8bit (this is reflected in the color gradations on the right of the plot as a key... 11-255, blue to red)
take a close look at whats happening here. This plot is showing that in some areas near the max PW (right side of graph X axis), the AFR is predominantly ABOVE the 14.1 AFR line. To further diagnose, looking at the color of those data points shows that they occur frequently at the upper ranges of TPS values. This tells us that at max TPS (255=red) there is STILL not enough fuel to deliver an AFR of 13.0:1 (which is what I like)
Now, im SURE engineers at Buell left enough headroom in the fueling to make sure this could never happen. As you would when designing any type of EFI system. Rule of thumb is that you dont ever wanna exceed 80% of the injector capacity.
The second plot was confirming what I was seeing in the first one, by setting up axis as follows:
X= AFR
Y= RPM
Z= Injector Pulsewidth
In this one, i wanted to compare with RPM to verify what I was feeling in the seat at those RPM was indeed happening internally. Here it becomes even more evident that above approx 4500RPM with full injector PW, the AFR still is way too lean. Were talkin WOT areas here. Unacceptable on these aircooled engines to be running at that mixture under those criteria.
So all this leads us to fuel supply. I decided to get into the fuel pump assembly and what i found was the culprit. Not sure how this may have happened, but it was a combination of a failing pump and what i discovered either a faulty fuel hose, or it had work through after age.
Here is the corrugated hose with a hole in it. this is downstream of the pump and upstream of the regulator.....the fuel was leaking back into the tank and losing volume...
I have since gone through and did fuel pump, strainer, hoses, orings, etc.
Once it was all back together, It was instantly recognized as fixed because ECM was trying to take a bunch of fuel out of the idle EGO to compensate for the fuel that was now NOT missing. HA!
So then it was time to retune with the WB, make all necessary adjustments, and then leave all the ECM functions disabled and run it like a carbureted bike. (this is my preference and I dont recommend it for tunes without WB or tuners without experience as I edited cold starts, ignition timing, etc.)
Well, hope that helps with using the software as a diagnostic tool. You should now be able to set up the plots to isolate data pertaining to any system suspected of faults.
More to come in other subjects......at some point.....
http://www.buelletinboard.com/forums/showthread.php?21477-Using-Datalogs-for-Diagnostics
on to business...
I promised a few people in a few places Id post the methods for advanced use of datalogs and interpreting data.
To get started, you need to be able to understand the functions of MegaLogViewer's (MLV) scatter plots and their corresponding axis (3 of them). Im not sure if the unregistered version is capable of graphing the plots, so if not, youre outta luck. Its a powerful tool for the minimal price, relative to other diagnostic/logging software that is in the industry. Not to mention, it has the field naming of inputs already ported to Buell ECMs.
For this example, ECMSpy was used to record the logs, and MLV to view them. Just because Ive dealt with this pairing of software for years now. Perhaps I will get into Spy for Mono and do some stuff with that, but since the current setup works for all practical purposes, I'll stick with that.
Most of you know that software setup is good for UP TO 2007 Buells. This particular example is on a 2007 XB12S, with an LC1...logging/tuning was done with the full WB output, NOT a simulated NB.
Ok, so the start of the problem was dead feeling under hard accel and high RPM. Naturally a Log and Tune session was called upon. This is where the issue first started to show up in terms of data analysis. A log file was generated, then analyzed with the corresponding maps. MLV then spit out a new fuel map. I noticed there were quite a few cells that were at and above the max 255 8bit value. This prompted me to install the actual AFR gauge so that I could watch the realtime AFR independently of what I was seeing in the logs. There was a couple reasons for this; I wanted to see if the data fed to the log via ECM was accurate (having MLV suggest values above that of the max injector pulsewidth made me question the data stream), and also to see what the AFR was doing under certain circumstances with enrichments, etc enabled/disabled.
Now thats out of the way.....
I changed any values the VE analyzer returned that were above 255, and made them 255. If You do not do this, Spy interprets the values above 255 and then attempts to manipulate the adjacent cells and it gets all jumbled up. It will show places in the maps that are "0" and move other values around. By maintaining a MAX of 255, it prevents this from happening.
Then those maps were burned, and the result was logged again. Once the datastream accuracy and signalwas verified, I was able to start plotting suspect paramters in MLV's scatter plots.
It was evident shortly thereafter what the problem was.
Here are the graphics of the plots....(excuse the horrible image compression).
For the first plot, I set up the axis as follows:
X= rear cyl pulsewidth (in ms....the units doesnt matter, we are concerned with its duty cycle,..)
Y= AFR as measured by LC1 WB
Z= TPS 8bit (this is reflected in the color gradations on the right of the plot as a key... 11-255, blue to red)
take a close look at whats happening here. This plot is showing that in some areas near the max PW (right side of graph X axis), the AFR is predominantly ABOVE the 14.1 AFR line. To further diagnose, looking at the color of those data points shows that they occur frequently at the upper ranges of TPS values. This tells us that at max TPS (255=red) there is STILL not enough fuel to deliver an AFR of 13.0:1 (which is what I like)
Now, im SURE engineers at Buell left enough headroom in the fueling to make sure this could never happen. As you would when designing any type of EFI system. Rule of thumb is that you dont ever wanna exceed 80% of the injector capacity.
The second plot was confirming what I was seeing in the first one, by setting up axis as follows:
X= AFR
Y= RPM
Z= Injector Pulsewidth
In this one, i wanted to compare with RPM to verify what I was feeling in the seat at those RPM was indeed happening internally. Here it becomes even more evident that above approx 4500RPM with full injector PW, the AFR still is way too lean. Were talkin WOT areas here. Unacceptable on these aircooled engines to be running at that mixture under those criteria.
So all this leads us to fuel supply. I decided to get into the fuel pump assembly and what i found was the culprit. Not sure how this may have happened, but it was a combination of a failing pump and what i discovered either a faulty fuel hose, or it had work through after age.
Here is the corrugated hose with a hole in it. this is downstream of the pump and upstream of the regulator.....the fuel was leaking back into the tank and losing volume...
I have since gone through and did fuel pump, strainer, hoses, orings, etc.
Once it was all back together, It was instantly recognized as fixed because ECM was trying to take a bunch of fuel out of the idle EGO to compensate for the fuel that was now NOT missing. HA!
So then it was time to retune with the WB, make all necessary adjustments, and then leave all the ECM functions disabled and run it like a carbureted bike. (this is my preference and I dont recommend it for tunes without WB or tuners without experience as I edited cold starts, ignition timing, etc.)
Well, hope that helps with using the software as a diagnostic tool. You should now be able to set up the plots to isolate data pertaining to any system suspected of faults.
More to come in other subjects......at some point.....
total F#ckery........good points all....but F#ckery.....need more alcohol to achieve enlightenment
