This invention relates to a method and apparatus for validating items of money, such as coins or banknotes. It is known when validating coins to perform, two or more independent tests on the coin, and to determine that the coin is an authentic coin of a specific type or denomination only if all the test results equal or come close to the results expected for a coin of that type. For example, some known validators have inductive coils which generate electromagnetic fields. By determining the influence of a coin on those fields the circuit is capable of deriving independent measurements which are predominantly determined by the thickness, the diameter and the material content of the coins. A coin is deemed authentic only if all three measurements indicate a coin of the same type.

This is represented graphically in Figure 1, in which each of the three orthogonal axes P _lf P ₂ and P ₃ represent the three independent measurements. For a coin of type A, the measurement P _] _ is expected to fall within a range (or window) W _A1 , which lies within the upper and lower limits U _A1 and L _A1 . Similarly the properties P ₂ and P ₃ are expected to lie within the

ranges _A2 ^{and W} A ₃ ' respectively. If all three measurements lie within the respective windows, the coin is deemed to be an acceptable coin of type A. In these circumstances, the measurements will lie within an acceptance region indicated at _A in Figure 1.

In Figure 1, the acceptance region R _A is three dimensional, but of course it may be two dimensional or may have more than three dimensions depending upon the number of independent measurements made on the coin.

Clearly, a coin validator which is arranged to validate more than one type of coin would have different acceptance regions R _B , R _c , etc. , for different coin types B, C, etc. The techniques used to determine authenticity vary. For example, each coin property measurement can be compared against stored upper and lower limit values defining the acceptance windows. Alternatively, each measurement may be checked to determine whether it is within a predetermined tolerance of a specific value. Alternatively, each measurement may be checked to determine whether it is equal to a specific value, in which case the permitted deviation of the measurement from an expected value is determined by the tolerance of the circuitry. GB-A-1 405 937 discloses circuitry in which the

tolerance is determined by ^" the selection of the stages of a digital counter which are decoded when the count representing the measurement is checked.

In a coin validator which is intended for validating a plurality of coin types or denominations each measurement can be checked against the respective range for every coin type before reaching the decision as to whether a tested coin is authentic, and if so the denomination of the coin. Alternatively, one of the tests could be used for pre-classifying the coin so that subsequent test measurements are only checked against the windows for the coin types determined by the pre-classification step. For example, in GB-A-1 405 937, a first test provisionally classifies the coin into one of three types, in dependence upon the count reached by a counter. The counter is then caused to count down at a rate which is determined by the results of the pre-classification test. If the final count is equal to a predetermined number (e.g. zero) , the coin is determined to be a valid coin of the type determined in the pre-classification test.

In the prior art, each acceptance window is always predetermined before the test is carried out.

Some validators have means for adjusting the acceptance windows. The purpose of the adjustment is to either increase the proportion of valid coins which

are determined to be acceptable (by increasing the size of the acceptance window) or to reduce the number of counterfeit coins which are erroneously deemed to be valid (by reducing the size of the acceptance window) . Adjustment of the window is carried out either manually, or automatically (e.g. as in EP-A-0155126) . In any event, the result of the window adjustment is that the upper and lower limits of the acceptance window are predetermined. However, by reducing the acceptance windows in order to avoid accepting counterfeit coins, it is possible that genuine coins will then be found to be invalid. Conversely, by increasing the acceptance windows to ensure that a maximum number of genuine coins are found to be valid, more counterfeit coins may also be determined to be valid. The consequence is that adjustment of windows may have adverse effects as well as beneficial effects, and may not increase the "acceptance ratio" (i.e. the ratio of the percentage of valid coins accepted to the percentage of counterfeit coins accepted) , or may only increase this ratio by a small amount.

In the field of banknote validation, measurements are also compared with acceptance regions generally of the form shown in Figure 1. Similar problems thus arise when modifying the acceptance windows to try to

avoid accepting counterfeit notes or rejecting genuine notes.

According to the present invention there is provided a method of 'Validating items of money comprising deriving at least two different measurements of a tested item, determining whether each measurement lies within a respective range associated with a particular money type, and producing a signal indicating that money of that type has been tested if all measurements fall within the respective ranges for that type, characterised in that the respective range for at least one of the measurements varies in dependence on at least one other measurement. The reference to "different measurements" is intended to indicate the measurement of different physical characteristics of the tested item, as distinct from merely taking the same measurement at different times to indicate a single physical characteristic or combination of such characteristics. For example, in GB-A-1 405 937, and in several other prior art arrangements, the time taken for a coin to travel between two points is measured. Although this could be regarded as taking two time measurements and subtracting the difference, the purpose is simply to obtain a single measurement determined by a particular

combination of physical characteristics, and therefore this does not represent "different measurements" as this is understood in the present case. Similarly, it is known to take two successive measurements dependent on the position of a coin with respect to a sensor as the coin passes the sensor, and then to take the difference between those two measurements. Again, this difference would represent a single measurement determined -by a single combination of physical characteristics (e.g. a variation in the -surface contour of the coin) .

In many circumstances, using the invention enables selection of windows which result in an improved acceptance ratio. For example, it may be found empirically that measurements Ε-± ^{and p} ₂ ^of valid- oney items of type A tend to lie within ranges W _A1 and _A2 respectively. However, it may also be found empirically that genuine items having a large value P- _j _ are unlikely also to have a large value P ₂ . Using the techniques of the invention, the upper limit of range ^W A2 ^{can be made} smaller when large values of P_ are detected. This would not significantly affect the number of valid items which are erroneously rejected, but would cause counterfeit items which may have large values of P- _] _ and P ₂ to be rejected.

The invention can be carried out in many ways.

Some examples are:

(1) A plurality of windows (W' _A1 , " _A1 , etc.) may be stored for a single property measurement P- _| _ of a single money type A. The window to be used may be selected on the basis of a different property measurement, e.g. P ₂ .

(2) The property measurements could be compared with an acceptance region for a known type of counterfeit money, and the tested item rejected if the properties are found to lie within this acceptance region. If the acceptance region overlaps the acceptance region for a genuine item, this means that the effective acceptance region for the genuine item is reduced by the overlap between its normal region and the acceptance region for the counterfeit. As will be explained more fully below, the consequence of this is that one or more acceptance windows defining the acceptance region for the money are effectively reduced as a consequence of having found a particular combination of property measurements. (3) Two or more property measurements may be combined in order to derive a value which is

a predetermined function of these measurements, and the result may be compared with a predetermined acceptance window. Because the derived value is a function of two measurements, it will be understood that the permitted range of values for each measurement will be dependent upon the other measurement(s) . The invention also extends to money validating apparatus arranged to operate in accordance with a method of the invention.

Arrangements embodying the invention will now be described by way of example with reference to the accompanying drawings, in which: Figure 1 schematically illustrates an acceptance region in a conventional validator;

Figure 2 is a schematic diagram of a coin validator in accordance with the present invention;

Figure 3 illustrates by way of example a table stored in a memory of the validator of Figure 2, the table defining acceptance regions;

Figure 4 schematically illustrates an acceptance region for the validator of Figure 2;

Figure 5 is a flowchart illustrating one possible method of operation of the validator of Figure 2;

Figure 6 illustrates an alternative method of

operation ;

Figure 7 illustrates an acceptance region in a modified embodiment;

Figure 8 is a flowchart of the operation of the modified embodiment;

Figure 9 illustrates an acceptance region in a further modification of the embodiment of Figure 2; Figure 10 is a flowchart of the operation of the modification of Figure 9; Figure 11 is a graph showing the distribution of measurements of a plurality of coins of the same type; and

Figure 12 illustrates an acceptance region in a still further modification of the embodiment of Figure 2.

The coin testing apparatus 2 shown schematically in Figure 2 has a set of coin sensors indicated at 4. Each of these is operable to measure a different property of a coin inserted in the apparatus, in a manner which is in itself well known. Each sensor provides a signal indicating the measured value of the respective parameter on one of a set of output lines indicated at 6.

An LSI 8 receives these signals. The LSI 8 contains a read-only memory storing an operating program which controls the way in which the apparatus

operates. Instead of an LSI, a standard microprocessor may be used. The LSI is operable to compare each measured value received on a respective one of the input lines 6 with upper and lower limit values stored in predetermined locations in a PROM 10. The PROM 10 could be any other type of memory circuit, and could be formed of a single or several integrated circuits, or may be combined with the LSI 8 (or microprocessor) into a single integrated circuit. The LSI 8, which operates in response to timing signals produced by a clock 12, is operable to address the PROM 10 by supplying address signals on an address bus 14. The LSI also provides a "PROM-enable" signal on line 16 to enable the PROM. In response to the addressing operation, a limit value is delivered from the PROM 10 to the LSI 8 via a data bus 18.

By way of example, one embodiment of the invention may comprise three sensors, for respectively measuring the conductivity, thickness and diameter of inserted coins. Each sensor comprises one or more coils in a self-oscillating circuit. In the case of the diameter and thickness sensors, a change in the inductance of each coil caused by the proximity of an inserted coin causes the frequency of the oscillator to alter, whereby a digital representation of the

respective property of the coin can be derived. In the case of the conductivity sensor, a change in the Q of the coil caused by the proximity of an inserted coin causes the voltage across the coil to alter, whereby a digital output representative of conductivity of the coin may be derived. Although the structure, positioning and orientation of each coil, and the frequency of the voltage applied thereto, are so arranged that the coil provides an output predominantly dependent upon a particular one of the properties of conductivity, diameter and thickness, it will be appreciated that each measurement will be affected to some extent by other coin properties.

The apparatus so far described corresponds to that disclosed in GB-A-2094008. In that apparatus, on insertion of a coin, the measurements produced by the three sensors 4 are compared with the values stored in the region of the PROM 10 shown in Figure 3. The thickness measurement is compared with the twelve values, representing the limits of six ranges for the respective coins A to F, in the row marked P _] _ in Figure 3. If the measured thickness value lies within the upper and lower limits of the thickness range for a particular coin (e.g. if it lies between the upper and lower limits U _A1 and L _A1 for the coin A) , then the thickness test for that coin has been passed.

Similarly, the diameter measurement is compared with the twelve upper and lower limit values in the row P ₂ , and the conductivity measurement is compared with the limit values in the row marked P ₃ . If and only if all the measured values fall within the stored ranges for a particular coin denomination which the apparatus is designed to accept, the LSI 8 produces an ACCEPT signal on one of a group of output lines 24, and a further signal on another of the output lines 24 to indicate the denomination of the coin being tested. The validator has an accept gate (not shown) which adopts one of two different states depending upon whether the ACCEPT signal is generated, so that all tested coins deemed genuine are directed along an accept path and all other tested items along another path.

The validator of GB-A-2094008 has acceptance regions, defined by the values stored in PROM 10, generally of the form shown in Figure 1. In the present embodiment of the invention, however, one of the six acceptance regions has the form shown at R _A in Figure 4. This differs from the region of Figure 1 in that it has been reduced by the volume shown at r _A . Thus, any received items having properties falling within the volume r _A will not be accepted by the validator. Assuming that it is found statistically

that there is a fairly high likelihood of counterfeit coins having properties lying within r _A , and a fairly remote possibility of genuine coins of type A having properties lying within this region, then the acceptance ratio is improved.

The acceptance regions R _B , R _c , etc. , each have the form shown in Figure 1, although if desired each could be modified to the form shown in Figure 4.

One possible way of operating the validator is explained below with reference to Figure 5. At step

50, the LSI takes all three of the measurements P-L, P ₂ and P ₃ . At step 51, the program proceeds to check whether the measurement P- _j _ is within the acceptance range indicated at ' _A1 in Figure 4. This is defined by the upper and lower limits U _A1 and L _A1 stored in the PROM 10, shown in Figure 3. If the measurement P-JL lies outside this range, the program proceeds as indicated as step 52 to check whether the measurements ^p _l ' ^p ₂ ^{and p} ₃ ^are appropriate for any of the other coin types B, C, etc.

Otherwise, at step 53, the program checks whether the measurement P ₂ lies within the respective range _A2 , and then at step 54 whether the measurement P ₃ lies within the respective range _A3 . If all three property measurements lie within the respective ranges for the coin type A, the program proceeds to step 55,

wherein the program checks whether the property measurement P- _j _ is less than or equal to a predetermined value P'-^ shown in Figure 4. If so, this indicates that the property measurements lie within the non-shaded region of R _A , and the coin is deemed acceptable. Accordingly, the program proceeds to step 56 where the appropriate signals indicating a valid coin of denomination A are issued.

If Ε>_ ≥ P'i then at step 57 the program checks whether P ₃ < P' ₃ . If so, then the property measurements have been found to lie within the shaded region shown in Figure 4, and the coin is deemed acceptable. Accordingly, the program proceeds to step 56. However, if P ₃ > P' ₃ , the property measurements have been found to lie within the region r _A , and the inserted item is therefore deemed not to be a coin of type A. Accordingly, the program proceeds to step 52.

Thus, the permissible window range for the property P ₃ depends upon whether or not the measurement P-^ is greater than or less than a predetermined value ι? - _ . Similarly, the range for P-L depends upon whether or not P ₃ is greater than or less than P' ₃ . With prior art arrangements having acceptance regions as shown in Figure 1, it would be possible to reduce the acceptance window W' _A1 for

property "P^ to W" _A1 . However, the modified range would be applicable for all values of P ₃ , thereby resulting in an acceptance region corresponding to the non-shaded portion of R _A . In Figure 4, the acceptance region also includes the shaded volume, so that rejection of genuine coins is less likely to occur.

Figure 6 is a flowchart illustrating an alternative technique for achieving the acceptance region shown in Figure 4. At step 60, the property measurements P _lr P ₂ and P ₃ are taken. At step 61, the property measurement P ₃ is compared with a predetermined value P' ₃ . If P ₃ is greater than P' _3/ the program proceeds to step 62; otherwise the program proceeds to step 63. At step 62, the window range W _A ^ for property measurement P-t_ is set equal to W" _A1 , and at step 63, the window is set equal to W' _A1 . The PROM 10 may be arranged to store two sets of limits U' _] _,

^L Αl' ^U "AI ^{and L} 'Α _l ' ^ ⁿ P ^{lace of tne} single set U _A1 and L _A1 in Figure 3, so that the two window ranges ^W ' _A1 and W" _A1 can be derived.

At step 64, the property measurement P- _j _ is compared with the appropriate window range determined at step 62 or 63, and if it is found to fall outside this range, the program proceeds to step 65. Thereafter, the program proceeds to check whether the property measurements are appropriate for the

remaining coins B, C, etc.

Otherwise, the program checks to determine whether property P ₂ lies within the associated window W _A2 at step 66, and then at step 67 checks whether property measurement P ₃ lies within the range W _A3 . If all three properties lie within the respective ranges, then the program proceeds to step 68, where the signals indicating acceptance of a genuine coin of denomination A are issued. Figure 7 shows the acceptance region R _A for a coin of type A in a validator according to a modified embodiment of the present invention. Figure 7 also shows a region R _N for a non-genuine coin. The validator is arranged to operate so that, if the property measurements lie within the region JJ, which has the same form as the region R _A in Figure 1, then the coin is deemed to be non-genuine and is rejected. It will be noted that the regions R _N and R _A in Figure 7 overlap. Any coins whose properties lie within the overlap region r _A will be deemed non-genuine and rejected. Accordingly, this reduces the effective acceptance region R _A for coin A by the overlap volume ^r A-

One technique for accomplishing this is described with reference to the flowchart of Figure 8. At step

80, the measurements P _lf P ₂ and P ₃ are taken. At step

81, the measurement P- _j _ is compared with each of the window regions ^W _A1 ,. W _B1 , ... etc. for the respective coins. The property measurement is also compared with a window W _N1 defining the upper and lower limits for property value P_ of the non-genuine acceptance region R _N . The window W _N1 can be defined by upper and lower limits stored in the PROM 10 in a similar manner to the upper and lower limits for the genuine coin denominations. Similarly, the upper and lower limits for the other properties are stored to define windows 2 ^{and W} N3-

During the course of step 81, if the property measurement P_ is found to lie within any of the respective windows ^W _A1 , ^W _B1 , ... or W _N1 , then an associated flag F _A1 , F _B1 , ..., F _N1 is set. Otherwise, the associated flags remain in a cleared state.

At step 82, the property measurement P ₂ is compared with respective windows W _A2 , W _B2 , ... , WJJ ₂ , to control the states of respective flags F _A2 , F _B2 , • • • i FJ 2•

Similarly, at step 83, the property measurement P ₃ is compared with respective window ranges W _A3 , W _B3 , ... , W _N3 , to control the states of respective flags

^F A3' ^F B3 i ' ^• ' ' ^F N3 ^• At step 84, the states of the flags F _N1 , F _N2 and

F _N3 are checked. If all are set, this indicates that

the properties have been found to lie within region R _N , and the program proceeds to step 85, which causes the inserted item to be rejected as a non-genuine coin. Otherwise, the program proceeds to check whether the measured properties fall within the windows for valid coins. First, at step 86, the flags F _A1 , F _A2 and F _A3 are checked. If .all are set, this indicates that the properties have been found to lie within region R _A (excluding overlap region r _A ) , and the program proceeds to step 87, where -signals indicating a genuine coin of denomination A are issued. Otherwise, the program proceeds to step 88 to check for a coin of denomination B in a similar way. If the properties lie within the respective acceptance region R _B , the appropriate signals are issued at step 89. Otherwise, the program continues to check the other coin denominations until it reaches step 90, which checks for a coin of denomination F. If the properties are found to lie within the respective acceptance region R _F , the appropriate signals are issued at step 91. Otherwise, at step 92, the program produces a signal indicating that the inserted item is to be rejected.

If desired, there could be other non-genuine coin regions R _N , which may overlap the same acceptance region R _A and/or other acceptance regions R _B , ... Rp.

Appropriate flags would be selectively set in steps 81 to 83. The states of these flags could be checked as in step 84, and if all the flags for any particular non-genuine coin region are set, the program causes a reject signal to be issued as at step 85.

In the above arrangement, the property measurements are compared with a region R _N to determine the presence of a non-genuine coin before checking the regions R _A , etc. However, if desired, the program could operate as a conventional validator by checking the regions R _A , R _B , etc. initially, and only if an acceptable coin is found (or possibly only if an acceptable coin of a specific denomination is found) check whether the properties fall within one or more regions R _N .

It will be noted that the technique of Figure 8 involves checking each property measurement against the respective windows for every denomination before determining which coin denomination has been received. In Figures 5 and 6, each property is checked against a range for a particular denomination, and the ranges for other denominations are checked only if the coin fails the test for that denomination. Obviously, each embodiment may employ either of these techniques, or indeed employ another sequence of operations.

Figure 9 shows the acceptance region R _A in a still further embodiment of the invention. The acceptance region R _A is similar to that shown in Figure 1 except that it has been reduced by the volume indicated at r _A at one corner. The volume r _A is defined by the interception of the region R _A and a plane indicated at PL.

One possible technique for achieving the acceptance region shown in Figure 9 is described with reference to Figure 10. At step 100, the property measurements P- _j _, P ₂ and P ₃ are taken. At step 102, the program checks to determine whether the following conditions are met: ^c l ^p l ^{+ c} 2 ^p 2 + ^C 3 ^P 3 + c ₄ < ⁰ , where c _l7 c ₂ , c ₃ and c ₄ are predetermined coefficients stored in a memory (e.g. the PROM 10) of the validator. If the conditions are not met, this indicates that the property measurements define a point which is located on the side S _] _ of the plane PL shown in Figure 9, and therefore the program proceeds to step 104, where the property measurements are checked against the acceptance regions for coin denominations B, C, etc. in the conventional way.

Otherwise, the program proceeds to step 105, where the property measurements are compared with the acceptance region R _A , in the normal _, way. This step will be

reached only if the property measurements lie on the side S ₂ of the plane PL. If the measurements are found to lie within the region R _A , the program proceeds to step 106, where the signals indicating receipt of genuine coin of denomination A are issued. Otherwise, the program proceeds to step 104 to check for other denominations.

In the examples given above, the reductions r _A in the unmodified acceptance region R _A are located at a corner or along an edge of the region R _A . This is not essential. It may in some circumstances be desirable to locate the region r _A closer to the centre of the region R _A , or towards the centre of a surface thereof. For example, referring to Figure 1, the reduction region r _A could be in the form of a trough extending along the centre of one of the surfaces defining the region R _A . This may be of use in validating coins which produce different measurements depending upon their orientation within the validator when being tested, e.g. depending upon whether a coin is inserted with its "heads" side on the left or right. Such measurements may be grouped in one or two major areas depending upon orientation, so that properties which are found to lie in a central region indicate that the tested item is unlikely to be genuine.

In all the above embodiments, the boundaries of the acceptance region R _A are planar. It will be appreciated that they could have any configuration. This applies also to any non-acceptance regions R _N which may be used, such as in the embodiment of Figs. 7 and 8. In the embodiment of Figures 9 and 10, non- planar boundaries could be achieved by using a non-linear equation at step 102. Examples of other possible equations are: C _I P-L + c ₂ P ₂ + c ₃ P ₃ + c ₄ + c ₅ - Pi < 0, ^p l ^p 2 ≤ ^k ' where c-^ to c ₅ and k are predetermined values.

Obviously, two or more such equations may be used. In any of the described embodiments, it is possible to modify as many of the coin acceptance regions R _A , R _B ... R _F from the general form shown in Figure 1 as desired. In addition, any of the acceptance regions may be reduced by more than one of the volumes r _A . In the Figure 4 example wherein the unmodified acceptance region R _A is reduced by the region r _A in one corner thereof, it could additionally be reduced by other volumes located in separate positions. Similarly, in Figure 7, there could be other non-genuine coin regions R _N overlapping the acceptance region R _A in different positions, and in

Figure 9 other surfaces could intersect the acceptance region R _A to define additional non- acceptance regions r _A .

In the above embodiments, the effective acceptance region is defined by sets of windows ^'

(representing the unmodified region R _A ) together with additional parameters representing the reduction r _A in that region. However, it is not essential that the unmodified window limits be employed. Instead, the entire effective acceptance region R _A can be defined by, for example, formulae such as those used in the embodiment of Figures 9 and 10.

One example of this will be described with reference to Figures 11 and 12. Referring to Figure 11, this shows the distribution of two measurements of a plurality of coins of the same type passing through the same validator. The measurements M- _] _ and M ₂ are represented by respective axes of the graph of Figure 11. I represents the idle measurement, i.e. the values Mη_ and M ₂ obtained when no coin is present in the validator. The points P represent the measurements of the respective coins. It will be noted that although the positions of the points vary substantially, they are all grouped around a line L-j_, and within a region bounded by lines L ₂ and L ₃ . This grouping may be due to the relationship between the

properties measured by measurements M- _j _ and M ₂ , or may be just an empirically observed result of statistical analysis.

It is possible, therefore, to test for the presence of a genuine coin by determining whether the measurements M _] _ and M ₂ of the coin lie within the boundaries L ₂ and L ₃ . In the present embodiment, this is done by calculating further measurements I-± and P ₂ , such that • l>ι_ represents the amount by which the measurement _ exceeds the idle value - of that measurement, and P ₂ represents the amount by which M ₂ falls below the idle value. The following test is then performed:

L _L ≤ £2 < U _L , P-L where L _L and U _L are respectively predetermined lower and upper limits, corresponding to lines L ₃ and L ₂ .

This results in an acceptance region R _A occupying the area between the inclined lines shown in Figure 12. This arrangement imposes no limits on the absolute values of Ε ₁ and P ₂ . In practice, it may be desirable to impose such limits, for example by testing for ^P 1L ≤ ^p l ≤ lU' where P _1L and P _1Tj are respectively lower and upper predetermined limits. This will result in the

acceptance region R _A occupying only the shaded region in Figure 12.

It will be understood that the steps used to carry out this technique can correspond to those conventionally used in validators, except for the calculation of_ ^ρ _2 which is carried out before the ^p l resulting value is checked against window limits.

The references throughout the specification to windows or ranges are intended to encompass ranges with a lower limit of zero or with an upper limit of infinity. That is to say, a property measurement can be deemed to be within an associated range merely by determining whether it lies above (or below) a particular value. In the embodiment of Figures 7 and 8, this also applies to the range R _N for the non-genuine coin. The program can be arranged to operate so that the property measurements are .checked against the range R _N only if they have already been found to lie within the range R _A (including the region r _A ) . In these circumstances, the upper limits of the windows defining the acceptance region R _N shown in Figure 7 are of no value, and therefore the windows defining the region R _N could be considered to extend to infinity.

References herein to coins are intended to encompass also tokens and other coin-like items.

Although the preceding description relates to the field of coin validation, it will be understood that the techniques are similarly applicable to banknote validation.