Quantum QT310 DATA SHEET

LQ

P

ROGRAMMABLE

 Single channel digital advanced capacitance sensor IC
 Spread spectrum burst modulation for high EMI rejection
 Full autocal capability
 User programmable via cloning process
 Internal eeprom storage of user setups, cal data
 Variable drift compensation & recalibration times
 BG and OBJ cal modes for learn-by-example
 Sync pins for daisy-chaining or noise suppression
 Variable gain via Cs capacitor change
 Selectable output polarity, high or low
 Toggle mode (optional via setups)
 Push-pull output
 Completely programmable output behavior

via cloning process from a PC

 HeartBeat™ health indicator (can be disabled)

APPLICATIONS

ROX

™

QT310

QP

C

APACITANCE SENSOR

IC

 Fluid level sensors
 Industrial panels

This device requires only a few external passive parts to operate. It uses spread-spectrum burst modulation to dramatically
reduce interference problems.

The QT310 charge-transfer (“QT’”) touch sensor IC is a self-contained digital IC capable of detecting proximity, touch, or fluid
level when connected to a corresponding type of electrode. It projects sense fields through almost any dielectric, like glass,
plastic, stone, ceramic, and wood. It can also turn metal-bearing objects into intrinsic sensors, making them respond to
proximity or touch. This capability coupled with its ability to self calibrate continuously or to have fixed calibration by example
can lead to entirely new product concepts.

It is designed specifically for advanced human interfaces like control panels and appliances or anywhere a mechanical switch
or button may be found; it can also be used for material sensing and control applications, and for point-level fluid sensing.

The ability to daisy-chain permits electrodes from two or more QT310’s to be adjacent to each other without interference. The
burst rate can be programmed to a wide variety of settings, allowing the designer to trade off power consumption for response
time.

The IC’s RISC core employs signal processing techniques pioneered by Quantum; these are specifically designed to make
the device survive real-world challenges, such as ‘stuck sensor’ conditions and signal drift. All operating parameters can be
user-altered via Quantum’s cloning process to alter sensitivity, drift compensation rate, max on-duration, output polarity,
calibration mode, Heartbeat™ feature, and toggle mode. The settings are permanently stored in onboard eeprom.

The Quantum-pioneered HeartBeat™ signal is also included, allowing a host controller to monitor the health of the QT310
continuously if desired.

By using Quantum’s advanced, patented charge transfer principle, the QT310 delivers a level of performance clearly superior
to older technologies yet is highly cost-effective.

 Appliance controls
 Security systems

 Access controls
 Micro-switch replacement

 Material detection
 Toys & games

LQ

AVAILABLE OPTIONS

A

Copyright © 2002 QRG Ltd QT310/R1.03 21.09.03

8-PIN DIPSOICT

QT310-D-00C to +700C

-QT310-IS-400C to +850C

Table 1-1 Pin Descriptions

5Vd

c

FunctionNamePin

Ext Cal, latch clear input/CAL_CLR1

Sync Output/SYNC_O2
Sense 1 lineSNS13

Negative supply (ground)VSS4

Sense 2 lineSNS25

Sync Input/SYNC_I6

Detection outputOUT7

Positive supplyVDD8

Alternate Pin Functions for Cloning

Serial clone data clockSCK3

Serial clone data outSDO6

Serial clone data inSDI7

1 - OVERVIEW

The QT310 is a digital burst mode charge-transfer (QT)
sensor designed for touch controls, level sensing and
proximity sensing; it includes all hardware and signal
processing functions necessary to provide stable sensing
under a wide variety of changing conditions. Only one low
cost sampling capacitor is required for operation.

A unique aspect of the QT310 is the ability of the designer to
‘clone’ a wide range of user-defined setups into the part’s
eeprom during development and in production. Cloned setups
can dramatically alter the behavior of the part. For production,
the parts can be cloned in-circuit or can be procured from
Quantum pre-cloned.

Figure 1-1 shows the basic QT310 circuit using the device,
with a conventional output drive and power supply
connections.

1.1 BASIC OPERATION

The QT310 employs bursts of charge-transfer cycles to
acquire its signal. Burst mode permits power consumption in
the microamp range, dramatically reduces RF emissions,

Result

SNS1

r
e

l

Sta rt

l
o

r
t

n
o

C
t
s

r

Done

u
B

ingle-Slope
S

Switched Capacitor ADC

Charge

Amp

Cs Cx

SNS2

lowers susceptibility to EMI, and yet permits excellent
response time. Internally the signals are digitally processed to
reject impulse noise, using a 'consensus' filter which requires
several consecutive confirmations of a detection before the
output is activated.

A unique cloning process allows the internal eeprom of the
device to be programmed to permit unique combinations of
sensing and processing functions.

+2to

100nF

8

VDD

12

/CAL

SYNC_O

6

SYNC_I

7

OUT SNS2

VSS

4

SNS1

3

5

C

4.7nF

ELECTRO DE

s

C

Calibration

10K

10K

Figure 1-1 Basic QT310 circuit

1.2 ELECTRODE DRIVE

1.2.1 S

The IC implements direct-to-digital capacitance acquisition
using the charge-transfer method, in a process that is better
understood as a capacitance-to-digital converter (CDC). The
QT switches and charge measurement functions are all
internal to the IC (Figure 1-2).

The CDC treats sampling capacitor Cs as a floating store of
accumulated charge which is switched between the sense
pins; as a result, the sense electrode can be connected to
either pin with no performance difference. In both cases the
rule Cs >> Cx must be observed for proper operation. The
polarity of the charge build-up across Cs during a burst is the
same in either case. Typical values of Cs range from 10nF to
200nF.

Larger values of Cx cause charge to be transferred into Cs
more rapidly, reducing available resolution and resulting in
lower gain. Conversely, larger values of Cs reduce the rise of
differential voltage across it, increasing available resolution
and raising gain. The value of Cs can thus be increased to
allow larger values of Cx to be tolerated (Figures 5-1 to 5-2).

As Cx increases, the length of the burst decreases resulting in
lower signal numbers.

The electrode should always be connected to SNS1;
connections to SNS2 are also possible but this can cause the
signal to be susceptible to noise.

It is important to limit the amount of stray Cx capacitance on
both SNS terminals, especially if the Cx load is already large.

WITCHING OPERATION

x

Figure 1-2 Internal Switching

LQ

2 QT310/R1.03 21.09.03

This can be accomplished by minimising trace lengths and
widths.

1.2.2 C

The PCB traces, wiring, and any components associated with
or in contact with SNS1 and SNS2 will become touch
sensitive and should be treated with caution to limit the touch
area to the desired location.

Multiple electrodes can be connected, for example to create a
control button on both sides of an object, however it is
impossible for the sensor to distinguish between the two
electrodes.

ONNECTION TO ELECTRODE

Figure 1-3 Mesh Electrode Geometry

1.3.2 K

Like all capacitance sensors, the QT310 relies on Kirchoff’s
Current Law (Figure 1-4) to detect the change in capacitance
of the electrode. This law as applied to capacitive sensing
requires that the sensor’s field current must complete a loop,
returning back to its source in order for capacitance to be
sensed. Although most designers relate to Kirchoff’s law with
regard to hardwired circuits, it applies equally to capacitive
field flows. By implication it requires that the signal ground
and the target object must both be coupled together in some
manner for a capacitive sensor to operate properly. Note that
there is no need to provide actual hardwired ground
connections; capacitive coupling to ground (Cx1) is always
sufficient, even if the coupling might seem very tenuous. For
example, powering the sensor via an isolated transformer will
provide ample ground coupling, since there is capacitance
between the windings and/or the transformer core, and from
the power wiring itself directly to 'local earth'. Even when
battery powered, just the physical size of the PCB and the
object into which the electronics is embedded will generally
be enough to couple a few picofarads back to local earth.

The implications of Kirchoff’s law can be most visibly
demonstrated by observing the E3B eval board’s sensitivity
change between laying the board on a table versus holding
the board in your hand by it’s batteries. The effect can also be
observed by holding the board by the electrode ‘Sensor1’,
letting it recalibrate, then touching the battery end; the board
will work quite well in this mode.

IRCHOFF’S CURRENT LAW

1.2.3 B

The acquisition process occurs in bursts (Figure 1-7) of
variable length, in accordance with the single-slope CDC
method. The burst length depends on the values of Cs and
Cx. Longer burst lengths result in higher gains and more
sensitivity for a given threshold setting, but consume more
average power and are slower.

Burst mode operation acts to lower average power while
providing a great deal of signal averaging inherent in the CDC
process, making the signal acquisition process more robust.

The QT method is a very low impedance method of sensing
as it loads Cx directly into a very large capacitor (Cs). This
results in very low levels of RF susceptibility.

URST MODE OPERATION

1.3 ELECTRODE DESIGN

1.3.1 E

There is no restriction on the shape of the electrode; in most
cases common sense and a little experimentation can result
in a good electrode design. The QT310 will operate equally
well with a long, thin electrode as with a round or square one;
even random shapes are acceptable. The electrode can also
be a 3-dimensional surface or object. Sensitivity is related to
electrode surface area, orientation with respect to the object
being sensed, object composition, and the ground coupling
quality of both the sensor circuit and the sensed object.
Smaller electrodes have less sensitivity than large ones.

If a relatively large electrode surfaces are desired, and if tests
show that an electrode has a high Cx capacitance that
reduces the sensitivity or prevents proper operation, the
electrode can be made into a mesh (Figure 1-3) which will
have a lower Cx than a solid electrode area.

LECTRODE GEOMETRY AND SIZE

1.3.3 V

When detecting human contact (e.g. a fingertip), grounding of
the person is never required, nor is it necessary to touch an
exposed metal electrode. The human body naturally has
several hundred picofarads of ‘free space’ capacitance to the
local environment (Cx3 in Figure 1-4), which is more than two
orders of magnitude greater than that required to create a
return path to the QT310 via earth. The QT310's PCB
however can be physically quite small, so there may be little
‘free space’ coupling (Cx1 in Figure 1-4) between it and the
environment to complete the return path. If the QT310 circuit
ground cannot be grounded via the supply connections, then
a ‘virtual capacitive ground’ may be required to increase
return coupling.

IRTUAL CAPACITIVE GROUNDS

Figure 1-4 Kirchoff’s Current Law

LQ

3 QT310/R1.03 21.09.03

Figure 1-5 Shielding Against Fringe Fields

A ‘virtual capacitive ground’ can be created by connecting the
QT310’s own circuit ground to:

(1) A nearby piece of metal or metallized housing;
(2) A floating conductive ground plane;
(3) A fastener to a supporting structure;
(4) A larger electronic device (to which its output might be

connected anyway).

Because the QT310 operates at a relatively low frequency,
about 500kHz, even long inductive wiring back to ground will
usually work fine.

Free-floating ground planes such as metal foils should
maximise exposed surface area in a flat plane if possible. A
square of metal foil will have little effect if it is rolled up or
crumpled into a ball. Virtual ground planes are more effective
and can be made smaller if they are physically bonded to
other surfaces, for example a wall or floor.

of the QT310, sensitivity can be high enough (depending on
Cx and Cs) that 'walk-by' signals are a concern; if this is a
problem, then some form of rear shielding may be required.

1.4 SENSITIVITY ADJUSTMENTS

There are three variables which influence sensitivity:

1. Cs (sampling capacitor)

2. Cx (unknown capacitance)

3. Signal threshold value

There is also a sensitivity dependence of the whole device on
Vdd. Cs and Cx effects are covered in Section 1.2.1.

The threshold setting can be adjusted independently from 1 to
255 counts of signal swing (Section 2.3).

Note that sensitivity is also a function of other things like
electrode size, shape, and orientation, the composition and
aspect of the object to be sensed, the thickness and
composition of any overlaying panel material, and the degree
of mutual coupling of the sensor circuit and the object (usually
via the local environment, or an actual galvanic connection).

Threshold levels of less than 5 counts in BG mode are not
advised; if this is the case, raise Cs so that the threshold can
also be increased.

1.4.1 I

NCREASING SENSITIVITY

In some cases it may be desirable to greatly increase
sensitivity, for example when using the sensor with very thick
panels having a low dielectric constant, or when sensing low
capacitance objects.

Sensitivity can be increased by using a bigger electrode,
reducing panel thickness, or altering panel composition.
Increasing electrode size can have diminishing returns, as
high values of Cx load will also reduce sensor gain (Figures
5-1 and 5-2). The value of Cs also has a dramatic effect on
sensitivity, and this can be increased in value up to a limit.

Increasing electrode surface area will not substantially
increase sensitivity if its area is already larger than the object
to be detected. The panel or other intervening material can be
made thinner, but again there are diminishing rewards for

1.3.4 F

The electrode can be prevented from sensing in undesired
directions with the assistance of metal shielding connected to
circuit ground (Figure 1-5). For example, on flat surfaces, the
field can spread laterally and create a larger touch area than
desired. To stop field spreading, it is only necessary to
surround the touch electrode on all sides with a ring of metal
connected to circuit ground; the ring can be on the same or
opposite side from the electrode. The ring will kill field
spreading from that point outwards.

If one side of the panel to which the electrode is fixed has
moving traffic near it, these objects can cause inadvertent
detections. This is called ‘walk-by’ and is caused by the fact
that the fields radiate from either surface of the electrode
equally well. Again, shielding in the form of a metal sheet or
foil connected to circuit ground will prevent walk-by; putting a
small air gap between the grounded shield and the electrode
will keep the value of Cx lower and is encouraged. In the case

IELD SHAPING

LQ

Figure 1-6 Burst Detail

4 QT310/R1.03 21.09.03

Figure 1-7 Burst when SC is set to 1

(Observed using a 750K resistor in series with probe)

Figure 1-8 Burst when SC is set to 0 (no sleep cycles)

(Observed using a 750K resistor in series with probe)

doing so. Panel material can also be changed to one having a
higher dielectric constant, which will help propagate the field.
Locally adding some conductive material to the panel
(conductive materials essentially have an infinite dielectric
constant) will also help; for example, adding carbon or metal
fibers to a plastic panel will greatly increase frontal field
strength, even if the fiber density is too low to make the
plastic electrically conductive.

1.4.2 D

In some cases the circuit may be too sensitive, even with high
signal threshold values. In this case gain can be lowered by
making the electrode smaller, using sparse mesh with a high
space-to-conductor ratio (Figure 1-3), and most importantly by
decreasing Cs. Adding Cx capacitance will also decrease
sensitivity.

It is also possible to reduce sensitivity by making a capacitive
divider with Cx by adding a low-value capacitor in series with
the electrode wire.

ECREASING SENSITIVITY

1.5 TIMING

Figure 1-7 and 1-8 shows the basic timing parameters of the
QT310. The basic QT310 timing parameters are:

Tbd Burst duration (1.5.1)
Tbs Burst spacing (1.5.2)
Tsc Sleep Cycle duration (1.5.2)
Tmod Max On-Duration (1.5.3)
Tdet Detection response time (1.5.4)

1.5.1 B

The burst duration depends on the values of Cs and Cx, and
to a lesser extend, Vdd. The burst is normally composed of
hundreds of charge-transfer cycles (Figure 1-6) operating at
about 240kHz. This frequency varies by about ±7% during the
burst in a spread-spectrum modulation pattern. See Section

3.5.2 page 13 for more information on spread-spectrum.

URST FREQUENCY AND DURATION

The number of pulses in a burst and hence its duration
increases with Cs and decreases with Cx.

1.5.2 B

Between acquisition bursts, the device can go into a low
power sleep mode. The duration of this is a multiple of Tsc,
the basic sleep cycle time. Tsc depends heavily on Vdd as
shown in Figure 5-4, page 16. The parameter SC calls out
how many of these cycles are used. More SC means lower
power but also slower response time.

Tbs is the spacing from the start of one burst to the start of
the next. This timing depends on the burst length Tbd and the
dead time between bursts, i.e. Tsc.

The resulting timing of Tbs is:

If SC = 0, the device never sleeps between bursts (example:
Figure 1-8). In this case the value of Tsc is fixed at about

2.25ms, but this time is not spent in Sleep mode and maximal
power is consumed.

if SC >> 0 (example: SC=15), the device will spend most of its
time in sleep mode and will consume very little power, but it
will be much slower to respond.

By selecting a supply voltage and a value for SC, it is possible
to fine-tune the circuit for the desired speed / power trade-off.

1.5.3 MAX ON-D

The Max On-Duration is the amount of time required for
sensor to recalibrate itself when continuously detecting. This
parameter is user-settable by changing MOD and SC (see
Section 2.6).

Tmod restarts if the sensor becomes inactive before the end
of the Max On Duration period.

URST SPACING

Tbs = Tbd + (SC x Tsc) where SC > 0

-or­Tbs = Tbd + 2.25ms where SC = 0

: TBS, T

URATION

, T

SC

MOD

LQ

5 QT310/R1.03 21.09.03

1.5.4 R

Response time Tdet from the onset of detection to the OUT
pin becoming active depends on:

If the control bit DIS is normal (0), then Tdet depends on the
rate at which the bursts are acquiring, and the value of DIT. A
DIT number of bursts must confirm the detection before the
OUT line becomes active:

If DIS is set to ‘fast’, then Tdet is computed as:

Quantum’s QT3View software calculates an estimate of
response time based on this formula.

ESPONSE TIME

Tbs Burst spacing (Section 1.5.2)
DIT Detection Integrator Target (user setting)
DIS Detect Integration Speed (user setting)
Tbd Burst duration (if DIS is set to ‘fast’)

Tdet = Tbs x DIT (normal DIS)

Tdet = (SC x Tsc) + (DIT x (Tbd + 2.25ms)) (fast DIS)

, T

DET

1.6 EXTERNAL RECALIBRATION

The /CAL_CLR pin can be used to recalibrate the sensor on
demand. A low pulse of at least Tbs (burst spacing) duration
is require to initiate a recalibration. The calibration occurs just
after /CAL_CLR returns high.

In BG1 mode (Section 2.8.4), the calibration data is not stored
in EEPROM, and the part will recalibrate after each power up.

In BG1 mode, if the device has been set for Toggle Latch
output mode, the /CAL_CLR pin becomes an output reset
control and the part cannot be recalibrated via /CAL_CLR.
However the part can be recalibrated by powering it down and
back up again (Section 2.7.3).

In BG2 mode, the calibration data is stored in EEPROM, and
the part will not recalibrate after power up, using instead the
stored calibration data. The internal eeprom has a life
expectancy of 100,000 erase/write cycles.

In OBJ mode, the part stores the calibration data into
EEPROM and the part will not recalibrate after power up,
using instead the stored calibration data.

In both BG2 and OBJ mode, the device must be calibrated
using the /CAL_CLR input, or the calibration data can be set
via cloning process, otherwise the calibration data will be
invalid.

are crowded together with a rep rate that depends entirely on
the burst lengths (Section 1.5.1).

Response time, drift compensation rate, max on-duration, and
power consumption are all affected by this parameter. A high
value of SC will allow the device to consume very low power
but it will also be very slow.

2.2 DRIFT COMPENSATION (PDC, NDC)

Signal drift can occur because of changes in Cx, Cs, Vdd,
electrode contamination and ageing effects. It is important to
compensate for drift, otherwise false detections and sensitivity
shifts can occur.

Drift compensation is performed by making the signal’s
reference level slowly track the raw signal while no detection
is in effect. The rate of adjustment must be performed slowly,
otherwise legitimate detections could be affected. The device
compensates using a slew-rate limited change to the signal
reference level; the threshold and hysteresis points are slaved
to this reference.

Once an object is detected, drift compensation stops since a
legitimate signal should not cause the reference to change.

Positive and negative drift compensation rates (PDC, NDC)
can be set to different values (Figure 2-1). This is invaluable
for permitting a more rapid reference recovery after the device
has recalibrated while an object was present and then
removed.

Positive drift occurs when the Cx slowly increases. Negative
drift occurs when Cx slowly decreases (see Section 2.8.1).

PDC+1 sets the number of burst spacings, Tbs, that
determines the interval of drift compensation, where:

Tbs = Tbd + (SC x Tsc)

where SC > 0 (Section 1.5.2)

-or-

Tbs = Tbd + 2.25ms

where SC = 0 (Section 1.5.2)

Example: PDC = 9, (user setting)

Tbs = 100ms

then

Tpdc = (9+1) x 100ms = 1 sec

NDC operates in exactly the same way as PDC.

2 - Control & Processing

All acquisition functions are digitally controlled and
can be altered via the cloning process.

Signals are processed using 16 bit integers, using
Quantum-pioneered algorithms specifically
designed to provide for high survivability.

2.1 SLEEP CYCLES (SC)

Range: 0..255; Default: 1
Affects speed & power of entire device.

Refer to Section 1.5.2 for more information on the
effect of Sleep Cycles.

SC changes the number of intervals Tsc
separating two consecutive burst (Figure 1-7 and
1-8). SC = 0 disables sleep intervals and bursts

LQ

Figure 2-1 Drift Compensation

6 QT310/R1.03 21.09.03