linux/Documentation/input/ntrig.txt
<<
>>
Prefs
   1N-Trig touchscreen Driver
   2-------------------------
   3        Copyright (c) 2008-2010 Rafi Rubin <rafi@seas.upenn.edu>
   4        Copyright (c) 2009-2010 Stephane Chatty
   5
   6This driver provides support for N-Trig pen and multi-touch sensors.  Single
   7and multi-touch events are translated to the appropriate protocols for
   8the hid and input systems.  Pen events are sufficiently hid compliant and
   9are left to the hid core.  The driver also provides additional filtering
  10and utility functions accessible with sysfs and module parameters.
  11
  12This driver has been reported to work properly with multiple N-Trig devices
  13attached.
  14
  15
  16Parameters
  17----------
  18
  19Note: values set at load time are global and will apply to all applicable
  20devices.  Adjusting parameters with sysfs will override the load time values,
  21but only for that one device.
  22
  23The following parameters are used to configure filters to reduce noise:
  24
  25activate_slack          number of fingers to ignore before processing events
  26
  27activation_height       size threshold to activate immediately
  28activation_width
  29
  30min_height              size threshold bellow which fingers are ignored
  31min_width               both to decide activation and during activity
  32
  33deactivate_slack        the number of "no contact" frames to ignore before
  34                        propagating the end of activity events
  35
  36When the last finger is removed from the device, it sends a number of empty
  37frames.  By holding off on deactivation for a few frames we can tolerate false
  38erroneous disconnects, where the sensor may mistakenly not detect a finger that
  39is still present.  Thus deactivate_slack addresses problems where a users might
  40see breaks in lines during drawing, or drop an object during a long drag.
  41
  42
  43Additional sysfs items
  44----------------------
  45
  46These nodes just provide easy access to the ranges reported by the device.
  47sensor_logical_height   the range for positions reported during activity
  48sensor_logical_width
  49
  50sensor_physical_height  internal ranges not used for normal events but
  51sensor_physical_width   useful for tuning
  52
  53All N-Trig devices with product id of 1 report events in the ranges of
  54X: 0-9600
  55Y: 0-7200
  56However not all of these devices have the same physical dimensions.  Most
  57seem to be 12" sensors (Dell Latitude XT and XT2 and the HP TX2), and
  58at least one model (Dell Studio 17) has a 17" sensor.  The ratio of physical
  59to logical sizes is used to adjust the size based filter parameters.
  60
  61
  62Filtering
  63---------
  64
  65With the release of the early multi-touch firmwares it became increasingly
  66obvious that these sensors were prone to erroneous events.  Users reported
  67seeing both inappropriately dropped contact and ghosts, contacts reported
  68where no finger was actually touching the screen.
  69
  70Deactivation slack helps prevent dropped contact for single touch use, but does
  71not address the problem of dropping one of more contacts while other contacts
  72are still active.  Drops in the multi-touch context require additional
  73processing and should be handled in tandem with tacking.
  74
  75As observed ghost contacts are similar to actual use of the sensor, but they
  76seem to have different profiles.  Ghost activity typically shows up as small
  77short lived touches.  As such, I assume that the longer the continuous stream
  78of events the more likely those events are from a real contact, and that the
  79larger the size of each contact the more likely it is real.  Balancing the
  80goals of preventing ghosts and accepting real events quickly (to minimize
  81user observable latency), the filter accumulates confidence for incoming
  82events until it hits thresholds and begins propagating.  In the interest in
  83minimizing stored state as well as the cost of operations to make a decision,
  84I've kept that decision simple.
  85
  86Time is measured in terms of the number of fingers reported, not frames since
  87the probability of multiple simultaneous ghosts is expected to drop off
  88dramatically with increasing numbers.  Rather than accumulate weight as a
  89function of size, I just use it as a binary threshold.  A sufficiently large
  90contact immediately overrides the waiting period and leads to activation.
  91
  92Setting the activation size thresholds to large values will result in deciding
  93primarily on activation slack.  If you see longer lived ghosts, turning up the
  94activation slack while reducing the size thresholds may suffice to eliminate
  95the ghosts while keeping the screen quite responsive to firm taps.
  96
  97Contacts continue to be filtered with min_height and min_width even after
  98the initial activation filter is satisfied.  The intent is to provide
  99a mechanism for filtering out ghosts in the form of an extra finger while
 100you actually are using the screen.  In practice this sort of ghost has
 101been far less problematic or relatively rare and I've left the defaults
 102set to 0 for both parameters, effectively turning off that filter.
 103
 104I don't know what the optimal values are for these filters.  If the defaults
 105don't work for you, please play with the parameters.  If you do find other
 106values more comfortable, I would appreciate feedback.
 107
 108The calibration of these devices does drift over time.  If ghosts or contact
 109dropping worsen and interfere with the normal usage of your device, try
 110recalibrating it.
 111
 112
 113Calibration
 114-----------
 115
 116The N-Trig windows tools provide calibration and testing routines.  Also an
 117unofficial unsupported set of user space tools including a calibrator is
 118available at:
 119http://code.launchpad.net/~rafi-seas/+junk/ntrig_calib
 120
 121
 122Tracking
 123--------
 124
 125As of yet, all tested N-Trig firmwares do not track fingers.  When multiple
 126contacts are active they seem to be sorted primarily by Y position.
 127
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.