mrbill said:
teklektik said:
Anyhow - some things to try ...
Thanks for your suggestions. I kept a few of them ... Maybe I'm just being too fussy.
Well, glad we made a little progress. There are definitely some oddities with PAS, so I don't think you are being too fussy - anyhow - it's good to get feedback of all kinds...
Anyhow, I've had some overrun issues myself but they occurred in very specific circumstances - not as broadly as your experiences. So, here's a couple more thoughts.
First, I am wondering if you are using the CA display as an indication of overrun or the actual sound of the motors. If using the display, it's a good plan to adjust
Pref->Averaging = 0.02 sec (short as possible) so there is no delay in the display reflecting the actual readings. This might be worth a try if you haven't already, just to be sure that the actual CA activity is distinct from other mechanical noise cues.
mrbill said:
teklektik said:
The PAS start/stop thresholds have a noticeable effect on the startup/shutdown times and might be worth a little tinkering...
I spent quite a bit of time settling on +65 watts as the start threshold. Lower power led to spurious "surging" when soft-pedaling--the Thun BB sensor indicates brief spikes of power often when soft-pedaling or when spinning the cranks under no load--, and higher power increased the start delay too much. I still get an occasional surge of power when soft-pedaling or when not intending to add power, but these incidents are infrequent enough that I can live with them.
Here we seem to have gotten our wires crossed. I was referring to
PAS->StrtThrsh and
PAS->StopThrsh, not
Trq->AsstStart.
This start and stop rpm stuff can be a little hard to figure out. The Guide suggests a Start rpm that is lower than the Stop rpm and has a little table of suggested rpms for different pole counts. Actually, the start rpm can be set to higher rpm than the Stop and things will still work. In fact, you can set these all over the place and the PAS will still operate although certain weirdnesses can crop up in different situations. So here's the deal: Justin had some ideas about acceptable time delays for determining starting and stopping (0.75s for startup detection and 0.5s for shutdown detection). At the end of the day, these times map into the PAS pole inter-pulse period. The Guide table and CA built-in defaults show the rpms that are required to achieve these suggested inter-pulse delays.
However, after messing with the settings on the bike, I'm not entirely certain that this 'suggested detect period' is a better means to pick good responsive settings than 'rpm as a percentage of cadence' - to a certain extent, it defeats the advantage of having additional magnets on the wheel to get quidker response. On my bike with a 16 magnet PAS wheel I use Start/Stop rpms of 20/24. I have a nominal cadence close to 70, so these rpm settings are tripping the PAS logic at less than 1/3 of normal cadence which seems to give the CA a pretty early cue on shutdown and an acceptable lag in start-up. The start delay on resuming pedaling while rolling is very quick. With a relatively high start rpm setting, there are no false engagements while fiddling the pedals at low or walking speeds.
- I'm thinking you might try these 20/24 or similar settings instead of the Guide recommendations - this might noticeably shorten your overrun times. You might lower the Start rpm if it seems too demanding, but starting in a sufficiently low gear so initial cadence is high or goosing the throttle on getaway should offset a higher Start setting...
It might be worth noting again that having the Start/Stop values far apart can lead to occasional surging when stopping as minor cadence fluctuations in this 'start/stop overlap region' cause brief detection snafus. The closer these parameters are together, the quicker a rider will transition the hiccup range.
mrbill said:
Side note: Setting the Aux Pot to 0% in any PAS mode does not completely shut off PAS-mediated motor power. I do occasionally see the motor spinning unloaded and occasionally get surging. I would call this a bug, since setting the pot to 0% should effectively turn off the selected PAS mode.
I cannot duplicate this on the bench (my bike isn't wired to be able to try this), but there are certain failure modes that are difficult to achieve with my little
simulator thingie. I'm wondering if raising
Aux->MinAuxIn a tenth of a volt will help ensure that things are really off in all situations.
mrbill said:
teklektik said:
These modes rely on Watts limiting, ... However, you might try switching to Current throttle, increasing the up ramping V/sec values to get them a little more out of the picture, then using AGain to get the desired throttle response and WGain to control the PAS response. ...Just a thought...
I tried lowering AGain but found throttle and PAS response to be damped too much. I tried raising both AGain and WGain, but this lead to too much drivetrain "slam" and underdamped initial power application. These along with the Speed gain parameters were set according to your Unofficial User Guide--thank you for this effort, btw--, and I think they're pretty well dialed in for this motor/controller/drivetrain/bike.
Okay - I was going after minimizing the ramp times here by taking over some of the damping with the Gain settings. Seems like that didn't work out for you.
So lastly, you have mentioned surging a couple of times - perhaps you can leave the present ramping and
AGain in place, but reduce
WGain to about 8-10. Since only PAS will be using Watts limiting, this should smooth out the PAS operation and damp PAS power surging. I run with
WGain at 8 and others have gone as low as 5 - this is by way of example and may not apply to your build or tastes. This should not affect throttle or any other operation - just make sure your
PLim->MaxPower is slightly above current limits (
SLim->MaxCurrent) you might incur so the overdamped Watts limiting doesn't come into play in 'throttle mode driving'.
