`
`
`=FCYPRESS
`
`PERFOR A
`
`CYPRESS SEMICONDUCTOR CORPORATION
`Internal Correspondence
`
`Date: 3/2/2010
`Ta:
`Jeff Dahlin (JVY)
`Author: Michael Hills (HFM)
`Author File#: HFM#061
`TMA200 Water Rejection Baselining Using SelfCap
`(draft)
`
`Ww: 1009
`
`Subject:
`
`Category:
`Distribution: EBX, TATE, ELG, IVY
`
`
`
`Summary:
`Waiter rejection was implemented for the Lenovo Rocket and LG Miniproject on
`TMA300 by adding a self-cap scan to help manage baseline updating.
`Maintaining a good baseline is the first problem with water on a panel using a
`mutual cap scan. Depending on the panel and finger thresholds, this alone can
`in some cases provide water rejection and/or wetfinger tracking, but other panel
`designs can have very poor wet finger tracking and in some cases even false
`touches from water on the panel even if the baseline algorithm is robust. The
`primary goal of the methods discussed in this meme js to provide a robust
`baseline so that when thewater is removed from the panel it wil resume normal
`operation without having to reset the PSoC.
`
`Problem Background:
`There are a coupledistinct problems with water on a panel that is scanned with
`mutual cap.
`
`Thefirst problem is that water creates a negative signal on a mutual cap scan.
`See Figure (Need to add figure}. That makesit difficult to create a robust
`baseline update algorithm.
`If the panelis initialized with wateror a finger on the
`panel, there must be the possibility fo reset all the baselines oncethis is
`removed or moved. The baseline must be able to update overtime to
`accommodate shifts from temperature, but if water accumulates or goes away
`slowly due to condensation or evaporation, the baselines should not be updated.
`Using mutual cap scan it is practically impossible to distinguish between the
`situation where a PSoC is started with a finger on the panel and it Is then
`removed, or itis started without a finger and a water drop falls on the panel. The
`waterproof method in VICK#331 and implemented in PSoC Designer 5.0 SP 6.4
`uses the assumption that the negative signal from wateris less than the positive
`signal from a finger to distinguish the twoevents. That assumption seemsto be
`
`Page i of 5
`
`
`
`true on the DVK, but does coversituations of a small finger af startup. That
`assuniption was not true on the Lenove Rocket single solid diamond panelor the
`LG Mini panel. On both of these, the negative signal from water could be fess
`than, equal to, or even larger thanthe largest signal from a finger.
`
`Even with the baseline algorithm solidified, the level of water rejection /
`waterproof performance can vary significantly based on the design of the panel.
`For example, a large drop of water on the panel can create a positive mutual cap
`signal if the edge of the drop Is positioned correctly in relation to the sensor. The
`magnitude or the positive signal depends on the panel design. Figure {Need to
`add figure} will show theprofile of a water drop with positive signals near the
`edge. A filter that ignores any positive signal next to a negative signal can
`prevent this kind of false touch, but that can also cause a finger to be rejected in
`certain cases with poor grounding of the phone(signal disparity with phone
`sitting on @ box).
`in addition, finger tracking with water on the screen is also
`quite variable. On the Single Solid Diamond (SSD) Lenovo panel, finger tracking
`through a large puddle worked relatively well with just the baseline improvement,
`but for the LG mini panel, and the updated Dual Solid Diamond (DSD) panelfor
`Lenovo,it did not work weil at all. On the Lenovo SSD panel, a puddie around a
`finger caused the signal to spread significantly, but the peak signal wasstill
`located under the finger, and there was usually not a large increasein the signal
`around the edge of the puddle. This allowed reasonably good finger tracking
`with the existing centroid algorithm. This was not tested with poor grounding
`(5D), and itis likely that this case would have had poorfingertracking with a
`puddle on the screen. On the Lenovo DSD panel and the LG panel, even with
`good grounding, the signal under the finger was often not the strongest. The
`strongest signals would be around the edge of the puddle. Finger tracking in this
`situation would require modifications to the centroid aigorithm similar to the large
`finger tracking solutions discussed in VICK#335 and VICK#359,or cornbining
`the mutual cap scan datawith the self-cap scan data to find the finger positions.
`
`This memo anly addresses the Baseline problem.
`
`' General Method Description:
`
`The following two sections discuss the algorithms from a conceptual standpoint.
`Details of implementing the self-cap scan are discussedin a later section {notin
`this draft}. For the baseline improvement, two somewhatdistinct methods were
`used for Lenovo Rocket and for LG Mini. There were advantages and
`disadvantages for both methods. Hopefully an even better method, possibly a
`hybrid of the two, can be developed.
`
`General Method Used for Lenovo:
`
`For Lenovo, a self-cap scan was performed immediately after each mutual cap
`scan. A baseline for each self-cap sensor is maintained. For the original single
`
`Page 2 of 5
`
`
`
`solid diamond Lenove Panel the scan was performed with an inverted shield. A
`normal shield will ideally cause waterto have no effect on the self-cap reading.
`Without a shield, wateris detected, but not as strongly as a finger. With an
`inverted shield, the signals from water on the panel are similar in strength to a
`finger. See Figure {Need fo add figure}. The data from the water scan was used
`in two ways.
`
`First, if self-cap detects a signal (this could indicate either water or a finger on
`the panel), baseline updating is stopped. The existing mutual cap baseline
`algorithm is modified so that a signal below the negative noise threshold locks
`the baseline instead of resetting the baseline. These two changes prevent
`moisture or water drops from causing the mutual cap baseline to decrease,
`resulting in a false touch when the water is removed. See Figure {Need to add
`figure}.
`
`Second, if the self-cap signal falls below the self-cap baseline by more than
`NoiseThreshold, all of the baselines are reset.
`If the PSoC is powered on when
`there is a finger or water drop on the panel, this allows the baseline to be reset
`to a lower or higher value when the finger or water is removed. Note that there
`is additional logic so that the reset isn’t performed until there are no otherfingers
`detected on the screen by the self-cap scan to ensure that the panelis reset fo a
`good value. This does have a shoricaming in that if the panel is powered on
`with several water droplets, then some of them are wiped off, there could be
`faise touches there had previously been water drops until all of the drops have
`been removed.
`
`This method is very good when working properly, because it keeps the baseline
`near the correct value all of the time.
`{ft could be improved slightly in terms of
`how the baseline reset is done. As described here, if a finger is on the screen at
`power on, the finger will be tracked as if moves around, but there will be a dead
`spot where it first touched until the fingerislifted from the panel.
`It would
`probably be possible to reset the baselines on part of the screen as soon as the
`finger is not there. There could be some risk however in making this robust. As
`implemented, normal operation resumes as soon asthe fingeris picked up. The
`biggest dangerin this methodis that if the mutual cap baseline gets off, if won't
`recover. For example, the self-cap threshald for detecting water or a fingeris too
`high, 4 small amount of moisture that isn't detected by the self-cap could cause
`the mutual cap baseline to decrease.
`ff it gets below the Noise threshold, that
`baseline will lock and never update again, but will be hypersensitive. Similary,if
`a single sensor is repeatedly activated, or a finger approaches very slowly, the
`mutual cap baseline can increase.
`ff it increases more than NoiseThreshold,
`when the user stops tapping, the baseline will be locked and there will be an
`insensitive spat on the screen. Theoretically, if the baseline update rate and
`noise thresholds for the self-cap and mutual cap scan are exactly matched, this
`won't be a problem, but practically, self-cap and mutual cap sensing seem to
`have different sensitivities to different types of activation. So if you gotthe
`
`Fage 3 of 5
`
`
`
`thresholds exactly matched for a 9mm brass finger, they might not be matched
`for a 6mm water droplet. The LG testing did a fot of repeated tapping and
`expased these flaws. These flaws were made worse because of the workable
`noise thresholds were limited because of the noisy panel. At first we scoffed at
`the concept of this test since it didn't seem like a real world scenario, but it is not
`unlikely that some game form the app store will require repeated tapping in one
`location. After playingfor a while, the user may become somewhat sweaty,
`making this kind of scenario somewhat more likely to occur.
`
`There is also the limitation that the baseline update stops while a finger is on the
`screen.
`{lf the phone was in a hot car, then the user picked it up and started
`using if actively while a coal breeze is blowing outside the car, the ideal baseline
`could shift fairly rapidly and not update properly because the baseline is locked
`as long as the useris touching it.
`
`There were additional problems encountered due to a loss of sensitivity to water
`when Lenovo switched fromthe Single Solid Diamond to Dual Solid Diamond
`‘TO pattern.
`
`General Method used for LG:
`To get around theselimitations, a different method was used for LG. Rather
`than locking the baseline whena finger is present, the logic is inverted and the
`baseline is reset continuously as long as there is not a finger present. The scan
`method was changedto use a waterproof shield instead of the inverted shield.
`This allows the baselineto follow to negative numbers, but it willalways recover
`once the water is gone because the baseline will be reset. There are several
`practical cifficuities with this methad, Oneis that as a fingeris being placed on
`the screen, the mutual cap may havea slight signal increase before the fingeris
`detected by the self cap scan, so the baseline may get reset to a slightly higher
`than ideal value. Thatis combated somewhat by doingthe self-cap scan right
`before the mutual cap scan, but when no fingeris detected with self-cap, the
`baseline is reset with the previous mutual cap reading. That has the unfortunate
`side effect thai the baseline gets bumpedalittle high whena fingeris lifted.
`That might be prevented somewhat by requiring at least a couple cycles with no
`finger detected in self-cap before resetting the baseline.
`
`Orie issue found with the waterproof self-cap scan was that a water drop that
`was partially in the active area of the [TO and partially over the traces on the
`edge would create a much stronger signal that a water drop that was entirely
`within the active area. That is mostlikely caused because the drop couples to
`the grounded shield near the edge of the panel rather than coupling only to the
`driven shield inside the active area.
`
`On further consideration it seems like it would have been good fo continue using
`the inverted shield to detect water. That would prevent the baseline from being
`
`Fage 4of5
`
`
`
`| seern to remernber
`reset continuously when there was water on the screen.
`that there was some reason for using the waterproof self-cap scan, but | can’t
`think of any good reason now. One marginal reasonis that if there is a large
`water drop on the panel, there can be positive signals near the edge of the water
`depending on placement.
`If the baseline gets reset, this positive signal will be
`baselined out.
`[ call this a marginal reason because in order to avoid other
`problems, the threshold had to be set very low onthe self-cap so the mutual-cap
`baseline is not reset every cycle. This means that a water drop the had a
`positive signal at the edge could still cause a false touch for several cycles until
`the baseline was reset.
`
`Onesignificant side effect of using a waterproof self-cap scan and resetting
`whenevera touch is not detected is that it easily creates additional touches when
`there is water on the screen and auser fries to useit,
`For example:
`1} Place a drop on the screen. The baseline will be reset to a lower value
`2) Place a finger on the screen somewhere else. This will lock the baseline.
`3) Move the finger through the water drop, wiping it away.
`4) Now that the drop is removed, the signal at that location will return to
`normal, but the baseline will stil be low, causing a false extra touch. The
`finger will be detected as a second touch.
`Clearly that is a wet finger tracking issue that may not need to be supported for
`some customers, but if some degree of wet finger tracking is desired, this
`probiem must be resolved. Changing the self-cap scan to use inverted shield for
`water detection seemslike It could improvethis situation significantly because
`the baseline would lock as soon as the first water drop was detected. This
`hasn't been tried yet and there may be some fatal flaw that | didn’t think of.
`
`Self Cap Scanning with TMA300:
`
`The TMA300 device presents some challenges for implementing the self-cap
`scan. This was simplified somewhat by scanning only rows or columns. A self-
`cap scan of all sensors was not necessary since only finger detection, not
`position was required, There may be sorne advantages of performing a full self-
`cap scan of rows and columns.
`
`Note that the problemsin self-cap scanning discussed here can probably be
`greatly reduced by using series resistors on the sensors. This memo does not
`discuss what resistor values would be appropriate or what side effects (if any)
`this would have an the mutual cap scan.
`
`Thefirst problem for self-cap scanningis that the peak current through the CCH
`is high.
`
`<to be continued>
`
`
`
`
`Fennel To,
`
`savttoss
`ONOaswot
`
`
`"eee
`
`PERFOR &
`
`CYPRESS SEMICONDUCTOR CORPORATION
`infernal Correspondence
`
`WW: 7009
`
`Date: 3/2/2010
`To:
`Jeff Dahlin JVY}
`Author: Michael Hills (HFM)
`Author File#: HFM#0614
`Subject: TMA300 Water Rejection Baselining Using SelfCap
`(draft)
`|
`Category:
`Distribution: EBX, TATE, ELG, JVY
`
`
`Summary:
`Water rejection was implemented for the Lenove Rocket and LG Mini project on
`TMA300 by adding a self-cap scan to help manage baseline updating.
`Maintaining a good baselineis the first problem with water on a panel using a
`mutual cap scan. Depending on the panel and finger thresholds, this alone can
`in some cases provide water rejection and/or wetfinger tracking, but other panel
`designs can have very poor wetfinger tracking and in some cases even false
`touches from water on the panel even if the baseline algorithm is robust. The
`primary goal of the methods discussed in this memois to provide a robust
`baseline so that when the water is removed from the panelif will resume normal
`operation without having to reset the PSoC.
`
`Problem Background:
`There are a couple distinct problems with water on a panel that is scanned with
`mutual cap.
`
`The first problem is that water creates a negative signal on a mutual cap scan.
`See Figure {Need to add figure}. That makes it difficult to create a robust
`baseline update algorithm.
`if the panelisinitialized with wateror a finger on the
`panel, there must be the possibility to reset all the baselines oncethis is
`removedor moved. The baseline must be able to update overtime to
`accommodate shifts from temperature, but if water accumulates or goes away
`slowly due to condensation or evaporation, the baselines should not be updated.
`Using mutual cap scan it is practically impossible to distinguish between the
`situation where a PSoC is started with a finger an the pane! and it is then
`removed, or it is started without a finger and a water dropfalls on the panel. The
`waterproof method in VICK#331 and implemented in PSoC Designer 4.0 SP 6.1
`uses the assumption that the negative signal from water is less than the positive
`signal from a finger to distinguish the two events. That assumption seems to be
`
`Page tof 5
`
`
`
`true on the DVK, but does coversituations of a smail finger at startup. That
`assumption was not frue on the Lenove Rocket single solid diamond panel or the
`LG Mini panel. On both of these, the negative signal from water could be less
`than, equal to, or even largerthan the largest signal fromafinger.
`
`Even with the baseline aigorithrnsolidified, the level of water rejection /
`waterproof performance can vary significantly based on the design of the panel.
`For example, a large drop of water on the panel can create a positive mutual cap
`signal if the edge of the drop Is positioned correctly in relation fo the sensor. The
`magnitude or the positive signal depends on the panel design. Figure {Need to
`add figure} will show the profile of a water drop with positive signafs near the
`edge. A filter that ignores any positive signal next to a negative signal can
`prevent this kind of false touch, but that can also cause a finger to be rejected In
`certain cases with poor grounding of the phone (signal disparity with phone
`sitting on a box).
`In addition, finger tracking with water on the screen is also
`guite variable. On the Single Solid Diamond (SSD) Lenovo panel, fingertracking
`through a large puddle workedrelatively well with just the baseline improvement,
`but for the LG mini panel, and the updated Dual Solid Diamond (DSD} panelfor
`Lenovo, if did not work well at all, On the Lenove SSD panel, a puddle around a
`finger causedthe signal to spread significantly, but the peak signal was still
`located under the finger, and there was usually not a large increase in the signal
`around the edge of the puddle. This allowed reasonably good finger tracking
`with the existing centroid algorithm. This was not tested with poor grounding
`(SD), and it is tkely that this case would have had poorfingertracking with a
`puddie on the screen. On the Lenovo DSD panel and the LG panel, even with
`good grounding, the signal underthe finger was often not the strongest. The
`strongest signals would be around the edge of the puddle. Finger tracking in this
`situation would require modifications to the centroid algorithm similar to the large
`finger tracking solutions discussed in VICK#335 and VICK#359, or combining
`the mutual cap scan data with the self-cap scan data to find the fingerpositions.
`
`This memo only addresses the Baseline problem.
`
`General Method Description:
`
`The following two sections discuss the algorithms from a conceptual standpoint.
`Details of implementing the self-cap scan are discussed in a later section {not in
`this draft}. For the baseline improvement, two somewhat distinct methods were
`used for Lenovo Rocket and for LG Mini. There were advantages and
`disadvantages for both methods. Hopefully an even better method, possibly a
`hybrid of the two, can be develaped,
`
`General Method Used for Lenove:
`
`For Lenovo, a self-cap scan was performed immediately after each mutual cap
`scan, A baseline for each self-cap sensor is maintained. For the original single
`
`Page 2 of 5
`
`
`
`solid diamond Lenovo Panel the scan was performed with an inverted shield. A
`normal shield will ideally cause water to have no effect on the self-cap reading.
`Without a shield, water is detected, but not as strongly as a finger. With an
`inverted shield, the signals from water on the panel are similar in strength to a
`finger. See Figure {Need fo add figure}. The data from the water scan was used
`in two ways.
`
`First, if self-cap detects a signal(this could indicate either water or a finger on
`the panel), baseline updating is stopped. The existing mutual cap baseline
`algorithm is modified so that a signal below the negative noise threshold locks
`the baseline instead of resetting the baseline. These two changes prevent
`moisture or water draps frorn causing the mutual cap baseline to decrease,
`resulting in a false touch when the water is removed. See Figure {Need to add
`fiqure}.
`
`Second, if the self-cap signal falls below the self-cap baseline by more than
`NoiseThreshold, all of the baselines are reset.
`If the PSoC is powered on when
`there is @ finger or water drop on the panel, this allows the baseline to be reset
`fo a lower or higher value when the finger or water is removed. Note that there
`is additional logic so that the reset isn’t performed until there are no otherfingers
`detected on the screen bythe self-cap scanto ensure that the panel is reset to a
`goad value. This does have a shortcoming in that if the panel is powered on
`with several water droplets, then sorne of them are wiped off, there could be
`false touches there had previously been water drops until all of the drops have
`been removed.
`
`This method is very good when working properly, because it keeps the baseline
`near the correct value ail of the time.
`{{ could be improved slightly in terms of
`how the baseline reset is done, As described here, if a finger is on the screen at
`power on, the finger will be tracked as if moves around, but there will be a dead
`spot where it first touched until the fingeris lifted from the panel.
`it would
`probably be possible to reset the baselines on part of the screen as soon as the
`finger is not there. There could be some risk however in making this robust, As
`implemented, normal operation resumes as soon as the finger is picked up. The
`biggest dangerin this methadis thatif the mutual cap baseline gets off, it won't
`recover. For example, the self-cap threshald for detecting water or a fingeris too
`high, a small amount of moisture that isn’t detected by the self-cap could cause
`the mutual cap baseline to decrease.
`If it gets below the Noise threshold, that
`baseline will lock and never update again, but will be hypersensitive. Similary, if
`a single sensor is repeatediy activated, or a finger approaches very slowly, the
`mutual cap baseline can increase.
`If it increases more than NoiseThreshold,
`when the user stops tapping, the basefine will be locked and there will be an
`insensitive spot on the screen. Theoretically, if the baseline update rate and
`noise thresholds for theself-cap and mutual cap scan are exactly maiched, this
`won't be a problem, but practically, self-cap and mutual cap sensing seem to
`have different sensitivities to different types of activation. So if you got the
`
`Page 3 af 5
`
`
`
`thresholds exactly matched for a 9mmbrass finger, they might not be matched
`for a 6mm water droplet. The LG testing did a lot of repeated tapping and
`exposed these flaws. These flaws were made worse because of the workable
`noise thresholds were limited because of the noisy panel. At first we scoffed at
`the concept of this test since it didn’t seemlike a real world scenario, but itis not
`unlikely that some game form the app store will require repeated tapping in one
`location. After playing for a while, the user may become somewhat sweaty,
`making this kind of scenario samewhat morelikely to occur.
`
`There is also the limitation that the baseline update stops while a finger is on the
`screen.
`if the phone was in a hot car, then the user picked it up and started
`using it actively while a cool breeze is blowing outside the car, the ideal baseline
`could shift fairly rapidly and not update properly because the baseline is locked
`as long as the useris touching Ht.
`
`There were additional problems encountered due to a loss of sensitivity to water
`when Lenovo switched from the Single Solid Diamond to Dual Solid Diamond
`ITO pattern,
`
`General Method used for LG:
`To get around these fimitations, a different method was used for LG. Rather
`than locking the baseline when a finger is present, the logic is inverted and the
`baseline is reset continuously as lang as there is not a finger present. The scan
`method was changed to use a waterproof shield instead of the inverted shield.
`This allows the baseline to follow to negative numbers, but if will always recover
`once the water is gone because the baseline will be reset. There are several
`practical difficulties with this method. One is that as a finger is being placed on
`the screen, the mutual cap may havea slight signal increase before the fingeris
`detected by the self cap scan, so the baseline may get reset to a slightly higher
`than ideal value. That is combated somewhat by doing the self-cap scan right
`before the mutual cap scan, but when ne finger is detected with seff-cap, the
`baseline is reset with the previous mutual cap reading. That has the unfortunate
`side effect that the baseline gets bumpeda little high when a finger Is lifted.
`That might be prevented somewhat by requiring at least a couple cycles with no
`finger detected In self-cap before resetting the baseline.
`
`One issue found with the waterproof self-cap scan was that a water drop that
`was partially in the active area of the [TO and partially over the traces on the
`edge would create a much stronger signal that a water drop that was entirely
`within the active area. That is most likely caused because the drop couples to
`the grounded shield near the edge of the panel rather than coupling only to the
`driven shield inside the active area.
`
`On further consideration it seemslike it would have been goad te continue using
`the inverted shield to detect water, That would prevent the baseline from being
`
`Page 4 of 5
`
`
`
`| seem to remember
`reset continuously when there was water on the screen.
`that there was some reason for using the waterproof self-cap scan, but | can’t
`think of any geod reason now. One marginal reason is thatif there is a large
`water drop on the panel, there can be positive signals near the edge of the water
`depending on placement.
`{f the baseline gets reset, this positive signal will be
`baselined out.
`| call this a marginal reason becausein order to avoid other
`problems, the threshold had to be set very low onthe self-cap so the mutual-cap
`baseline is nat reset every cycle. This means that a water drop the had a
`positive signal at the edge could still cause a false touch for several cycles until
`the baseline was reset.
`
`Onesignificant side effect of using a waterproof self-cap scan and resetting
`whenever a touch is not detectedIs that it easily creates additional touches when
`there is water on the screen and a usertries to use if.
`For example:
`1) Place a drop onthe screen. The baseline will be reset to a lower value
`2) Place a finger on the screen somewhere else. This will lock the baseline.
`3} Move the finger through the water drop, wiping it away.
`4) Nowthat the drop is removed, the signal at that location will return to
`normal, but the baseline wil stil be low, causing a false extra touch. The
`finger will be detected as a second touch.
`Clearly that is a wetfinger tracking issue that may not need to be supported for
`some customers, but if some degree of wet finger tracking is desired, this
`problem must be resolved. Changing the self-cap scan to use inverted shield for
`water detection seems like it could improve this situation significantly because
`the baseline would lock as soonas thefirst water drop was detected. This
`hasn't been tried yet and there may be some fatal flaw that | didn’t think of.
`
`Self Cap Scanning with TMA300:
`
`The TMA300 device presents some challenges for implementing the self-cap
`scan. This was simplified somewhat by scanning only rows or columns. A self-
`cap scan of all sensors was not necessary since onlyfinger detection, not
`position wasrequired. There may be some advantages of performing a full self-
`cap scan of rows and columns.
`
`Note that the problemsin self-cap scanning discussed here can probably be
`greatly reduced by using series resistors on the sensors. This memo does not
`discuss what resistor values would be appropriate or what side effects (if any)
`this would have on the mutual cap scan.
`
`The first problem for self-cap scanningIs that the peak current through the CCIt
`is high.
`
`<to be continued>
`
`Page 5 of 5
`
`
`
`
`0 gir
`co
`r2 CYPRESS
`
`‘wengosoooanatite
`
`PER FDR as
`
`CYPRESS SEMICONDUCTOR CORPORATION
`internal Correspondence
`
`Date: 3/8/2010
`Te: Eric Blom(EBX)
`Author: Michael Hills (HFM)
`Author File#: HFM#O62
`Subject: TTSP TMA300 Water Rejection Extension
`Gategory:
`Distribution: EBX, JVY
`
`WW: 2017010
`
`Summary:
`This outlines the planfor the preliminary version of the TMA300 TTSP water
`rejection extension.
`
`Background:
`Using a TX-Rx all points scan (mutual cap scan), water creates a negative
`signal. That makes i difficult to create a robust baseline update algorithm. A
`typical scenario is that when wateris on the panel, the baseline updates to a
`lower level than so when the wateris wiped off, the signal rises and is reported
`as a false touch that never goes away.
`
`A self-cap scanrespondsdifferently than a rnutual cap scan. Depending on the
`shield settings, water can cause almost no changein the signal, or if can cause
`a changein the same direction as a touch. By using a self-cap scan in addition
`to the mutual cap scan it is possible to make modify the baseline update
`algorithm to be more robust.
`
`See HFM#0614 for more in depth discussion.
`
`TTSP Extention:
`The TTSP extension will do the following:
`
`1) Implement a self-cap scan. This will be configurable to allow using different
`rnodes such as waterproof shield, inverted shield, increased IDAC gain, single
`ended measurement, etc. These options will be in place since options like
`inverted shield may ideally provide better waterproofing, but may result in
`currents tao high for TMA300 to handle, so other options may be necessary.
`
`Page tof?
`
`
`
`2} Override the standard baselining algorithm in the TMA3C0 user module with a
`baseline algorithm that takes the self-cap scan data into account. This will also
`have options to configure it to work with the different types of self-cap ecan.
`
`3) Include preliminary versions offiltering that may be useful for eliminating faise
`touches from positive signals fro water that can occurin some cases on a mutual
`cap scan.
`
`integration with TTSP and UM:
`The currently known hooks are listed below. More may be used for the complete
`implementation.
`
`Overloads:
`Replace TTSMain_TrueTouchStart to perforrn initialization/calibration of selfcap
`scan,
`
`Replace TTSScanSensors with a function that runs the self cap scan in addition
`to the mutual cap scan. Also, rather than callingTTUM_ScanAlSensors, it will
`call TTUM_CustomScanAilSensors which will run the modified baseline update
`routine,
`
`
`
`There may be some minor changes in the TMA306 user module necessary to
`allow the hooks to work.
`Ideally these will be rolled into the final PD 5.1 user
`module, so none of these modifications will be needed.
`
`Limitations:
`During this first phase, the extension will require more work to configure than is
`ideal.
`Instructions will be included.
`
`some filtering to help prevent false touches may be needed on some panels.
`The first version of the extension may include some preliminary versions of
`thesefilters that can be turned on, buf that are highly inefficient and probably not
`suifable for customer use.
`
`Page 2 of 2
`
`
`
`
`
` SF CYPRESS |
`
`
`BE RPS Baa
`
`CYPRESS SEMICONDUCTOR CORPORATION
`internal Correspondence
`
`WW: 6878
`
`Bate: 5/4/2009
`To: Michele Paparella (MPU)
`Author: TOP(Tony Park}
`Author File#: TOP#34
`Subject: Water rejection in Krypton (TMG200}
`Gategory: Touch screen
`Distribution: Ho_apps,vud,ylim,ebx,
`
`Water rejection in Krypten
`
`Summary
`This documentation describes the way ofwater rejection in Krypton. Water rejection issue
`was initiated by Preston in Samsung mobile. This documentation includes what happens in
`PSoC with water drop to the panel and what is the possible solution in Krypton without
`shield,
`Newsuggested solution shows good water rejection cither finger touch down or lift-off, but
`accuracy of XY coordinate need to be further improved. Standardscanning methodand new
`water rejection scanning method must be usedboth at a same time in a project for water
`rejection. In this documentation, there are two scanning methods: I would call that “Water
`Rejection Scanning” for water rejection and “Standard Scanning” whichis in current UM.
`
`Test Setup
`lL. Panel: ITO LCD module from TPO
`2. Sensors: 9 cols and 8 rows
`3, PSoC: CY8CTMG200
`
`
`
`
`
`Seanning Methods
`1. Standard scanning method
`
` . 3
`
`yy
`4
`{Water on the panel}
`
`cage eaetimes sp
`5e
`4
`[Fingeron the panel]
`
`in standard scanning method, unused sensors are all grounded. Scanning sensoris
`being surrounded by ground plain. Water increases capacitance almost as much as
`
`finger touching, so we are not able to differentiate water from touching.
`
`[RAWcount Standard Scanning]
`Above standard scanning shows same RAWcount changes from either finger or water.
`
`
`
`
`
`
`
`
`
`
`
`
`TR eStiieae
`
`Differentlyfrom standard scanning method, in water rejection