S2R 1000 On road and on dyno Tech Edge Wideband data log results - Written 05/08
Summary: More S2R results and a comparison of on road and on dyno air/fuel logs from the Tech Edge Lambda controller and data logger and Dynojet dyno air/fuel probe.
I’ve been using a Tech Edge 3A2 Wideband unit (lambda controller and data logger) - http://wbo2.com/3a1/default.htm -and have been running it on our demo S2R 1000. All the information you see below has been interpreted using the TEWBlog software as developed by Justin ‘The TEWBlog guy'. The information has been copied out of the TEWBlog “matrix” function and into a MS Excel spreadsheet, where it is easily manipulated into graphs. The “matrix” function is only available once you’ve registered the TEWBlog software, which you do by giving Justin an insignificant amount of money to help compensate him for the time he spent developing it. If you want to use the output of the logger to develop maps or map corrections you need the “matrix” function, and besides that it’s a very nice way to display any logged data.
We took the S2R out and logged some WOT runs, with the set up being Moto One header pipe, Staintune mufflers, std air filter and std ECU with Lambda sensor disconnected. Then I took it to the dyno (hadn’t dynod this combo before) and ran it with both the Tech Edge unit and the Dynojet air/fuel probe. The first graph below compares the on-road Tech Edge log to the on-dyno Tech Edge log, with the values shown being the average values. The on-road logs had between 25 and 35 samples for each WOT RPM range, whereas the on-dyno results had between 8 and 13 (5 dyno runs in total).
Red is on-road, green is on-dyno. I’ve stopped the graphs at 8,375 RPM, meaning any info over 8,500 RPM I’ve disregarded. This is due to the rev limiter induced wackiness that I couldn’t be bothered taking out of the log file. This might sound lazy, but when you consider the unit logs 10 times a second and can log around 45 minutes or so (up to 1 Meg of internal memory) it’s a lot of info to look thru.
The way the “matrix” function interprets the data is that all samples within a specified RPM and throttle range are averaged to give a mean figure. These sections of RPM and throttle position can be set as desired, with up to 30 sections for each. A sample matrix is shown below. The RPM given across the top is the maximum RPM used for that averaging section. For example, 4,000 corresponds to the averaged data range from 3,750 to 4,000 RPM. Similarly for degrees of throttle opening, which increases as you go down the page. You can see the little white box at the top right which shows the info for a specific cell and the number of samples in that cell. In this case, 7 to 9 degrees throttle opening, 7,500 to 7,750 RPM, 0 samples.
To use this info you just use the “copy” function which you use to copy the matrix into a spreadsheet where you can manipulate the data or drop it into a Dynojet Power Commander file or a Rapid Bike file or use it to modify Ultimap 1.5M and U59 custom files using some spreadsheets I have developed.
This example is a good indication of what happens when you go out and do lots of WOT stuff and cruise in between. The log can be a cruel thing, especially at a race track for guys who think they’re really on it. You can’t hide from the data logger, and there have been a few occasions where people have been told to “open the throttle further for longer or you’re just wasting my time”.
Logging the commute to work and back shows how (unavoidably) little throttle opening you use in that scenario. Generally the hardest cells to get samples for are those in the centre and centre right of the matrix as shown, and it can be almost impossible to fill them in on a straight piece of road.
You can also show the minimum and maximum values logged, useful as a gauge of possible error, how accurate your results are or how much variation can exist out in the open environment. For example, at the lower RPM (up to 3,500 or so) where the enrichment for rapid throttle opening is still having an effect, you can get a wide range if for example you start some runs at 2,000 RPM and some at 3,000 RPM. Whereas by the time we get to 6,000 RPM there’s much less variation. Hitting the rev limiter can also lead to some quite wild variation as the fuel and/or ignition cut out and back in, so you need to be careful in using some data and the theoretical changes that might come from automated spreadsheet or software interpretation of it.
The red and orange curves are on-road min and max, the two green curves ondyno min and max. As I would expect the somewhat more stable and controlled dyno environment has given on average less variation. And what the first graph above really shows the important part to me is that the consistency between the on road average and on dyno average is very good. How this relates to bikes with truly functional ram air setups I don’t know from my experience the R1100S with Boxer Performance Induct would be a good (as in bad) case in point here.
For the next graph we add the Dynojet air/fuel trace to the averaged on-road and on-dyno graph. Remember this Dynojet trace of air/fuel is for a roll on acceleration run, not a held RPM run, so it mimics the on road situation well. Red is logged on-road average, green logged on-dyno average and blue is the Dynojet trace, read off the Dynojet Winpep software generated graph of Air/fuel vs RPM. You can see the Dynojet graph shows a somewhat consistant lag compared to the green on-dyno log.
If we move the Dynojet curve 250 RPM to the left you can see the comparison is better.
And if we move the Dynojet curve another 250 RPM to the left you can see the comparison is overall better again. When the air/fuel curves aren’t as flat as these curves are (often the case) the lag is more obvious. This 500 RPM lag is what I usually expect to see when using this particular dyno, so I’m happy with the result.
Next I’ll compare Dynojet air/fuel curves for our demo bike with the above set up to the customer bike featured in the previous S2R report, which has the same set up except for the Dobeck unit. That bike had the green pot set to 3 for correct idle mixture, so the whole map (as was set for dyno testing) is 0.32ms richer than our demo bike. The run shown for our demo bike is with the yellow pot set to 1 (no extra fuel). Although the fact our demo bike had the idle mixture about right without any playing once the Lambda sensor was disconnected should by rights mean that overall both bikes should have about the same mixture overall. Green is our demo with the Lambda sensor connected, red our demo with the Lambda sensor disconnected and blue the customer bike from the previous S2R report with the Lambda sensor disconnected.
Next I took the bike out and did the same runs as previously logged with the Tech Edge unit, this time without the air box lid, just the std paper air filter tied down as I had on the dyno previously. The other change here compared to the dyno runs with this configuration was that I had had the ECU out of the bike for a week or two, and this may have had the effect of clearing the adaption tables, assuming of course there is WOT adaption. This is something I don’t actually know about. Anyway, when I’d done this on the dyno previously it was richer than with the air box lid, which didn’t make any sense.
The next graph shows the average on road air/fuel from above (std air box lid) in green with these results for no air box lid in red. As you can see, it’s around 0.5 to 0.7 of a ratio leaner overall, less so at higher RPM. This is exactly what I would expect to see.
The next graph shows the above average air/fuel ratio plus the min and max as logged on road without an airbox lid. The number of samples at each data point ranged from 20 to 30 more at higher RPM as the acceleration had obviously dropped off by then. I’d say the wacky point at 8375 on the max curve could be a rev limiter induced thing.
So, I had to go off to the dyno again to confirm the above result (after having Wendy promised I wouldn’t dyno this bike again) just in case something had changed. It had. The green, blue and red curves are for this set up - Moto One header, Staintune mufflers, no air box lid and Lambda sensor disconnected from 3 different dyno sessions before the ECU was removed from the bike. As you can see, the consistency is very good. The yellow line is this last run, after the ECU was removed. As you can see, it’s clearly leaner across the range.
I also logged the dyno runs using the Tech Edge unit, and if you compare the yellow curve to the curves given by the Tech Edge unit for this combination, both on road and on dyno, you can see quite a difference. The two Tech Edge log curves match up quite well, but the Dynojet curve is clearly richer. The next graph shows the two Tech Edge unit logs on the road is blue, on the dyno is orange compared to the Dynojet air/fuel trace in red. The Dynojet trace is moved 500 RPM to the left as above. In this instance, the Dynojet trace is quite a bit richer. At least the on road and on dyno Tech Edge logs show good consistency (you know how much I like that).
I think I need to replicate this comparison with more bikes before drawing any strong conclusions, but what this means in terms of mapping and variations between the Dynojet trace and reality when running open air box lids I don’t know. Personally I tend to believe the Tech Edge result more than the Dynojet, but I can’t give a real reason why. Maybe it’s just a one off, I don’t know.