My First Real World Use Report for SW102 v0.6.8 & TSDZ2 v0.55, uncalibrated torque sensor
Background: Testing was solo on a dual trike, at windy temps colder than I normally ride (48°F, 8.9°C), so data collection was limited to 2.9 miles (4.7 km) of travel. All testing was done with the IGH locked on "5" which provides a 1 to 1 rpm setting. "0" Assist testing showed that I could comfortably generate 140W of power longer term and 160W for shorter intervals (both higher than I expect for an old man on a trike). A clamp-on amp meter at the battery was used to track power consumption, zero'd after power on.
The Story: Started out with the Assist settings at 0.020/0.040/0.060/etc. In Assist 1, the modest hill to leave home melted away, so stopped after 0.25 miles to adjust the settings to 0.010/0.020/0.030 (limited subsequent tests to levels 1-3, given the huge power being unleashed). Continued testing, and found that power was still flowing freely, as even pedaling down a very slight grade, human power was 40-60W and motor power power was 120-140W. Wow! Stopped again after another 0.25 miles and adjusted to 0.050/0.010/0.015. Next, on a slight uphill, human power went to 110-120W and motor power dropped to 60W or less. Huh? Clicked up to Assist 2 and that reversed that inverted trend. Eventually stopped and readjusted down & up multiple times with different conditions, settling on 0.012/0.024/0.036. These setting were retained for the rest of the testing.
At these settings on Assist 1, level ground was greatly over-assisted, while on a slight/modest hill, Assist confusingly dropped to half of human power from time to time, but, clicking to Assist 2 melted them away. On the big hill to climb back home, Assist 2 was doable, but a lot of work, BUT, clicking to Assist 3 caused this hill to melt away. The motor power indicator went to the vicinity of 750W, and I noted the amp meter with readings fluctuating around 16A. Finally, the motor casing was dead cold at every check along the route. I had a meeting coming up (plus no desire to be colder), but, if I had spent more time, I would have perhaps tried 0.006/0.012/0.018/0.024/0.036. Although, now that I think about it, I tried 0.007 on a mild hill and seemed to have felt no assist. 60°F (15.6°C) is in the forecast in a couple of days, so I'll collect more data then, perhaps dual, with particular attention to investigating that seeming inconsistency of power output.
Summary: First off, I didn't collect near as much power consumption data as planned, as I was uncomfortably cold, especially when I turned around into the wind, and, the apparent Assist inconsistency left me a bit befuddled, keeping my eyes off the amp reading and more focused on the Human vs Motor power variation. However, some significant observations were noted. Someone had reported: Assist settings above 0.200 seemed to reverse Assist level provided. As noted above, I never tested above an assist setting of 0.036 and I maxed out the power rating of the drive. Hhmmm? Someone had reported: Sometimes assist level seemed to go down when you pressed harder. Yep, definitely did, multiple times, depending on Assist level selected!!! Someone asked something like, ... Oooo, I can't remember: But, yes, on level to slight downward grades, I was almost traveling on the speed of thought!!! Too easy for exercise. Yes, I could flip to Assist 0, but, that adds to the ride management load.
Overall, I really like the addition of the three configurable info screens, BUT, I really miss the fact that the currently selected Assist level is not readily visible as I approach a hill (another ride management loading item). I really like the fact that the configuration screen requires a different key press, thus you can simply toggle between the info screens (was really important when going from config to info1 caused a joggle in the system, but, see note below.) Can hardly imaging effectively using 10 levels of Assist, much less 20, but, I see the point in casainho's split utilization. My main target is Level 1 - solo cruising, Level 2 - dual, leaving 3 and/or 4 for the somebody help me/us now situations. Generally, I let the tension build in the legs, so when I call for Assist, it is needed right now.
While preparing for the testing above yesterday, I did see the lights flicker a time or two when leaving the config screen to return to the main screen, but, no motor joggle. HOWEVER, today when powering on the display to collect info while typing this note, I experienced a short runaway, then, when leaving the config screen, a huge runaway. I'm glad the way I power on and change screens has my hand on a handle, as it had quite a few Watts of power behind it, thankfully momentarily.
Apologies, this note got longer as I typed and thought. Hopefully, it will provide some useful Developer/Tester insights.