Anelok: save the slider !

[Werner Almesberger]

Anelok's capacitive slider has made a terrible mistake - it has
caused embarrassment for its boss. Worse yet, the slider gives
no sign of accepting the blame and removing itself gracefully.
If Werner has learned anything from his work in China, it is
that such a transgression must not go unpunished.

Will the slider be destroyed by Werner's just anger ? He already
added button shortcuts (cursor up, left, and down, or the
corresponding vi cursor keys, for top, middle, and bottom "soft"
button) to the simulator ... and kinda likes the result.

https://gitlab.com/anelok/anelok/commit/cb9ae5a66a392c409955348bf73b7183de497efe

Time is running out for the slider ! Only you can save it !

The following quests await the champion who rushes to the
slider's aid:

- locate the Gloves of Fumbling [cursed], understand how they
were supposed to work, and discover the source of their
corruption;
(hint: fw/2014/touch.c)

- cleverly devise a way to remove the curse;

- implement and test your solution; (*)

- send your code to the Dark Oracle (Werner, wearing another of
his many faces), have its fitness judged, and have it stand
trial on Werner's devices.

- if necessary, repeat some of the above quests and use the EXP
you gained to reach a higher score.

(*) Courtesy of Xiangfu, a small number of caseless [1] and
pre-rework [2] prototypes of the CR2032 design are available.
If demand should exceed supply, a beauty contest will be held.

[1] Loading Image...

[2] http://downloads.qi-hardware.com/people/werner/anelok/tmp/boost-rework.pdf
Loading Image...
Loading Image...

In case you have not trained the skills needed to fulfill this quest,
consider using Word of Mouth to summon a suitable paladin.

- Werner

[Ron K. Jeffries]

Slider is OK concept, if it works. It's not clear to me what the proposed
alternative is or how it would be used.

My hunch is hardware stays the same with button areas on what is now a
sliver. That makes sense for cursor functions. I am baffled as to how vi
key bindings might be presented.

Your note is very clever.

Post by Werner Almesberger
Anelok's capacitive slider has made a terrible mistake - it has
caused embarrassment for its boss. Worse yet, the slider gives
no sign of accepting the blame and removing itself gracefully.
If Werner has learned anything from his work in China, it is
that such a transgression must not go unpunished.
Will the slider be destroyed by Werner's just anger ? He already
added button shortcuts (cursor up, left, and down, or the
corresponding vi cursor keys, for top, middle, and bottom "soft"
button) to the simulator ... and kinda likes the result.
https://gitlab.com/anelok/anelok/commit/cb9ae5a66a392c409955348bf73b7183de497efe
Time is running out for the slider ! Only you can save it !
The following quests await the champion who rushes to the
- locate the Gloves of Fumbling [cursed], understand how they
were supposed to work, and discover the source of their
corruption;
(hint: fw/2014/touch.c)
- cleverly devise a way to remove the curse;
- implement and test your solution; (*)
- send your code to the Dark Oracle (Werner, wearing another of
his many faces), have its fitness judged, and have it stand
trial on Werner's devices.
- if necessary, repeat some of the above quests and use the EXP
you gained to reach a higher score.
(*) Courtesy of Xiangfu, a small number of caseless [1] and
pre-rework [2] prototypes of the CR2032 design are available.
If demand should exceed supply, a beauty contest will be held.
[1]
http://downloads.qi-hardware.com/people/xiangfu/anelok/tmp/Anelok-3-brds.jpg
[2]
http://downloads.qi-hardware.com/people/werner/anelok/tmp/boost-rework.pdf
http://downloads.qi-hardware.com/people/werner/anelok/tmp/boost-rework-cables.jpg
http://downloads.qi-hardware.com/people/werner/anelok/tmp/boost-rework-cuts.jpg
In case you have not trained the skills needed to fulfill this quest,
consider using Word of Mouth to summon a suitable paladin.
- Werner
_______________________________________________
Qi Hardware Discussion List
http://lists.en.qi-hardware.com/mailman/listinfo/discussion

[Werner Almesberger]

Post by Ron K. Jeffries
Slider is OK concept, if it works.

Yes, that little "if" is the problem ...

Post by Ron K. Jeffries
It's not clear to me what the proposed
alternative is or how it would be used.

The alternative would be to add buttons (either micro switches
or domes) where the slider is now. With three buttons, you'd
have the same functionality as with the slider now, minus
sliding.

Comparison buttons (3 of them) vs. slider:

Attribute Slider Button
--------------------------------------- --------------- ---------------
BOM cost "free" moderate / high
Response to dirt / aging immune degrade / fail
Mechanical load distribution case top PCB
(Unintended) activation threshold low medium
Signal processing demanding easy (debounce)
Tactile response no yes
Power consumption (when idle) polling none
Gestures, single touch 5 3 (5 ?)
multi touch n/a <= 4, may be messy

To scroll through a list, you'd press "up" or "down" repeatedly.
(There can be no auto-repeat because long presses are already
used for other functions.)

For example, when on the accounts list, the buttons are currently
assigned to the following functions:

Button Tap Long press Very long press
------- ----------------------- --------------- ---------------
Top Previous entry (up) Setup (1, 2) Setup (2)
Middle unassigned Select entry Select entry
Bottom Next entry (down) Switch off Zap session

(1) Would become "back" once directories are implemented and
you're in a subdirectory.
(2) Or "Alert" if an asynchronous notification is pending.
(Press "A" in the simulator to generate a radio alert.)

If you make a sliding movement while pressing on the slider,
fw/ui/input.c switches to slider mode and generates N pixels
up/down information.

You can play with all this in the simulator.

With more than three buttons, be it real or virtual (e.g., by
combining top/bottom and middle), auto-repeat would also be
possible. Going > 3 buttons may be get messy, though, given the
limited space.

Post by Ron K. Jeffries
I am baffled as to how vi
key bindings might be presented.

Ah, that's for the simulator:

Cursor key vi key Anelok slider area
--------------- --------------- ------------------
Up K Top
Left L Middle
Down J Bottom

Post by Ron K. Jeffries
Your note is very clever.

Thanks ! :)

- Werner

[Dave Ball]

[repost - I sent the original off-list by mistake!]

Post by Werner Almesberger
Anelok's capacitive slider has made a terrible mistake - it has
caused embarrassment for its boss. Worse yet, the slider gives
no sign of accepting the blame and removing itself gracefully.

I won't pretend to understand how the capacitive slider is intended to
work, but reading here:

http://www.embedded.com/design/mcus-processors-and-socs/4415133/Adaptive-Kalman-filtering-for-better-buttons-slider-capacitive-sensing


I was interested in their description of a slider "A slider is a series
of touch buttons cascaded together with overlapping transitions and
designed specifically to track finger position and movement."

As I understand it, the current Anelok slider is actually a linear
slider, not a series of buttons. Are there some alternative layouts of
the PCB track that would make the slider less prone to atomspheric
variations?


Dave

[Werner Almesberger]

Post by Dave Ball
[repost - I sent the original off-list by mistake!]

Thanks ! :)

Post by Dave Ball
As I understand it, the current Anelok slider is actually a linear
slider, not a series of buttons.

Yes, it's a simple linear slider. Having multiple areas should
make it easier to associate a contact with one of the areas
(i.e., the one that produces the highest reading).

But I suspect that it could become difficult to translate this
into the linear position needed for sliding.

Post by Dave Ball
Are there some alternative layouts
of the PCB track that would make the slider less prone to
atomspheric variations?

There are lots of layouts ;-) For basic principles, I used
Freescale's AN3863:
http://cache.freescale.com/files/sensors/doc/app_note/AN3863.pdf

There, they make sliders using multiple areas. The problem is
that this uses a lot of channels = pins. Freescale themselves
picked a more economical approach for their FRDM board:
http://www.freescale.com/webapp/sps/site/prod_summary.jsp?code=FRDM-KL26Z

This has a dual Z shape, similar to Anelok's triple Z.

One possible variation would be to remove the ground traces
between the sensor plates, and let the MCU drive one plate to
ground while measuring the other. This would leave room to
increase the plate surface and may yield a stronger signal.
Noise immunity may suffer, though, given that the GPIO acts like
a ~100 Ohm series resistor when connecting the plate to ground.

Post by Dave Ball
http://www.embedded.com/design/mcus-processors-and-socs/4415133/Adaptive-Kalman-filtering-for-better-buttons-slider-capacitive-sensing

Since you mention Kalman filtering, I would indeed hope that
better signal processing could make a huge difference here.
Alas, this isn't really my field, and my contraption held
together (or not) by spit and duct tape shows it.

- Werner