72
`
`IEEE TRANSACTIONS ON MAN-MACHINE SYSTEMS, VOL. MMS-11, No. 1, MARCH 1970
`
`An Electrotactile Display
`
`ROBERT M. STRONG, memBER, IEEE, AND DONALD E. TROXEL, MEMBER,
`
`IEEE
`
`Abstract—An explorable electrotactile display has been con-
`structed and tested. A thus far neglected sensation was identified
`and has been shown to be more useful than the more common
`electrotactile sensations. Exploration of the surface of
`the elec-
`trotactile display elicits a sensation describable as texture. Ex-
`periments have indicated that the intensity of this texture sensation
`is due primarily to the peak applied voltage rather than to current
`density as is the case for the classical electrotactile sensation. For
`subjects employing the texture sensation, experimental results
`are given for approximate thresholds and for the effect of electrode
`area on these thresholds. A boundary-localization measurement
`is offered as a measure of the usefulness of the display for textured-
`area presentation, and form-separation measurements are given
`as a measure of usefulness for line-drawing presentations. A pro-
`posed model for the mechanism producing the texture sensation
`is offered as a guide for future experimentation and display-engi-
`neering development.
`
`INTRODUCTION
`
`E HAVE BEENinvestigating electrotactile ma-
`
`W trix displays for use as an information input to
`
`the blind. Previous work concerned with electro-
`tactile displays has, for the most part, used rather large
`electrodes firmly attached to various parts of the body
`such as the chest, back, and arms. Ourinvestigations have
`been concerned primarily with an array of fairly small
`electrodes, which the subjects have been able to actively
`search with their fingers much as they might search an
`array of mechanical tactile stimulators.
`
`STIMULATOR CHARACTERISTICS
`
`The array consisted of small electrodes 70 mils in di-
`ameter that were spaced on 100-mil centers (Fig. 1). The
`maximum extent of the array was 1.0 inch wide by
`1.8 inches long. The heel of the subject’s hand rested on
`the single-return electrode. The pulses applied to the
`subject via the matrix of electrodes were of the bipolar
`rectangular type [as later shown in Figure 8(a)| from a
`source whose output impedance was 200 000 ohms. In all
`of the experiments reported here, unless otherwise noted,
`the pulse was symmetrical with a halfwidth of 0.60 ms,
`and a repetition rate of 200 pps.
`In most of the experiments the subjects had control of
`the stimulus amplitude, andin all cases they could easily
`interrupt the stimuli by removing their fingers from the
`display.
`
`revised October 2, 1969.
`Manuscript received July 21, 1969;
`This work was supported principally by the National Institutes
`of Health under Grants 5P01 GM15006-03 and 5P01 GM14940-03,
`and in part by the Joint Services Electronics Program under Con-
`tract DA28-043-AMC-02536[E].
`The authors are with the Department of Electrical Engineer-
`ing and Research Laboratory of Electronics, Massachusetts In-
`stitute of Technology, Cambridge, Mass. 02139.
`
`
`
`Pigek: Stimulator array.
`
`SENSATIONS ELIcrIrEeD
`types of sensa-
`The subjects reported two distinct
`tions. The first of these is similar to the sensation that
`Gibson [1] has reported when he used electrodes on the
`fingers; that is, that the sensations appear to be deep
`within the finger, indeed often concentrated in the joints,
`and that the sensation seems to progress up the finger,
`involving more of the finger as the stimulus amplitude
`is increased. Subjects reported that this type of sensation
`wasrelatively unpleasant and that it did not subjectively
`offer much information about the presentation. This sen-
`sation apparently has a mechanism moredirectly related
`to the peak stimulus current than to the voltage applied
`between the electrodes. A completely different sensation
`was experienced by most subjects when their fingers were
`dry and had,
`therefore, a high skin resistance. If the
`subject brushed his finger lightly over the surface, the
`surface appeared to acquire a texture, which could be
`varied by varying the stimulus parameters.’
`This sensation bears many of the properties of an
`ordinary texture sensation,
`the most
`important being
`that it is a relatively small amplitude effect, and that it
`disappears in the absence of finger motion. The sensation
`does not seem to be at all unpleasant, and its qualities
`
`1 The possibility that such a sensation existed, and its prob-
`able mechanism were suggested by Malinckrodt et al. [9].
`Valve Exhibit 1037
`Valve Exhibit 1037
`Valve v. Immersion
`Valve v. Immersion
`
`

`

`STRONG AND TROXEL: ELECTROTACTILE DISPLAY
`
`73
`
`TABLE I
`THRESHOLD VARIATIONS
`
`Threshold Value (volts peak)
`
`
`JD LBDate RK
`
`
`
`can be varied quite a bit by changing either the pulse-
`repetition rate or the peak voltage. The mechanism for
`this “texture” sensation seems to be directly related to
`the peak stimulus voltage, not to the stimulus current.
`The typical peak stimulus current at
`the sensation
`threshold for the texture sensation was of the order of
`50 vA, and a wide variation was noted. The peak voltages
`at threshold varied considerably with such circumstances
`as electrode size (see Experiment 2) but were generally
`of the same magnitude as those needed to elicit the
`sensations of Gibson [2] and were often lower.
`The texture sensation exhibits two peculiarities, both
`apparently due to skin-resistance effects. The first is the
`need to “warm up” the finger. Subjects would typically
`spend 3 to 5 minutes making scanning motions over the
`display surface when attempting to first acquire the sen-
`sation at the beginning of a session. It is believed that
`this serves the dual purpose of drying the skin surface
`and permitting appropriate adjustment of
`the finger
`pressure. The second phenomenon was a tendency for
`the texture’ sensation to fail or become sporadic. This
`appears to occur on days when the subject is, for un-
`known reasons, unable to increase his skin resistance
`sufficiently.
`
`EXPERIMENTS
`
`Several brief experiments were performed on three sub-
`jects in order to determine in a rough manner the char-
`acteristics of such a display. First, the characteristics of
`the sensation itself were examined, and results are pre-
`sented for the threshold of texture sensation, and the
`variation of that
`threshold with electrode area. The
`second group of experimentsis related to a display using
`“textured areas” as the presentation element. This group
`includes measurements of the just-noticeable differences
`for amplitude and pulse-repetition rate and the localiz-
`ability of a boundary. Finally, the applicability of the
`display to the presentation of point and line figures was
`investigated by measuring the minimum spacing needed
`to permit the user to determine that two such figures are
`distinct and not part of a larger figure.
`The first two experiments presented involve threshold
`measurements. These thresholds were measured by de-
`tecting both the ascending and descending limits in
`alternate sequences until 15 pairs of values had been
`accumulated for each measurement. The measurements
`were repeated on successive sessions over a two-week
`period. Since the number of samples is small, these re-
`sults, like those of the experiments that appear later in
`the paper, should be taken as indications of the range in
`which the actual values of the parameters measured can
`be expected to fall, not as accurate measurements of the
`parameters themselves.
`
`Experiment 1—Texture Threshold
`
`Table I shows the variation of the threshold of sensa-
`tion over the three subjects and the range into which the
`thresholds fall for the texture sensation on a single
`
`25 /24
`
`49
`
`31
`
`20/36
`
`47/40
`
`November 23
`November 24
`November 25
`December 1
`December 2
`December 3
`December4
`December 5
`December 6
`26
`December 7
`21
`December 14
`37
`December 16
`
`
`29
`37
`
`26
`
`43/36
`
`pattern presentation. The pattern used is shown in Fig.
`2(a), and the surrounding electrodes were grounded. All
`values reported are measurements in volts of peak pulse
`amplitude.
`A remarkably wide variation exists between subjects.
`It should be noted also, however, that a large variation
`occurs for a single subject from one session to the next.
`Experiment 2—Effect of Electrode Area on Threshold
`A number of threshold measurements were made in
`order to determine the effect of the electrode area on the
`threshold of touch. Over a period of two weeks and at
`least four sessions for each subject, measurements were
`made of the thresholds for five different patterns of ex-
`cited electrodes. These patterns are shownin Fig. 2(b)-(f).
`Measurements of each threshold were made on at least
`two different days and the results averaged to arrive at
`the result presented. The results are presented as a plot
`of threshold versus an area parameter, measured in units
`of electrode area. In Fig. 3 are shown the averageresults
`for three subjects using the texture sensation.
`These results show a reduction in threshold with an
`increase in electrode area. This result appears to be con-
`trary to the findings of Gibson et al. [2] for the previously
`reported electrotactile sensations. Included on the graph
`is a curve of the form
`
`J
`T=V/A;
`
`the shape predicted in part by a model for the produc-
`tion of the texture sensation, which appears at the end
`of this paper.
`In the development of the textured area form of the
`display, the primary considerations have been the ability
`of the subject to distinguish between textures and to
`locate the boundary between two areas of different tex-
`tures in a situation similar to that expected in the use of
`an explorable display. The presentations for the experi-
`ments described below are,
`therefore, all based on the
`simultaneous presentation of two different textures on
`two disjoint but immediately adjacent areas of the dis-
`play array. In all cases the size of an area was maintained
`larger than the finger pad so that the subject could feel
`
`

`

`74
`
`IEEE TRANSACTIONS ON MAN-MACHINE SYSTEMS, MARCH 1970
`
`200
`°
`O0OO000000
`900
`°
`°
`0060
`°
`OOQOO00CO
`00
`REGION ||OQOQOQQOOOOO nes
`(a)
`(d)
`
`08000
`90000
`60000
`000
`
`(b)
`QQ
`0060
`0000
`oo
`
`oo
`°°
`
`(e)
`
`°
`
`REGION 2
`
`OOOOOO0OO
`
`© SouRcE |
`© SOURCE 2
`
`Electrode patterns for threshold measurements.
`
`Fig. 4. Presentations for JND experiments.
`
`TABLE II
`AMPLITUDE JusT NoTICEABLE DIFFERENCES
`
`
`
` Level JD LB RK
`
`
`
`
`
`Near Threshold
`JND (volts)
`Center (volts)
`Percent
`Midrange
`JND (volts)
`Center (volts)
`Percent
`Near Maximum
`6.04
`5.85
`3.25
`IND (volts)
`77.2
`64.1
`54.1
`Center (volts)
`
`
`
`6.0 10.8Percent 7.8
`
`1.32
`23.2
`5.7
`
`3.17
`43.4
`7.3
`
`1.77
`30.9
`5.7
`
`2.35
`38.6
`6.1
`
`1,20
`23.2
`5.2
`
`2.24
`39.6
`5.8
`
`
`
`e AVERAGE THRESHOLD
`(THREE SUBJECTS)
`
`°
`
`
`
`\Tee
`2/a
`
`Lt
`l
`L Lo L
`IL
`L
`i
`L
`L
`i
`4
`8
`12
`16
`20
`24
`28
`AREA (ELECTRODE AREA UNITS)
`
`r 40
`o>
`uo)
`¥ 30
`
`a 5o
`
`O 20
`xWw)ta
`== 10
`
`0
`
`0
`
`Fig. 3. Effect of area on texture threshold.
`
`the textures independently if he wished. The boundary
`between the two areas was always a straight line, and
`its location was always specified along an axis perpendic-
`ularto it (Fig. 4).
`
`Experiment 3—Just Noticeable Difference for Amplitude
`This measurement was made by fixing the amplitude
`of one area and varying the amplitude of the other. For
`each measurement 100 presentations were made. In each
`presentation the lower area was fixed at the center am-
`plitude and the upper area was fixed at an amplitude
`selected by a uniform random choice from a predetermined
`range. The stimuli thus differed only in amplitude and
`location. The experiment was performed three times, once
`each at 10 percent above the ascending threshold, at the
`“most comfortable” level selected by the subject, and at
`approximately 10 percent below the maximum level the
`subject would accept for long periods. The test was of the
`two-alternative-forced-choice type,
`the choices being
`“bottom is stronger” and ‘‘top is stronger.”
`The resulting data were treated by determining the
`percentage of the time the response “bottom” was given
`for a particular stimulus pair, and graphing the results.
`The reported just noticeable difference (JND) is then one
`half the interquartile range on that graph, or where per-
`centages are given, the percentage that this figure repre-
`sents of
`the center value of
`the same measurement
`(Table ID.
`
`Experiment 4—Just Noticeable Difference for Frequency
`In order to determine approximately the JND for
`frequency,
`the subjects were presented with the same
`situation described in the previous experiment, but with
`stimuli that differed only in pulse-repetition rate. In all,
`75 presentations were made for each measurement. In
`each presentation the lower area was fixed in frequency
`at 200 pps. Only one determination was made, and the
`range from which the samples were selected for the upper
`area was the entire useful frequency range of 100 to
`1000 pps.
`The same spatial presentation was used as in Experi-
`ment 3, and again the subject was permitted as much
`time as he wanted to make a decision. Decision times
`averaged about 10 secondsforall subjects. The presenta-
`tion amplitude was set at the subject’s most comfortable
`level, and the amplitudes of the two areas were identical.
`No attempt was made, however, to correct for the effect
`of pulse-repetition rate on the subjective strength, and so
`the stimuli were subjectively the same strength only
`when the frequencies were identical.
`The subjects were asked to designate which of the two
`areas was “coarser,” a term that had come into common
`use to designate frequency-based texture differences. The
`resulting data were treated as in Experiment 3 (Table IID).
`The results are consistent with their reports in other
`situations,
`though the JND appears to be somewhat
`larger than might have been hopedfor.
`
`

`

`STRONG AND TROXEL: ELECTROTACTILE DISPLAY
`
`75
`
`TABLE III
`TABLE IV
`ArEa-BounpARY LOCALIZATION
`Frequency Just NoriceaBLe DIFFERENCES
`
`
`Subject
`IND
`Percentage
`JD
`LB
`RK
`
`
`Near threshold
`38.0
`76 pps
`JD
`—1.86
`0.74
`0.16
`mean*
`LB
`83 pps
`41.5
`
`
`
`
`RK Error<variancet77 pps 38.5 0.14 1.09 2.02
`
`
`
`amplitudet
`22.9
`24.3
`15.4
`:
`Twice threshold
`~1.58
`0.21
`0.19
`Center frequency is 200 pps.
`mean
`
`Error4variance 0.38 0.25 2.48
`
`
`amplitude
`45.7
`48.5
`30.8
`Preferred amplitude
`The remaining three experiments in the textured-area
`0.20
`0.04
`1.94
`
`
`group were performed with the same stimulus arrange- Error¢variance 0.16 0.47 1.96
`
`
`amplitude
`27.0
`26.7
`23.1
`ment. In each case, two areas of the same form used in
`
`the JND experiments were used, but with the location of
`the transition from one stimulus region to the other
`variable. In each case,
`the stimuli remained constant
`throughout
`the experiment. Each measurement repre-
`sents 50 presentations. Each presentation had a number
`N of rows of pins in the upper-electrode group connected
`to one source, and the remainderof the pins in the array
`connected to a second source.
`The subject was permitted to explore the display for
`an unlimited period of time but was requested to respond
`quickly. He was asked to specify the number of rows in
`the upper area. The resulting data were, in each case,
`analyzed to derive the mean and variance of the localiza-
`tion error in one dimension.
`
`* Meansare in tenths of inches.
`} Variances are in hundredths of square inches.
`¢ Amplitudes are in volts.
`
`distinet boundary, the boundary in this case was more
`like an indistinct region than like a sharp transition. This
`remained true even for very large amplitude differences,
`only disappearing when the border was emphasized by
`grounding a row of pins. No experiment was performed
`to test this technique.
`The results appear in Table V. The amplitude values
`used are recorded as well as the mean and variance of the
`errors. As might be expected, these results show a some-
`what larger variance than the previous experiment, but
`it is not excessively large.
`
`Experiment 7—Frequency Boundary
`The previous boundary-localization experiment was
`also carried out with a 2-JND difference in frequency.
`The procedures were the same as was the method of data
`analysis. In this case, the frequencies and amplitude were
`taken from the results of the frequency JND experiment.
`The results appear as Table VI. With the notable ex-
`ception of the results of JD, which show small variance,
`these results are similar to those of the amplitude-bound-
`ary experiment. The subjects again reported that the
`boundary was not a distinct thing, but rather an indistinct
`region between twoareas of distinctly different texture.
`
`Experiment 8—Form-Separation Measurements
`The last group of measurements to be presented are
`related to point-and-line presentations. In each of the
`measurements given below the subject was presented
`with two figures whose separation was variable. The
`figures were all simple points and lines. The unexcited
`areas of the display were connected to ground. The sub-
`jects were premitted to select the amplitude that they
`felt most comfortable with, much as they would in using
`a point-and-line display.
`The subject was informed of the shape of the pattern,
`and was allowed to adjust the amplitude to a comfortable
`level. He was then instructed to report either that he
`could feel two distinct forms or that they were merged.
`It was impressed on the subject that a report of “separate”
`
`Experiment 5—Area-Boundary Localizations
`
`The boundary between an “on” and an “off” or
`grounded area was explored at three amplitudes, thresh-
`old plus 10 percent, twice that value, and at the level
`preferred by the subject as “best.” In each case, the
`upper portion of the array was excited with a standard
`pulse and the lower region of the array was grounded.
`The results are presented in Table IV. Decision times
`were on the order of 30 seconds.
`Subject RK reported having difficulty maintaining the
`sensation on the day these data were taken, and his skin
`resistance was abnormally low. It is presumed that this
`is the reason for the large variance that his results exhibit.
`
`Experiment 6—Amplitude Boundary
`
`In order to obtain an initial measurement of the ability
`of a subject to locate the boundary between two areas
`excited by signals that were identical except for differing
`amplitudes, an experiment was performed using the
`results of Experiment 3. Using the same stimuli as in
`the “best amplitude” case of that experiment, with the
`amplitudes differing by 2 times the measured JND, the
`procedure of Experiment 5 was repeated.
`At this level of difference all of the subjects felt that
`the difference in amplitudes was easily detectable. Any
`smaller difference, however, would elicit complaints that
`the difference was not always clear. Those subjects
`utilizing the texture sensation reported that unlike the
`situation of the previous experiment that provided a
`
`

`

`76
`
`IREE TRANSACTIONS ON MAN-MACHINE SYSTEMS, MARCH 1970
`
`TABLE V
`Two AREA-TRANSITION LocALIZATION—AMPLITUDE DIFFERENCES*
`
`Subject
`Center Amplitude (volts)
`Error
`Meant
`Variancet
`
`
`1.47
`—0.166
`39.6
`JD
`1.56
`—0.024
`43.4
`LB
`1.88
`0.106
`38.6
`RK
`
`
`* Localization of the boundary between two areas whoseexcitation
`differs by two JND’s in amplitude.
`+ Mean error is measured in tenths of inches.
`t Error variance is measured in hundredths of square inches,
`
`TABLE VI
`Two Arga-TRaAnsitTion LocaLizaTION—FREQUENCY DIFFERENCES*
`
`Error
`Subject
`Amplitude (volts)
`Meant
`Variance}
`
`
`0.078
`0.71
`39.6
`JD
`2.70
`0.311
`43.4
`LB
`3.67
`0.74
`38.6
`RK
`
`
`* Localization of the boundary between two areas whose excitation
`differs by two JND’sin frequency.
`+ Meanerror is measured in tenths of an inch.
`t Error variance is measured in hundredths of a square inch.
`
`was to be given only if they were felt to be separate, not
`if they “might”’ be separate.
`The data collected have been organized to show the
`percentage of responses “‘separate” that were elicited for
`each numberof grounded pins separating the twofigures,
`and an average over the subject set is also given. Since
`the measurement
`is relatively coarse, due to the con-
`straint of pin size in the available electrode array, the
`results are given in tabular form. Each set of results
`represents 75 presentations to each subject with separa-
`tion distances ranging from 0 to 5 pins. The presentations
`were made in a random order, with no attempt to con-
`strain the response time. Response times averaged about
`10 seconds.
`The patterns are shownin Fig. 5, and the corresponding
`results as Tables VII-IX. Note that the measurement
`given X is the number of unexcited electrodes between
`figures. Therefore, the minimum distance between elec-
`trode segments that are excited is 0.LX ++ 0.03 inch.
`
`PRoposED MECHANISM AND MopEL FoR TEXTURE
`SENSATION
`
`The mechanism that we propose is that an electrically
`induced variation in the vertical force between the sub-
`ject’s skin surface and the display electrode is converted
`by the friction mechanism into a variation in the lateral
`force (tangential to the skin surface) as the subject passes
`his finger across the electrode. This variation in lateral
`force then causes motions of the portions of skin in con-
`tact with the surface over the electrode to vary about the
`relatively constant motion of the whole finger. The re-
`
`L x2
`C00®00@000
`—{
`--O.l in
`
`(a)
`908000800
`
`o0ecoceDo
`x3
`(b)
`oooogosCo
`
`BQO00COBDO
`AP
`fae
`
`(c)
`
`Fig. 5. Presentations for form-separation test.
`
`TABLE VII
`SEPARATION OF Two Points
`
`
`Separation X
`(in point
`spacings) RK
`LB
`JD
`Average
`
`
`Subject Responses (percent)
`
`2.7
`8.3
`0
`0
`0
`5.5
`0
`16.6
`0
`i
`34.6
`58
`83.3
`12.5
`2
`93.1
`94
`88.2
`97
`3
`4
`100
`100
`100
`100
`5
`100
`100
`100
`100
`
`
`-
`
`Necessary separation is 3 points.
`
`TABLE VIII
`SEPARATION OF Two LINES
`
`
`Separation X
`(in point
`spacings) RK
`LB
`JD
`Avetage
`
`
`Subject Responses (percent)
`
`0
`0
`0
`0
`0
`0
`0
`0
`0
`1
`42.4
`42
`68.6
`16.6
`2
`95
`94
`100
`91
`3
`100
`100
`100
`100
`4
`
`
`
`
`100 100 1005 100
`
`Necessary separation is 3 points.
`
`TABLE IX
`SEPARATION OF A Porn’ FROM A LINE
`
`Separation X
`(in point
`spacings)
`RK
`LB
`JD
`Average
`
`
`Subject Responses (percent)
`
`5.5
`0
`0
`17
`0
`39
`0
`33
`83
`1
`96.5
`100
`98
`91
`2
`100
`100
`100
`100
`3
`100
`100
`100
`100
`4
`
`
`
`
`100 100 1005 100
`
`Necessary separation is 2 points.
`
`

`

`STRONG AND TROXEL: ELECTROTACTILE DISPLAY
`
`q7
`
`quired variation in vertical force is assumed to be gener-
`ated by electric-charge accumulations in the subject’s
`finger and on theelectrode surface.
`The following symbols will be used in the discussion.
`
`FINGER MOTION
`——_____
` SKIN
`fy
`SURFACE
`
`TABLE
`SURFACE
`
`ELECTRODES
`INSULATING
`PLASTIC
`
`Fig. 6. Applicable forces.
`
`values measured for mechanical thresholds in the tangen-
`tial direction are similar to those for vertical motion,
`and about 10 microns (about 4 X 10~* in) peak-to-peak.
`displacement [3], [5], [6].
`The value of mechanical impedance in the tangential
`direction has been measured as about 0.48 Ib/in [3], [4].?
`It is not practical to directly measure the coefficient of
`friction, as it depends on a great many factors including
`applied finger pressure, atmospheric humidity, and indi-
`vidual skin factors, especially perspiration. However,
`subjects indicate a coefficient 4, for the finger in contact
`with the dry electrode, of the order of 1, with a maximum
`variation in that judgment of less than a factor of 3 in
`either direction. This value was the result of simply
`asking the subjects to estimate the ratio of horizontal to
`vertical force in their finger motions. We would, there-
`fore, expect to require a vertical-force variation on the
`order of 2 X 107‘ Ib, peak to peak.
`The electrical model used, in the form of a pair of
`closely spaced parallel plates of the sameslice, is shown
`in Fig. 7. The electrically induced force between the two
`“plates” and the capacitance of the system are given by
`
`f.(t)
`
`AG
`
`F,
`
`Be
`
`f. = wf.
`Zn
`
`a(t)
`
`A
`
`C
`
`The vertical componentof force (pounds)
`between the skin and theelectrode.
`The variation with time of any function
`G(é) about its average value (in the units
`of GO).
`The force (pounds) applied by the sub-
`ject in the vertical direction. (It is as-
`sumed to be essentially constant).
`The coefficient of friction for the skin-
`display surface pair.
`in
`The total tangential force (pounds)
`steady-state finger motion.
`The mechanical stiffness, in pounds per
`inch, of the skin in the direction tangen-
`tial to the skin surface.”
`The position (inches) of the local skin
`surface with respect
`to a coordinate
`framefixed to the finger.
`The thickness (inches) of insulator layer ¢.
`T;
`The dielectric constant of insulator layer7.
`€;
`T; = T;/s; The effective thickness (inches) of ma-
`terial 7.
`The area of skin contact with the elec-
`trode.
`The capacitance (farads) of the skin sur-
`face in contact with the electrode.
`The instantaneous voltage applied to
`vo)
`__&Afv()]?_
`the subject.
`Ah = 34 fy)
`The peak voltage applied in the jth ex-
`V;
`perimental case.
`s (subscript) applies to the subject’s skin.
`p (subscript) applies to the plastic insulator.
`= 8.854 X 10°7'*f/m is the permittivity of free space.
`
`GA
`CRG:
`
`The pertinent forces are shown in Fig. 6. Mathemati-
`cally, this system can be described as follows:
`
`f. = F, + Af,
`
`fh =uf,
`
`Af, = u Af,
`
`Aa(t) = Af.()/Zn-
`
`Wewish to show that Az(é) is of the same magnitude
`as the motion required for mechanical stimulation of the
`skin in a tangential direction, at frequencies near the
`pulse frequency used to elicit. the texture sensation. The
`
`that the mechanical impedance
`[4]
`2Tt has been shown [3],
`of the skin, for small
`relative displacements,
`is essentially a
`spring in nature, with a constant stiffness in the range of dis-
`placements encountered in these experiments.
`
`In order to extract the parameters of this model, in
`particular the skin thickness, an experiment was per-
`formed. It consisted of measuring the threshold of sensa-
`tion for each of three conditions.
`The active electrode consisted of the entire right half
`of the display array, much larger than the subject’s finger-
`tip. This was done three times in scrambled order, and
`the results averaged to yield the values given in Table X.
`Note that only one subject was used to derive these values.
`These values were measured on only one subject, and
`only one day. While variations in the specific voltage.
`values do occur,
`the same calculated values were ob-
`
`3 This value was not measured on the finger pads. It was meas-
`ured for small excursions (less than 6 mils) and we can expect
`that the skin of the finger pad is not significantly stiffer over:
`small excursions than the skin of the upper arm where the meas-
`urement was made. Franke’s measurements unfortunately do not
`apply directly because of the large areas of contact and large
`excursions involved,
`though they indirectly support
`the figure-
`given [4].
`
`

`

`78
`
`LOW
`Teterion eM)
`
`weEsSII ee
`
`INSULATING PLASTIC
`PLASTIC '
`ELECTRODE
`
`(a)
`
`PLATE #
`(VERY LARGE AREA)
`Ts
`T
`
`és
`€p
`
`tELECTRODE
`
`(b)
`
`SKIN SURFACE LAYER
`INTERPOSED
`INSULATOR
`
`Fig. 7. Electrical model.
`
`TABLE X
`
`
`Condition
`Average Peak Voltage at Threshold (volts)
`
`
`17
`No insulator
`73
`0.5-mil insulator
`110
`1.0-mil insulator
`
`
`tained with another subject.* It should be noted that
`these figures also reflect a large difference in the coeffi-
`cient of friction between the insulated and uninsulated
`cases, so that the insulated and uninsulated cases cannot
`be directly compared.
`The insulator used was polyvinylidene chloride (Dow
`Saran); its pertinent properties are shown in Table XI.
`In the experimental situation there is a return electrode
`under the palm of the same hand in addition to the con-
`tact at the fingertip. The results of other experiments,
`not reported here, indicate that this electrode is usually
`quite wet from perspiration, and can thus be neglected
`when compared to the dry fingertip contact. Skin re-
`sistances measured in this experimental situation on the
`order of 2 X 10° ohms with no interposed insulator and
`dry finger electrodes, and 5000-10 000 ohms with both
`electrodes wet. No measurable current flows with the
`insulator between the finger and the electrode.
`It has been assumed that the tangential-force varia-
`tion on the skin is the sameat all three threshold meas-
`urements shown in Table X. The voltage pulse train
`applied to the user’s finger is shown in Fig. 8(a). The as-
`sumption is made that the time constant (RC product)
`of the skin is much smaller than the voltage pulsewidth,
`in order to guarantee the rectangular shape of the force
`pulse.
`The variation in tangential force is thus just the peak
`value of the electrically produced force, or
`
`yA Wmax)
`“he = 30 Ef) 2 May
`
`4Skin thicknesses computed using the model change consid-
`erably, but calculated forces do not.
`
`IERE TRANSACTIONS ON MAN-MACHINE SYSTEMS, MARCH 1970
`
`TABLE XI
`
`PROPERTIES OF THE INSULATOR
`
`
`Dielectric constant
`
`3.5-5.5
`(approximately 4.5 at 200 Hz)
`350 V/mil
`Dielectric strength
`> 10% O-em
`Resistivity
`
`'
`I
`~—5 millisec —4
`
`
`0.6
`
`millisec
`
`0.6
`
`
`millisec
`
`“Vmax
`
`
`
`max
`
`f
`
`(a)
`1
`aye
`!
`kx 5 millisec
`1.2
`millisec
`
`(b)
`
`Fig. 8. Voltage and pulse waveforms.
`
`giving
`
`6AM, max) =
`EA (v, max) -=
`6Alu; max) _
`“OP +h.) “2h+t,)
`“2h. +f.)
`
`Manipulation of the force functions for the two insu-
`lated cases, noting that 7,, = 27, and uw. = ps, can
`be made to give
`
`the
`thickness,
`For each 0.50 X 107° inch of physical
`insulating material used has an effective thickness of
`Tf, = 0.111 X 10% in = 7,,.
`
`This gives an effective thickness of the dry surface skin
`layer to be
`
`T, = 0.108 x 107°
`
`The variation in vertical force generated by the model,
`using this value of skin thickness and the threshold volt-
`age measured for the no-insulator case of the experi-
`mentis
`
`Af, = 93.3 X 10°° lb perelectrode point.
`
`Since the voltages needed to elicit sensation become
`extremely high for single-electrode presentations, the ex-
`periment used here was performed using an array of active
`electrodes much larger than the finger pad. This fact
`makes it difficult, however, to determine the actual con-
`tact area being used by the subject. The result is, there-
`fore, reported in terms of the force per electrode area,
`which is 3.846 X 10°° in. It is estimated that the maxi-
`mum possible contact area is about 5 points, and the
`minimum about 1 point.
`
`

`

`IEEE TRANSACTIONS ON MAN-MACHINE SYSTEMS, VOL. MMS-11, No. 1, MaRcH 1970
`
`79
`
`This value of force does not, of course, take into account
`the surface coefficient of friction. However, unless the
`subjects have consistently and repeatedly overestimated
`the coefficient of friction, the resultant tangential force
`would be greater than the expected force needed to cause
`sensation by the proposed mechanism.
`Earlier it was assumed that the RC time constant for
`the skin was small compared to the pulsewidth. To show
`that this is true, we calculate the value of this surface
`capacitor as
`
`C=
`
`
`eA
`T
`
`C = 8.56 pF per point electrode.
`
`the maximum value of capacitance is about
`Thus,
`45 pF, and the corresponding RC product, using the skin
`resistance for the case of two wet electrodes, becomes
`45 X 107° second. This value is clearly much smaller
`than the pulsewidth of 10°° second. Thus, we can ex-
`pect the full effect of the rectangular force pulse.
`
`CONCLUSION
`
`Theexistence of a texture effect produced by an elec-
`trical stimulator has been demonstrated. We have pre-
`sented a justification for a model of this effect in which a
`physical motion of the skin is caused by the potential
`difference between the clectrode and the interior side of
`the skin. The texture effect has been clearly distinguished
`from the usual type of electrical stimulation by noting a
`
`direct dependence of the perceived stimulus intensity and
`the applied voltage rather than the usual result of the
`stimulus intensity being a function of the applied current.
`Indeed, the use of an insulator between the electrode and
`skin produces no apparent change in the perceptual
`qualities of the texture effect, while the resulting current
`is several orders of magnitude lower than that normally
`required to elicit electrotactile sensations. An explorable
`tactile display has been built that utilizes this effect, and
`the results of preliminary experiments indicate that a
`person can resolve details presented by this electro-
`tactile display.
`
`REFERENCES
`{1] R.H. Gibson, private communication, April 1967.
`[2] —~, “Electrical stimulation of
`the skin senses,” Progress
`Rept. 4, Department of Psychology, University of Pittsburgh,
`Pittsburgh, Pa., February 1, 1967.
`[3] D. S. Alles, “Kinesthetic feedback system for amputees via
`the tactile sense,” Sc.D. dissertation. Department of Mechan-
`ical Engineering, Massachusetts Institute of Technology, Cam-
`bridge, 1968.
`[4] E. K. Franke, “Mechanical impedance of the surface of the
`human body,” J. Appl. Phystol., vol. 3, pp. 582-590, 1950-1951.
`[5] R. T. Verrillo, “Investigation of some parameters of
`the
`cutaneous threshold of vibration,” J. Acous. Soc. Am., vol.
`34,