Copyright 2010 ATMEL CONFIDENTIAL1QMatrixV1.00 Sept 2010Copyright 2010 ATMEL CONFIDENTIAL TopicsQMatrix BasicsGoalsEquivalent CircuitImportant TermsCharge Transfer BasicsSwitching SequenceMeasuring TechniqueCharge FlowHistorical Switching SequenceWhat Happens during Touch?Dual Slope Measurement VddDual Slope Measurement - CsDetection algorithm and ProcessingSensitivityCharge Transfer Waveform DetailIncomplete ChargingIO Cell and MCU CharacteristicsX-Y Line CouplingElectrode DesignCs does NOT change sensitivityIncreasing QMatrix SensitivitySampling Resistor (Rsmp)Internal OscillatorBurst Length (BL)SaturationEMCPower, Energy, PerformanceMultiple ChannelsKey MatrixMatrix Design BasicsScan SequenceX-line ConsiderationsY-line ConsiderationsFalse KeysFactors Influencing SensingMain FactorsDifferential DriftCommon Mode DriftNoise ProblemsNoise SolutionsQMatrix vs. QTouchAdvantages of QMatrix over QTouchSummaryQuick Chart2Copyright 2010 ATMEL CONFIDENTIALQMatrix BasicsCopyright 2010 ATMEL CONFIDENTIAL QMatrix BasicsGoalsEquivalent CircuitImportant TermsCharge Transfer BasicsSwitching SequenceMeasuring TechniqueCharge FlowHistorical Switching SequenceWhat Happens during Touch?Dual Slope Measurement VddDual Slope Measurement - Cs4Copyright 2010 ATMEL CONFIDENTIAL GoalsTo detect finger presence near electrodeTo get more keys per chip than QTouchTo get a more efficient ratio of keys to MCU IO pins than QTouchElectrode to be behind a dielectric panel so no direct galvanic connection to measuring circuitElectrode arrayMeasuring Circuit5Copyright 2010 ATMEL CONFIDENTIAL Equivalent CircuitGND is the local circuit returnEARTH is the “free space” returnFor simplification assume Cx Cp2, Cp1, and Cf Cx, CtWe are interested in Cx and Ct primarilyCp1: Parasitic IO pin capacitance to GND (1-2pF ?)Cp2: Wiring capacitance to GND (few pF)Cx: Electrode self coupling capacitance (2-10pF ?)Ct: Touch capacitance to earth (few pF)Cf: Coupling capacitance between circuit GND and earth (few pF)GNDGNDGNDEARTHVddCp1Cp2CfCtEARTHMCUCx6Copyright 2010 ATMEL CONFIDENTIAL Important TermsCx Self coupling capacitance of electrodeCs Sampling capacitorCt Touch capacitance to EarthX-Line The transmitter line used for charge transferY-Line The receiver line used for charge transferRsmp Sampling Resistor used to act like a current source after the burst phase of the acquisitionVcs The voltage accumulated during the burst phase of the acquisition7Copyright 2010 ATMEL CONFIDENTIAL Charge Transfer BasicsCs Cx : example Cs = 10nF, Cx = 10pFS1,2,3,4 are simple CMOS IO pins, switching between GND, Z and VddBy controlling S1,2,3,4 we can transfer charge into Cs through CxBy repeating this many times we can measure* Cx*normally just a relative measurement is important i.e. changes in CxCsGNDGNDVDDS1S2S3Sampling capacitorMCUCxGNDS4XYkY8Copyright 2010 ATMEL CONFIDENTIAL Switching Sequence#1 provides initial condition at start of acquisition (needs time)#2 & #3 rising edge on X pin pre-charges the X track and stray capacitance#4 drives (“pulls”) charge through Cx into Cs #5 & #6 Isolate charge on Cs#7 discharges Cx ready for next transferNote : break before make on I/Os can remove need for 2nd float state !#S1S2S3S4NOTES1CLOSEDOPENCLOSEDCLOSEDCx and Cs discharge2OPENOPENOPENCLOSEDFloat state3OPENCLOSEDOPENOPENPre-charge X-line4OPENCLOSEDCLOSEDOPENCharge transfer5OPENCLOSEDOPENOPENFloat state6CLOSEDCLOSEDOPENOPENIsolate Cs charge7CLOSEDOPENOPENCLOSEDDischarge CxrepeatCsGNDGNDVDDS1S2S3CxGNDS4XYkY9Copyright 2010 ATMEL CONFIDENTIAL Burst Switch State 1Provides initial condition at start of acquisition (also idle state)CsGNDGNDVDDS1S2S3CxGNDS4XYkY#S1S2S3S4NOTES1CLOSEDOPENCLOSEDCLOSEDCx and Cs discharge2OPENOPENOPENCLOSEDFloat state3OPENCLOSEDOPENOPENPre-charge X-line4OPENCLOSEDCLOSEDOPENCharge transfer5OPENCLOSEDOPENOPENFloat state6CLOSEDCLOSEDOPENOPENIsolate Cs charge7CLOSEDOPENOPENCLOSEDDischarge CxrepeatGNDCxGNDGNDCsGND10Copyright 2010 ATMEL CONFIDENTIAL QMatrix Basics: Burst Switch State 2,3Rising edge on X pin pre-charges the X track and all the stray capacitance on this line no charge transfers to Cs yetState #2 is present for the repeat loop from state #7 avoids simultaneous change of S1 and S2 that *could* cause charge to enter Cx#S1S2S3S4NOTES1CLOSEDOPENCLOSEDCLOSEDCx and Cs discharge2OPENOPENOPENCLOSEDFloat state3OPENCLOSEDOPENOPENPre-charge X-line4OPENCLOSEDCLOSEDOPENCharge transfer5OPENCLOSEDOPENOPENFloat state6CLOSEDCLOSEDOPENOPENIsolate Cs charge7CLOSEDOPENOPENCLOSEDDischarge CxrepeatCsGNDGNDVDDS1S2S3CxGNDS4XYkYCxCsVDDCxCsGND11Copyright 2010 ATMEL CONFIDENTIAL QMatrix Basics: Burst Switch State 4Bottom end of Cs is pulled down causing charge to flow through Cx into Cs this is the charge transfer stateThe length of this state is normally programmable and is known as the “Dwell Time”Longer dwell times can be needed if there is heavy stray capacitance on X and/or YCsGNDGNDVDDS1S2S3CxGNDS4XYkY#S1S2S3S4NOTES1CLOSEDOPENCLOSEDCLOSEDCx and Cs discharge2OPENOPENOPENCLOSEDFloat state3OPENCLOSEDOPENOPENPre-charge X-line4OPENCLOSEDCLOSEDOPENCharge transfer5OPENCLOSEDOPENOPENFloat state6CLOSEDCLOSEDOPENOPENIsolate Cs charge7CLOSEDOPENOPENCLOSEDDischarge CxrepeatCxCsVDDGNDQcs = QcxVCs1CxCsVDDGND12Copyright 2010 ATMEL CONFIDENTIAL QMatrix Basics: Burst Switch State 5The float state avoids simultaneous switching of S1 and S3 which cause charge loss in Cs#S1S2S3S4NOTES1CLOSEDOPENCLOSEDCLOSEDCx and Cs discharge2OPENOPENOPENCLOSEDFloat state3OPENCLOSEDOPENOPENPre-charge X-line4OPENCLOSEDCLOSEDOPENCharge transfer5OPENCLOSEDOPENOPENFloat state6CLOSEDCLOSEDOPENOPENIsolate Cs charge7CLOSEDOPENOPENCLOSEDDischarge CxrepeatCsGNDGNDVDDS1S2S3CxGNDS4XYkYVCs1CxCsVDD13Copyright 2010 ATMEL CONFIDENTIAL QMatrix Basics: Burst Switch State 6Isolate charge on Cs this charge is the all-important answer (or it will be !)CsGNDGNDVDDS1S2S3CxGNDS4XYkY#S1S2S3S4NOTES1CLOSEDOPENCLOSEDCLOSEDCx and Cs discharge2OPENOPENOPENCLOSEDFloat state3OPENCLOSEDOPENOPENPre-charge X-line4OPENCLOSEDCLOSEDOPENCharge transfer5OPENCLOSEDOPENOPENFloat state6CLOSEDCLOSEDOPENOPENIsolate Cs charge7CLOSEDOPENOPENCLOSEDDischarge CxrepeatCxCsVCs1VDDGND14Copyright 2010 ATMEL CONFIDENTIAL QMatrix Basics: Burst Switch State 7Discharges Cx ready for next transferThe Burst Length (number of charge pulses) in QMatrix is fixed by userSo the Charge Transfer loop above will repeat till a user specified value (e.g. 16)After that, the loop will exit and the Measurement Phase will beginCsGNDGNDVDDS1S2S3CxGNDS4XYkY#S1S2S3S4NOTES1CLOSEDOPENCLOSEDCLOSEDCx and Cs discharge2OPENOPENOPENCLOSEDFloat state3OPENCLOSEDOPENOPENPre-charge X-line4OPENCLOSEDCLOSEDOPENCharge transfer5OPENCLOSEDOPENOPENFloat state6CLOSEDCLOSEDOPENOPENIsolate Cs charge7CLOSEDOPENOPENCLOSEDDischarge CxrepeatCxCsVCs1GNDGND15Copyright 2010 ATMEL CONFIDENTIAL Measuring Technique - Measure PhaseMeasurement phase uses a resistor Rsmp, a comparator and a counter to measure the magnitude of Vcs that resulted from the burst (note Vcs is ve looking from Y pin)After burst; counter is started, S6 is opened and S5 is closed causing Vcs to ramp up towards Vdda capture register latches counter as Vcs crosses GNDThe duration of the counter is our SignalCsGNDGNDVDDS1S2S3MCUCxGNDS4XYkYRsmpSMPCAPTURECTRCLK+-VDDS5GNDS6-VCs1-VCs2-VCsn RsmpCxCsGNDGNDVDD16Copyright 2010 ATMEL CONFIDENTIAL Charge flowVCs = VCs1 + VCs2 + . + VCsn ; n is the Burst LengthGNDCxGNDGNDCsGND#1#1#7#7CxCsVCs1GNDGNDVCs2VCsn#2#2CxCsVDD#3#3CxCsGND#4#4CxCsVDDGND#5#5VCs1CxCsVDDCxCsVCs1VDDGND#6#6CxCsVCs1GNDGND#7#7VCs1#2#2#3#3#4#4CxCs#5#5VDDCxCsVCs1VDDGND#6#6VCs2CxCsVCs1GNDGND#7#7VCs2VCs2VCs1CxCsVDDVCs1CxCsVDDGNDVCs1CxCsGND17Copyright 2010 ATMEL CONFIDENTIAL Typical WaveformsVoltage across Cs builds up for every transfer cycle we performTypical voltage magnitude across Vcs increases in discrete stepsMeasuring between GND and Y/Yk we see the burst of pulsesDischarge time “t” to go from the terminal voltage to GND is measured with the counter and comparatorThe counter value is our measurement (Signal)VykTime0.25V|Vcs|TimeApproximately linear0.25VVsnsTime-0.25VtTerminal voltageCsGNDGNDVDDS1S2S3CxGNDS4XYkYVcs18Copyright 2010 ATMEL CONFIDENTIAL Historical Switching SequenceMany QRG QMatrix design have a slightly different switching schemethe bottom of Cs is clamped to GND first and *then* the X pin is driven highpushing charge through Cx and CsThis works OK, but with three small negatives:1.There is a short period where S2 and S3 are closedcausing some charge flow that was not intended (very small effect)2.The X line and its associated stray capacitance doesnt get pre-charged so it can need a longer Dwell Time to get full charge transfer3. It is harder to minimise the Dwell Timesome applications benefit greatly with noise immunity by making the Dwell Time as small as possible19Copyright 2010 ATMEL CONFIDENTIAL What Happens During Touch ?Now we bring a finger close to electrodeThis diverts charge from Cx, and so away from CsEach charge pulse (#4) now charges Cs by a smaller amount If each pulse deposits less charge, then Vcs rises slower Vcs reaches a smaller terminal voltage needs less time to discharge to GND so counter value gets lessThe Signal count change (delta) is proportional to CtE-fieldXYEARTHCtVcsTimeNo touchDeltaTimeTouchVcs20Copyright 2010 ATMEL CONFIDENTIAL Dual Slope Measurement - VddTimeSignal is still the sameTimeNormal VddVcsIncreased VddVcsThe QMatrix acquisition process uses a dual slope method, so a self-nulling benefit is seen (unlike QTouch)Effects of Vdd change:Increasing Vdd More charge is transfered each pulse Vcs rises faster Vcss terminal voltage is higher BUT Vcs is discharged through GPIO The GPIO is at Vdd So discharge slope is steeper as wellThe discharge duration is the SAMECsGNDGNDVDDS1S2S3MCUCxGNDS4XYkYRsmpSMPCAPTURECTRCLK+-VDDS5GNDS6QMatrix measurement is immune to Vdd changes21Copyright 2010 ATMEL CONFIDENTIAL Dual Slope Measurement - CsTimeSignal is still the sameTimeNormal CsVcsIncreased CsVcsEffects of Cs change:Increasing Cs less charge is transfered each pulse Vcs rises slower Vcss terminal voltage is lower BUT when Vcs is discharged RC constant is higher So discharge slower as wellThe discharge duration is the SAMEQMatrix measurement is immune to Cs changes22Copyright 2010 ATMEL CONFIDENTIALDetection Algorithm and ProcessingCopyright 2010 ATMEL CONFIDENTIAL Detection AlgorithmThe detection process is the same as QTouch, refer to QTouch slides for detailsCalibration Positive Error RecalibrationDetect Threshold (NTHR)Detect Hysteresis (HYST)Detection Integration (DI)Max On DurationAdjacent Key Suppression (AKS)24Copyright 2010 ATMEL CONFIDENTIALSensitivityCopyright 2010 ATMEL CONFIDENTIAL SensitivityCharge Transfer Waveform DetailIncomplete ChargingIO Cell and MCU CharacteristicsX-Y Line CouplingElectrode DesignCs does NOT change sensitivityIncreasing QMatrix SensitivitySampling Resistor (Rsmp)Internal OscillatorBurst Length (BL)Saturation26Copyright 2010 ATMEL CONFIDENTIAL Charge Transfer Waveform DetailTypical pulse time is 250ns to 2s (incl float)Typical time after pulse is 1 to 2sTypical overall cycle is 5 to 10s i.e. 200 to 100KHzTypical burst lengths range from 16 to 64 pulses i.e. 160 to 640sDischarge time is a function of Cs, Rsmp, key design etc, but is typically high 10s of s and uses a 4MHz counterThe terminal voltage is typically -75 mV to 150 mVDo not exceed -250mV160us to 640us typVcsTimeNo touch10s of us typ27Copyright 2010 ATMEL CONFIDENTIAL Incomplete ChargingThere is a RC time constant affecting electrode charging, due to capacitance (Cx, Ct, Cp) and series resistance (Rs or resistivity of material, e.g. ITO)Incomplete charging can cause erratic measurements, prevent by:Reducing capacitive and resistive loadingIncreasing charge pulse durationAlways check charge pulse shape with oscilloscope and coin method (explained in Touch Sensor Design Guide)WRONGIncomplete chargingIncomplete Charging = BAD!CORRECTComplete Charging = GOOD!Scope and Coin method28Copyright 2010 ATMEL CONFIDENTIAL IO Cell and MCU CharacteristicsY linesQMatrix works best for simple CMOS IO cells with low CioThe SNS side of the Y line is connected to a comparator with timer capture unitIO pin capacitance has a detrimental effect on Y and Yk so low capacitance on pins is goodSelect IOs with low stray C (osc/xtal pins are rarely any good !). X linesThere are virtually no other IO requirementsX lines are just simple CMOS outputs (they never float)Comparator Comparator input must be able to sense near GND without any phase inversion (or other odd behaviour !)29Copyright 2010 ATMEL CONFIDENTIAL X-Y Line CouplingQMatrix key design plays a large part in good sensitivityThe sensitivity is dominated by the XY coupling (gap and edges)Edge effect rather than surface area effect (like QTouch)Typical Key DesignsBasic Key DesignsXYXXYXX30Copyright 2010 ATMEL CONFIDENTIAL Electrode DesignGeneral guidelines:Keys should be bigger than 5x5mmIncrease interleaving between X and Y to increase sensitivityT/2 rule - XY gap should be nominally half the front panel thickness (T)Have Y lines as thin as possible to reduce noise pickupSurround the Y with X to contain the field For 1-layer designs its ok to split the X “ring” and make 2 connections to itThe Y electrode is sensitive like a QTouch electrode so similar layout rules applyT/2XYBAD!GOOD!31Copyright 2010 ATMEL CONFIDENTIAL Cs does NOT change sensitivityIncreasing Cs does NOT have the same effect as with QTouchIncreasing Cs, the terminal voltage gets smaller BUT the slope gets shallower, so net change is zeroSee Dual Slope Measurement slideIncreasing Cs is not the way to improve sensitivity for QMatrixCsGNDGNDVDDS1S2S3Increasing Cs does NOT improve QMatrix sensitivityMCUCxGNDS4XYkY32Copyright 2010 ATMEL CONFIDENTIAL Increasing QMatrix SensitivityThe sensitivity of a QMatrix sensor can be tuned in Hardware and SoftwareHardware parametersRsmp (Sampling Resistor)Internal Oscillator Software parametersBL (Burst Length)NTHR (Negative Threshold)33Copyright 2010 ATMEL CONFIDENTIAL Sampling Resistor (Rsmp)Increasing Rsmp, increases sensitivity and vice versadischarge slope shallower, takes longer giving higher numerical resolutionRsmp cannot be increased too farThe slope angle gets too lowDegrades the SNR of the measurementMeasurement time is made unnecessarily longRsmp rangeTypical 220 K 1 M 470 K is commonTimeSignal Resolution increasedTimeNormalVcsIncrease RsmpVcsCsGNDGNDVDDS1S2S3MCUCxGNDS4XYkYRsmpSMPCAPTURECTRCLK+-VDDS5GNDS634Copyright 2010 ATMEL CONFIDENTIAL Internal OscillatorIncreasing Internal Oscillator speed, increases sensitivityincreases resolution of Timer used to measure discharge durationThus sensitivity is increasedTypical QMatrix sensors utilises 4MHz Oscillator speedCan be increased to get better sensitivityDependant on AVR device usedWhen changing the Internal Oscillator, check that the Charge time is still sufficient to fully charge electrodeThe delay in most touch devices are implemented in terms of instructions35Copyright 2010 ATMEL CONFIDENTIAL Burst Length (BL)Increase the Burst Length (BL) to increase sensitivitygive higher terminal voltage thus longer discharge timehigher numerical resolution and SNRThe burst length cant be increased indefinitely.Vcs is negative and must be kept above approx 0.25V to avoid I/O diode conductionDiode conduction can cause saturation and other undesired effectsIf saturation is observed reduce burst length ORincrease Cs-0.25VTimeSignal Resolution increasedTimeNormalVcsIncrease BLVcs36Copyright 2010 ATMEL CONFIDENTIAL SaturationSaturation means the Cs charge fully and cannot receive any more chargeSaturation will reduce the sensitivity of the sensor significantlySaturation can be caused byBurst Length is too longCs is too smallY line is excessively loaded (by X line or Ground)Probing the Cs with a scope can give an indication of saturationIf not saturating, the charge accumulation curve should be almost linear37Copyright 2010 ATMEL CONFIDENTIALEMCCopyright 2010 ATMEL CONFIDENTIAL EMCSpread SpectrumSeries Resistors39Copyright 2010 ATMEL CONFIDENTIAL Spread SpectrumThe same techniques used for QTouch are also used for QMatrixRefer to the QTouch slides for details40Copyright 2010 ATMEL CONFIDENTIAL Series Resistors (Rx, Ry)The same techniques used for QTouch are also used for QMatrixRefer to the QTouch slides for detailsRx series resistor for X linesUsed mostly for reducing emissionsRy series resistor for Y linesUsed mostly for improving immunity41Copyright 2010 ATMEL CONFIDENTIALPower, Energy and PerformanceCopyright 2010 ATMEL CONFIDENTIAL Power, Energy, PerformanceThe same considerations for QTouch are also applicable for QMatrix. Refer to the QTouch slides for detailsQMatrix uses shorter burst than QTouchso for small key counts it can be much more energy efficientas QMatrix can complete the measurements in shorter timeIt is also possible to burst into multiple Y lines in parallelso greatly reducing the acquisition time and energy requirementsaving is only true for more than one Y line of course43Copyright 2010 ATMEL CONFIDENTIALMultiple ChannelsCopyright 2010 ATMEL CONFIDENTIAL Multiple ChannelsKey MatrixMatrix Design BasicsScan SequenceX-line ConsiderationsY-line ConsiderationsFalse Keys45Copyright 2010 ATMEL CONFIDENTIAL Key MatrixThe discussion so far has focused on a single channelThe goal of QMatrix is to sense more channelsThe diagram below shows an example matrix of 8 keysmany other configurations are possibleQMatrix, 8 channelsRequires 4X + 4Y = 8 pins Qtouch, 8 channelsRequires ( 2 * 8 ) = 16 pinsQtouch requires 2 pins per channel QMatrix uses less pins for a given number of channelsY0Y1X0X1X2X346Copyright 2010 ATMEL CONFIDENTIAL Matrix Design BasicsThere is always a trade-off in deciding the X-Y matrix organisation to get the desired number of keysY-lines need two pinsso are less efficientY-lines are receivers so must be routed with care (like QTouch electrode traces)Generally 1, 2 or 3 Y-lines are valuableX lines require only one pinX-lines are insensitive to touch/noise so they can be routed pretty much anywhere47Copyright 2010 ATMEL CONFIDENTIAL Scan SequenceThe traditional scanning sequence for QMatrix:grounds all X and Y lines (idle state)In turn, each X line is pulsedjust one Y line is activated to collect chargeSo scan sequence becomes X0Y0, X1Y0, X2Y0.X0Y1, X1Y1, X2Y1etcTotal scan time is a function of number of keys48Copyright 2010 ATMEL CONFIDENTIAL Multiple Channels: False KeysA QMatrix key is formed when a X and Y comes close togetherWhen X and Y lines run close to each other, a potential key is formedThis could look like a “false” keySolutions:Separate X and Y lines as far as possible, e.g. on different layer and different sidesRunning a GND line between X and Y lines solves the problemRunning Y-line behind X-line/Ground on non-touch side (X-line/Ground acts like a shield)Where X and Y lines must be crossed, doing so at 90o reduces the capacitive coupling between them so is advantageous49Copyright 2010 ATMEL CONFIDENTIALFactors Influencing SensingCopyright 2010 ATMEL CONFIDENTIAL Factors Influencing SensingMain FactorsDifferential DriftCommon Mode DriftNoise ProblemsNoise Solutions51Copyright 2010 ATMEL CONFIDENTIAL Main FactorsAs explained before, QMatrix uses a dual slope conversion, therefore it is virtually insensitive to changes in Vdd and CsThis means that basic acquisition results hardly vary with temperatureQMatrix is also less sensitive to moisture films, because;1.the electric fields are very “closed” and localised to the key area2.the QMatrix key is not effected by a floating conductor between X and Y3.the narrow* charge pulses cannot “see” through the water film to GND (water films look like a distributed low pass filter)*For appliance use it is beneficial to reduce the high charge pulse time for this reason. Figures less than a few hundred nanoseconds can be very beneficialbut see notes about pulses being too narrow !52Copyright 2010 ATMEL CONFIDENTIAL Differential Drift MethodSame as QTouch but less important because QMatrix is inherently more stableBecause QMatrix is so stable, the drift rates can be set to be very slow, mitigating most of the risks. For appliance use this may not be so appropriate (sudden condensation for example) but for most mobile applications it is the case.53Copyright 2010 ATMEL CONFIDENTIAL Common Mode DriftQMatrix does not need or use common mode driftQMatrix is largely immune to Vdd and Cs variations54Copyright 2010 ATMEL CONFIDENTIAL Noise ProblemsQMatrixuses relatively short bursts typically (typically 16 64 pulses for keys)accumulates small amounts of terminal voltage (75 mV 150 mV)Therefore, QMatrix is inherently more susceptible to noisecoupled into sensors at a frequency that is quite close to the sample frequency orat some low frequency like mains noise (50/60Hz) There is little inherent averaging effect across the charge transfer pulsesunlike QTouch that uses much longer bursts (200 pulses typically for keys)result can show up as noisy signals55Copyright 2010 ATMEL CONFIDENTIAL Noise SolutionsIncreasing the Burst Length can helpMore inherent averagingBut at expense of reduced response timeSynchronising the burst start to the noise can helpe.g. mains 50/60Hzusing a zero crossing detector or large resistor to an IO pinSynchronization is used to overcome noisy appliance, e.g. induction hobsUsing self shielding electrode designsThis works on the principle that the X electrodes are virtually immune to noise/loadingSo the electrode can be designed so that the X shields the Y from the noise sourceFor example, the 2 layer flooded-X designs can shield from noise coming from underneath the sensorAdditional filtering can be appliedDue to the fast measurement timesMore measurements can be made and more filters implemented56Copyright 2010 ATMEL CONFIDENTIALQMatrix vs QTouchCopyright 2010 ATMEL CONFIDENTIAL QMatrix vs. QTouchAdvantages of QMatrix over QTouchQMatrix vs. QTouch SummaryQMatrix vs. QTouch Quick Chart58Copyright 2010 ATMEL CONFIDENTIAL Advantages Over QTouchThe acquisition process uses a dual slope method, so a self nulling benefit is seen (unlike QTouch):1.If Vdd increases, Vcs increases more per-pulse, but then the slope rate increases by the same factor. Opposite is also obviously true for decreases in Vdd2.If Cs decreases then Vcs increases more per-pulse, but then the slope rate increases by the same factor. Opposite is also obviously true for decreases in CsConversion times are generally much shorter than QTouch Faster scan times and lower energy consumptionE-field is “closed” between X and Y making the active touch area very easy to control and insensitive to nearby circuitsMore efficient pin usage allows much larger key countsOnly one X line is active at once, the rest are grounded so noise immunity is improved59Copyright 2010 ATMEL CONFIDENTIAL QMatrix vs QTouch - SummaryQTouch: plusSimple electrode designSmaller code sizeFewer chip resources usedGood for ProximityQTouch: minusAbove a few channels, needs more pins-2 pins per channel-EVK2080A: 16 pins for 8 channelsLonger burst times-Higher power consumptionTiming more variable-Burst Length dependant design, environment and touch stateQMatrix: plusAbove a few channels, needs fewer pins-X by Y needs (X + 2Y + 1) pins-EVK2080B: 4 x 2, 9 pins for 8 channelsBetter moisture toleranceShorter bursts-Lower power consumptionPredictable timingTouch-sensitive area more localised-Less sensitive to nearby circuitryLocalised field more defined key areaQMatrix: minusMore complex electrode designLarger code sizeMore chip resources used-Timer counter, comparatorCan be more noise-sensitive-Short burst duration-Tiny accumulated VcsLocalised fields not ideal for proximity60Copyright 2010 ATMEL CONFIDENTIAL Quick ChartQuick Comparison Table: Scores out of 1061Copyright 2010 ATMEL CONFIDENTIALSummaryCopyright 2010 ATMEL CONFIDENTIAL SummaryCx is measured with a burst of pulses driven through Cx into a sampling capacitor Cs, typically a few nFCx is a deliberately formed capacitance between row and column electrodesThe burst is a fixed length often 16 to 64 pulsesThe terminal voltage on Cs is measured (a ve voltage)A dual slope method is used for the measurement making the technique highly stableWhole process is driven by firmware, uses a comparator and a counter/capture circuitHigh numbers of keys can be scanned very quickly63