Touchscreen Interface Tester

Would this really be necessary? If it's the screen he's wanting to test
that would need a mechanical apparatus, though I would assume these
screens have already been well tested in target environments. But if
it's an application interfaced to the touch screen, it seems to be that
could be tested by imitating the data going across the interface (often
just a serial stream). The application would have no way of telling the
signals apart.

The screen should not permit invalid data sequences and never transmit
them, but you could assume those might be a possibility, and write a
trap in the application anyway. You could also manually sit there for a
half hour and poke at it yourself, capturing the data. Mix this in with
randomized data sequences.

Otherwise I think you're going to have to build something, and it won't
be cheap and it won't be very fast.

-- Gordon


BRW wrote:

Thanks for the reply Gordon.

The application he's testing is written explicitly for the touchscreen
using COTS software that interfaces with the COTS touchscreen system.
There is no easily accessable "port" (neither software nor hardware)
between the actual touchscreen and the application.  He doesn't want a
partial test of a hacked system.  What is desired is a true "end-to-
end" test from the user's perspective.  It's the only way to be sure
of the robustness of the application as a whole.

I'm glad you mentioned speed since I didn't cover that.  It doesn't
have to be fast.  One touchscreen press every 3-5 seconds would be
fine.  The idea would be to run the automated test all night or even
several days.

BRW

On Wed, 16 Jul 2008 17:43:40 -0700 (PDT), BRW



Make an x/y bridge/trolly operated by two servos (use drawer
rails and rollers from lowes/homedepot) with a third servo
mounted on the trolly to do the screen touching (use a servo
setup similar to below). Probably all components needed for
~$100. From my tinkering, I got ~425 discrete positions from a
standard servo with ~190 deg rotation range.

http://www.geocities.com/zoomkat/pix/servotest.jpg

Since you'd have to buy the hardware, then invest time and energy to
program it, why not find a college kid or two willing to do this for
$8-10 an hour. If you're really wanting a user's perspective for testing
it needs to be human, or else a very expensive and complicated test rig
that will simulate human responses - mistakes in positioning, too light
or too forceful of touches, and everything else a mechanical simulation
probably won't do.

At $10 an hour you can hire a couple of kids and get 30-40 hours of
testing. Splurge and buy them some Pepsi and pizza. Make that part of
the test - how your buddy's software does with a messy screen!

Not to take away a job for our robot friends, but some things are still
better (cheaper) done by humans. Robots are generally only
cost-effective over the long term, say simulating months or years or
testing.

-- Gordon


BRW wrote:

BTW, thanks for the article this month in N&V (or was it Servo?) about
surplus suppliers.  There were some great sources that I didn't know
about.

BRW

    A Lynxmotion 5 robot arm, at $279, should do the job.
The accuracy is moderate but good enough for touch-screen
testing.

    That's the plastic version. You might have to get the
aluminum and ball bearing version, at $452.39, or the robot
may wear out before the touch screen does.

     http://www.lynxmotion.com

                    John Nagle

Damn, I threw this out a month ago because I couldn't get any tackers:
Free: Roland A3 size DXY-980 Plotter.
Mind you, it was in Australia.

Don...




--
Don McKenzie

Site Map:            http://www.dontronics.com/sitemap
E-Mail Contact Page: http://www.dontronics.com/email

Xbee Wireless Modules, and low cost Interface Boards.
http://www.dontronics-shop.com/xbee-boards.html

On Wed, 16 Jul 2008 14:38:57 -0700 (PDT), BRW


   How many different touch positions does he need to test? You could
put a solenoid at each position. Solenoids are pretty cheap, easy to
drive, etc.