This invention relates primarily to improvements in apparatus and techniques for sensing the presence or identity of a taggant in a coin , token or object of value, and thereby sorting coins, tokens or objects of value. Some aspects of the invention are also applicable to sensing for the presence or identity of a taggant in a bank note or the likes, such as a polymer based bank note or a paper based bank note.

In particular, but not exclusively, the invention is useful in coin discriminators and in methods of discriminating between genuine coins and reject coins. Coin discriminators may be relatively large machines of the type used to process coins in bulk, or may be relatively small, e.g. hand-held machines of the type used to process individual coins, or any machine in between. Although the invention is described herein primarily in relation to coins, the skilled person would appreciate that it can also be applied to bank notes or the like, making changes as appropriate as would be apparent to one skilled in the art reading this disclosure. The term 'coin' is used herein to include any type of monetary token or token having value, metal currency, plastic or non-metallic token, a counterfeit coin, a component of a composite coin, or a washer for example. The coin may be disc shaped or generally disc shaped, or may be any other desired shape of generally planar form, such as square, oblong or oval for example. Construction of the coin may also vary from single material disks, to multiple material disks, for example bi-metallic/ bi- material and tri-metallic/ tri-material designs. Coin discriminators are used for measuring different characteristics of a coin in order to determine its type, e.g. its denomination, currency or authenticity. Various dimensional, electric and magnetic characteristics are measured for this purpose, such as the diameter and thickness of the coin, its electrical conductivity, its magnetic permeability, and its surface and/or edge pattern, e.g. its edge knurling. Coin discriminators are commonly used in coin handling machines, such as coin counting machines, coin sorting machines, vending machines, gaming machines, etc. Examples of previously known coin handling machines are for instance disclosed in WO 97/07485 and WO 87/07742. There is an evolving problem in the art of how to quickly, reliably and robustly discriminate between genuine and reject coins. Increasingly there is a need to be able to do this in connection with a variety of different coins of different denominations and different currencies. The present invention stems from some work involved in attempting to alleviate these problems. The term 'reject' is used herein to include any fake, bogus or unwanted coin (such as a foreign coin for example), and thus includes, but is not limited to, any counterfeit coin, slug, any damaged coin, or coin with a defect. Some high value items, such as bank notes and coins, may have material added to them that is used to confirm they are genuine. One such type of material is a taggant. Taggants generally comprise a chemical marker, but may comprise a physical marker. Taggants as described herein absorb light, or other electromagnetic radiation, and then re-emit electromagnetic radiation. The skilled person would understand that other taggant types may be used. Taggants are usually added to coins during manufacturing of the coins, and are generally invisible to the human eye.

The main type of taggant currently used in the art is called a down-converter. Down- converter taggants absorb relatively high energy photons, such as ultraviolet, and emit lower energy photons, such as visible light, or infra-red. Another type of taggant is an up-converter. Up-converter taggants are less efficient since they must absorb multiple relatively low energy photons to emit a single higher energy photon.

The most common down-converter is the phosphor coating the inside of a fluorescence strip light. This absorbs ultraviolet from the electrical discharge in the tube and converts the energy to visible light.

There is a time delay between absorbing the higher energy photon and emitting the lower energy photon. This effect is called fluorescence. The time delay can be relatively long, for example "glow in the dark" plastics and paints; or can be a tiny fraction of a second. The time taken for the emitted light to decrease by 50% is called the half-life. With taggants there is a big difference between the brightness of the incident light and the re-emitted light. This difference can be as large as a factor of lxlO ⁹ , making taggant detection difficult.

According to a first aspect of the invention, there is provided a method of sorting one or more coins comprising a taggant. The method comprises illuminating the or each coin using a light source, the light source being arranged to emit radiation of a wavelength which excites the taggant; and detecting radiation fluoresced by the taggant using a detector. Advantageously, a known taggant signature can be recognised and used to identify coins of a particular kind or denomination, and/or to identify reject items which are not coins.

Optionally, the method may include detecting at least one of:

i. dirt on the detector or light source;

ii. dirt between the light source and the coin; and

iii. dirt between the coin and the detector.

Advantageously, this may allow a user or operator to be alerted when cleaning is needed, and/or may trigger an automated cleaning process. Prompt removal of dirt may improve accuracy of the sorting method.

In examples in which dirt is detected, detecting dirt may comprise measuring light reflected from a window between the detector and the coin and/or the light source and the coin.

Optionally, the method may include calibrating the detection. The detection may be calibrated by:

providing a reference item having a taggant of a known level or profile;

processing the reference item to obtain a taggant level or profile signal at the detector; and

comparing the taggant level or profile signal against a known or assumed taggant level or profile signal associated with the reference item. Advantageously, calibration may improve accuracy of sorting and consistency between different sorting machines/between different implementations of the method.

In embodiments wherein a reference item is used for calibration, the taggant in the reference item may be protected by at least one of the following methods:

(i) being inserted into a pocket or indentation in the coin;

(ii) being contained in a layer within the coin; and

(iii) being enclosed within an outer rim of metal. Optionally, the light source may be arranged to be switched on and off, and the method may further comprise synchronising one or more light source switching times with detection times.

Advantageously, this may allow detection to be performed in lower light conditions due to the time difference between absorption and re -emission of photons by the taggant, so making detection easier due to reduced levels of background light.

According to a second aspect of the invention, there is provided a coin counting or processing machine arranged to sense the presence or identity of a taggant in a coin, the machine comprising a light source and a detector. The light source is arranged to emit radiation of a wavelength which excites the taggant and the detector is arranged to detect radiation of a wavelength which is fluoresced by the taggant.

Optionally, the coin counting or processing machine may comprise a detection filter arranged such that light must pass through the detection filter to reach the detector, the detection filter being arranged to only allow passage of radiation at or around the taggant' s fluorescence wavelength.

Advantageously, the detection filter may reduce the incidence of stray environmental light and/or other light not useful in the detection process on the detector.

Optionally, the coin counting or processing machine may comprise an excitation filter arranged such that light from the light source must pass through the excitation filter to reach a coin, the excitation filter being arranged to only allow passage of radiation at or around the taggant' s excitation wavelength. Advantageously, the excitation filter may reduce the incidence of light of wavelengths which are not useful in the excitation process on the coin, thereby reducing the amount of reflected light.

Optionally, the detector of the coin counting or processing machine may be arranged to detect at least one of:

i. dirt on the coin, detector or light source;

ii. dirt between the light source and the coin; and

iii. dirt between the coin and the detector.

Optionally, the light source and detector of the coin counting or processing machine may be arranged sequentially such that the emission and detection of radiation occur in different places as a coin moves through the machine.

Optionally, the light source may be arranged to be switched on and off, and wherein light source switching times and detection times are synchronised.

Optionally, the coin counting or processing machine may be arranged to allow detection of radiation to occur when the light source is switched off, such that the emission and detection of radiation occur at different times.

Optionally, the machine is provided with a reference item for machine calibration, the reference item having a taggant of a known level or profile. When a reference item is used, the taggant in the reference item may optionally be at least one of the following:

(i) inserted into a pocket or indentation in the coin;

(ii) contained in a layer within the coin; and

(iii) enclosed within an outer rim of metal.

Optionally, the detector may be arranged to determine a colour of the coin using light reflected from the coin. Features described in relation to one of the above aspects of the invention may be applied, mutatis mutandis, to the other aspect of the invention. Further, the features described may be applied to the or each aspect in any combination. There now follows, by way of example only, a detailed description of embodiments of the present invention with reference to the accompanying drawings in which:

Figure 1 shows a schematic view of a detection system according to the invention;

Figure 2 shows a schematic view of a second detection system according to the invention;

Figure 3 shows a schematic view of a third detection system according to the invention;

Figure 4 shows an illustrative example of the effect of ambient light on recorded taggant signal; Figure 5 shows circuitry used in an embodiment of the invention, including a secondary integrating feedback loop ;

Figure 6 show circuitry used in an embodiment of the invention, including a transimpedance amplifier;

Figure 7 shows an illustrative example of the correction of the effect of ambient light provided by the circuitry of Figure 6 ;

Figures 8a and 8b show a schematic view of light emittance and reflection and coin movement in an embodiment ;

Figures 9a and 9b show a schematic view of the effect of dirt on light reflection; Figure 10 shows a schematic view of a detection system according to the invention in which a reference sensor is used ;

Figures 11a to 11c show schematic views of three different coin constructions which may be used according to the invention;

Figures 12a and 12b show two examples of how a taggant profile might look for coins of two different constructions according to the invention; Figure 13 shows light-emitting diode (LED) temperature management according to the invention;

Figures 14 shows a schematic view of a detection system according to the invention in which filters are used; and

Figure 15 illustrates the use of multiple sensors and a synchronisation signal .

In the figures, like reference numerals are used to reference like components, with 806 and 1306 being the LEDs shown in Figures 8 and 13 respectively, for example.

In this invention, one or more techniques are used to aid detection. These include techniques such as:

1. Optical filtering;

2. Sensing at a different place ; and

3. Sensing at a different time.

Optical filtering uses the different wavelengths of the incident and emitted light to detect the small emitted signal. For example, if the incident light is ultraviolet and the emitted light is green, at the detector, a filter that blocks ultraviolet but allows green light through can be used. This technique works best when there is a big wavelength difference between the incident and emitted light.

Sensing in a different place relies on relative movement between the light source, detector and/or the tagged coin. Either the light source, the detector, the item being checked (e.g. a coin) or a combination of more than one of these might move. For example, glow in the dark materials are best seen by taking them from sunlight into a darkened room. Sensing at a different time relies on turning the light source on and off. When the incident light is turned off, the taggant signal decays at a rate in accordance with its half-life. Relative movement of the light source, detector and/or the tagged coin may or may not be used in addition to sensing at different times. A practical taggant sensor may use more than one of the above techniques.

Details of various methods for taggant signal analysis are discussed herein.

The invention disclosed herein is not to be limited by the precise method chosen for extracting a taggant signal.

Embodiments of the invention may use the same general technique of checking the decay of the taggant emission some time after the excitation has been removed. This may be measured at a fixed point in distance as the coin to be tested moves from the excitation area to the detection area at a pre -determined rate. Alternatively, it may be that the coin does not move and the excitation is de -energised before measuring a fixed time afterwards. In some cases, a combination of movement and de -energisation and re-energisation may be used within the same apparatus. In any case, the emissions from the taggant may be determined by a single absolute measurement, or a multitude of measurements to determine a decay or slope rate.

No matter how the or each measurement is obtained, measurements may be obtained more than once as a coin is moved relative to a sensor (or sensor moved relative to the coin). This allows a profile to be built up over time. The number of samples in the profile may depend on the taggant excitation and decay times, the sensor response time, relative movement and other factors in the detection system.

Taggant sensor implementations

A typical taggant sensing arrangement includes a light source and a detector. When looking for genuine coins containing a taggant with a known response (i.e. genuine coins that emit a particular response signal in response to excitation), the light source must emit light at a wavelength that is suitable to excite the taggant. For example, if it is known that a particular taggant that has been applied to a coin is excited by infra-red (IR) light, then a light source emitting relatively narrow band infra-red light can be used. In other embodiments, a wide band light source might be used, e.g. a white light source. In the former example (with a narrow IR source), a filter may not be needed (but could be used regardless to reduce stray light in the system).

In embodiments with one or more filters, an excitation filter may be used. An excitation filter is generally a high quality optical glass filter. Such filters are commonly used in fluorescence microscopy and spectroscopy for selection of the excitation wavelength of light from a light source. Most excitation filters select light of relatively short wavelengths from the light source, as only the shorter wavelength (higher frequency) radiation may have sufficient energy to cause the taggant to fluoresce. The skilled person would understand that the choice of filter, and of wavelength, depends on the taggant(s) to be detected.

In the example of using a wide band light source, a broader band source may be used in conjunction with a filter to excite a known taggant. The filter may reduce excess light noise, allowing through only light of around the wavelength required to make the taggant fluoresce.

The detector of this invention is suitable to detect electromagnetic radiation at a wavelength that is known to be emitted by the excited taggant. The detector may be a narrow band detector in some embodiments. In other embodiments, the detector might be a wider band detector and is used with a suitable filter, i.e. a filter that allows passage of radiation at the taggant emission wavelength, and preferably (substantially) not at other wavelengths.

The presence, or otherwise, of a taggant can be ascertained based on the absolute response values obtained at the detector. The detector signal can be processed such that, if it is above a threshold value, then it is determined that a taggant is present; and if it is below that threshold value then it is determined that a taggant is not present. Due to difficulties associated with taggant processing (discussed in more detail below), in some embodiments, there may also be a 'grey' or 'indeterminate' area, perhaps defined by two threshold values, between which there is uncertainty as to whether a coin contains a taggant or not. If a measurement is deemed to be indeterminate a coin may be rejected; or may be recirculated through the sensing system for repeat processing; or may be required to pass a higher level of test, or require a greater certainty of authentication test result in another aspect of its nature. For example, a genuine coin may also be known to be of a particular size or weight, or shape, or have certain acoustic response properties - one or more of these other properties may also be measured to test authenticity and the determined results for the specific coin may be required to meet higher thresholds for greater certainty if its taggant authentication is indeterminate. As would be understood by the skilled reader, in other embodiments, different combinations of light source, detector and appropriate filter arrangements can be selected according to specific excitation and response properties of a taggant.

In some embodiments of the invention, more than one taggant might be applied to a coin. Each taggant might have a different excitation and response profile. Different light sources may be provided to excite each taggant. Alternatively, a white light source may be provided to ensure that each taggant is activated. Multiple de tector arrangements may be provided to detect different frequencies. Alternatively, a multispectral photodiode (i.e. a photodiode array with selected band pass filters over each segment) may detect and be able to distinguish between radiation of different wavelengths. Alternatively a multispectral filter arrangement may be placed before a standard detector to detect a signature response from a known taggant combination.

In some embodiments, it is envisaged to identify a taggant by measuring its decay rate or half-life. The process for doing this is described in more detail below.

1. Optical filtering

In one embodiment, a taggant that produces a strong infra-red signal from short wavelength blue light could be sensed using a blue LED light source, and a photodiode enclosed in a plastic that blocks visible light. Figure 1 shows this arrangement, with suitable components. The LED (e.g. Optek Technology OVLGB0C6B9) produces a narrow beam of bright blue light. This shines on the taggant and causes it to emit infra-red. In this example, the detector is an Osram SFH213-FA photodiode. The Osram photodiode is not sensitive to blue light, but responds to the infra-red from the taggant.

The skilled person would understand that the wavelength of excitation or emission is taggant-dependent and could vary widely between different example implementation. Further, the choice of system components may depend on multiple factors including the type and amount of taggant used, the accuracy needed and cost. The example s listed above are for reference only and are in no way intended to be limiting to the scope of the invention. In this example, the narrow beam light source illuminates a first area of a coin, whose surface has been minted with a taggant. This causes the taggant to re -radiate infra-red in all directions. In order to detect a very small taggant signal, it is desirable to capture as much of this re-radiation as possible. This embodiment uses a large area detector mounted as close as possible to the illuminated area. In other embodiments a light collection mechanism (e.g. comprising a suitable arrangement of one or more lenses and one or more mirrors) may be used to gather and guide radiation to a detector located at a (more) remote location.

2. Sensing in a different place

If the taggant absorbs and emits radiation at similar wavelengths (e.g. infra -red), then it is difficult to filter out the incident light while allowing the emitted light to pass. A solution to this is to move the taggant from the light source to the detector in a time that is comparable to the half -life. Figure 2 schematically shows an emitter and receiver with a screen therebetween, so that no light can pass directly from one to the other. The taggant covered object is moving fast enough so that it is still "glowing" when it passes under the detector.

In a known sensing machine, in which there is bulk processing of coins for determining whether they are authentic or not, there is usually a well -defined path and speed of travel for a coin through a sensing region, and so the inventors have determined an optimum distance for a particular coin travelling through a particular machine such that the detector detects emitted radiation from the same region of the coin that was illuminated by the light source.

This type of sensor and the one detailed above in section 1 are the simplest and cheapest to implement in electronics. The light is on all the time and the taggant reading is just the current through the photodiode. Note, in other embodiment s, the photodiode may be replaced by a photo-transistor, which would provide a larger, but less repeatable, signal.

3. Sensing at a different time

If the taggant has the similar wavelength problem (i.e. it is excited at, and emits at a similar wavelength) and cannot be moved quickly compared to its half -life, then in some embodiments, this invention arranges to sense its emission by checking if it is still "glowing" after the light source has been switched off.

Figure 3 shows a light emitter which is switched on and off. Typically, the light source on-time must be, at least, a few (e.g. about 6, 7, 8, 9 or 10 in some examples) half-lives of the taggant. When the emitter is switched on, it will overload the photodiode amplifier. The amplifier is selected to recover quickly compared to the taggant half-life. After the recovery time the still glowing taggant can be detected.

The electronics for this type of sensor is more complex, because the light has to be switched at a rate defined by the half -life and the amplifier has to recover from an overload of several orders of magnitude. In some embodiments where there is more than one taggant to be sensed, then there will be more than one half -life to be taken into consideration. Usually in a sensing machine, in which there is bulk processing of coins for determining whether they are authentic or not, there will be one or more other sensors (as indicated above). According to this invention, in some embodiments, the light source switching time and sensor read time are synchronised, or one or both are coordinated with the operation of other sensors (e.g. inductive sensors) to reduce or avoid interference between different sensing systems. Ambient light

In some embodiments, a sensor according to this invention might have a problem with ambient light, e.g. sunlight and/or artificial light. In one embodiment of the invention, the inside of a machine that checks for taggant on currency, can be made dark. However, this may not be possible on a hand held detector, for example.

A known problem for taggant sensors in commercial applications is the lack of control over the environment. Even for embodiments wherein the sensing apparatus is designed to be kept in the dark, for example using a cover, there is no guarantee that the cover will be closed properly. As the machine requires an aperture into which to feed coins, it is also possible that ambient light can enter through that aperture.

In some embodiments, a method of removing the effects of ambient light before any measurements are taken is implemented.

In a particular embodiment, the ambient light compensation mechanism comprises a transimpedance amplifier (i.e. a current-to-voltage converter which can be used to amplify current output).

If the total contribution of light pollution falls within the sensing range of the transimpedance amplifier, it may be possible to simply take a base reading as the system starts to subtract from any subsequent measurements. An issue with taggant sensing is that the gain of the transimpedance amplifier is often so high that it does not take much stray light in order to saturate it, or at least make the baseline high enough that the measurement range is compromised. This can be seen in Figure 4, where the measurement with the higher ambient light shows a loss of measurement range due to the operational amplifier output reaching its positive output voltage limit. In other embodiments, other mechanisms for reducing the amount of ambient light reaching the detector may be provided. For example, in one embodiment, a coin discriminating machine comprises a belt used to convey coins through a sensing region in which the taggant detector is located. The taggant detector may be placed on an opposite side of the belt relative to incoming ambient light, e.g. the detector is located underneath the belt, such that ambient light from outside the machine is blocked from reaching the detector by the belt. In some such embodiments, an added advantageous effect of the same feature is that the light source may be mounted in close proximity to the detector (due to a requirement that a coin must travel from the source to the detector within a short time), e.g. on the same circuit board, and the light source is typically bright, so the light source is also located on an opposite side of the belt relative to incoming ambient light - this results in less light pollution from the bright light source leaving the machine. A light source used with this invention often needs to be very bright, and may be at such a level that light therefrom could cause discomfort or damage to a human eye. Instead, careful placement of the light source relative to the belt reduces stray light.

An ambient light compensation mechanism may be provided in some embodiments. The ambient light compensation mechanism may process detector readings to compensate for an ambient light level. The ambient light level may be determined externally and sent to the compensation mechanism in some embodiments. In other embodiments, the compensation mechanism includes an ambient light level detector. In some embodiments, the ambient light level detector monitors the ambient light level continuously or in real time. The compensation mechanism may then continuously process and adjust readings from the detector based on the ambient light level. This is a particularly useful feature in some embodiment as one machine according to this invention may be installed in very different conditions compared to another machine according to this invention. For example, a vending machine may be placed inside a dark, windowless room, or may be placed outdoors in direct sunlight. In some embodiments, instead of, or in addition to, measuring the absolute signal response level from the taggant, the sensor is arranged to look for, or measure, changes in the signal response level from the taggant, e.g. by measuring changes in the detector photodiode output produced by the taggant. Monitoring the change in the photodiode current rather than the absolute level can help to mitigate the effects of ambient light, as the decay rate (instead of the absolute response level) can then being used to identify the taggant. However, 100Hz flicker from artificial light is a problem for taggants whose half-lives last a similar time. In these cases keeping the sensor in the dark is the best approach. In embodiments where the decay rate is measured, there are various options for implementing the sensor circuitry. In one example embodiment, a light source is used to excite taggant on a first region of a coin. Subsequently, while the taggant is emitting radiation, the emitted radiation is measured at a first detector, and then further subsequently the emitted radiation is measured at a second detector. In this example, the detectors may be arranged along a path upon which the coin is arranged to travel in a predictable manner, i.e. such that the detectors are measuring the taggant response from the first region of the coin. In other embodiments the, or each of the light source, first detector and second detector may move such that predictable relative movement is provided between the sensor components and the coin being tested. In yet further embodiments, the same detector can be used to measure the emitted radiation at the first and second stages. In further embodiments one or more subsequent measurements can be made to obtain more detailed decay profile information.

In some embodiments, the ambient light detector may be a dedicated, distinct diode. In other embodiments it may be the same diode used to detect the taggant response.

In some embodiments, a secondary integrating feedback loop as shown in Figure 5 is used to reduce the contribution of circuit offset voltage errors and environmental factors. Whilst the switch 502 is closed, this secondary operational amplifier adjusts its output (which forms the offset voltage) and thus the output voltage of the main amplifier until it is stable at the reference voltage, which in this case is ground. When a measurement is need, the switch 502 is opened and the integrating capacitor 504 holds the error voltage constant, which is automatically added or subtracted from the measured value as appropriate. The error voltage can be positive or negative.

The skilled person would understand that the feedback loop shown in Figure 5 is not directly applicable to a circuit using a transimpedance amplifier. In a transimpedance amplifier, the operational amplifier provides an output current (its output voltage converted to a current by the feedback resistor Rl) to match th at of the photo-diode current to leave the net photo-diode voltage at zero. Thus the output voltage is directly proportional to the current generated by the photo -diode. Adding or subtracting an offset would change the biasing conditions of the diode and thus its operating characteristics. It would also still be limited in terms of adjustment range. One way to overcome this issue is to use the integrating error amplifier technique to generate a current to directly sum with the photo-diode current as shown in Figure 6. Using this configuration, the biasing conditions remain the same and only the additional photo current due to ambient light is supplied by the error amplifier. The circuitry shown in Figure 6 can therefore be used to reduce the effect of ambi ent light.

With the switch 602 closed, the photo-current created by the ambient light is turned to an error current by driving the gate of the transistor with the integrating amplifier 610. This in turn allows a current to flow into the photo -diode until it is large enough to reduce the output of the the transimpedance amplifier (formed by resistor Rl and the operational amplifier 620) to the reference voltage, which in this case is ground again.

Opening the switch 602 causes the integrator 610 to hold the error voltage as in the previous implementation, thus keeping the offset current constant for the measurement. Any additional light sensed by the photo -diode 606 acts to increase the current flow which causes the output voltage of the transimpedance ampl ifier 620, Rl to increase from the reference voltage. The skilled person would understand that the circuit 600 as presented in Figure 6 is used simply to illustrate the main components in such a system - the precise circuit shown may be prone to oscillations; changes to reduce the gain bandwidth and increase stability may therefore be made . Figure 7 shows the operation of the ambient light correction circuit of Figure 6 to address the problem demonstrated in Figure 4.

Another application of this circuit 600 is the removal of the effect of dark currents from a measurement. Dark currents can be thought of as the current that ' leaks' through the photo-diode 606 even when there is no light. This can become significant in very high gain applications where it appears as an offset in the output signal. In at least some embodiments, it also varies with temperature to a much greater degree than the photo-current. In one particular diode tested, this variation is approximately 40% / °C for the dark current versus 0.18% / °C for the photo -current. Dirt detection (on the light source or detector or both)

The inventors have realised that another sensor issue is dirt. Existing optical sensor designs have poor tolerance to the build-up of dirt. With a coin or other token, dirt can take many different forms. In a minting environment, it can be in the form of metallic dust or press lubricating fluids. In a commercial environment, it can be anything from finger grease to chewing gum.

This problem is particular to sensors for sensing taggant in coins because of the low signal response levels for some taggants. Due to the taggant response signal being small, it can be hidden by e.g. thermal and pick-up noise in the electronics. Dirt is less of an issue at the light source because the light source is generally bright and intended to saturate the taggant such that it builds up quickly/in a reasonable time to substantially its maximum response level. In order to keep the response signal above the noise level, it can be useful to maximise incident light falling on the target (e.g. tagganted coin) and the maximum light to be emitted.

If the detector or target (e.g. the tagganted coin) is partially or wholly covered by dirt, then the response signal is reduced. As previously stated, the response signal can be small, and thus difficult to measure. Dirt on the target is unpredictable (some coins are dirtier than others), and can set a virtual detection limit within the sensor system. Dirt on the emitter and detector is something that can be sensed and, optionally, a request for cleaning can be made.

In some embodiments, a light source dirt detector, arranged to detect dirt between the light source and the target, is provided. In other embodiments a response detector dirt detector, arranged to detect dirt between the target and the response detector, is provided. In other embodiments, both dirt detectors are provided.

In response to detecting dirt at one, the other or both detectors, specific actions may be performed. For example, an alarm may be provided to a manual operator of the machine suggesting that cleaning is required. The alarm may specify which component is dirty, and so indicate which component needs cleaning. Alternatively, the specific action comprises cleaning a component (e.g. a surface of the light source or a surface of the response detector, or an associated cover of either of those) that is determined to be dirty. Some of the dirt issues may therefore be correctable to a certain level by cleaning. The skilled person would understand that some dirt substances are easier to remove than others, with chewing gum, for example, often presenting problems.

In some embodiments, cleaning a component comprises, in response to detecting that component is dirty, any one or more of:

• dispensing a cleaner (e.g. a cleaning fluid) on to a target that is due to travel through a sensing region past the component, such that the cleaner is relayed to the component without otherwise interrupting operation of the machine - the cleaner may be dispensed according to the component to be cleaned or the amount of dirt detected (e.g. a different cleaner may be used for different components) ;

• blowing air (or other gas) at a dirty component.

In other embodiments, one or more cleaning actions may be carried out periodically or on an ad-hoc basis regardless of whether dirt is detected or not.

The problem with prior art designs is that, without visual inspection or measuring from a reference coin, the dirt problem may not be known about until coins with should be returning a taggant signal are rejected for not having one.

One issue is how best to detect the dirt contaminant. Dirt can affect a sensor system by:

• attenuating the excitation energy, thus reducing the response from the taggant;

· attenuating the response; or

• a combination of the two problems listed above.

There are several methods of detecting the presence of contamination disclosed herein; the skilled person would select an appropriate method for the system at hand.

Figures 8 A and 8B show a cross section of an LED 806 and photo -diode based sensor 812 during normal use. The LED 806 excites the taggant in the coin 820 and the response from the taggant is measured after the LED 806 is switched off. During this time, the coin has moved slightly. A dirt detection system arranged to detect dirt on the LED 806 or sensor 812 should be used when the coin is not over the sensor 808, ideally when the system is starting, or in a known gap between consecutive coins.

Reflected energy measurement

In a pulsed taggant sensor, a common window 814 may be used to allow the passage of light from the LED 806 to excite the taggant and emitted light from the taggant travelling to the sensor 812. This also means a constant signal path and the possibility of detecting the amount of energy reflected back from contamination on th e window 812.

In a first instance, as shown in Figure 9a, the sensor 812 is clean so most of the light from the LED 806 passes through the window 814 and only a little energy is reflected back to the sensor 812, which produces a current proportional to the reflected energy. This value is stored for later use.

In a second instance, as shown in Figure 9b, there is significant contamination on the window 814 which reflects more of the light from the LED 806 directly to the sensor 812. This increased reflection produces a bigger current. The initial 'clean' current value can be subtracted to determine the increased level of reflection thus determine the contamination level.

Once the contamination level is known, this can be compared to a fixed threshold value to trigger a warning to the operator that the sensor 812 and/or window 814 needs cleaning. If an internal cleaning mechanism as described above is available, it may trigger a clean operation. It may also be possible to provide some correction of measured taggant values to reduce the frequency of cleaning.

Those skilled in the art will be able to determine that whilst the presented dirt detection scheme can be implemented using the taggant sensor components, it may require reconfiguration of the taggant sensor circuit due to the LED 806 saturating the photo diode 812 in normal use (the LED being the light source in this example) . An example of a simple change to rectify this situation would be to reduce the current used to drive the LED 806, optionally using the circuitry already in place. This may be as simple as changing the reference voltage used for the current source/sink circuit.

Another method would be to reduce the gain in the photo -diode amplifier, which can be more difficult when using a transimpedance amplifier as is common with photo- diode applications. Any switching circuitry added to change gain resistor values can add leakage currents and distort the measurement of low signal levels, such as those exhibited by taggants. Changing the gain on a secondary amplifier may be easier, or it may be practical to measure the output of the transimpedance amplifier directly for the 'low gain' measurements. This assumes that the transimpedance amplifier is not saturated. The skilled person would understand that a combination of the two techniques described above could be used.

Reference sensor

Alternative or additional embodiments may involve the use of a second sensor 1022 installed opposite the main taggant sensor 1012, as is shown in Figure 10. The method of analysis is largely the same : observing how a known value changes over time to detect the presence of dirt. However, in these embodiments the transmission of light through any contamination is tested, rather than just a reflection. The skilled person would understand that this may be a more robust method in cases where contaminants may not just be solid matter which can reflect light. In at least some such embodiments, the second sensor 1022, which is used as a reference sensor, is set up differently from the main sensor; having a lower light output and a less sensitive photo-diode amplifier. An advantage of this method is that it can be applied to sensors where the light source is continuously on and the movement of the coin is used as the "shutter".

In some embodiments, the taggant sensing system includes a light source controller arranged to vary the intensity of light emitted by the light source. Optionally, the light source intensity is varied in response to dirt detection at the light source , such that if dirt is detected, then intensity of emitted light is increased. This increase may be a gradual increase or a stepped (e.g. one or more steps) increase.

In some embodiments, the taggant sensing system includes a response detector controller arranged to adjust the detected signal (e.g. by applying an offset thereto). Optionally, the response detector adjustment is varied in response to dirt detection at the response detector, such that if dirt is detected, then appropriate adjustment is applied. For example, the applied signal offset may be a gradual increase/decrease in response to dirt level, or a stepped (e.g. one or more steps) increase.

Therefore, it is possible in some embodiments to provide a two -stage, or multi-stage, dirt response mechanism - a first stage of dirt response is to adjust control settings (e.g. light source control settings) or signal processing settings (e.g. by adjusting the detected signal) or both to compensate for dirt, while a second stage of dirt response is to initiate at least one cleaning action.

In this document, where detecting dirt on/cleaning of a component is referred to, it will be appreciated that the same process can be carried out to achieve the same goal on a cover or lens of the component, e.g. a light source cover/lens, or a response detector cover/lens. The skilled person would appreciate that, in general, direct contact with components is best avoided to reduce disturbance or damage to the system and that, as such, customers may generally be advised only to clean a cover or lens, and not the underlying component. Taggant measurement

A range of applications exist for taggant detector machines. For example, a simple currency checker (e.g. a handheld machine) can provide a yes or no determination. Does this sample have taggant or not? Taggants can be made with a wide range of properties. A more sophisticated sensor might measure one or more of:

· incident wavelength for peak emission;

• wavelength of the emitted light;

• amplitude of the emitted light; and

• half-life of the emission decay. The first two of the above can be checked by changing the colour of the incident light and filters before the photo-detector. The amplitude measurement requires a reference light level against which the signal from the taggant may be compared. This calibration reference may be from a known light source or a target with a known level of taggant. In both cases dirt on the sensor or target should be removed.

The half-life is a simple time measurement - how long does it take for the amplitude to decay by a certain amount (e.g. drop by 50%)? The advantage of using the half -life as a security feature is that it is not affected by a small covering of dirt.

Taggant coverage

A target, such as a high value item, such as a coin, may not have a uniform covering of taggant, but have it concentrated in specific places. Thus, the location of the taggant is used as an additional or alternative security feature in some embodiments of this invention. In order to use this feature, the invention in one embodiment comprises making a number of taggant response measurements at different locations on the item. Using the different measurements, a taggant profile of taggant across a target is constructed. This might be a one dimensional profile (taggant level across e.g. a length, width or diameter of a target), a two dimensional profile (taggant level across a particular area of the target) or a three dimensional profile (a combination of the above). This can confirm that the taggant is present where it should be, or is not present where it should not be, or both. For example, we may see two taggant peaks with a gap in the middle in a one dimensional profile of a target coin in which taggant is concentrated at the rim of the coin.

In different embodiments, a profile is made up of two measurements. In other embodiments, the profile may be made up of any number of measurements. For example, a response signal may be measured about 10, 15 or 20 times across a single coin being tested. The same data may be used to generate different profiles. For example, after gathering 20 data values for a single coin, different ' dense' (e.g. made up of 20 data values) or 'sparse' (e.g. made up of 4 values) profiles may be generated, and used for different, or the same, purposes. For example, one type of profile (e.g. sparse) may be used for determining whether a coin passes an initial threshold authentication test, and if the result is inconclusive or indeterminate, then another type of profile (e.g. dense) may be used for confirming whether the coin passes a further threshold authentication test.

In embodiments, where more than one taggant response data value is measured, the mean, median, mode average, or any other measure of taggant level can be used in determining whether a target contains taggant. The measure of taggant level may be compared to a reference level, e.g. a reference level that is stored in a computer memory, to determine whether a target contains taggant, or a combination of taggants, and is genuine.

In some embodiments, the individual response profile of a particular taggant or combination of taggants across a particular coin may be considered to be the signature of that coin-taggant combination. The signature may be affected by, amongst other things, the material used to make the coin, the process used to make the coin, the process used to add the taggant to the coin, the taggant type, amount and concentration used, wear and tear on the coin.

Taggant averaging

The size of the taggant signal from something like a coin can be variable. This is due to the picture embossed on the coin, sometimes the surface is pointing towards the detector and sometimes away from it. The variability can be reduced by illuminating a large area and taking an average result, or by taking many readings as the item moves over the detector and averaging the readings. Mixed taggants

A pure sample of taggant will show an exponential decay of emitted light. This gives a specific half-life. It is possible to mix two, or more, taggants with different half-lives in the same sample/target. This might produce a different decay profile, such as a slower than exponential decay, giving the appearance of the half -life increasing as the decay progresses. By taking a number of samples of the decay waveform, and applying relevant mathematics (involving Fourier transforms as is well known in the art of signal processing), it is possible to work out a distribution of half -lives in the sample. This distribution of half -lives is used as another security feature in some embodiments of this invention. Colour sensing

Taggant sensing consists of illuminating a target with a particular wavelength light (i.e. particular colour, for visible light) and sensing returned light levels (reflected and/or emitted light) . Detecting different types of taggant can require different wavelengths of illumination. In some embodiments, if reflected levels are measured while different wavelengths are used to illuminate the target, then an approximation of the colour of the target can be deduced too.

In this way, colour and taggant sensors may be combined to increase the overall efficiency of a coin discriminator - the electronics required to reliably verify that a target, e.g. an item of currency is not counterfeit can be provided in a more condensed, and reliable form.

Calibration

The inventors have realised that a particular difficulty exists in calibrating multiple coin/token discriminators for sensing the same taggant in the same coin.

A taggant sensor installed in one machine may give different readings to another due to manufacturing tolerances in the sensor system. This could cause problems as one machine may repeatedly accept coins that another repeatedly rejects. In order that multiple machines can be configured to accept or reject coins based on taggant readings the ability to reliably set an acceptance threshold is desirable. The following approaches may be used individually or in any combination: 1. Absolute values from a sensor

A sensor can be adjusted during the manufacturing process to give the same output response as its predecessors, e.g. by using a reference material or coin to give a target for adjustment. However, this approach will not compensate for any influence the rest of the machine may have on the sensor' s operation through physical or electrical means. This approach may also be less effective where the taggant sensor is integrated into a larger sensor system or the sensor relies upon the movement of the coin. It also means that any changes to the sensor due to aging cannot be compensated for.

2. Teach by example This approach is common in coin sorting machines where examples of the desired coin set are run through the machine in a teaching mode. The machine software then generates a series of acceptance windows based upon the measured values. There can be a lot of variance in a coin set over its life (for example due to scratches, dints or rust) - the sample used for teaching should therefore include a range of coins of different ages, as teaching on a set of new coins only could lead to a lot of rejects of older but valid coins. This approach generally does not provide a way of directly comparing one machine to another. 3. Normalisation using a calibration coin

SCAN COIN machines differ from many other prior art machines in that they use a custom set of very well defined coins and a process of calibration , amongst other distinctions. As the characteristics of the coins are known, various points of interest can be picked and used as reference points in order to calculate calibration constants used to perform normalisation on other coins. In this way, readings from one machine can be directly compared with other machines of the same type, and often also with machines of other types. This approach also means that acceptance windows can be the same for all machines of the same type.

One current problem with using a taggant sensor in a machine with a normalised sensor system is that there is currently no method of producing a significant number of coins with a consistent taggant concentration to be used as a calibration reference. This means taggant sensor calibration must take place outside of the regular calibration procedure and involves significant human interaction.

Various methods which could be used to produce such a coin outside of the regular coin production process are described below . The skilled person would understand that coins produced using one or more of these methods may offer increased protection to the taggant contained therein, so increasing the durability of the taggant and so the reliability and longevity of the fluorescence signal generated from the coin.

3.1. Taggant material in pocket As the taggant concentration in metallic plating can be variable, it may be more economical to produce a coin with an indentation or pocket, which can then be back filled or printed with a taggant loaded paint/ink/polymer. Additionally or alternatively, a taggant loader ' sticker' could be inserted into such an indentation or pocket. An example of this is shown in Figure 1 1 A.

In at least some coins using this method of construction, the taggant loaded material is located away from the surface of the coin. It is therefore less likely to be subject to wear or contamination which would affect the value it gives. In some examples, a transparent, substantially or partially transparent or translucent cover may be bonded over the taggant material to provide greater protection.

3.2. Taggant material in moulded coin

An alternative construction shown in Figure 1 1B involves the encapsulation of a taggant carrier in an optically transparent , partially or substantially transparent or translucent medium. In the example shown in Figure 1 1B , the target area is recessed from the surface so as to protect the target area from excess scratching which may degrade the signal over time The 'coin' could be formed by encapsulating the taggant carrying material in a polymer such as polycarbonate, which is typically used in moulded optical components. In some examples, the material may be bonded between two pre-moulded halves. A toughened glass or the likes could be used instead of, or in addition to, a polymer.

An advantage to making a coin by this method is that, assuming perfect taggant distribution in the taggant carrier, the coin would automatically give the same taggant reading on both sides. Currently there is much greater control when it comes to using taggant loaded ink than in any comparable plating process , making the insertion of a separate taggant carrier more likely to give accurate and repeatable results. The skilled person would appreciate that plating processes may improve and that the invention is not to be limited to non-plated coins. 3.3. Taggant material in moulded coin with metal outer

One problem with a transparent coin is that it may be difficult for some machines to detect the presence of a coin. The inductive sensors of some embodiments may not work and the optical detection techniques employed in many embodiments typically rely upon light from a collimated source being blocked by the coin to create a shadow on a linear array sensor. These traditional techniques for identifying the presence of a coin are unlikely to work if the coin is transparent - the taggant material may well not attenuate the light sufficiently to trigger such optical detectors .

Constructing a coin using a metal outer section and an optically transparent taggant carrier in the centre, as shown in Figure 11C, is one possible approach for overcoming this issue. The use of a metal outer section may also make the coin more hard wearing. In some examples, the centre section may be inserted and coined using a modified bi- metallic production process, with the centre section (in this example, a disk) optionally being supplied with a protective layer on both sides to perform height matching with the outer section and to protect the surface. Alternatively, the centre could be bonded into the 'ring' of the outer section using a suitable adhesive, or a circlip or the likes could be used to fix the centre to the outer section (in this example, a ring).

The skilled person would understand that a coin may not be circular, and may instead be square, hexagonal, octagonal, or any other shape. The shapes of the centre and/or outer sections may vary accordingly.

Embodiments of the invention include a calibration method for calibrating a coin discriminator machine comprising providing a reference token/coin having taggant of a known level or profile, processing the reference coin in the machine such that the machine outputs a taggant level or profile signal, and calibrating the machine by comparing the taggant level or profile signal against a known or assumed taggant level or profile signal associated with the reference token/coin.

For example, the reference token may contain a known amount of taggant at a particular location (e.g. at a particular radius on a circular, or near-circular, or regular polygonal) coin. The reference token optionally comprises a polymeric or plastic material. Alternatively, the reference token comprises at least an outer polymeric or plastic material layer. The reference token optionally comprises a polymeric or plastic material surrounding a paper note, or part of a paper note. The inventors have realised that there are problems associated with using a metallic reference coin. Metallic reference coins are more prone to damage or wear and tear that affects the taggant level or profile, and so renders the reference coin ineffective or non -ideal for use as a reference coin for calibration purposes (since the mere act of processing the token/coin through a coin discriminator machine causes an undesirable level of wear and tear).

In some embodiments, the reference token comprises the desirable hard wearing properties (e.g. at least an outer polymeric or plastic material layer) only on one side, or on a portion of one side, since the reference coin can be carefully placed into a coin discriminator machine in a particular orientation (i.e. with the desired 'correct' side facing the detector).

Multi-point detection

Although some methods may always look for a taggant in the centre of the coin, the ability to examine multiple measurements for different areas of a single coin gives rise to some opportunities.

Figures 12A and 12B show two examples of how a taggant profile might look for coins of two different constructions where the taggant may be present acros s the entire surface of the coin (Figure 12A), or just present on the rim of the coin (Figure 12B). The skilled person would understand that many other constructions may be used, and may produce results different from these. In real-world examples, the profile may rarely or never be as consistent as those shown in Figures 12A and 12B , and may vary in absolute amplitude, or relatively across the surface of the coin. There may be instances where the taggant is present on the coin but its absolute position cannot be pre-determined. Techniques that can be employed to determine the presence of a taggant in these instances can be any one or more of the following, or the likes:

• detection of a single measurement above a pre-determined threshold anywhere on a coin surface;

· detection of a number of measurements greater than a pre -determined figure above a pre-determined threshold anywhere on a coin surface;

• detection of a number of sequential measurements greater than a pre determined figure above a pre-determined threshold anywhere on a coin surface; and/or

· detection of an average of a number of measurements with the result being above a pre-determined threshold.

Profile detection

In some instances, it may be desirable to look at the relative levels of tag gant response or the number of taggant features across a coin instead of an absolute value.

In such cases, the taggant would be validated based on the shape of the profile rather than an absolute measurement in any single point. For a relative measuremen t, measurements may still need to be over a set threshold in order to be considered. For example, if the difference between specific peak and trough values were to be considered, the measured trough value may need to be of a level that could easily be distinguished from background noise to be used.

Further, the difference between the trough and peak values would need to be enough to not be reasonably attributable to noise to be used.

Averaging from distance

Some coin detection systems are restricted in the sample time or number of samples that can be taken as a coin passes. In these instances, it may be more practical to take an average of the taggant response in a larger area as the coin passes. This can be done by, for example, using a lens system to match the size of the desired sample area to the detector, or the detector can be mounted further from the coin. The latter technique can be at a disadvantage as the detectable signal drops by a cubic factor as the detector is moved further from the coin. However this techniqure may be the cheapest, depending upon packaging constraints. Initial experiments show that there are photo diodes available with wide viewing angles which would not require the detector to be moved too far from the coin surface to capture a much greater area. This can also give greater freedom in the positioning of a sensor system in a machine. Compensating for factors other than dirt

In some embodiments, the sensor circuitry may be modified to compensate for factors such a temperature, or humidity. This is to take in to account that different machines according to this invention might be installed in quite different conditions, e.g. a parking meter, which might be only rarely/never serviced, could be installed in different locations ranging from about -40 to +50 degrees C. The different conditions might affect operation of the light source and/or detector.

In embodiments for which there may be significant variation in operating temperature, not only should the chosen taggant sensor(s) be able to cope with the expected temperature range, but any temperature drift of the components used should also be considered.

For example, as the temperature of an LED increases , the amount of light it emits generally reduces significantly. At 60°C some power LEDs can only give out 50% of the luminous flux emitted at 25°C. This has a significant impact on the energy returned by the taggant

Further, the dominant wavelength of an LED may vary with temperature. For most currently known LEDs, the variation can be up to 4nm over the range 25°C to 80°C. Even a relatively small percentage change in wavelength can have a marked effect, especially for any taggant which is highly selective in the excitation wavelen gths.

Photo-diode sensitivity is another factor that may change with temperature. Typical values range from 0.1 to 0.2% per 1°C temperature change. One or more of these temperature variations, or the likes, can have a significant effect on the taggant emissions and/or on sensor data collected.

In many embodiments, the temperature of the sensor may be measured and a correction factor applied as appropriate. A correction factor could be applied through circuitry or in software.

In some embodiments, a correction factor may not be sufficient. In such embodiments, a method of controlling the temperature of the sensor components may be applied so that the variations in temperature are reduced or do not occur. Temperature control methods may be applied instead of, or in addition to, correction factor methods.

In some embodiments, a heater, a cooler, or both are provided to heat/cool components of the sensor. This heating or cooling may be in response to a determination (e.g. made by a thermistor, which forms part of the sensor) that heating or cooling is required.

In the embodiment shown in Figure 13, a heater (heating element 1302) and a temperature sensor 1304 are installed in close proximity to the LED 1306. The temperature sensor's output and a closed loop system are used to control the power into the heater 1302. When the system heats up through regular use, less power is applied to the heater 1302 to keep the temperature constant, substantially constant, or at least within a specified range.

The heater 1302 could be a commercially available ceramic element added to the back of the printed circuit board (PCB) 1308 during assembly, a bank of resistors or even a copper heating track formed in one or more of the copper layers of the PCB 1308. The thermally conductive area 1310 could consist of an array of vias in the PCB 1308, a metal slug insert, or the PCB 1308 could be based on an aluminium core or the likes.

Various modifications may be made to this invention without departing from its scope. For example, different detector and light source arrangements will be apparent to the skilled person following the teaching provided by this document. It will be apparent that the disclosed sensing system can be tuned to work with taggants, or c ombinations of taggants, that have a known, or determinable, response profile.

Filters used to customise a generic taggant sensor

Different taggants may be excited with different wavelengths and produce a response in different wavelengths, or in multiple wavelengths. The excitation wavelength(s) can be selected to be appropriate for the taggant to be detected and still produced in quantities that make it economically viable. In some embodiments, it may be acceptable simply to confirm that there is a taggant present with a response within the bandwidth of a typical photo -diode. However, in other embodiments there may be a need to be more precise. In such embodiments, optical band pass filters can be used to ensure that only the signal of interest is measured.

In embodiments wherein there is a need for filtering of the received signal, a wide band light source fitted with a band pass filter according to need can be used. In this way, a generic sensor can be produced for low to medium volume applications in quantities the same as existing coin detecting sensors. Customisation can be carried out at assembly time. An example of use of such filters is shown in Figure 14.

One known wide band light source is a white LED. Most white LEDs do not provide the perfect white light spectrum, but would generally be good enough for these purposes, as, in most embodiments, only a relatively small portion of the spectrum would be selected to excite the taggant.

One area of caution when using fluorescent light sources (currently, white LEDs are generally blue LEDs exciting a yellow phosphor) is that they do not turn off as quickly as a "plain" LED (i.e. an LED with no excited phosphor) . Typical switching times are around 1 to 50 μβ, so should not present a significant problem with taggants having half-lives of hundreds of μδ. A shutter covering the light source may be used instead of or in addition to a switch to achieve the same effect.

Integration with other sensors In the coin sorting machine using the sensor of many embodiments, a taggant sensor cannot be expected to work in isolation of other elements either in the complete coin detection system, or in the wider machine. Unfortunately, the return energy given off by the taggant is generally small and requires significant amplification. This makes such systems prone to noise picked up, for example through either magnetic or capacitive means via the large area photo -diodes typically employed. There are instances where the taggant sensor can be modified to reduce the effect of such interference; for example, changing the type or increasing the bandwidth of the operational amplifiers in the photo-diode circuit to recover from known interference in less time. However, the noise may not be able to be removed completely.

In such cases the signal from a taggant sensor alone may look noisy. Use of a control system which can control the sample time and period can reduce or remove noise issues, as described below with respect to Figure 15.

In the embodiment of Figure 15, multiple inductive sensors are operating in a sensor system as shown by the inductive sensor trigger signals. Interference can be seen on the taggant sensor output. In this embodiment, a synchronisation signal can also be seen. The skilled person would understand that this signal could also exist virtually inside a microcontroller.

The sync signal denotes when the microcontroller should take measurements from the taggant sensor. It is set to occur in the sensor driving sequence to give the maximum time for the taggant sensor to recover from the effects of noise, which can include damped oscillations which continue after any external noise energy has gone. This sequence may be repeated more than once as a coin passes.