Forum
Why no friction unit?! - Printable Version

+- Forum (https://copynotes.be/shift4me/forum)
+-- Forum: Discussion (https://copynotes.be/shift4me/forum/forum-6.html)
+--- Forum: Technical (https://copynotes.be/shift4me/forum/forum-7.html)
+---- Forum: Alternative builds (https://copynotes.be/shift4me/forum/forum-14.html)
+---- Thread: Why no friction unit?! (/thread-31.html)



Why no friction unit?! - Balor - 09-29-2020

Or am I missing it from your build files?

Friction of gears is not anywhere close to keeping the cable from overcoming it and servo has to maintain a constant current draw just to keep it in place! Small wonder you only get a few hours of runtime... I've tested that personally with a wattmeter, and I am using a 35kg servo!

I'm using a similar principle electronic shifting for quite some time, using a nodemcu with a modified motor shield, 2s lipo battery, and a 3d printed housing just for the servo - the shield with a box "lives" in a frame bag.

I also have a blynk "module" to setup shifting servo angles wirelessly, but I reflash it with wifi off because it drains a ton of power.

I also has 3d printed shifting buttons, but frankly the I'll need to redesign it...

My main goal was not automatic shifting actually, but management of 3 cassete clusters in one transmission - on a recumbent bike that is supposed to be faired and require wide range AND small steps between gears... but I've simplified the design to just two cassettes and it turns out manual shifting of 72 speeds is not confusing at all actually Smile
Still, it can be quite useful for an automatic electric transmission that can provide lots of torque at high efficiency AND at wide range of speeds.

Btw, have you thought about torque-sensing BB integration? It will be better than just cadence! Interfaces with arduino neatly and costs just 65$ -though I've yet to play with it, to busy finalizing my bike frame.
[Image: AHan0VZh.jpg]
I'll share the design files once I brush them up a bit...


RE: Why no friction unit?! - Jan Oelbrandt - 09-29-2020

(09-29-2020, 12:53 AM)Balor Wrote: Or am I missing it from your build files?

Friction of gears is not anywhere close to keeping the cable from overcoming it and servo has to maintain a constant current draw just to keep it in place! Small wonder you only get a few hours of runtime... 
No that's right, the servo is pulling directly on the spring-returned cable. But I think the ratio of power consumption versus action radius is satisfying enough (5AA 2200mAh batteries lasting +10 hours. The two other AA's are just there to prevent the Arduino from powering-off due to voltage drop below 5V - I know this can be done better, see this thread)
But I agree, to add a friction unit would be a good idea to release tension on the servo so that it would consume less current.
(09-29-2020, 12:53 AM)Balor Wrote: have you thought about torque-sensing BB integration? It will be better than just cadence!
Yes I have. But while I developed Shift4Me, I always tried to make the design as simple and easy-to-install (and modify settings) as I could. That's the reason I never went for (amongst other options) torque sensing. But feel free to implement it!

btw: nice wheels you got there! Cool


RE: Why no friction unit?! - Balor - 09-30-2020

That will depend a lot on what gears you are in. At higher gears friction of servo by itself might be enough.
At lower gears drain is significantly higher, and may, in fact, burn the servo after a long climb in lowerst gear (despite it not actually helping you to climb that hill, lol).
I recommend you to use a wattmeter (there are ones for arduino) and see for yourself.
It depends on strength of derailer tension spring though... an other idea is "balanced" shifting, with a spring that counteract tension of the return spring.

Also, the is "servo jitter" phenomenon that cause additional drain on the battery when servo is heavily loaded. Have you tried tackling it?


RE: Why no friction unit?! - Ed_R - 10-04-2020

Interesting setup you have there Balor, would like to know more.
The servo pulling against a spring is something that has bothered me to, must surely under some conditions result in constant current draw by the servo. Fortunately enviolo hubs use a 2 cable system so has never been a problem for me. Although they can pull a little towards underdrive under load.

You mention torque sensing which is interesting because it's exactly where I went with my automatic shifter. Got me into building homemade power meters. For me having my cadence vary based on power output is the ultimate fully automatic shifting.

“servo jitter” I was going to look into turning the PWM signal to the servo on and off. Apparently a servo only tries to maintain a position when its being fed a valid PWM signal, kind of obvious really. In my case with the 2 cables I thought it might be more efficient if the servo is only driven when it needs to make a significant movement. Can waste some energy trying to move the last fraction of a degree.


RE: Why no friction unit?! - Fesh - 01-14-2021

(09-29-2020, 12:53 AM)Balor Wrote: I'm using a similar principle electronic shifting for quite some time, using a nodemcu with a modified motor shield, 2s lipo battery,
...
I'll share the design files once I brush them up a bit...

Hi Balor,

I'd like to build a shifter with 2s1p 18650 cells and an Arduino Nano 33 BLE (so I can adjust values via iPhone.app).
Can you tell me more about your friction unit?

Greetings from Vienna,
Marc


RE: Why no friction unit?! - kanejoyce - 07-12-2021

That is heavily dependent on the gears you are in. At higher gears, servo friction may be sufficient.
Drain is considerably higher in lower gears, and may even burn the servo after a lengthy climb in the lowest gear (despite it not actually helping you to climb that hill, lol).
I propose that you use a wattmeter (there are arduino-compatible wattmeters) and test it for yourself.
However, it is dependent on the strength of the derailer tension spring... another option is "balanced" shifting, which uses a spring to offset the tension of the return spring.
slope game
five nights at freddy's