Robust user interface requirement

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
I have a UI requirement that's slightly unusual. Ideally it would be a touch
screen, but it will be 'touched' / poked in a big hurry, and maybe in a  
that would damage a normal screen at times. It's a control interface for
live music performance, and sometimes a number of changes have to be made
quickly, with the hands having to move from the instrument keyboard to the  
interface and back in a very short time. Up to now I've used normal  
pushbuttons and keypads, but the flexibility is a bit limited.

Is anyone aware of something that might suit?


Re: Robust user interface requirement
On 17/11/2013 8:04 PM, Bruce Varley wrote:
Quoted text here. Click to load it

No ideas except a suggestion - is there anything used in gaming /  
pinball / arcade games that may suit? They'd take a hammering.

Re: Robust user interface requirement
On Sun, 17 Nov 2013 20:04:51 +0800, Bruce Varley wrote:

Quoted text here. Click to load it

Can you put buttons around the edges of the screen instead?  You get  
quick use, lots of flexibility, and real buttons.

Tim Wescott
Control system and signal processing consulting
We've slightly trimmed the long signature. Click to see the full one.
Re: Robust user interface requirement
On Sun, 17 Nov 2013 10:16:05 -0600, Tim Wescott wrote:

Quoted text here. Click to load it

Soft keys!  That's what they're called.  Something like this is what I  
have in mind:

It's used on many many pieces of test equipment, and it works great.  
You'll need to be diligent about having a reasonable menu structure, but  
you needed to do that anyway.


Tim Wescott
Wescott Design Services
We've slightly trimmed the long signature. Click to see the full one.
Re: Robust user interface requirement
snipped-for-privacy@seemywebsite.please says...
Quoted text here. Click to load it

I did an interface for data collection on fishing boats that
used configurable LCD prompts and capacitive touch sensors
behind a 1/8" polycarbonate panel.  It was waterproof and,
with a bit of tuning, even able to register touches from
users wearing rubber gloves.  The waterproof part might
be useful if the performances will occur in bars where
bottles may be thrown by dissatisfied customers!

Mark Borgerson

Re: Robust user interface requirement
Bruce Varley wrote:

Quoted text here. Click to load it

Hi Bruce,

When you say "poked" is that with a human finger?

Does a screen need to display information in order for a selection to be  
made. How big an area do you need and does it have to be very rigidly  
mounted? How changeable is the selection infoirmation?

So far the screen on my Acer Iconia has withstood some heavy pokes from some  
people who have accidentally jabbed at it when identifying something of  
interest. However, I am not certain it could withstand someone who is very  
heavy handed. I would have thought most musicians were dexterous enough to  
be able to use such a device though.

We've slightly trimmed the long signature. Click to see the full one.
Re: Robust user interface requirement
Hi Paul,

On 11/17/2013 1:57 PM, Paul E Bennett wrote:
Quoted text here. Click to load it


Quoted text here. Click to load it

I think the issue is that the operator wouldn't have *time*/luxury to
do this in a leisurely fashion.  The equipment may serve a real, needed
purpose -- but, it's a distraction from PLAYING MUSIC.

Consider most musical instruments require both hands to play (harmonica
requiring none?) and usually some amount of concentration and visual
"attention".  Looking up from a keyboard (organ) to see which button
you want to press (i.e., identify its location in space) AND then
pressing it -- without skipping a beat in your performance probably
means you *throw* your hand at it instead of finely positioning your
finer as you might on a computer keyboard.

Sort of like the SCAN buttons on car radios... just slap it and
let*it* find the next station -- instead of having to focus your
attention on that task.

Re: Robust user interface requirement
Don Y wrote:

Quoted text here. Click to load it

I was referring to a touch-screen slate which, in touch screen mode, has  
proved reasonably robust. The Acer Iconia might not suit Bruce's  
price/performeance point though. I am sure other touch screens might be  
similarly robust and provide reasonable sized areas for menu selection by  

We've slightly trimmed the long signature. Click to see the full one.
Re: Robust user interface requirement
Hi Paul,

On 11/18/2013 2:08 AM, Paul E Bennett wrote:
Quoted text here. Click to load it

Understood.  My point was that the "actuation time" (how long the
"finger" is in a "sense-able" position might be too brief for many
technologies to "see" reliably.

If his users are so "hurried" that they can't take the time to
"deliberately" actuate the screen, they probably would be even
more annoyed to notice that their attempt had not been registered
(perhaps not noticing until after they had SUCCESSFULLY invoked
some other function) and that they would have to adjust their
actions on the fly to re-touch the original control.

IMO, that was the "real problem" Bruce is facing (moreso than

Re: Robust user interface requirement
Don Y wrote:

Quoted text here. Click to load it

Looking at Bruce's original post, he has accomplished the task with normal  
pushbuttons and keypads before. I have noticed that my touch screen seems to  
be at least as quick as that method and may just give the versatility. I was  
just not sure the price/performance point would be right for him. The  
qustion would be regarding re-purposing a touch screen device to act within  
reasonable time for this function and give the fleibility he was after.  

We've slightly trimmed the long signature. Click to see the full one.
Re: Robust user interface requirement
Hi Paul,

On 11/18/2013 11:14 AM, Paul E Bennett wrote:
Quoted text here. Click to load it

Yes, but they are mechanical devices.  Push it hard/fast/slow/soft and
those metal contacts still come together -- regardless of how briefly.
(depends on how you scan them).

Quoted text here. Click to load it

I assumed he was looking to OEM something -- using components and
not repurposing an existing device.  Hence my litany on different
touchPANEL technologies.

I've had very bad experience with touch pad response times.

PDA's (resistive) and older phones seem to be "adequate" when used with
a stylus (as long as your "jab" doesn't "ding" the touch overlay).
But, require more "deliberate" effort with a fingertip.  Even seemingly
hard "taps" get ignored if too brief.

SWMBO has a habit of "stabbing" the "YES" button on the signature
pads at checkouts ("Is this amount correct?") and having to do so
repeatedly (not that she is missing the "box" but, rather, her
actions are too "staccatic").  I have dismissed this as related to
the amount of abuse these devices must typically encounter!

And, some assistive technology devices (press here to say "Please
bring me some water" and here to say "I need to go to the bathroom")
that are *very* difficult to coerce into "seeing" a button press.
I've assumed this was a consequence of needing the display to be
extraordinarily robust given motor skill limitations of its users.

I have (large LCD) panels from three different manufacturers with SAW
technology that also seem to want deliberate actions.

This tablet PC uses an active digitizer yet the pen often fails
to register a click.  I've grown wary of letting others use it
as they invariably start STABBING the display thinking that
will make the click happen more reliably (in fact, all it will
probably do is crack some ferrites!)

OTOH, I've used some touchPADS (i.e., the small things in keyboards)
that would register a click if your hand hovered above it!  :<

As I can't peer inside at their implementations, I can't tell
whether these are inherent limitations of the technology,
"filters" added to reduce false positives or otherwise driven
by the applications.

I figured the photointerrupter approach to be the least risk
up front:  beam gets broken, you see the finger's location

Bruce will have to sort out what his practical concerns are.

Re: Robust user interface requirement
Den søndag den 17. november 2013 21.57.37 UTC+1 skrev Paul E Bennett:
Quoted text here. Click to load it
Quoted text here. Click to load it
Quoted text here. Click to load it
Quoted text here. Click to load it
Quoted text here. Click to load it
Quoted text here. Click to load it

yep modern capacitive touch works through glass, I've read that some work t
hrough at least 6mm

another way I remember from way back, to to have a stiff plate such as glas
on three force sensors and sorta "triangulate" the point of touch  


Re: Robust user interface requirement
Hi Lasse,

On 11/17/2013 3:15 PM, wrote:

Quoted text here. Click to load it

For the OP, I think the problem will be how quickly it can be sensed
and how stable the "finger" can be in that time interval.

Quoted text here. Click to load it

Someone made a (CRT) "monitor base" many years ago that did exactly
this.  Set your monitor *on* it, calibrate and you have a touch monitor.

I suspect this would be bad for the OP solely based on the mechanics
involved -- monitor jiggling as it is "struck", etc.

Re: Robust user interface requirement
Hi Bruce,

On 11/17/2013 5:04 AM, Bruce Varley wrote:
Quoted text here. Click to load it

But, that "poke" wouldn't damage a CRT or LCD (with a lexan face)?
I.e., you're not worrying about drummer taking a swipe at it with a

How fast is fast?  And, is it "fast" because it needs to be "of short
duration"?  Or, because the operator is taking a careless *jab* at it?
Presumably, its a *set* of controls that the user needs to be able
to freely and independantly access?  Not just "move to the next
preprogrammed preset"

Quoted text here. Click to load it

But, the time to press a pushbutton isn;t necessarily short -- depends
on the mechanics of the button and how reliably it stays (?) pushed
(are these momentary or latching/toggle?)

Quoted text here. Click to load it

You want a relegendable button, effectively?

Does the legend have to be graphic?  Multichrome?  etc.
How precisely can you expect users to *find* the button?  I.e.,
does the size need to be variable as well?  (of course, you have very
little flexibility with your existing mechanical buttons in this
regard.  But, are all your mechanical buttons the same size?
Nothing the equivalent of an EMERGENCY STOP??)

How many buttons in what sort of space (area)?  Any size/weight/cost
constraints (naw, those things never come into play!  :>)

There are lots of choices for touch screen technologies.
[I'll restrict my comments to these, initially.  There are
also *other* approaches you might want to investigate]

Most technolgies assume that you have a firm surface against which the
"actuator" (finger, tylus) can operate.  I.e., a *typical* LCD monitor
has a flimsy plasic "film" so pressing on it will actually damage the
display!  Most originate in the CRT days so you had a significant bit
of glass that the user could interact with mechanically without
damaging the display.

If the points on the display surface are relatively few, you an use a
bezel of photoemitter/detector pairs.  Beams of (Ir) light IN FRONT OF
the display surface.  Watch to see which *2* (or more -- for folks with
fat fingers :> ) beams are interrupted -- one X ,one Y -- and you
know where the finger has touched the screen.

These are relatively easy to calibrate -- the emitters/detectors
have fixed locations so *they* require no calibration (though they
must be aligned properly with their mates).  You do, however, need a
way to ensure the image *behind* the array "fits" its dimensions
(cuz your software expects pixel (x,y) to be "behind" a particular
set of emitter/detector pairs).  For a CRT, this means adjusting
height and width in the monitor (or, comparable scaling in the
video generator).  For an LCD, it's not an issue!  Just ensure the
registration of the touch panel wrt the display itself (left/right,

The downside is this approach costs more emitter/detector pairs to
get finer "precision" in "finger reporting".  Conversely, if you
don;t need to "see" fingers in some portion of the display,
you can omit the emitter/detectors that would otherwise be needed to
cover that part of the display!  "Speak now or forever hold your peace"

This can be power hungry, relative to other approaches.  One way to
conserve power is to multiplex the drives.  But, that increases the  
amount of time it takes to reliably *see* an finger.  (if you have 50
emitter detector pairs on each axis, muxing can save A LOT of power!)
So, you tend not to get *big* savings because you don;t want to
increase the "scan time" for the entire array.

[You can, however, easily do "X detect" followed by "Y detect" and cut
power in half for very little sacrifice in response time *or*
hardware cost.]

Another optical variation uses a fixed, rotating light source that
sweeps the screen.  Less power in the optics -- but now you have
mechanism to attend to.  And, slower responses.

In the first approach, failure of *a* detector or emitter causes the
loss of a row (or column) sense.  Depending on where your buttons are
(and, if they can be relocated dynamically), you may lose some number
of buttons related to that dead row/column.  With the rotating beacon
approach ("Cyclops" was one commercialized name), failure of the
beacon (optics *or* mechanics) meant the entire detector was shot.

[I offer this as I suspect most performers aren't keen on repairing
kit during performances -- though they may have the ability to do so!]

You can also optically detect the deflection of a clear membrane
suspended *above* the display surface.  Hunt for the "bright spots"
on each axis to indicate the points of maximum stress, etc.

You can also require the finger to be "active".  The classic case was  
the lightpen -- a photodetector tethered to the display in a pen body.
Essentially, when it "saw" the scanning electron beam from the CRT, a
snapshot of the "address" being accessed by the crt controller was made.
This could be interpreted to determine which row/colum *pixel*
was being seen at the time.

Of course, reports happened at the frame/field rate of the display;
you couldn't be guaranteed to "see" anything until all pixels had had
a chance to "shine".

Similarly, you can have a light source *in* the finger (ET phone home)
that sensors in the display's *frame* detect.

The same is true of active digitizing pens, etc.

I imagine you *don't* want the user to have something in his hand to
ineract -- even if it was *permanently* in his hand for the duration
of the gig!

Touchpads in PDAs used a membrane overlay.  Two conductive transparent
sheets of plastic held apart by invisible separators.  Pressure on the
outermost deflected it so that it would come into contact with the
inner (closer to display) surface.  The "adjacent" surfaces of each
membrane were coated with a resistive material.  A potenial would be
impressed across one membrane.  The voltage drop at any point X%
from the edge of that membrane would be X% of the impressed potential.
(imagine an infinite number of identical resistors in series).

The *other* membrane would be used to *sense* the potential at the
point of contact -- the entire membrane would feed an "A/DC" that
was configured ratiometricly wrt the excitation voltage on the
other membrane.  I.e., 000000 meant the contact point was exactly
at the "lowest potential end" of the membrane while 1111111 meant it
was at the highest potential end of the membrane.

[pictures would be *so* much easier to explain!]

Repeat the process exciting the membrane along the other axis.  Combine
these two A/DC readings and you have a spot in 2-space.  Speed of A/DC
determines how quickly you can resolve location on an axis.  Number of
bits in A/DC determines how many positions you can resolve ALONG that

These were popular because they were inexpensive to produce.  Esp
in those small sizes (imagine, for example, trying to sense position
on a small display with *optics*!)

Problem is, they interfere with visibility -- decreasing light output
(or "throughput" for transflexive displays) and making things a bit
fuzzier.  And, as they represented a genuine surface *above* the
display, they were susceptible to damage -- the (separated!) membranes
were typically softer than the surfaces that supported them so they  
would get gouged, etc.

In your case, as the membrane is expected to deflect, too abrupt a
motion can cause it to "bounce back" -- before you get around to
noticing it (you would need to do your own controller to ensure you
poll fast enough)

The same sort of approach can be done capacitively.  There are different
approaches to capacitive detectors.  E.g., each of the previously
mentioned techniques will successfully detect a drumstick used to
actuate a virtual button.  Capacitive detectors tend to rely on the
human body *in* the circuit.  Gloved hands can cause problems
(as would a drumstick, etc).

I'm particularly fond of SAW (surface acoustic wave) technology.
Essentially, a "sound" (inaudible) is coupled to the glass (CRT
invariably).  Your finger (or other material) contacting the glass
changes the characteristics of the "sound" when it reaches the
detectors ("microphones").  This lets the controller determine
the interference is located.  It's biggest advantage is that
it also lets you detect (relative) *pressure*, not just location.

But, anything that changes the characteristics of the "acoustic
conductor" affects its accuracy.  E.g., get it wet and it behaves
differently than when dry!  (It's also not immediately obvious
to users that a given pressure from their *thumb* may be interpreted
entirely different than the same pressure from their index finger!
Or, *your* index finger!)

Again, in your application, this is a slower technology.  The
frequency of the "sound" effectively sets a polling rate (can't
see the result until you've got a signal to recognize!) as does
propagation delay through the glass (1-2 ms just for the signal
to get across the surface!)

(sigh)  I've probably forgotten a technology or three...

*If* you can afford the cost (materials & labor) *and* have a
low enough "resolution", I suspect the emitter/detector array
will give you the quickest RELIABLE scan.  (you don't want a
performer to have to repeat the action because it wasn't
recognized the first time!)  It's not as elegant as many of
the newer approaches but you asked for robust, not elegant!

If these buttons were just presets or effect in/out, you
might also consider allowing them to relocate themselves
(or, be relocated by the performer) in the event a detector
fails, has BBQ sace spilled on it, etc.  The Show Must Go On.

Or, even a semicustom design:  imagine strips of flexible plastic
each fed by an emitter on one end and sensed by a detector on
the other end.  When the strip is deflected *anywhere* along
its legth/width, the light output conducted through that strip
drops off...

(sigh)  Lots of word.  Hopefully some will help.

You might also want to consider other input modalities, depending
on the choices/controls you are trying to provide...

Re: Robust user interface requirement
2013-11-17 13:04, Bruce Varley skrev:
Quoted text here. Click to load it
You can have capacitive touch buttons.
Work like a touch screen, but no mechanics, just electronics and a  
sensor which is just a certain layout on a PCB.

Ulf Samuelsson

Quoted text here. Click to load it

Site Timeline