Bug? ERG mode quits adjusting resistance down

[Lunchtime0614]

First - is the discussion the right place to report this kind of thing??

So this is only my second ride with the SS2K but in both rides I was able to slow my cadence make my power be off the target and as expected it turned the knob to get me back. But then if I increased my cadence to be way over the target it wouldn't bring the resistance back down.

Odd that both times this started well into the ride over halfway. One was a 60 min ride and todays was 45 min.

I didn't have UDP logging on but I used my phone to capture some stuff I thought was odd in the logs via the web...sorry for the cutoff images - iPhone mini.

These are in order of how they happened.

[doudar]

There's not much log to go by, but if I had to guess, you're hitting the automatic low end stop of the powertable.

As you're riding, the power/resistance lookup table is built. One of the things it's calculating is where the 50w resistance level is. This 50w level is assumed to be the lowest resistance available in any mode (Sim or ERG). If this minimum resistance is reached (looks like 314 in this case), the SS2K won't adjust any lower. This position is adjusted (calculated using assumptions for a generic powercurve) for 90rpm cadence. If you're pedaling faster than 90 rpm, your power is going to be higher than 50w.

There's also a maximum limit (Max Bike Brake Watts) that's adjustable on the settings page. It's function is exactly the same as the min function only it's currently user configurable.

The reason both of these exist is to provide some end-stop limit protection to prevent the stepper motor getting some wild command and then grinding against the hard stops on your bike (or potentially unthreading the knob in some cases).

I know @eMadman has built a couple custom firmwares with the minimum set lower. It's easy to change on a custom binary - it's just MIN_WATTS in settings.h

Potentially we could add yet another setting on the settings page for a minimum as well - I'm hesitant to do so as I'm trying to keep it as simple as possible.

[Lunchtime0614]

Yeah I get it - I'll record UDP next time but it's no way hitting the 50w lower end. The ride didn't call for anything less than 120w.

And when I raised my cadence I was putting out 250w plus and it wasn't going down.

This is hard to see...but here is the ride showing the power graph - top chart.

[Lunchtime0614]

Look at the end where I held the power well above the target line

[doudar]

A little more log should help .

I'm fairly confident it's hitting the limit for some reason - we'll have to figure out why it might not be getting calculated correctly in your case.

It should be self correcting though and it looks like you were having problems for minutes, so it could be something else entirely.

[Lunchtime0614]

Agreed - I knew this wasn't much to go on but I wanted to put this out there before I forgot about what was happening. My training schedule has me doing another ride Monday at the latest - but it's a VO2 max ride that might not give me the opportunity (some stead state) pedaling to record the log better. I suppose I can try to get it to do it in a short amount of time - usually it seems to only do this well into the ride...I actually tried to get it to do it in the beginning of this ride during the 5min warmup at 120 but it didn't do it.

At the end when it started doing this I thought maybe it was just in my head and I wasn't giving it enough time to adjust so I purposely road above the target to see if it would hold at 220-240 instead of adjusting down...that's what you see at the end in the screen shot.

In the end - I'll get more data posted.

[Lunchtime0614]

@doudar - I had 30 min free time today so I decided to see if I could easily and consistently replicate the issue...I am.

I was never below a 60 cadence and I was able to ramp my cadence up to 100 or so and hold it there for a few min and the SS2K never attempted to adjust the resistance down. I did this both on android and iphone with the trainer day app. Attaching both UDP logs below.

It also seemed like I could make it not want to come back up to the target as well...but I didn't fully test that. I'm wondering if at some point I hit a limit ever so slightly and then it doesn't recover back to normal or something.

Android:
NotAdjustingDown-Android.csv

iPhone:
NotAdjustingDown-iphone.csv

[doudar]

Yes! Thank you!

[MarkusSchneider]

Hi all,
The Watts were old log is logged from the ERG Task. Log is written when the power value is not updated since the last cycle.
Increasing the cycle speed will not help.
I think the problem is the calculation of the lower boarder.
Value is calculate with a cadence of 60 and watts 50. If cadence is higher also watts are higher on the same setPoint.

I can look at this tomorrow an create a binary without check of the lower boarder to see if this will solve the problem.

[Lunchtime0614]

@MarkusSchneider - thanks! I don't claim to understand everything that you and @doudar discuss here...but given this comment

Value is calculate with a cadence of 60 and watts 50. If cadence is higher also watts are higher on the same setPoint.

why would SS2K not adjust the resistance down when I'm pedaling at a cadence of 100, power output from the pedals is in the 200 watt range and the app is calling for 120 watts in erg mode?

[MarkusSchneider]

@htims05 Good Point 😉.
The ratio of setpoint / cadence / watts depends on your bike. At the moment we have a fix factor to take cadence in account.
And the ratio is also not linear (at least on my bike).
In future we can try to calculate this factor.

Normally it is not a big issue because ERG Mode adjusts the differences for the setpoint form the power table and the measured power.
Only when the lower or upper boarder is checked it is a problem because this check is made against a fix setpoint calculated at 90 RPM.

The check of the lower boarder should prevent the spinner to run to far until the knob reaches its end.
If this happend you hear a nice cracking noise 😁

[eMadman]

I have my lower erg wattage fork which has a lower setpoint - I don't have your and Anthony's strength and needed it lower. This commit has the main changes in it, I think -

3ecbc6a

Min set point is 35w instead of 50, and I'm riding with a lower power table increment since my 130-140w FTP effectively means I only get 3 erg table entries... the increment might be too low for a stronger rider.

The fork is a bit of a mess and needs to be updated with the latest from Develop - likely won't have time to look into that till tomorrow evening, but my settings.h might be useful to compile a fresh firmware from.

[MarkusSchneider]

@htims05
here is a firmware without the check of the lower boarder.
Please check if this solves your problem.

@doudar
I've commented this line:

SmartSpin2k/src/ERG_Mode.cpp

Line 114 in 12331a3

rtConfig.setMinStep(_return);

firmware.zip

[Lunchtime0614]

here's a new binary built with the latest libraries.

There's a chance.... :)

Make sure to unzip first: firmware.zip

@doudar - does this have Markus' change he made included?

[Lunchtime0614]

@doudar - here's your firmware with the latest binaries. This one seems good! I couldn't replicate the issue that I spoke of above - I've tested it only once though. Same app (trainer day) same ride as before - which was to give a 120Watt target. The only thing I was able to make it do is that I quickly increased my Cadence up to 120 and held it there....it started to bring the resistance down but then stopped around 200....maybe I hit an upper limit on cadence? It's in the later half of the ride - circled in red below...once I slowed the cadence down everything went back to working as it was earlier in the ride.

Aside from the issue I mentioned above at 120 cadence - the only thing I possibly could say is that it was somewhat slow to bring the resistance up to 120 when I slowed the cadence....now maybe that's the setting I have on my SS2k, but it seems slower than if the target changes or if I speed up the cadence and it needs to bring the resistance down.

Android_LatestBinaries_ERG_120wTarget.csv

[doudar]

Good news! Hopefully it works over a few more rides for you.

You could try increasing the erg sensitivity to make changes not related to target changes faster.

[Lunchtime0614]

Yeah this was just me testing. I probably rode for 10 min. I have a longer ride planned for tomorrow.

[Lunchtime0614]

@doudar - I did a full ride with the new binary firmware. I didn't record the log, since it was a 45min training ride I didn't think we wanted 45min of a log.

That said my rides have a 120 Watt target warmup and cool down...and then in the middle the ride us up/down all over depending on the ride.

When I first started this ride I was thinking that it was back to it's old tricks not adjusting, but it was close enough to the 120 to not worry about it (like 113ish) and I have a cadence of 80-90. Then during the main part of the ride I found myself thinking this is working perfectly now...increasing resistance as the target calls for it, lowering etc...I could feel the wheel making very small adjustments up and down. Then I get to the cool down at 120 again...and keeping a 80-90 cadence it doesn't want to come down. I can easily keep it at 140-150 and this was me not even trying to do it. You can see this in the beginning and end of the graph below (moreso at the end).

So it feels - oddly enough like there's some sort of issue at 120 - I mean it makes no sense and of course this is one ride, so we probably don't need to make a knee jerk reaction until I get more data.

The only other thing I'll throw out there - is Trainer Day has a power smoothing function - as shown here - https://forums.trainerday.com/t/our-new-power-smoothing-slider-out-june-27th-28th/735 - so Does that have any impact on the SS2K or does this really only affect what's shown on the app screen (I suspect the latter)? Edit: I turned off power smoothing near the end of the ride thinking maybe that had something to do with it...it was turned off well before the 120W cooldown at about the 31 min mark where it drops the connection - seems like it does that if I change a setting in the trainer day app mid-ride, but then reconnects fine..

[MarkusSchneider]

@htims05 @doudar Would it make sense to set the lower limit to 35 and to add an setting to disable the check of the low and high incline check?
If you like the idea I can add this setting to development.

[Lunchtime0614]

@MarkusSchneider - I'll let that decision be up to you and @doudar. I'll certainly test it if it's decided so.

[doudar]

@htims05 @doudar Would it make sense to set the lower limit to 35 and to add an setting to disable the check of the low and high incline check?

If you like the idea I can add this setting to development.

@MarkusSchneider
I think best would be to just implement an adjustable minimum brake user setting. We could do both, I'm just a little concerned about the json string getting too long and causing radio time issues.

Along the radio time issues, what would really help is figuring out why sometimes the log messages are sent twice. That could really really help . I've looked into it, I can't figure it out.

[MarkusSchneider]

Along the radio time issues, what would really help is figuring out why sometimes the log messages are sent twice. That could really really help . I've looked into it, I can't figure it out.

I've also reviewed the code and could not find the problem. Have you also seen this error?
Looks like the UPD Appender is registered twice.

If this error happens, are also the messages in the debug page duplicated?
Only way to nail down the problem is to add some extra logs and record the serial log.

About what kind of radio issues you are concerned? Sending the settings form HTTP Server to the Browser via REST request?
Or do you expect a deadlock if the transport of the message takes to long blocking BLE communication?

We can break the settings page into smaller chunks e.g WIFI Settings, BIKE Settings, ... and store the settings in different files.

[doudar]

About what kind of radio issues you are concerned? Sending the settings form HTTP Server to the Browser via REST request? Or do you expect a deadlock if the transport of the message takes to long blocking BLE communication?

You nailed it - Deadlock if it's not released in time for BLE communications.

I've also reviewed the code and could not find the problem. Have you also seen this error? Looks like the UPD Appender is registered twice.

It even appears in the serial log. UDP on/off setting doesn't seem to affect it. More lines are doubled on mine than are normal but it's seemingly random.

We can break the settings page into smaller chunks e.g WIFI Settings, BIKE Settings, ... and store the settings in different files.

Best would be to send via websockets, if not for the IOS captive portal issues. Otherwise, yes, breaking up the JSON could work as well. I'm mostly concerned about rapid transmissions of http information, file size directly isn't an issue.

[doudar]

@MarkusSchneider , another idea to save space, radio time and add more functionality would be if(internetConnected), redirect to a more advanced, full featured UI on GitHub or elsewhere that communicates with the local SS2K via websockets.

[Lunchtime0614]

A full ride log here - It worked flawlessly for the ride - granted the ride didn't have much variation in resistance....but at the end again - during the 120w cooldown i was able to keep the watts well over 120 for a few min. I started by slowly lowering my cadence and it was keeping me at 120...then I decided to slowly increase the cadence....that's when it quit turning the knob to reduce resistance.

Full45MinRide.csv

[doudar]

Newest release is in the OTA with all the changes from this thread. Your unit should automatically update if it's connected to the internet. Feedback is greatly appreciated!

[eMadman]

You might want it to be configurable. The ic4 over reports wattage. 35 is likely to crash the knob to zero. The low wattage changes I'm using with mine only work because I have a crank power meter

…

On Mon., Jul. 4, 2022, 12:35 a.m. Markus Schneider, < ***@***.***> wrote: @htims05 <https://github.com/htims05> @doudar <https://github.com/doudar> Would it make sense to set the lower limit to 35 and to add an setting to disable the check of the low and high incline check? If you like the idea I can add this setting to development. — Reply to this email directly, view it on GitHub <#348 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AF4U2PGQ6G7TN4PWFAVCO2LVSJSYFANCNFSM52HIHGLA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[doudar]

No, it doesn’t have the limits commented out.

On Sun, Jul 3, 2022 at 1:31 PM htims05 ***@***.***> wrote: here's a new binary built with the latest libraries. There's a chance.... :) Make sure to unzip first: firmware.zip <https://github.com/doudar/SmartSpin2k/files/9033705/firmware.zip> @doudar <https://github.com/doudar> - does this have Markus' change he made included? — Reply to this email directly, view it on GitHub <#348 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEEO2KAZ4TY6UVH4IGBZTCDVSHL65ANCNFSM52HIHGLA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- *----------------------* *Anthony Doud*

[doudar]

This should be fixed by #391