anelok: touch slider idea

Werner Almesberger werner at
Tue Dec 3 16:51:59 EST 2013

Nick brought up an interesting idea: what if Anelok had a slider ?
We discussed touch sensor ideas before, but more in the style of
an iPod-like wheel.

Now, that display has a large area at its bottom for its circuit,
which isn't visible on the outside. There may be just enough room
for a sensor.

Q1: would it fit ?

The vertical stacking of the case is as follows:

- 0.8 mm window (nominal - varies due to machining tolerances),
- 0.3 mm airgap (so that a bezel can be formed, for stopping the
  masking fluid),
- 1.0 mm bezel (constrains the OLED module laterally),

and so on.

An FPC carrying the sensor could go into the airgap. A thin PCB
(0.4 mm) would also fit if the airgap was marginally larger. This
airgap would have to be separated from the one under the window
(we want to be able to have paint above the sensor) by a small
ridge. Some 0.8 mm plus 0.2 mm of tolerance should do.

Sandwiching the OLED between sensor and main PCB would be a more
interesting exercise. The OLED is already on the fumbly side (with
its fragile FPC) and this would make it worse.

Q2: could it replace the wheel ?

The wheel is not too dissimilar from a one-dimensional slider with
a button. On a slider, we could in fact have multiple virtual
buttons. E.g., consider the arrangement on

"V.A" is the visible display area, i.e., the area that's meant to
be looked at. The display matrix is a bit smaller.

The OLED module ends some 5.2 mm below the bottom border of the
visible area, and there's also a bit more room (1.6 mm) to bend
the FPC.

The display width (about 35 mm) is roughly as far as I can swipe a
thumb while holding Anelok without getting the feeling I'm heading
for early RSI. The sensor could be a bit wider than the display,
to accommodate extra buttons.

The drawing shows five zones: left and right button (red), left
and right swipe touchdown, and center. The touchdown zones could
only be used to initiate a swipe (and no tap), the button and
center zones would do the opposite. (But see below.)

Q3: how would the signals be interpreted ?

A think a sensor like the slider on the FRDM-KL25Z board basically
produces one variable (theirs uses two channels, but they're
correlated): let's assume there is a low value when the slider is
not touched, a minimum value (well above the "idle" value) on the
left side, and a high value on the right side.

A few examples:

1) Center tap. Reading jumps from idle to center, then drops again.

2) Left tap. The same concept, just the value is a lower.

3) Tap in touchdown area. We ignore that.

4) Drag with stop. Value jumps to that of the touchdown position,
   then changes to that of the lift-off position. Since we stop,
   the value lingers for a moment at the lift-off position and
   the rate of change may also decrease before stopping.

5) Drag inside touchdown area. This is the same as a tap and is

6) Another drag, this time with inertia. When quickly flipping
   through items, one may want to have a "fast forward" mode.
   This could be selected by lifting the finger off while still

   Anelok could then continue scrolling (or whatever the movement
   does) while the thumb returns to the touchdown area, and
   scrolling speed could even be accumulated, as if this was a
   wheel with inertia.

   This could be detected by the value falling off rapidly after
   reaching the lift-off position.

   Not sure if this concept can be implemented properly and whether
   it would even make sense. But it may be an idea worth trying.

7) What do we do if a swipe begins in the left or right button
   area ? Not sure. Maybe we should still accept it, i.e., as if
   the touchdown areas extended all the way to the ends of the

I think this could work as an input device. It would allow making
the device more compact and it would eliminate the wheel, which is
expensive, a sourcing risk, and also constrains the somewhat
delicate vertical stacking.

Q4: is it even possible ?

I don't know. Capacitative sensors in general like to be large.
This one would be tiny. They also like to be far from fields.
This one would sit on the glass containing part of the display's

I guess someone would have to try and see if this works.

A fairly simple prototype could be made with the OLED, an MKL25
(in 32-QFN, for easier soldering), and a FRDM-KL25Z for
programming and to supply power. The sensor could me made from a
suitably thin PCB.

Single-sided 0.4 mm PCBs are readily available, e.g.,

Double-sided would be better since it would allow for a proper
ground plane.

One could also use the FRDM-KL25Z directly to drive the OLED and
the sensor, but that may yield a different (and probably worse)
analog performance. But it may be good enough for a first quick

The KL25 and beyond have a built-in touch sensor circuit that
takes care of most of the legwork.

A driver for the sensor on FRDM-KL25Z can be found here:

Q5: would it feel right ?

That would have to be explored :-)

Opinions ?

- Werner

More information about the discussion mailing list