US 3222668 A
Description (OCR text may contain errors)
De'c. 7, 1965 B. LIPPEL 3,222,668
CAPACITIVE comma Filed Aug. 16, 1961 4 Sheets-Sheet 1 FIG-.3
IN VENTOR BERNARD LIPDEL.
ATTORNEYS Dec. 7, 1965 B. LIPPEL 3,222,668
CAPACITIVE CODER Filed Aug. 16, 1961 4 Sheets-Sheet 2 Isa-.5
TO l- G-HEZ E GHT DG-VT STAGEs r 63 g t :85- 64 AND 69 0 B 68 AN D e4 AND 1 AMC 82 macaw) 1 INVENTOR BEENAED LlPPEL.
ATTORNEYS Dec. 7, 1965 B. LIPPEL 3,222,663
CAPACITIVE CODER Filed Aug. 16. 1961 4 Sheets-Sheet 5 INVENTOR. Belzmmzo LIPPEL.
ATTOZ M EY'S .LOL
Dec. 7, 1965 B. LIPPEL 3,222,653
GAPACITIVE CODER Filed Aug. 16, 1961 4 Sheets-Sheet 4 INVENTOR. BEZNAZD L.| DPEL KZM? ATTORNEYS .mV M" SK United States Patent 3,222,668 CAPACITIVE CODER Bernard Lippel, 39 Fairway Ave., West Long Branch, NJ. Filed Aug. 16, 1961, Ser. No. 131,910 18 Claims. (Cl. 340--347) The present invention relates to analog-to-digital converters (also called digital transducers, coders, encoders, digitalizers or digitizers) for producing representations of physical positions expressed with binary digits. The position or input to be represented or encoded is usually the angle of a shaft, but in some instances it may be the amount of a linear displacement.
It is preferable in the majority of applications that a definite digital indication of position be made independent of the past history of motion, for which purpose the device of the invention is of a class which reads an internal scale automatically, in a manner analogous to manual reading of the marked scales associated with protractors or yardsticlcs. Analog-to-digital conversion de vices of said class are variously known as coded pattern devices, code wheels, code disks, or reading-type digitizers.
The two principal types of prior-art coded pattern devices are, first, the commutator type, having brushes running on electrical contacts and, second, the optical type, employing a pattern of transparent and opaque areas, a source of light, a slit and an array of photocells. The theory and construction of such prior devices are abundantly described in standard textbooks (for example, Richards, Digital Computer Components and Circuits, pp. 467-482 and Susskind, Notes on Analog-Digital Conversion Techniques, pp. 6-40 through 6-73), and will therefore be discussed only briefly herein.
Basically, the commutator type coder employs a separate commutator track for each binary digit of the representation and at least one brush associate-d therewith, and the optical form employs equivalent optical elements. A principal design problem is to insure correct and orderly change of the numerical indication without transition errors so that when the input angle of the angle encoder is varied smoothly, the digital representation takes on all required successive numerical values and no incorrect values are interpolated therebetween, even momentarily.
Well-known means for prevention of transition errors are described in the above references and include use of unit-distance codes (such as Gray code or unit-distance binary-decimal codes) and logical detenting schemes illustrated by the double-brush and V-scan varities.
It is also shown in the references that the prior art devices which encode linear position are straightforward modifications of the analogous devices for angle coding, so that both types of instrument are disclosed when only the angle encoders are fully discussed. In the case of the present invention, a similar relation exists; wherefore I shall, for the sake of brevity and clarity, henceforth describe the invention as it applies to angle coders.
The brush-and-comrnutator coders of the prior art experience friction between brushes and commutator which may be objectionable in some cases because it introduces a frictional torque or drag upon the rotatable input. A more serious consequence of the friction is that the brushes and commutator are subject to a high degree of wear which materially shortens the life of the instrument. Although coders of the optical or photoelectric type are not subject to such frictional forces and wear of the coded pattern, they are for their part relatively expensive, depend on light sources having limited life, and have other shortcomings. They are consequently, rarely selected in preference to the commutator for the cases in which superior resolution capability of the optical devices would not be utilized.
Encoders of the subject invention, provide accuracy 3,222,668 Patented Dec. 7, 1965 and resolution comparable to (or better than) that provided by prior art commutator coders in an instrument of comparable size; have simple and very rugged construction; are free from the friction and wear of commutatorand-brush devices; and may be constructed very economically owing in part to liberal manufacturing tolerances.
The invention will be described in detail as applied to capacitative angle coders, in which changes in capacity occur between stationary and movable electrode patterns. It will, however, be clear to those skilled in the art that the invention is equally suitable to other embodiments, and in particular to inductive angle coders in which the mutual inductances between stationary and movable loops or coils are varied likewise.
The capacitative angle coders of the present invention produce one of two clearly distinguishable states of capacity for each binary digit signal of the output. In one state (which may be considered arbitrarily to indicate the binary digitm bi-t 0), the added capacity is low in relation to stray capacity, and in the other state (which may then be considered to represent the bit l), the added capacity is appreciably larger than the stray capacity. The encoders of this invention therefore permit easy distinction between 0 and l signal-s, even when the stray capacity added to both states is significant.
Furthermore, either the high or the low capacity, but not an intermediate value, is presented to each signal output terminal, except possibly for one bit position (associated with the least significant digit of the output) wherein equivo'cation cannot cause transition errors. Finally, these advantages are obtained in a device of moderate size and convenient shape, regardless of whether the output number is expressed with the radix two, or by bit-coded representation of ordinary decimal numbers or other natural numbers basically expressed in terms of radixes which are not powers of two.
A capacitative coder could, in theory, be constructed by positioning probe electrodes close to, but not touching, the conductive surface of a coded commutator in place of the brushes of the prior art. The largest permissible area and extent of each such electrode corresponds to the maximum permissible size and extent of the contact area of the replaced brush; this is rarely more than .01 square inch in present commutator devices. Such probes, being capacitatively coupled to conducting surfaces of the coded commutator, would be therefore, in theory, equivalent to the brushes for making and breaking the circuit paths for high-frequency electrical energy; however, low-frequency or direct-current energy (which are equally suitable for brushes) would not be suitable.
It is known to those skilled in the art that the specific capacitative sensing arrangement described above is not practical unless the coder be made many times as large as a brush-type coder of equivalent accuracy and resolu tion. For example, suppose that a probe electrode is made .01 square inch in area and positioned only 0.002 inch from the conductive surface of a commutator track: then the capaicty between probe and commutator may be computed to be:
C=0.22 .-=0.22= 1.1 micromicrofarads separation .002
This, therefore, is the maximum change in capacity which will result when the commutator moves to bring alternately conductive and non-conductive elements of each track opposite the probe. A much larger stray capacity will ordinarily reside in the circuit wiring to the probe electrodes, and at the inputs of the external devices which may be connected thereto. There-fore, it will be very diffi-cult in the assumed case to distinguish clearly between 0 and 1 bit indications; and if a more generous spacing than .002 inch is provided between the probe and the commutator, the capacitance change will be even less, as indicated by the formula. It is therefore not practical to use such simple 'capacitative take-offs, unless large-area probes are provided, which in turn requires that the size of the commutator for angle encoding shall be made many times that needed for a conductive brush coder indicating the same number of digits, in which a brush area of .01 square inch is entirely suitable.
For the specific case where the output numbers are expressed with the radix two, Speller, Patent No. 2,873,- 440, shows a many-fold increase in capacitance, obtained by providing, in effect, as many identical probe electrodes as there are conductive areas in each bit track of the commutator. The probes are spaced uniformly around the track and connected in parallel electrically. He shows further, the use of V-scan logical detenting to avoid transition errors.
V-scan permits that, when the size of each of the K elemental probes associated with a binary digit having weight 2 is A, the next higher binary digit may then be obtained with probes only K/2 in number but having each an area proportional to 2A, and so one. The total electrode area for each typical ring of probes is therefore proportional to KA, and the l-capacity signal generated by cooperation of a rotor with the probes is, therefore, generally the same for every bit of the radix-two number.
A very serious disadvantage of the above-mentioned prior art capacitative coder arises from a requirement for each bit-defining track on the movable rotor (as well as each bit track component of the stator) to include two interleaved electrode areas each of which is insulated from the other and from all remaining rotor tracks; two separate capacitative slip-ring connections to the rotor are therefore provided for each binary digit encoded. Another serious disadvantage of the cited art is that it cannot provide analog-to-digital conversion using .a basic number base which is not an integral power of two.
In the capacitative coders of the present invention, a similar use is made of multiple stator probes connected in parallel, but all electrode surfaces of the rotor are at the same electrical potential. Only one electrical connection is therefore required to the rotor, such as may conveniently be made through the rotor shaft upon which the movable pattern electrodes are fixed mechanically.
In one form of the present invention output numbers are expressed in a basic radix not an integral power of two, exemplified by bit-coded decimal outputs, also called binary-decimal outputs. Bit-coded decimal numbering is discussed in Susskind, page 3-5 et seq. I have discovered that, with the aid of certain reduced-zone codes (whereby one stator track serves for the generation of more than one bit), embodiments of the present invention arranged for binary-decimal output can be constructed to have the same capacitative bit-signal variation as radix-two coders having substantially the same accuracy, resolution and physical size. For example, either a straight binary (radix two) coder which resolves a circle into 128 parts (represented by seven-bit numbers) or a decimal coder, which resolves a circle into 100 parts using eight bits to represent two decimal numerals) may be constructed with substantially the same size and volume and will then provide the same strength capacitance signals for each bit. Without such a reduced-zone coding scheme, it would be necessary for the decimal coders to be made considerably larger in size then equivalent binary coders.
For bit-coded decimal encoders, I prefer a two-zone four-bit code according to which the ten decimal numerals are represented by ten distinctive groups of four bits each. Specific suitable codes are also arranged in such a manner that one rotor track of the digitizer and its associated stator take-off elements generate a first bit in each decimal place, and a second pattern track cooperates with triple stator take-off means to generate the three remaining bits. In the case of said first bit, the rotor track and the array of cooperating stator elements (which will be referred to henceforth as the stator track) are similar to those in the embodiments of my invention for radix two providing to as much capacitance change in relation to track area. This will henceforth be referred to as 80100% standard capacitance. In the case of the three remaining hits, the number of individual probe elements in any parallel-connected group must be reduced in relation to the size of individual elements, as a result of which a theoretical maximum of only 40% standard capacitance may be observed in such a track having the same zonal area as a first bit track. However, advantage is taken of the wide spacing between elements to interleave on the one stator track three independent groups of parallel-connected probes sufficient for all three bits, thereby more than overcoming the reduction.
In practice, the three-bit stator and rotor tracks of a decimal coder may be constructed with larger zonal area than the one-bit tracks. Thus, provision of 2.5 times the standard area ideally provides for the latter three bits, 2.5 40%, or 100% of standard capacitance and the total electrode area devoted to all four bits of a decimal digit may easily be made less than four times standard. Therefore, the average pattern area per bit is less than that required for a radix-two coder which provides the same value of capacitance change for each bit. However, when ideal coders of equivalent accuracy and resolution are compared, the pure binary form is actually slightly more compact in relation to capacitance signals, because more bits are required for the binary-decimal embodiment to indicate with specified resolution.
Another desirable feature of my invention, not found in the prior art, derives from the use of a structure similar to that of a variable air-capacitor in which parallelstacked rotor plates are interleaved between stator plates. The equivalent of each pattern trac is provided by a rotor section containing one or more similar rotor plates which cooperate with an interleaved stator section, and the number of plates in each such section (or track) may be adjusted to provide approximately the same capacitance signal for each output bit. The interleaved structure also provides larger capacitance signal in relation to volume and in relation to stray capacity than can be obtained with a structure having tracks all in one plane. Finally, I may construct the rotor entirely of metal, cutting away sectors from rotor plates to provide the rotor pattern, and fastening the various plates to a metal shaft.
Various forms of my invention employ V-scan or other forms of logical detenting. The referenced textbooks de scribe V-scan selection logic for pattern-reading digitizers operating with the radix two, which is also explained clearly and in great detail in Barkers British patent application No. 650,913, published Mar. 7, 1951. In accordance with the V-scan scheme, each bit of the binary number, except the least significant bit, is generated by alternative leading or lagging reading mechanisms, selected according only to whether a 0 or 1 value has been obtained for the preceding bit output, and all selections are made in accordance with a simple chain of control, progressing from the least significant bit of the binary number to the most significant in the order of significance.
My copending application, Ser. No. 746,712, now abandoned, entitled Cathode Ray Coding Tube and Circuit, explains that said V-scan scheme may be regarded as a special case of a more general technique, therein referred to as logical selection and herein and sometimes elsewhere called logical detenting, which serves to avoid ambiguity errors in reading-type coders not necessarily restricted to radix-two numbering. Various forms of logical detenting may be used to obtain tolerance advantages for manufacture or operation of the instruments. Said copending application illustrates or describes in detail other types of logical detenting (not restricted to radix two) in which the alternative reading mechanisms are not necessarily leading and lagging takeolf means; may utilize alternative pattern tracks as well as alternative take-offs; and may utilize branching control instead of simple chain control. With branching control the choice of alternative reading mechanisms for more than one subsequent bit may depend on the value obtained for one bit. The Barker patent application also shows, for radix-two coding, the use of dual alternative pattern tracks, having individual take-oils as an alternative to V-scan with common tracks.
The class of bit-coded numbering systems which permit chain selection are known as monotonic codes, but branching selection can optionally be used (if desired) with such codes. On the other hand, codes which permit branching selection do not necessarily lend themselves to chain selection.
Ziserman in US. Patent No. 2,873,442 shows a modification of Speller, Patent No. 2,873,440, in which leading and lagging brushes cooperate with common pattern tracks, but wherein a form of branching control is used to provide for bit-coded numbers expressed in a base which is not a power of two. (A separate pattern track is used for each bit, four tracks being required for each decimal digit of a decimal output and both rotor and stator tracks are divided as in Speller).
Finally, the use of large" and small alternative take-offs (instead of leading and lagging) in cooperation with a common track, was also described by the inventor in the I.R.E. Transactions on Electronic Computers, vol. EC-4, No. 4, pp. 158 and 159. The same reference also discloses the use of two-zone decimal codes and a branching control scheme for logical detenting therewith. The code Q of copending application No. 746,712 now abandoned is a particular example of the codes of said publication, and code P of the application is an improvement thereof having particular usefulness in the instant capacitative decimal coders.
The present invention is hereinafter illustrated for radix-two numbering by a preferred form of capacitative coder in which stator sections are divided into lea-ding and lagging sub-sections cooperating With single rotor sections to provide output bits in accordance with the V-scan selection technique. It will be clear to persons skilled in the art that this form of my invention can also be constructed with dual rotor tracks cooperating individually with the dual alternative stator subsections, or with branching control of the selected sections.
For bit-coded numbers in which the radix is not a power of two, the invention is illustrated by means of a capacitative bit-coded decimal coder, using a particular 2-track code for 4-bit representation of the various decimal numerals. The actual code shown in Code P of the aforementioned copending application, which differs from Code Q in that it is better suited to a coder construction with leading and lagging stator subsections. It will be clear to persons skilled in the art that when separate alternative rotor tracks are used for each separate stator subsection, a code of the type illustrated by Q of the copending application may be more preferable for reasons explained therein. It will alsobe clear to such persons, that various other bit-coding schemes for decimal numbers for which suitable detenting logics are known (for example, the logics shown in the copending application) may be employed for capacitative decimal encoding using dual rotor tracks for each bit; however, unless reduced track-count codes are also used, such coders will be larger than the preferred decimal coder herein described, or else will provide less capacitance variation for the bit signals.
Finally, although the various improvement over prior art incorporated in the present invention apply to inductive transfer of signals between physically separated but juxtaposed rotor and stator conductor configurations as well as to capacitative transfer, the invention is, for the sake of brevity and conciseness, described only as it applies to capacitative coding. It should be, understood that I may, within the full spirit of this invention, readily transmute principles of design and operation from capacitative to inductive forms, in accordance with well-understood principles of equivalency and dualism.
It is therefore one object of the present invention to provide a simple, compact, position digitizer which will operate for longer periods of time Without care and attention than has hitherto been possible.
Another object of the present invention is to provide a digitizer wherein output bit signals are induced by capacitative or inductive action.
A further object of the present invention is to provide a capacitative coder having a fixed stator and a movable rotor, wherein all capacitor electrodes which relate to the rotor are at the same electrical potential, so as to facilitate connection of said rotor electrodes to external electric circuits, and to minimize the effects of stray capacities.
It is still another object of the present invention to provide a capacitative coder for analog-to-digital con version which may be fabricated conveniently and economically.
It is another object of the present invention to provide a capacitative coder in which only those capacitative electrode elements associated with the least significant digit of the numerical output have to be fabricated with highest accuracy, and wherein the stator and rotor elements associated with output digits of increasing order of significance may be fabricated with correspondingly greater tolerances.
A further object of the present invention is to avoid reading ambiguities by the method of generalized logical detenting; according to which equivocation requiring arbitrary decision between 0 and 1 bit values occurs in connection with at most one bit, associated with the least significant output digit, and the arbitrary decision does not introduce sensible errors as distinguished from quantizin g effects.
It is yet another object of the present invention to provide a capacitative position coder with which input position is measured by the averaged effect on many cooperating stator and rotor electrodes arranged for the mutual cancellation of errors of fabrication and assembly.
It is another object of the invention to provide a capacitative coder which may be constructed either with small axial cross-section and long axial length, or alternatively with larger axial cross section and shorter length, to suit a variety of needs.
It is a further object of the present invention to provide analog-position-to-digital conversion wherein the numerical outputs represent input positions in a natural numbering system, using positional notation not necessarily based on the radix two, and wherein the n digits of the many numbering system are each represented by a unique arrangement of bits.
A further object of one form of the invention furnishing number outputs scaled with a radix not an integral power of two is to operate with approximately the same efficiency and convenience as coders for the radix two, permitting approximately the same sized instrument for equivalent accuracy, resolution, and bit signal strength.
A further object is to permit the use of standardized rotor and stator sections, such as can be tandem-assembled in different numbers and combinations of types to vary the resolution and/ or the range of positions encoded.
The above and still further objects, features and advantages of the present invention will become apparent upon consideration of the following detailed description of specific embodiments thereof, especially when taken in conjunction with the accompanying drawings, where- 1n:
FIG. 1 shows the three most significant bit stages of a capacitative transducer suitable for a radix-two digitizer.
FIG. 2A shows, for the same transducer, a rotor leaf of the most-significant-bit stage and FIG. 2B shows a cooperating stator leaf.
FIGS. 3A and 4A are rotor leaves, and FIGS. 3B and 4B are stator leaves, respectively associated with the two neXt-most-significant bits.
FIG. shows a radix-two digitizer in diagrammatic form, including the three least significant bit stages of the transducer of FIG. 1, and associated electronic circuits.
FIG. 6 is analogous to FIG. 1 but shows a complete capacitative transducer suitable for generating decimal numbers ranging from 00 to 99, using a bit-coded decimal notation.
FIGS. 7, 8, 9 and show rotor leaves (A, on the left) and corresponding stator leaves (B, on the right) for the four sections of FIG. 6.
FIG. 8C shows a portion of FIG. 8B in detail.
FIG. 11 is a diagrammatic development analogous to FIG. 5, showing the operation of a complete encoder including the three transducer stages shown in FIGS. 7, 8 and 9 and portions of the stage shown in FIG. 10.
FIG. 12 shows a dual-purpose rotor leaf 12A and the corresponding stator leaf 12B, wherein the outer portions generate the initial bit of a digitizer output and the central portion is associated with a higher-order bit.
Referring now to FIGURE 1 of the accompanying drawings, there is illustrated a capacitative transducer, including for each bit a separate stage or section corresponding to the individual tracks of prior art. The transducer is combined with pulse circuitry and logical gating networks, not here shown, to provide a scale-of-two embodiment of the present invention. Only three sections are illustrated, namely, a most-significant-digit rotor-stator section 1, a neXt-most-significant rotor-stator section 2 and a third-most-significant rotor-stator section 3. The sections 1, 2 and 3 are individually illustrated in FIGURES 2, 3 and 4 respectively. Section 1 of the rotor consists of one or more semicircular electrodes 4, better illustrated in FIGURE 2A, secured to a conductive shaft 6 which constitutes the shaft for all of the sections of the rotor. In the specific embodiment illustrated, the shaft 6 is provided in section 1 with three semi-circular electrodes 4, which are interleaved between four stator leaves as shown in FIGURE 1. The stator leaves of section 1 are shown better in FIGURE 28. Only one leaf is evident in this view, the other three leaves being identical and directly behind. Each leaf is shown to consist of two contiguous but separate plates or pattern elements, 7 and 8, arranged to overlap two adjacent 90 sectors of the circle traced 1 out by rotation of the rotor sector 4, which circle is indicated by a dotted line. The four stator plates 7 are held in place by metallic bars 9 and 11 which provide electrical interconnection between said plates as well as providing a support therefor. The four stator plates 8 are insulated from 7 and are likewise supported by and secured to longitudinally extending bars 12 and 13 which also serve to electrically interconnect the plates. All plates 7 and 8 are notched at their adjacent inner corners as at 14 in FIGURE 2B so as to provide a clearance between these plates and the shaft 6.
The second rotor-stator section 2 likewise comprises one or more butterfly-shaped rotor electrodes 16 secured to the shaft 6. Three electrodes are actually shown in FIG- URE 1. Each of the rotor-electrodes includes two radially extending 90 sectors 17 and 18 disposed at 180 with respect to one another. In the figures as drawn, the sectors 17, being viewed along the longitudinal axis, overlap the left-hand portions of the rotors 4 of the section 1 and the sectors 18 overlap the cutaway areas, but the clockwise edges of 17 and 18 line up with the radial edges of 4. In practice any relative orientation of the various rotor sections may be used provided that the relation between corresponding stator sections is suitable thereto.
The three rotor electrodes 16 are shown interleaved with four stator leaves, each of which contains four leaf segments or pattern elements 19, 21, 22 and 23, each of which includes a 45 sector as shown in FIGURE 3B. The segments 19 and 21 are disposed immediately adjacent one another with a narrow insulating gap therebetween and the combination of the two is disposed at an angle of 180 with respect to a second pair of contiguous leaf segments 22 and 23. The segments 19, 21, 22 and 23 are each provided with apertures for receiving two posts for supporting the segments and for interconnecting like segments. FIGURES 2, 3 and 4 are arranged relative to FIGURE 1 so that FIGURE 1 is seen by looking at the left edges of the apparatus illustrated in FIGURES 2, 3 and 4. Therefore, in FIGURE 1, only four of the eight posts can be seen. Specifically, the posts 24 and 26 are associated with sector 19 and posts 27 and 28 are associated with sector 21. It will be noted that the angular position of stator section 2 with respect to stator section 1 is shown such that the segments 19 and 21 are centrally located with respect to the sectors 7 and 8. (In practice, other orientations may be used to correspond with other rotor section positioning.)
The section 3 of the transducer comprises rotor leaves 29 each having four equal-size, equally-spaced radial sectors 30, 31, 32 and 33. The sector 30 is shown with an upper edge generally aligned with the upper edge of the segment 17 of the rotor of FIGURE 3 and the whole of it overlaps the sector 17. This relationship defines the conventional position of the other three sectors. Each stator leaf of the section 3 includes eight segments 34, 35, 36, 3'7, 33, 39, 40 and 41. Each of the segments is shaped to include a sector which constitutes approximately one-sixteenth of a circle, and the segments or elements are grouped in pairs such as 34-35, 36-37, etc., with only a narrow gap between the two members of each pair. The pairs are uniformly spaced with between center lines of the pairs. The stator pairs 34-35 have the gap therebetween generally aligned with the gap between the stator pairs 19 and 21, while the stator pair 38-39 has the gap therebetween generally aligned with the gap between the stator pairs 22 and 23. The gaps alluded to here are the narrow gaps subsisting between the stator segments associated as pairs.
The stator section 3 is provided with eight rods, only four of which are illustrated in FIGURE 1 since each rod has a corresponding rod exactly aligned therewith in the direction of the view of FIGURE 1. The leaf segments 41 are supported and electrically connected together by a rod 42 while the segments 40 have a corresponding rod. The sectors 34 are interconnected and supported by a rod 43, the elements 35 by a rod 44 and the sectors 36 by a rod '46. The sectors 39, 38 and 37 have rods connecting them together which correspond respectively to the rods 43, 44 and 46.
In FIGURE 4, the actual electrical interconnections between the various stator pattern elements are illustrated, and it is seen that alternate elements are interconnected so that the sectors 34, 36, 38 and 40 are all interconnected to form one electrical group while 35, 37, 39 and 41 are interconnected to form a second electrical group. More generally, the stator section associated with each typical binary digit is divided into two sub-sections, further identified below as the A and B stator sub-sections, each preferably consisting of a plurality of leaf segments connected in parallel. Thus, four segments 7 are interconnected to comprise sub-section 1A of stator section 1 and the four remaining segments 8 comprise sub-section 1B. In stator section 2, eight elements are tied together electrically to form a sub-section 2A. These are the sectors 19 and 23 on each of four stator leaves. The remaining leaf segments, groups 21 and 22, comprise subsection 2B. In like manner, sub-section 3A of the stator includes the groups labelled 34, 36, 38 and 40 and subsection 313 includes 35, 37, 39 and 41.
The number of transducer sections required is the same 9 as the number of bits in the binary number to be provided for indication of angular measurement. The construction of rotor and stator sections associated with only the three most significant bits is shown in FIGURES 1, 2, 3 and 4, but normally a larger number of sections will be required. The necessary arrangement of each succeeding section after the third may be inferred by extension of the progression of arrangements shown for the three most significant bits. The required arrangement may also be summarized and generalized as follows: For the nth section (which generates the binary digit, Whose angular weight is 360 2-), the rotor consists of 2 equal sector blades and the space between blades is ideally the same as the blade width. The associated stator section leaves generally have sector-shaped electrodes equal to the associated rotor blades in angular size and separation, but each divided by a radial cleft into two equal half-sectors; all clockwise halves of tthe divided electrodes are connected electrically in parallel to form an A sub-section of the stator (in accordance with the convention of FIG- ,URES 2, 3 and 4) and the remaining (counterclockwise) members are likewise connected to form a B sub-section.
Except for the stator section associated with the least significant bit, A and B divisions of each stator section are insulated from one another, from remaining sections of the stator, and from the rotor. In the special case of the least significant bit, the stator section is not typical, insofar as A and B sub-sections are also connected together, but the composite stator section is still insulated from other sections and from the rotor. In the case of the least significant section, it is preferable to combine the connected A and B segments of each stator pair into a single larger stator electrode, omitting the dividing space, which is especially convenient in view of the fact that the finest electrode pattern elements are found in this section. The preferred stator-rotor arrangement for the least significant bit is not shown specifically for the radixtwo digitizer of FIGURES 1 through 4 of the accompanying drawings, inasmuch as the general construction (for a different number of radial teeth) is shown in FIGURE 7. In the case of a radix-two digitizer providing n binary digits, the number of blades 93 on each leaf of the rotor 92 (of FIGURE 7) and the number of electrodes (or internal teeth) '100 on each leaf of the stator 94 should be 2 Referring now specifically to FIGURE of the accompanying drawings, there is illustrated diagrammatically a system employing the transducer whose three most significant digit stages are illustrated in FIGURES 1 through 4 combined with suitable external circuitry in an analogto-digital converter employing logical detenting. In this figure the portions of the c-apacitative transducer shown are a section 47 for the least significant binary digit, a section 48 for the next-to-least significant digit and a section 49 for the next most significant digit which is the last illustrated. The section 47 includes equally displaced rotor sectors 93, all electrically interconnected and stator elements 100 also electrically interconnected in one group. The section 48 includes the plurality of interconnected and equal-sized rotor elements 53 which are twice the angular width of the rotor element 93 of the section 47 and spaced twice as far apart. The section 48 also includes a plurality of pairs of stator elements 54 including A and B elements in each pair. The A members are connected together via a lead 56 which also serves as an input lead to an AND NOT or inhibitor gate 57. The B segments of the stator elements 54 are connected to gether via a lead 58 which also serves as one input lead to an AND gate 59. The gates 57 and 59 outputs are inputs to an OR gate 61 having an output lead 62 serving as one output lead for the conversion system. The rotor of the section 49 comprises a plurality of rotor elements 63 which are twice the angular width of the elements 53 and therefore four times the angular width of the elements 93. Section 49 is provided with a plurality of stator pairs 64, each comprising an A and B member. The A members are connected together by a lead 66 which also serves as an input lead to an inhibitor gate 67. The B members are connected together by a lead 68 which also serves as an input lead to an AND gate 69. The gates 67 and 69 feed an OR gate 71 having an output lead 72 constituting another output lead of the system. All of the rotor elements 93, 53 and 63 are connected together electrically and, in the physical embodiment of the invention, are connected together by means of the metallic input shaft to which they are affixed. The rotor sectors are also connected via a lead 73 to a pulse source 74 which also applies pulse energy across a resistor 76. The resistor 76 is provided with a variable tap 77 to fix the fraction of the pulse amplitude provided via a lead 78 to an amplitude discriminator 79. The amplitude discriminator is also connected via a lead 81 to the stator elements 100 of the least-significantbit transducer section 47. The amplitude discriminator 79 compares pulses on the lead 81 with the height of the reference pulses on the lead 78 and if a pulse on the lead 81 is greater than the height of a contemporaneous pulse on the lead 78, an output pulse appears on a lead 82. If the pulse energy on the lead 81 is of smaller amplitude than the pulse energy on the lead 78, then no pulse is produced on the lead 82. The lead 82 constitutesthe output lead for the least significant binary digit D of the apparatus. The pulse source may provide either a single pulse or a continuous succession of electrical pulses and either pulse polarity can be accommodated by suitable design of the discriminator and gates.
The pulses appearing on the lead 82 are also connected via a lead 83 as second input pulse to the AND gate 59 and as an inhibitor input to the AND NOT gate 57. To complete the description of the circuitry, the output lead 62 of the OR gate 61 is the bit output D and also is connected via a lead 84 to the second input of the AND gate 69 and the inhibitor input of the inhibitor gate 67. Similarly, the lead 72 (if the output bit D which it provides is not the most significant digit of the transducer) is connected via a lead 85 to the AND gate and the inhibitor gate of the next highest order gating circuit which are not shown in FIGURE 5.
In operation of the apparatus, a pulse generated by the source 74 is applied simultaneously to all of the rotor segments 93, 53 and 63, from which it may be coupled to the A elements of each stator section, the B elements, both or neither, depending upon the relative position of rotor and stator electrode elements. If we assume the particular relationship between the stator assembly and rotor assembly illustrated by FIGURES 1 through 5 (which relationship exists during the transition from the full-scale angular position represented by 111 l1lall 1sto the zero-angle position represented by 000 000- all US), it is seen that the B elements of stator section 64 are in juxtaposition with the rotor electrodes 63, so that a large pulse is developed on lead 68. At the same time, the A elements are entirely between rotor electrodes and are therefore coupled to the rotor relatively loosely, so that the pulse developed on lead 66 is negligible, being well below the pulse amplitude needed to actuate the AND NOT gate 67, even when no inhiibtor gate is applied thereto. With regard to the section 48 of the transducer, the B elements likewise overlap their associated rotor segments 53 and the A elements do not. However, the undivided stator elements of the least significant section 47 overlap their associated rotor elements 93 by one-half. Therefore, the worst possible condition of operation for the least significant digit is illustrated in FIGURE 5. In the illustrated arrangement, the amplitude discriminator is capable of producing either a one or zero (depending upon whether the amplitude of the pulse developed on lead 81 is distinguishably greater or less than the amplitude of the pulse developed on the lead 78) but no other output. Either output condition is possible where the degree of overlap is exactly one-half. Assume that an output pulse is produced by the amplitude discriminator; then the least significant digit D1 is represented as a 1. This pulse is applied to the inhibitor leads of the inhibitor gate 57 and therefore, blocks this gate.
In the particular rotor position illustrated, no pulse appears on lead 56 to traverse the gate in any event but, in other situations, to 'be illustrated subsequently, pulses might be applied to the gates 57 and/or 67 from their associated input leads. The pulse appearing on the lead 83 is also supplied to the AND gate 5-9 and passes a pulse developed on the lead 58 by the B segments of the transducer sections 48.
The pulse passed by the gate 59 is also passe-d by the OR gate 61 and therefore appears on lead 62 (to indicate D =1) and also on lead 84 to activate gate 69 while inhibiting gate 67. Inasmuch as a pulse is coupled to the B elements of stator section 64 for the indicated rotor position and appears on lead '66, said pulse passes through the activated AND gate 69 and the OR gate 71 to appear on leads 7'2 and 85. The pulse on output lead '72 indi cates D '=l and the pulse on lead 85 activates one gate and inhibits a second gate respectively connected to the B and A outputs of the next-more-significant stator section, in the same manner as the control pulses on leads 83 and =84. (In case 49 is the most significant bit section of the transducer, the connection 85 is, of course, omitted.) The output D D D is therefore 111. The other situation which may arise with equal likelihood under the conditions illustrated is that the amplitude discriminator 79 determines that the pulse on the lead 8 1 is of smaller amplitude than the pulse on the lead 7 8 and, in consequence, a pulse does not appear on the lead 83. This then represents a for the digit location D Gating pulses are not applied to the gates 59 and 69 and therefore pulses are not passed by these gates. Inhibiting pulses are not applied to the gates 57 and 67; however, since substantially no pulse energy is coupled to the A segments of the sections 48 and 49, the open inhibitor gates do not pass pulses. Consequently, no pulses appear on the leads S2, 62 and W2, and the output D D D is in this case 000. It will be seen that in a digitizer having it stages the binary number output indication for zero degrees input, as illustrated, will always be either n ls or n 'Os, which are successive numbers modulo 2 in the natural scale of two. The choice between the two possible output indications is :made entirely by the amplitude discriminator 79, which acts directly for the least significant bit D, and indirectly for the other bits.
Considering another situation, let it be aussumed that the rotor elements are shifted slightly to the left as viewed in FIGURE so that the rotor elements 93 are aligned with the stator elements 109 and the rotor elements 53 overlap half of the A elements of the stator section 54. The rotor elements 63 will for their part overlap one fourth of the A elements of the stator section 64. The partial overlaps give rise to a condition that would not specifically determine the number generated if the A elements were active. However, with the rotor 93 and stator aligned, a pulse or 1 signal is produced upon the output lead '82 which insures that the A sections of the apparatus are not connected to the outputs and only the B sections are active in the same way as discussed above. Under the conditions set forth, all B sections remain maximally coupled to their rotor electrodes and the apparatus produces all ls. This is the same condition which arises if the discriminator 79 produces an output pulse when the situation is as illustrated in FIGURE 5.
Considering the opposite condition, when the rotor assembly is shifted the same amount to the right, the rotor elements 93 are completely misaligned with respect to the elements. In this case, it is the B elements of stator sections 54 and 64 which are partially overlapped and have partial 1 signals coupled thereto, the A elements being completely removed from the associated rotor electrodes. However, inasmuch as the rotor and stator elements 9'3 and of the least significant section 47 are now completely misaligned, D =O. Only the elements 54A are therefore viewed by the system (through the AND :NOT gate 57 and the OR gate 61) to ascertain that D =0. The O-signal on lead 84 likewise results in the viewing of stator elements 64A, and not 64B, for the D output; and so on for additional bit sections not shown in FIGURE 5. It is seen therefore that the same condition of all US arises as when the discriminator 79 fails to produce an output during the critical transition situation illustrated :by FIGURE 5 as shown.
Finally, it will be seen that the automatic elimination of partially overlapped stator elements illustrated in the above examples applies for further shifts of the rotor assembly to the right or left, up to the next transition points where the signals on lead 81 again approximate the signals on lead 78. While rotor shift is rightward within the whole of an interval approaching in size half the spacing between centers of electrode elements 93 (or 100) of the rotor (or stator), the least significant bit supplied by the discriminator is O, and the full output D D D consequently is 000 and when shift is to the left within an equal interval, the D output is always 1 and the binary output number is 111 This is exactly the operation of a digitizer which properly quantizes and encodes into natural binary numbers.
It will now be clear to those skilled in the art that, although only the transition from zero to full-scale reading has been shown in the drawing and explained in detail, the system of FIGURE 5 correctly quantizes and encodes any other position into natural binary code, as a consequence of the successive doubling of the size of the electrodes in successive sections. In effect, the finest electrodes 93 and 100 of the least significant section 47 and the associated discriminator 79 resolve and distinguish between all successive positions; and the larger electrodes of the succeeding sections, 48, 49, etc., play no part in resolving adjacent readings but have a role in generating the binary digits put out by the system. It has been shown that by use of split stator electrodes and a system of pulse gates, all sections (48, 49, etc.) higher in order than the least significant function without amplitude discriminators, the output 'bit signal indicated for the next less significant bit being, in each case, sufficient to remove equivocation in a manner consistent with the output of one discriminator 79.
Thus far, the detailed description of my invention has been devoted to the so-called pure binary form, whose output is a number expressed in the conventional scale of two. However, the invention applies also to coders whose output numbers are expressed in other radixes (ten or sixty, for example), using bit-coded forms of the numbers.
We may refer to numbers expressed by means of the radix n and conventional positional notation as n-ary numbers. Such numbers are usually written with n different numerals; however, only two numerals (conventionally 0 and 1) may be used, if a distinctive and unambiguous arrangement of 0s and/or ls is substituted for each of the n numerals. Such a substitutive numbering system is herein called a bit-coded n-ary number code. Bit-coded decimal codes, in particular, are well-known in the electronic computer art. Treatments thereof in standard texts (e.g. Chapter 6 of Richards Arithmetic Operations in Digital Computers) are readily extended to radixes other than ten and are illustrative of the general principle.
The radix ten is the most important radix encountered in practice which is not an integral power of two, although sixty is also prevalent for measurement of angle and time, and other radixes, such as twelve, are occasionally encountered. The accompanying FIGURES 6 through 11 accordingly illustrate specifically a bit-codeddecimal capacitative digitizer and persons skilled in the 13 art will readily perceive, with the aid of the explanations below, how my invention can be applied to other number bases (not necessarily ten, two, or an integral power of two).
TABLE I Table I is sometimes referred to as the codiset of the particular bid-coded-decimal notation employed in the digitizer of FIGURES 6 through 11. This is also the notation of Code P of copending application Ser. No. 746,712. Said codiset records a definitive arrangement of four bits chosen to represent each of the ten decimal digits 0, 1, 2, 8 and 9 and is illustrative of a variety of codisets especially suitable for forms of my invention providing for decimal or other radix-es which are not integral powers of two. To illustrate the use of Table I, we may assume that the decimal number 251 is generated by the decimal digitizer of FIGURE 11. Said digitizer puts out a sequence of twelve bits, namely 001001110001. Let the sequence of bits be divided into groups of four, thus: 0010, 0111, 0001; Table I now lists the decimal numeral which should be substituted for each group of four bits set apart by commas. The group 0010 corresponds to 2, 0111 corresponds to and 0001 corresponds to 1. Hence the decimal number 251 has been represented. When other embodiments of my invention employ codisets requiring more or less than four bits for each digit of the number base, output bits should, of course, be separated correspondingly into groups of more or less than four hits, after which each group is independently recodable with the aid of the applicable codiset to indicate the number as it is commonly written for the radix involved.
For convenience the four bits corresponding to a code for any decimal digit M are referred to as D, C, B and A and the code is written DCBA in Table I and elsewhere herein.
Where M is the units-place decimal digit, it may also be called M and the component bits will be called D C etc.; in the tens decimal place the notation M D etc., will be employed, and so on, the subscript always showing the decimal-place weight.
Any number, such as ten, which is not an integral power of two has an odd factor which remains after successive division by two; in the case of the base number ten said odd factor is five. In the preferred construction ofthe decimal transducer, shown in FIGURES 6-11, the A bit of each decimal digit is produced by means of an individual rotor-stator section similar to the bit sections of the radix-two embodiment, but five different combinations of the remaining bits B, C, and D of each decimal place are produced by cooperation of a common rotor section with B, C, and D stator elements interleaved within the same stator section.
FIGURE 6 shows a particular transducer suitable for readings from 00 to 99 inclusive. A is determined entirely by section 89; B C and D are all determined in section 88; A is determined in section 87 and B C and D in section 86.
The rotors of all of the sections 86 through 89 are physically supported on, and electrically connected to, a metallic shaft 91 to which the input rotary motion is applied. One leaf of the rotor 92 of the section 89 of the transducer is illustrated in FIGURE 7A and constitutes in the illustrated embodiment fifty radially extending teeth 93 of equal angular width, 3.6 or slightly less, and spaced 7.2 apart. The angular width of the teeth 93 will generally be made less than 3.6 to compensate for fringing of the electrostatic field at the radial tooth edges. (The best angular width depends upon the separation between rotor and stator leaves and the narrowest linear width of the teeth, and will be discussed more fully below.)
A stator 94 illustrated in FIGURE 7B comprises a ring plate 296 from which project inward, fifty teeth also having a radial width of approximately 3.6 with 7.2 spacing between centers. In this section of the apparatus, the teeth or spokes may be cut or etched into metallic plates 296 which are separated by conducting spacers 96, so that all of the teeth 100 are electrically connected together. In the other sections, such as 86 through 88, various groups of teeth must be electrically isolated from one another and the bodies or spacers 96 which hold the teeth or electrodes of the stator in place while separating the several stator leaves should be made of insulating material.
Referring now to FIGURES 8A and 8B of the accompanying drawings, there is illustrated a rotor leaf and a stator leaf, respectively, of the section 88 of the transducer. The rotor leaf 97 includes ten equally spaced radial teeth 98. The angular width of a tooth '98 is approximately 14.4 and the space therebetween is therefore approximately 21.6". The stator leaf 99 includes ten groups 101, each comprised of six fingers 102. The width of an entire group is 2l.6. The radial fingers 102 of each group are electrically insulated from one another by narrow fissures and, in order to ease fabrication, the stator plate may constitute a printed circuit panel with the conductive patterns placed thereupon to provide the individual fingers 102 of the ten groups of stator elements 101. Each finger 102 of each group is, by means not shown in FIGURE 8B, electrically connected to the finger of each of the other groups having a corresponding position therein. Thus, in FIGURE 8B, the clockwisemost finger 102 of each stator group is connected to the corresponding clockwisemost element 102 of each of the other stator groups, while all next-counterclockwise fingers are likewise interconnected, and so on.
Referring now specifically to FIGURE 9 of the accompanying drawings, there is illustrated a stator-rotor arrangement of the section 87 of the transducer. Section 87 is provided with a rotor, FIGURE 9A, having on each leaf 103 five radially extending fingers 104 displaced equally about the rotor. The elements 104 have a width of approximately 36 and the spaces therebetween are therefore approximately 36. As shown in FIGURE 9B the section 87 also is provided with stator leaves 106 comprising five pairs of stator elements 107. Each such pair 107 of stator elements comprises two radially extending fingers 108 having a radial width of approximately 14.4 and spaced apart so as to leave a 7 .2 space between the two elements 108 of each stator pair 107.
The angle between the radial outer edges of the entire pair is then the sum of 14.4 taken twice and 7.2, which is 360. The angular space between adjacent pairs is therefore likewise 36, there being five pairs and five 36 spaces. (It is a feature of my invention that liberal constructional tolerances apply, most particularly to the various angular dimensions. Where design-center values are specified herein with exactness, it should be understood that the stated values may in practice be provided only approximately in most cases, in the interest of convenience 0r economy).
The rotor and stators of the section 86, as illustrated in FIGURES 10A and B, respectively, of the accompanying drawings, comprise a rotor with leaves 109 each having a radial extension 111 of a radial width of approximately 144. Each stator leaf 112 secured to or printed upon'the insulating spacer 96 is provided with six radial 15 fingers 113 each having a radial width of approximately 36 with insulating clefts therebetween; and between the first and sixth fingers there remains a vacant sector approximately 144.
A complete digitizer system utilizing the transducer structure of FIGURES 6 through 10 is illustrated schematically in FIGURE 11 of the accompanying drawings. In this figure the various stator and rotor elements and groups thereof are designated with the same reference numerals as in FIGURES 6 through 10, and lower-case letter designations referring to induced pulse signals are applied to each group of similar and electrically interconnected stator elements to assist the explanation which will follow. Specifically, stator section 94 is designated a indicating that the initial bit component (A of the units-place decimal digit is induced therein, and the six individual fingers 102 of each stator group 101 are designated b b 0 d and d the subscript indicating that they are all associated with the units-place digit and the letters indicating that they receive signals associated with B, C and D bit components. The left and right members 108 of the stator pairs 107 are designated as a and a whereas the fingers 113 are designated b and 12 through ri and a' wherein the subscript 10 indicates that their function relates to the tens-place decimal digit.
The rotor sections 92, 97, 103 and 111 are all connected together mechanically (through the shaft not shown) so as to rotate together with rotation of the input shaft 91, and also electrically to each receive interrogating pulses from a pulse generator 74 similar to that previously shown in FIGURE for a radix-two digitizer. The resistor 76, the tap or slider 77 and the discriminator 79 are also included in FIGURE 11 as in FIGURE 5, so that a reference pulse signal on the lead 78 is compared with the a signal induced on the stator 94 and the signal lead 121 attached thereto. The amplitude discriminator 79 produces a pulse signal on the output bus 122 whenever the pulse voltage on the lead 121 is greater than the reference pulse voltage on the lead 78. (It should be noted at this point that both binary and decimal forms of the apparatus may utilize positive pulses and gate circuits suitable therefor, or else negative pulses and appropriate gating technique may equally well be employed).
A voltage appearing on the bus 122 appears at the A bit output terminal 123, inasmuch as it is the A bit signal of the units place more specifically designated A The bus is also connected to six gates, such that pulses (or more generally 1 signals) appearing thereon enable the three AND gates 127, 128 and 129, while simultaneously inhibiting the AND NOT gates 124, 125 and 126. The b fingers of the stator in transducer section 88 are connected to inhibitor gate 124 via the lead 131, while the adjacent b fingers are connected to the AND gate 127 via the lead 132. Either gate 127 or gate 124 will therefore be enabled, but not both, when an interrogating pulse is put out by the generator 74, according as whether or not the A bit signal put out by the discriminator 79 is a 1 (pulse signal by convention) or a 0 (no pulse). The outputs of gates 124 and 127 are both connected to the terminal 139 through the OR gate 138 and serve as the B bit output. It will be seen that when A is 0 (no pulse on 122) inhibitor gate 124 is enabled and AND 127 is disabled, so that terminal 139 is connected to the b fingers and disconnected from the b fingers; but when A is 1, the appearance of a pulse on the bus 122 reverses the conditions, so that b is disconnected from terminal 139 and [2 connected. In like fashion, the inhibit gate 125 is paired with the AND gate 128 and the two are connected through the OR gate 141 to the terminal 142, so that c is connected to the output 142 when A is 0, and 0 when A is 1, In exactly the same fashion either d, or d are connected to the terminal 144 according only to whether A is 0 or 1. The D signal on terminal 144, however, is also applied to the lead 146 which controls another inhibitor gateAND gate pair (147 and 148 re- 11:3 spectively) associated with the OR gate 149 to serve in the fashion explained above as a transfer switch for connecting the terminal 151 to either the a group or the am group of the stator fingers 108 in transducer section 87 in accordance with said binary D signal.
The bit signal A10, Which when D is 0 is read from that half of the stator fingers 108 designated a and when I), is 1 is read from the remaining half designated a appears at the terminal 151 and the bus 154 which are connected to the output of OR gate 149. The bus 154 is connected to the control inputs of three AND gates and three AND NOT gates to select either those fingers designated 1710', C10 and 11 01' else b o, C10 and dm to be connected to the B C and D terminals, exactly as the bit signal on the bus 122 selects either primed or unprimed fingers to the output terminals 139, 142 and 144 which relate to the B, C, and D bit components of the units decimal place.
It should be understood that lower-case letters are used herein to designate both electrically paralleled groups of stator electrodes and the signals capacitatively induced thereon, whereas upper-case letters refer to the digitizer bit-output terminals and also the binary signals thereon.
When the rotor assembly, consisting of the mechanically coupled rotor sections 92, 97, 103 and 111 is positioned as shown in FIGURE 11 relative to the stator, the pulse signals capacitatively induced from the pulse generator 74 onto the lead 121 will be neither fullamplitude signals denoting 1 nor minimum-amplitude signals denoting 0, but halfway inbetween. It then devolves upon the discriminator 79, acting by comparison with the reference signal on lead 78, to provide standard 0 or 1 pulse signals, without equivocation, at the A terminal 123, exactly as explained previously for the D pulse signal of the system of FIGURE 5. Furthermore, just as the D signal of FIGURE 5 appears also on the control bus 83, the corresponding A signals, in the instance of the FIG- URE 11 system, appears likewise on the control bus 122. It will be seen, then, that when A is chosen to be 0, the b 0 and d finger groups are connected respectively (through the units-place gates controlled by the bus 122) to the terminals 139, 142 and 144; and when A; is chosen to be 1, the b 0 and d finger groups are so connected. In either case, since negligible capacitative coupling occurs from the rotor 97 to any of the fingers 101 in the particular rotor position illustrated, the units-place output bit code D C B A can be only 0000 or 0001. At the same time, the signal D on bus 146 controlling the selection transfer switch consisting of gates 147 and 148 in conjunction with OR gate 149 is 0 for either discriminator choice. A is therefore obtained from the a stator fingers, which are seen to be removed from the proximity of the associated rotor fingers, and the signal at the output terminal 151 is 0. Said signal is also the control signal on the bus 154.
FIGURE 11 does not show the transducer developed over sufficient length to indicate fully the relationship between stator and rotor fingers in the case of the fourth section 86. This relationship is more readily indicated in FIGURE 10, which has therefore been drawn to show the rotor 111 and the stator fingers in the same relative position which they have in FIGURE 11, and the FIGURE 11 letter notations have been placed hereon. Specifically shown is the fact that the rotor overlaps the d fingers but none of the other five finger groups of the stator section.
Referring again to FIGURE 11, it will be seen that, inasmuch as the bus 154 ha a 0 control signal the d fingers are not selected, the selection being b c and d The tens-place decimal digit D C B A is therefore 0000, and the two-place bit-coded decimal number D A is either 0000 0000 or 0000 0001 according to the choice made by the discriminator 79. Reference to Table I shows that these codes indicate decimal numbers which are written in conventional fashion as or 01, respectively.
Now consider the effect when the rotor assembly, including the rigidly coupled sections 92, 97, 103 and 111, is rotated approximately 1.8", corresponding in FIGURE 11 to a left-ward rotor shift of half the width, more or less, of one of the rotor teeth 93, whereby the rotor teeth 93 now overlap the stator teeth 100 very little. Under such conditions the pulse signal on the lead 121 is clearly less than the reference signal on lead 78, whereby the output of the discriminator 79 is clearly 0. It will be evident that the 0 output persists without change throughout the entirety of the 3.6 interval of rotor positions defined on the right by the half-overlap rotor position actually drawn in FIGURE 11, in which the rotor overlaps equally the stator teeth 100 and the spaces therebetween, and on the left by the next half-overlap rotor position. Each such 3.6 interval within which the A reading does not change may be referred to as a quantum interval, or quantum, for this particular digitizer; the system operates to insure that remaining bit outputs never change in the interior of any quantum.
Directing attention now to sections 88 and 87 of the transducer when the entire rotor assembly is moved back and forth Within the first quantum interval (defined by FIGURE 11 and a small leftward movement of the rotor), it will be seen that the d group of stator fingers may overlap the rotor teeth 98 in varying degrees, ranging from substantially no overlap to substantially full overlap. However, inasmuch as A remains 0, the d, fingers and not the d fingers, are connected to the output and no variations are observed at the output terminal 144. All of the remaining electrode elements of sections 88 and 87 preserve their overlap relationships substantially without change. The case of section 86 may be examined better by reference to FIGURE which corresponds to FIGURE 11 with respect to the indicated rotor position. Fingers d are overlapped fully in said position but the remaining fingers not at all. Recalling that the six fingers shown are spaced apart at ten-quantum intervals, it is evident that motion of one quantum counterclockwise can produce a maximum of 10% overlap of the d fingers, which were previously not overlapped, but the overlap conditions of the other five groups of fingers cannot change. In practice, the insulating spaces between stator electrode fingers (including the space between the d and d finger of each leaf) may be made appreciably large in relation to the quantum interval, and/ or the angular width of the rotor tooth 111 may also be reduced, so that the overlap of d is reduced substantially below 10%, or eliminated entirely. In any event, the 0 signal induced with even full 10% overlap is treated by the subsequent gating circuits in the same manner as an ideal O-signal, since a pulse amplitude of approximately 10% of the ideal l-signal pulse is disregarded by most conventional gating circuits, i.e., treated as no pulse.
It has been shown that the discriminator output is 0 and the bit-code output of the digitizer is 0000 0000 for all positions within the applicable quantum interval. In like manner the next larger output reading 0000 0001 whose generation has been traced in detail also applies throughout the next adjacent quantum interval in the direction chosen to be increasing. In general, any output remains unchanged throughout a quantum interval, and, regardless of the amount of rotor displacement relative to the stator, the system operates so that (except for A the initial bit of the units decimal place) no bit signal is ever read out from a stator electrode in a seriously overlapped condition. To avoid equivocation in A the discriminator 79 is constructed to be bistable, i.e. only standard "0 or 1 outputs are possible.
To further demonstrate the proper operation of the coding system, we shall finally consider the case when the rotor as seen in FIGURE 11 is moved more than one but less than two quanta to the left, i.e. in the direction of decreasing output numbers. Recalling that the actual position shown is midway between the angle denoted 0000 0000, having value zero, and 0000 0001, having value one, We see that the case considered refers to the quantum position for which the output should be 1001 1001. (This output, whose value is ninety-nine, should occur because there are only one hundred quantum angles provided for.)
Imagine, therefore, a leftward motion of the integrally connected rotor sections 92, 97, 103 and 111 by one-andone half widths of the teeth 93, more or less. The rotor teeth 93 then fully overlap the stator teeth 100, and the A bit is l. The rotor teeth 98 of section 88 completely overlap the al fingers and half overlap the d fingers, but none other in the stator groups 101. The primed fingers, specifically d are disconnected from the code output ter minals 139, 142 and 144 when A is 1. Instead, the completely overlapped d fingers and completely non-overlapped b and c finger groups are connected to the output, resulting in a reading DCBA=1001 for the units place.
Leftward displacement of the rotor 103 of section 87 by approximately one-and-a-half quanta continues the overlapping of the a finger but introduces a very slight overlapping of the non-overlapped a finger in each stator pair 107. Since D is 1, however, the 1signal on lead 146 insures that A and not A is utilized. The A bit output remains therefore 1. From FIGURES l0 and 11, it can be seen that partial overlap in section 86 may occur only in the d fingers, which are not used, owing to the l-signal on lead 154. Because d is overlapped, and h and 0 are entirely free of the rotor, DCBA is 1001 also for the tens-place decimal digit. The digitizer has therefore been shown to count backwards from 01 to 00 to 99, with the decimal numerals bit-coded in accordance with the codiset of Table I.
In general, because there is only one set of stator electrodes for the A bit, which may therefore be interrogated in the partially overlapped condition, resulting in a partial signal on lead 121, the discriminator 79 is provided to regenerate a standard 0 or 1 signal on the bus 122 and the terminal 123. In the case of all other bits, however, including the A bits of higher decimal places, the capacitor electrodes are never connected to the output terminals when the signal induced thereon are not clearly 0 signals or 1 signals. (This is equivalent to a statement that logical detenting preceding from A is provided for said remaining bits.)
It will be seen that the operation of the decimal digitizer of FIGURE 11 depends upon the fact that the radix, ten, is the product of five and two. Each digit is quantized in a one-zone quinary stage (e.g. 88 or 86) but each quinary quantum level is divided by two in a stage (e.g. 89 or 87) generally similar to the repeated similar stages of the radixtwo embodiment.
The coder of FIGURES 71l provides one-hundred levels by means of an A stage 89, followed by a B C D stage 88, followed by an A stage 87, followed by a B C D stage 86. To provide two hundred levels, one provides an additional A stage and on all units and tens-place 1eaves doubles the number of teeth (halving angular dimensions of all teeth or fingers and of the spaces therebetween). To provide one thousand counts, an additional B C D stage is required and the number of elements on the electrode patterns of FIG- URES 7-10 would be multiplied tenfold, but this is equivalent to replacing the stator 94 with a split electrode stator (like that of section 87) and adding a finer graduated stage for the units place. It will be evident, that the linear development of FIGURE 11 applies to the first two decimal digits of all the above cases, and that the total number of counts may conveniently be any reasonable power of ten or twice such power of ten, e.g. 100, 200, 1000, 2000.
To encode on a scale of sixty, we may likewise factor the radix thus: 60=5 3 2 2. The factor five, being 19 prime, is obtained by scale-of-five sections with one-zone coding, already illustrated with sections 88 and 36 of FIG- URE 11. Each factor two calls for an individual bit section similar to those shown for scale-of-two coding and for a factor three a one-track section generating three different combinations of two bits is provided.
Inasmuch as the scale of sixty is ordinarily used, however, in the form of decimal'coded-sexagisimal numbers, (i.e. instead of sixty different symbols, the decimal numbers to 59 inclusive are written) it is preferable to rearrange the order of the factors thus: (3x2) (5 2). The factors in the left-hand parentheses are then treated as a scale-of-six stage composed of two sections and providing the tens-place digit of the reading of, say, minutes, running from 0 to 5, inclusive. The right-hand parenthesis may then be viewed as a decade stage of two sections, exactly as in one decade of a decimal coder (except that the full scale length in quanta will be a power of sixty rather than a power of ten) and provides the units place of the numbers between 00 and 59.
The fringing effect at the radial edges of sector-shaped teeth or fingers must be considered whenever the width of the teeth or of the space between teeth is small in comparison with the separation between rotor and stator electrodes. This will most often occur in the rotor and stator sections having the finest teeth (such as sections 89 or 88 of FIGURES 7 and 8 or section 47 or 48 of FIGURE 5). It will be evident on FIGURE 7 that the rotor teeth 93 and the stator teeth 1% may be ineffective where they extend so close to the center that the linear separation between teeth is not larger than the spacing between rotor leaves 92 and stator leaves 94, as shown in FIGURE 6. On the other hand, in the grosser transducer bit sections, whose electrodes have large angular dimension (e.g. the sections of FIGURES 2, 3 or 10), the fringing effect is not serious, and portions of the leaves near the center may be used almost as effectively as the peripheral areas. To provide a properly operating transducer constructed in a minimum volume, I may therefore utilize combination rotor and combination stator leaves whose general nature is illustrated by FIGURES 12A and 12B respectively. On such leaves patterns having fine detail are confined to the outermost circular zones and the interior Zones are utilized for grosser patterns, associated with higher order stages of the digitizer.
Referring specifically to FIGURE 12A, a rotor leaf is shown having, at the outer edge, thirty-two teeth 41, such as would be suitable for the least significant bit generation in a scale-of-two coder reading with six binary digits. Within the toothed structure, two 90 sectors are also cut out, so that the remaining portions 17 and 18 therebetween function exactly like the rotor blades 17 and 18 of FIGURE 3A. Although the finest bit section is shown at the outside in FIGURE 12A and the second most significant section (2 of FIGURE 1) is shown centrally, the outer ring may also be utilized for the second-finest bit generation pattern, or even less fine patterns, and the central portion may alternatively carry the electrode pattern of any other sufficiently gross stage (for example the sections 1 and 3 rotor patterns, shown in FIGURES 2A and 4A).
The specific dual-stage stator leaf shown in FIGURE 1213 will be seen to be arranged for cooperation with the specific rotor leaf of FIGURE 12A in a six-bit scale-oftwo coder, inasmuch as thirty-two teeth 42 are arranged to oppose the rotor teeth 41 and the central sectors 1%, 21', 22 and 23 are functionally equivalent to 19, 21, 22 and 23 of FIGURE 3B.
In practice, it may be convenient to place the fine teeth 41 and 42 on only a few of the rotor and stator leaves provided for the accompanying higher-order bit; or alternatively, the finest teeth 41 and 42 may be divided between combination rotor and stator leaves providing more than one of the gross transducer sections. The principal consideration is that the electrode areas and the number 2Q of rotor and stator leaves for each bit should be so related that roughly equal capacitative signals are obtained for all output bits. Likewise, it should be noted that FIG- URE 6 (which shows no combination leaves), shows for a decimal coder, more stator and rotor leaves for the sections 86 and 88 (whose electrode area is divided among three output bits) than for the section 87 (which provides only one bit). The section 89 is shown to have more leaves than 87 to compensate for the fringing effect. In an embodiment of the decimal coder shown in FIG- URES 611 which uses combination leaves, the section 89 would be omitted as a distinct rotor-stator unit and the toothed structures of FIGURES 7A and 7B would be combined with one or more of the sections 86, 87 and 88. If, say, 87 is the only section so combined, then the electrode areas and the number of leaves in modified section 87 would be adjusted to give a a and a signals of the same order of magnitude as the remaining binary signals generated in the transducer.
A typical two-track six-place codiset, suitable for the tens place of my capacitative digitizer when operating in a decimal-coded-sexagisimal system is given by Table II.
Six combinations of the bits group GFA are used, obtained by combining each of three values of GP (i.e. ()1, l1 and 10) with 0 and 1 values of A. The bit A is seen to be generable in the general manner shown for A of FIGURE 11; the bits F and G are seen to be generable in a single transducer section having in the stator g and g fingers suitably displaced from f and f fingers; and the bits group GFA is obtained by logical detenting with external gates in a manner entirely analogous to the system of FIGURE 11. The units-place digit associated with GFA takes on ten different values, and the codiset of Table I may be used therefore, so that the units decimal digit (of the count of sixty minutes, etc.) will be DCBA obtained as in FIGURE 11.
For the scale of sixty it may be convenient to employ in the two decimal places special codisets, such as the following:
TABLE III OOHl-H-HOOCJO HOHOHOHCHO Units-place numerals are then indicated by four-bit groups KJI-IA which can be generated by the system of FIGURE 11 with only the angular relations changed between the four sections of the transducer of FIGURE 6. Six tens-place numerals are fully defined by the three-bit groups JHA, and a modification of FIGURES 6-11 provides such groups, inasmuch as JHA is a two track codiset with A alternating and IH repeated. Furthermore, the six Arabic numeral indicated by JHA can be found on the ten-place codiset KIHA by placing a O on the left, to form OIHA. This may be convenient for typewriter print out and other purposes.
As a final example of how the factoring method can be applied to binary-coding any even valued scale, consider the modulus 360, which is the scale on which whole number of degrees are measured. Suppose it is desired to measure any angle in degrees. Although my invention cannot conveniently handle the conventional notation in which integral numbers of degrees go from 000 to 359 (owing to the fact that the tens place goes from through 9 thirty-five times and only from 0 to on the thirtysixth), it can easily indicate quadrants I, II, III and IV and measure the angle (going up to 90) counterclockwise (or, if desired, clockwise) from the quadrant axes. The quadrant will readily be seen to require only section 1 and section 2 of FIGURE 1, giving a two-bit quadrant number. To measure within the 90 quadrants, we note that 90:(3 3) (5 2). The units place is therefore generable by using the A and BCD sections shown in FIGURES 6 through 11, and also discussed for sexagisimal modulus. The tens place has nine possible digit values (0 through 8) which may be generated by two 3-place one-track sections similar to the GF stages for the codiset of Table II. The corresponding codiest may then be written:
TABLE IV Four bits will of course be used to indicate the tens place of degrees, and four more the units place.
While I have described and illustrated certain specific embodiments of my invention, it will be clear that variations of the details of construction which are specifically illustrated and described may be resorted to without departing from the true spirit and scope of the invention as defined in the appended claims.
What I claim is: i
1. A capacitive transducer for an analog position-todigital converter comprising a plurality of capacitor sections, each of said sections comprising a plurality of rotor and stator elements on interleaved parallel planes, an input shaft, all of said rotor elements being electrically directly connected to and supported on said shaft so that said rotor elements are all at the same potential, said rotor elements of the section employed to generate a first bit including a plurality of equally spaced, equal-size radial sectors, said stator elements of the section employed to generate the first bit including radially extending, equally spaced and equal size sectors at least substantially equal in number to the number of rotor sectors of the same section, said stator elements employed to produce the further bits including a plurality of radially extending, equally spaced and equal-size sectors, said last-mentioned sectors being divided into plural segments electrically insulated from one another.
2. The transducer according to claim 1 wherein said converter is employed to generate a binary number and wherein the number of rotor elements in a plane is equal to 2" where 2 -1 is the highest number the transducer can encode and x is equal to one for the first bit section and increases by one for each change in the order of the digit section.
3. An analog position-to-digital converter including a capacitive commutator including a plurality of capacitor sections, each of said sections comprising a plurality of rotor and stator elements on interleaved parallel planes, an input shaft, all of said rotor elements being electrically directly connected to and supported on said shaft so that said rotor elements are all at the same potential, said rotor elements of the section employed to generate a first bit including a plurality of equally spaced, equal-size radial sectors, said stat-or elements of the section employed to generate the first bit including radially extending, equally spaced and equal-size sectors at least substantially equal in number to the number of rotor sectors of the same section, said stator elements employed to produce the successively higher order bits having a plurality of radially extending, equally spaced and equal-size sectors, said last-mentioned sectors being divided into plural segments electrically insulated from one another, a pulse source for applying pulses to said rotor elements, an amplitude discriminator for producing a pulse when the amplitude of the pulse developed on the stator of the first bit section is above a predetermined magnitude, a plurality of AND gates and inhibitor gates, means connecting one segment of a stator sector to an AND gate, means connecting another segment of the stator sector to an inhibitor gate and means responsive to the generation of a pulse by said discriminator for determining which of said gates pass pulses.
4. The combination according to claim 3 wherein said means for determining comprises means for opening said AND gate and blocking said inhibitor gates when a pulse is passed by the gates associated with the next lower order section.
5. The combination according to claim 3 wherein said means for determining comprises means for opening plural said AND gates and blocking all of said inhibitor gates upon the production of a pulse by said amplitude discriminator.
6. An analo-g-to-digital converter including a first means, a second means, an output circuit; said first and second means being movable one with respect to the other; said first means including an extended array of first electrically conducting elements all connected together electrically so as to be at the same potential; said second means including an extended array including component groups of second conducting elements, elements of each group being connected together so as to be at the same potential and insulated from any other group; means for applying an electrical signal simultaneously to all of said first elements, thereby to induce various signals on the several groups of said second means, each induced signal corresponding to an effect averaged over all elements of a group; said output circuit being connected at one point to 2k1 groups of said second means, k being an integer, said output circuit including switching means to select the signals on only k of said groups for transmission to a second point arranged such that signals transmitted to said second point generally correspond to two specific types designated "0 and 1 bit signals and, with the possible exception of a first signal, all signals not clearly distinguishable as 0 or 1 signals being among those not transmitted to said second point; wherein said k signals transmitted to said second point are included in a bit-coded number representation which measures the relative displacement between said first and second members.
7. The analog-to-digital converter of claim 6 arr;nged to provide output thereof bit signals for the representation of an N-ary digit, N being a radix; further arranged to provide specific bit signals corresponding to each prime factor of N, h bit signals being provided for the factor j when 1' is not greater than 2, j and h likewise being integers, and concatenation of the specific bits corresponding to all prime factors of N providing the bit-coded representation of said N-ary digit; wherein the specific bits corresponding to a prime factor are induced in a set of groups of said second means cooperating with a set of elements of said first means and arranged so that each said group cooperates with each said element in said first means.
8. The combination according to claim 6 wherein said first and second means are capacitively coupled.
9. The combination according to claim 7 wherein said first and second means are capacitively coupled.
10. An analog-to-digital converter including a first means, a second means, and means for moving one of said first and second means as a unit relative to the other as a unit, said first means including a spatially extended array of first elements, conductive means maintaining all of said first elements at essentially a common potential, said second means including an extended array including groups of second conducting elements, means for main taining elements of each group equi-potential, means maintaining each of said groups in electrically insulated relation with respect to each of the other of said groups, said first elements being spatially realted in signal inducing relation with respect to said second elements, elements of each of said groups being in unique configuration relative to elements of each of the other groups for induction of signal from said first elements on a digitally coded basis, ambiguities existing in the induced signal for certain relative positions of said first and second means, and switching means connected to said second groups for removing said ambiguities.
11. The combination according to claim 10 wherein the inducing relation is capacitive only.
12. The combination according to claim 11 wherein relative areas and locations of said groups are arranged to provide bit-coding of said relative motion.
13. The combination according to claim 12 wherein said relative motion is relative rotation.
14. An analog-to-digital converter including a first means, a second means, means for moving one of said first and second means as a unit relative to the other as a unit, said first means including a spatially extending array of first elements, conductive means maintaining said first elements at essentially a common potential, said second means including an extended array including groups of second conducting elements, means for maintaining elements of each group equi-potential, means maintaining each of said groups in electrically insulated relation with respect to each of the other of said groups, said first elements being spatially in signal inducing relation with respect to said second elements, elements of each of said groups having a unique geometrical configuration relative to elements of each of the other groups 24 for induction of signal from said first elements on a digitally coded basis.
15. The combination according to claim 14 wherein the inducing relation is capacitive only.
16. The combination according to claim 14 wherein relative areas and locations of said groups are arranged to provide bit-coding of said relative motion.
17. The combination according to claim '14 wherein said relative motion is relative rotation.
18. An analog-to-digital converter including a first means, a second means, an output circuit; means for moving one of said first and second means as a unit in angular relation to the other as a unit; said first means including an extended array of first electrically conducting elements all connected together electrically so as to be at the same potential, said second means including an extended array including component groups of second conducting elements, elements of each group being connected together so as to be at the same potential and insulated from any other group; means for applying an electrical signal simultaneously to all of said first elements; elements of each of said groups being in unique geometrical configuration with respect to elements of each of the other groups for induction of signal from said first elements on a digitally coded basis, each induced signal corresponding to an eifect averaged over all elements of a group, said output circuit being connected at one point to 2k1 groups of said second means, where lc is the number of bits in each digitally coded output signal.
References Cited by the Examiner UNITED STATES PATENTS 2,474,646 6/1949 Behringer 317254 2,873,440 2/1959 Speller 340-347 2,873,441 2/1959 Miller 340-347 2,873,442 2/1959 Ziserman 340-347 2,913,646 11/1959 Noyes 317254 2,930,033 3/1960 Webb 340347 2,966,670 12/1960 Foss 340347 2,976,528 3/1961 Greunke et a1 340-347 3,020,533 2/1962 Schaefer et al 340-347 MALCOLM A. MORRISON, Primary Examiner.