Had the bike on my brake dyno this afternoon. The sole task was to run autotune and check results. First it was hard to start despite having the previous good tune loaded. Based on recent experience I flashed the ECU and loaded the exact same tune. The engine fired up first try afterwards. I still have no explanation for this, I just know that it is effective.
What I have discovered is that autotune has it's own VE & TPS tables stored in the ECU along with a record of which areas have been tuned. That record is in a table which has 20 throttle position columns and a range of engine speed rows. I have my system setup to use the VE table, which is load versus rpm, at anything less than 98% throttle. The record of successful autotune is therefore not totally relevant to how I am using the system. The record of successful autotuning for me would be better off being a table with a similar matrix to my VE table. Having said that it does work well. It is only the record of which areas have completed tuning that is a mismatch. The autotune VE table is updated over a larger area than is indicated by the tuning complete table. Flashing the ECU does not erase these tables. These tables are in no way able to be manually edited but they can be exported to a CSV file. From there it seems, I haven't done it and tried running the engine, you could copy the values to a VE table and run the system in 'manual tuned data" mode using that autotune data. If autotune is "on" you can alter the VE and TPS tables all you like and it will have no effect because the ECU is using the autotune versions of VE & TPS tables. The regular tables would only become effective if autotune was disabled AND you switched back to manual tune data.
What I have found is that the ECU is autotuning to richer than stoichiometric despite the Eco/Power mode switch being in the "0" or Eco position where it should be aiming for lambda = 1. My ALM gauge is trending towards 13.?:1 or slightly rich. I think this is due to a need to adjust a table which defines the output of the ALM to the ECU as being various values of lambda.
I haven't completely sorted clearing the tables and loading a new default, whether it loads the VE & TPS tables from the ECU or whether it loads a "default" set of tables from within the ECU. At some point a few days ago I ended up with autotune tables which were nothing I created. My system is supposed to be for a GPz250 so I think the "default" tables may be suitable as a start for one of them. I will sort that out.
When autotuning we were able to load the engine up at certain revs and load and hold it reasonably constant. Reading the TPS screen gauge and the MAP I could pick which cell would be tuned and filled with green. I could also tell which area of the autotune VE table would be altered. The autotune software varied a range of cells about the particular cell that was active. This made for a 3d chart that had no odd peaks or troughs. More time would see it looking pretty swoopy.
The dyno has not much inertia. I built it to use the brake so the bike is pretty much free revving in this video. Towards the end I loaded it to 1 bar manifold pressure at about 12,000rpm. It doesn't sound like it but you can see the gauges and also the tuned areas in the chart.
https://youtu.be/_5HMp6co4EA
Bookmarks