enigma-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Enigma-devel] TiltStick, USB based tilt sensor


From: Tacvek
Subject: Re: [Enigma-devel] TiltStick, USB based tilt sensor
Date: Thu, 7 Feb 2008 12:50:20 -0500


----- Original Message ----- From: "Till Harbaum / Lists" <address@hidden>
To: "Tacvek" <address@hidden>
Cc: <address@hidden>
Sent: Tuesday, February 05, 2008 2:46 PM
Subject: Re: [Enigma-devel] TiltStick, USB based tilt sensor


Hi,

i was fiddling around with that some time ago and it didn't work.
But the first tiltsticks at that time were pretty unprecise. This has
changed and using jscal is now eliminating the death zone.

There's still a problem left: The sensivity of the tiltstick ranges from
-2G to +2G, but by tilting you can only reach +/- 1G (at least on
this planet). So using jscal will result in the max/min range mapping
being limited to +/- 1G where the device itself can do +/-2G.

It just seems to me like it should be possible to instruct the kernel to
use the raw input values as the output values with no nullzone. I'm betting that feeding the correct sequence to "jscal -s <...>" would do the trick, but I have no idea what the correct sequence is.

Now you bring up an intersting point though. What exactly does this device output? It is a three axis accellerometer. The normal way to use those as a true tilt sensor actually combine the acelleration data of two axes (the normally vertical one and one of the other two) arctan(Vert/horiz) gives an angle of tilt. That would really not care about the maximum acceleration used. It sounds like you are actually outputting the real acceleration. Obviously ingoring the normally veritcal axis, and using the horizontal ones in enigma can work reasonably well, but technically you are still measuring acelleration, allowing things like "nudging" to work, right? That is actually mor interesting than just tilt sensing, and would be useful in other programs (think the nudge command in pinball among other things.)

Anyway, thanks for pointing me to jscal again.

But there's still one reason left to use the event interface: I am doing
this to use it on the Nokia internet tablets. These unfortunately don't
include any joystick routines in the kernel at all. So my only way to
to get the TiltStick to work there is with the event interface.

Well that is a valid reason. If it is reasonable possible, installing the interface as a kernel module would be preferable, but I don't know if that code can be installed as a module, or if it must be compiled in.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]