Touchscreen requirements

Bert Freudenberg bert at
Tue Jun 8 09:03:18 EDT 2010

On 08.06.2010, at 07:44, John Watlington wrote:

> OLPC is looking to add multi-touch to our interface over
> the next year --- it is certainly necessary for a tablet.
> But "multi-touch" describes a huge range of parameters.
> Before taking a first pass at a spec. document, I'd like to
> stir up some discussion.   Here are the parameters that
> I think should be specified.   Feel free to comment on them
> as well as suggest others!
> Cheers,
> wad

Thanks for inviting discussion!

I recently started testing Etoys on the Apple iPad - not the least to get an idea of what needs to be done to it for an XO tablet. Multi-touch is definitely "cool", e.g. you can move around multiple Etoys objects simultaneously. If you are interested, I can send you a test version (pm me). 

Etoys uses the Sugar tool bar on all platforms. That works very well, the buttons are large enough even for my fingers.

However, many other Etoys UI elements are much too small for touching. They're finicky even on the XO screen, but with a pointer you can at least hit them. That's rather hard with a finger (though the Apple touch screen is amazingly precise). A stylus would be preferable, but then, two-finger gestures are awkward with a stylus.

It would help if the tablet had a 4:3 ratio. Ideally 1200x900 pixels just like the XO, but at least preserve the ratio. The iPad has 1024x768 pixels and I made Etoys use a virtual 1200x900 screen hw-scaled to the actual screen size, looks very nice.

> -----------------
> 1) Number of simultaneous touches:
> The number of simultaneous touches that can be tracked.
> For W7, this is two.   I believe OLPC is looking for more.

IMHO multi-touch is almost essential on a tablet. Two would be enough IMHO, more are nice.

What's "W7"?

> 2) Behavior when number of simultaneous touches is exceeded:
> If the number of simultaneous touches is exceeded, what happens ?
> I suggest that the "oldest" touch be forgotten and no longer tracked,
> but have seen other behaviors as well.

I'd rather keep tracking the first touches and ignore additional ones. Then accidental touches wouldn't interrupt the current interaction.

> 3) Palm rejection:
> A number of vendors include "palm rejection" algorithms in their
> controllers.   I'm not sure how I feel about this --- I would prefer to
> push this information higher in the stack before discarding it...
> 4) "Sensor size":
>  This applies to multizone resistive touchscreens, which may be
> thought about as a number of small touchscreens, each capable of
> a single touch.    Two touches cannot be detected in any one zone,
> so this affects how close buttons which might be pressed simultaneously
> (think piano keys) can be placed to one another.   W7 specs 1 in. x1 in. max.
> I believe this needs to be closer to 1 cm x 1cm max.

I'm imagining to use a two-finger tap to invoke meta operations, like the right touchpad button in Etoys on the XO. For this it would need to detect two kid's fingers form one ...

> 5) Resolution:
> Do we need to have a touch resolution equal to the screen size ?

I have no real experience with this yet, but it needs to be high enough to allow drawing. 

> 6) Scan rate:
>  The number of times a second that the touch controller
> can identify and report a touch.    W7 specifies 50 Hz minimum,
> which seems a little high.
> 7) Robustness:
>  This is usually specified as the number of presses in one spot with a contact
> area of either 8mm (finger) or 0.8mm (stylus).   Industry standard for resistive
> (single or multizone) seems to be around 80K, which is too low for our needs
> (we try to reach a 2000 day lifetime).  But the one vendor supporting 250K
> touches was unusable by a bare finger (needed fingernail or stylus).
> 8) UV resistance:
>  Since this touchscreen is on top of a sunlight readable display, it will need
> to be UV resistant.   Our current standard (for the display) is no significant
> change (>5%) in optical properties after 4000 hours of full sunlight UV
> irradiation.
> 9) Humidity, temperature:
>  Same as the XO:  operation from 0 to 50C, in RH up to 95%.
> Regards,
> wad

No opinion on these issues, yet ...

- Bert -

More information about the Devel mailing list