Professional Documents
Culture Documents
The Electroquasistatics of The Capacitive Touch Panel: Abstract
The Electroquasistatics of The Capacitive Touch Panel: Abstract
INTRODUCTION
Field Equations
Consider the two-dimensional coated panels shown in Fig.
1. Sinusoidal voltage sources can be applied to the conductive
bars, as shown. If the frequency is low, quasi-static conditions
apply. In the plane of the coating, there is no free charge.
Ignoring the effect of capacitive leakage to structures away
from the coating, conservation of charge requires
V . J =O.
(1)
J = aE.
(3)
0093-9994/90/0500-0529$01.OO
v . ( a V 4 ) = 0.
O
1990 IEEE
(4)
530
TABLE I
MEASURED TOUCH IMPEDANCES, TYPICAL OPERATOR, RESISTORCAPACITOR CIRCUIT MODEL
Resistance
(Q)
Capacitance
(PF)
1400
1400
16
160
IO00
470
Condition
Surgical glove (series RC model)
Typical chair, no nearby metal surfaces
(series RC model)
Other hand on grounded metal surface
(parallel RC model)
Fig. 1. Touch panel horizontal measurement sequence. (a) Uniform horizontal field. (b) Uniform potential.
If
024 = 0.
(5)
Computation of Position
A human operator touching a surface will provide a leakage
path to ground of some impedance Z 1 . While conditions of
a touch are not predictable, several circumstances have been
measured experimentally to obtain typical values. These are
shown in Table I. The impedance is capacitive, as opposed to
inductive, under all conditions.
If a touch is applied to the panel, as in Fig. l(a), it will
draw a touch current I t x , with leading phase angle, from the
sinusoidal voltage sources. This current is simply the sum
of the currents injected by the supplies at the left and right
sides, and so is readily measured. Since 2, is unknown, a
second measurement will be required to allow calculation of
the horizontal touch position. Such a measurement is shown
in Fig.
Only the "Itage
have changed' Under
is constant Over time and very large
the assumption that
compared to the panel resistivity, and referring to Fig. 1,
Touch does not perturb the voltage
conditions on the panel due to the very
high impedance of the touch path to ground.
I,, =x v / z ,
(6)
I [ , = v/z,.
(7)
I1
Fig. 2
U
Fig. 3.
1) Switch off all top and bottom contacts. Connect all left
side contacts to -V and all right side contacts to +V. A
uniform horizontal field is created. Measure touch current
It, = Ileft + Iright .
2) Connect all left side contacts to +V and all right side
contacts to +V. Measure touch current I t , = Ileft Iright.
Compute x = Zt,/Itz.
3) Switch off the left and right sides. Connect all top side
contacts to +V and all bottom contacts to -V. A uniform
= Itop
+
vertical field is created. Measure touch current ItY
53 1
[---I--
i
+v+*v
117.
x=-I
RI
t 8)
Ib=o
k Ax
Irz
- V - ZIzRt
(1 + x ) R / 2
V - ItzRl
$-
(1 - x ) R / 2
(9)
532
-..t ..
. . I . ..
. J . . . .
. .I .
.L....
I .
_I
.
. I . . . . .I .
. I . . . . _I .
. I . . . . .l .
. I . . . ._I.
. I . . . 1. .
. I . . . . I .
. I . . . I..
. L . . . .
I .
1.
I.
.L....
. I . . . . J .
. I . . . . I .
. I . . . . 1.
.I.....
. I _ ..
I .
..I .
. L . . . .
. L . . . .
1 .
I .
I.
I .
. . . .
I . . . . .
....
c.
_ .
.\..
g>;:
, . I _ ..
. . . .
. L _ . ""
. . l . .
. .L . . . " .
' .
','
. . .1 '
I .
I..
..f.
. . I - '
. ..
. 1I
: < : : / . :I:; :: : 1 : :I
..
J . .
. . J . .
..J . .
..L
. . I
. . I
Fig. 6. Touch panel equipotentials, five electrodes per side. Contour interval: 0.05. Conductor voltages: 1, 0.
would be expected. The effect of this is a crowding of potential lines near the electrodes, indicating a significant local
potential drop. This crowding of potential indicates that higher
electrode voltages will be needed to create a given field near
the panel center, compared to the voltages needed with bar
contacts.
Thus the electrode spacing determines the position accuracy
of the panel near its edges, and the electric field strength near
the panel center. Clearly, more is better in this case; higher
fields mean higher current and thus better noise immunity,
while close electrode spacings enhance accuracy.
Most video monitors do not display information very close
to their edges. A touch panel overlay would normally be larger
than the display itself for ease in mounting. Thus touch information is of little utility within perhaps 2 cm of the panel edge.
Electrodes spaced 4 or 5 cm apart will provide 1% position
accuracy throughout the actual display surface. On a typical
monitor, this means that five to six electrodes are needed per
side.
The choice of 100 kHz also helps to avoid stray capacitive current flow from the CRT faceplate. The bulk of leakage
energy from the CRT occurs at harmonics of the beam flyback frequency, with some additional energy at a wide range
of video frequencies. Many computer monitors use a flyback
pulse frequency near 15.75 kHz, which is approximately a
standard television rate. The touch panel sensing circuitry
must present a low impedance to CRT frequencies.
The touch current is sensed as the sum of injected currents
into the two panel sides with a precision current-to-voltage
converter [6]. While currents can be sensed with good accuracy, a precise division process is difficult to implement in an
analog circuit. Instead, the sensed current is passed through
an automatic gain control (AGC) amplifier and converted to
digital form. Digital division avoids any further loss of accuracy.
The objective of the touch panel system is to detect
impedance changes at the panel surface of as little as 10 pF,
based on Table I data. Such a small change is not much larger
than leakage current changes that result from thermal effects,
the physical mounting of the panel, the placement of the display monitor relative to metal objects, dirt and debris on the
panel surface, or similar subtle causes. These effects cannot
be predicted for an end-use environment, so the system must
adapt to conditions. The process of distinguishing among effects and adapting to those that do not involve a touch contact is
difficult and ambiguous. As a result, higher touch impedances
tend to bring lower accuracy. Thick gloves are almost impossible to distinguish from leakage effects and thus cannot be
used with present capacitive touch panel techniques.
A TOUCHPANELSYSTEM
System Block Diagram
The electroquasistatic capacitive touch panel system has
been implemented in a production system, with a block diagram shown in Fig. 7. A host computer interacts with an
electronic touch panel control circuit, requesting and receiving information about touch position. The touch panel control
circuit includes a microprocessor for digital computation, data
communication, and control of the analog sensing circuitry.
Major components include a controllable sine-wave oscillator for driving the panel, a transformer to isolate panel drive
currents, a switch bank to connect oscillator signals to the
panel contacts, current detection and processing circuitry, and
the touch panel itself. The oscillator allows the operating frequency to be adjusted to avoid video signals from the display
unit. Such signals are capacitively coupled to the panel and
must not be mistaken for touch currents. A capacitive balance
network is connected to the transformer. It is set so that with
no touch, no current appears at the current detector. This
balance circuit plays the key role in adjusting for stray leakage impedances. The switch bank is critical to panel operation: it must be fast, have low on-state impedance, have high
off-state impedance, and not be damaged by spark discharges
from a fingertip. Analog switch integrated circuits meeting
the impedance requirements (Siliconix DG405) were used.
The switch bank is protected by a set of diode clamps, which
shunt the high currents seen during a discharge. The current
Computer
Processing
Communications
Position Ooto
Frequency
Oscdlator
Controller
Circuit
Fig. 7.
1) Monitor the panel in all four field configurations. If currents are below a threshold, assume no touch and adjust the
balance circuit for zero detected current. This process continues until current exceeds a threshold.
2 ) When sufficient current is detected, place the panel in
the horizontal z state, and set the AGC gain as necessary.
3) Measure I , , (horizontal).
4) Place the panel in the x state, and measure I l XCompute
.
x = I r x /Itz.
5) Place the panel in the z state (vertical), and measure I t z
(vertical).
6 ) Place the panel in they state, and measure I r y . Compute
Y =I t y l I t z .
7) Place the panel in the horizontal z state, and recheck I t ,
to validate the earlier result.
8) Filter and format the data. Send x,y , and z data to the
host as requested. (The meaning of z data is described later.)
9) Continue the measurement process in steps 2-8 until the
current goes below the touch threshold.
10) Return to the monitoring process of step 1.
This entire process allows position to be recomputed in less
than 20 ms for a typical touch impedance. The AGC gain and
speed depend on touch current amplitude, so that high touch
impedances give slower operation.
The Z Dimension
Capacitive touch panels share with the acoustic touch panel
[ 2 ] the possibility of a third measurement dimension. The
magnitude of I t , reflects the level of touch impedance. This
measured current is formatted and reported to the host computer in the form of a z component, which has a high value
for low current and vice versa. Since x and y data always indicate touch position, the added information is quite useful. For
example, a large value of z indicates that there is no active
533
System Accuracy
The two-dimensional model described predicts that measured position should not be affected by panel resistivity or by
touch impedance. The model assumes a uniform panel coating. A real panel is not uniform; its electric fields will be
distorted by resistivity variations as well as by the electrode
geometry. Limitations caused by leakage current levels and
noise effects also will be evident in the real system.
The system implementation described here has shown performance in keeping with the simplified theory. It is possible
to maintain a position accuracy on the order of 1 7 even
~
in the
face of other effects in a real system, unless the operator is
wearing thick gloves. The system here has proved to be workable on at least those panels with resistivities between 100 and
300 n/U, and those in which resistivity has local and global
variations below 10%.
Speed Issu es
A detection speed of 20 ms is faster than human reaction
time. In most situations, the operator perceives instant response of touch position to any finger movement. This is very
important in interactive graphics applications. The speed is
not high by computer standards. Indeed, one advantage is that
the host computer is not overwhelmed by a fast stream of
incoming data.
Speed is a function of touch impedance time constant, as
well as of the actual circuit measurement process. The slowest
such time constants seen in our tests are less than 2 ps and thus
do not limit speed. Analog switches can easily handle switch
rates of several kilohertz and thus make the time-multiplex
panel connection scheme feasible. The process involves thousands of switch operations each minute- far more than would
be reasonable with electromechanical relays.
Safety Issues
The capacitive touch panel connects a human operator to
an electrical source and so must be considered for possible
hazards. Two issues are the possibility of operator connection
to some hazardous source and the actual touch current itself.
Connection to dangerous sources inside a display unit can be
avoided by proper design and arrangement of components,
and by effective grounding of the panel control circuit board.
The circuit is protected so that it is never more than 20 V
off ground, thus indirect panel hazards are the same as those
associated with outer cases of any appliance or instrument.
The touch current is another matter. This current is simply the leakage through a person touching a low-voltage ac
534
source. The highest source voltage is 9 V rms in this implementation. The source is connected to an operator through
panel and circuit resistances, which total at least 100 0. The
resulting touch currents are no more than a few milliamperes
under any conditions. Such voltages and currents are below
those seen on many grounded appliance and instrument cabinets. In fact, the current levels are low enough to meet the
stringent requirements for patient-connected medical gear in
Underwriters Laboratories standard UL 544 [7]. To avoid any
added hazard, the panel source is programmed to shut off temporarily if the current exceeds some maximum limit.
FUTURE
EFFORTS
The electroquasistatic field sensing technique has been successful and is relatively straightforward. Future work will include efforts to reduce the number of electrodes through geometric mapping or other means. The field maps are predictable and permit this kind of mathematical manipulation,
but the calculations involved must not slow system response.
Other efforts will seek to improve performance in noisy environments by increasing the distinction between touch signals
and !eakage current drift effects. Although the present 20-ms
measurement time is faster than a human can react, it is possible, in applications such as keypad simulation, to have a touch
contact time briefer than that. In the future, touch panel systems which can measure the touch location within 5 ms will
be desirable. Touch panel control circuitry can, in principle,
be implemented on custom integrated circuits, which will lead
to very low system costs.
CONCLUSION
The capacitive touch panel interacts with a human finger in
a manner described by electroquasistatic field equations. This
insight has allowed design of a fast, precise, rugged touch
input device for video displays. This concept has been implemented in a touch panel system now in production.
ACKNOWLEDGMENT
The assistance of Roger McCoy, who gathered much of
the experimental data referred to in this paper, is gratefully
acknowledged.
REFERENCES
[I]