DEWHIRST, Stephen (9 Maypole Road, Broseley Wood, Shropshire TF12 5QH, GB)
HOWELLS, Geoffrey (The Old School House, Richards CastleLudlow, Shropshire SY8 4EQ, GB)
DEWHIRST, Stephen (9 Maypole Road, Broseley Wood, Shropshire TF12 5QH, GB)
| CLAIMS
1. A coin discriminator adapted to measure the diameters of a predetermined range of coins (11) and comprising a coin conveying means (30, 31) which defines a coin path (9; 34) along which coins in single file and orientated substantially in a common plane are caused to pass, in use of the discriminator, a linear optical sensor array (12) positioned on one side of the coin path and extending transversely of the direction of coin movement along the path, and illumination means (35) positioned on the other side of the coin path and opposed to the optical sensor array (12) , and wherein the length of the optical sensor array (12) and the position of the sensor array relative to the coin path (9; 34) are such that for at least the largest diameter coins of said predetermined range the coins will obscure the entire sensor array and some will brush thereagainst.
2. A coin discriminator adapted to measure the diameters of a predetermined range of coins comprising a rotatable coin gripping means (30, 31) which grips one edge (33) of a coin (11) and which carries gripped coins in single file between a linear optical sensor array (12) and an opposing illumination means (35) , the sensor array extending radially with respect to the rotational axis of the coin gripping means, processing means (36) responsive to the output of the sensor array (12) and configured to measure that angle (A) through which the coin gripping means has turned during which the coin blocks at least one pixel of the sensor array, and which measures that further angle (B) through which the coin gripping means turns during which the coin completely covers the sensor, and which calculates the coin diameter (25) from the first and further angles.
3. A coin discriminator adapted to measure the diameters of a predetermined range of coins and comprising a rotatable coin gripping means (30, 31) which grips one edge (33) of a coin (12) and which carries gripped coins in single file between a linear optical sensor array and an opposing illumination means (35) , the sensor array extending substantially radially with respect to the rotational axis of the coin gripping means, processing means (36) responsive to the output of the sensor array and configured to measure the angle (A) through which the coin gripping means has turned during which the coin blocks at least one pixel of the sensor array, and which measures the radial location of the pixels that are first covered and/or uncovered by the coin, and which computes the diameter (25) of the coin from those measurements. |
COIN DISCRIMINATOR
Field of the Invention
This invention relates to a coin discriminator and particularly to a coin diameter sensor.
The term 'coin' is used herein to include any type of disc, such as a token, a counterfeit coin, a component of a composite coin, or a washer.
Background to the Invention
Existing methods of optically measuring a coin' s diameter have been described in the following patent specifications:
US 5,542,520 Beisel et al.
US 5,076,414 Kimoto (Laurel)
US 6,552,781 Redeker (Zimmermann)
US 6,729,461 Brandle et al
In essence, they consist of a linear optical sensor, 12, as shown in Figure 1 , which is partly covered by a coin, 11 , as it moves along a datum edge, 10 to follow coin path 9. The optical sensor may be digital consisting of a line of pixels which are either in the coin's shadow or illuminated, or the optical sensor could be analog, in which the output voltage measures how much of the sensor has light shining on it.
The present invention is concerned with two problems which we have encountered with this type of coin diameter measurement:
1. When coins move along the datum, they push dirt and coin dust out of the way. This produces a "high tide" mark above the datum at the diameter of the largest coin. This high tide mark blocks light from reaching the sensor, causing an error in diameter measurement and/or requiring the sensor to be cleaned.
2. The coin my not be exactly touching the datum. If the coin were 0.5mm above the datum, it could appear 0.5mm bigger in diameter.
Summaries of the Invention
According to one aspect of the invention we provide a coin discriminator adapted to measure the diameters of a predetermined range of coins and comprising a coin conveying means which defines a coin path along which coins in single file and orientated substantially in a common plane are caused to pass, in use of the discriminator, a linear optical sensor array positioned on one side of the coin path and extending transversely of the direction of coin movement along the path, and illumination means positioned on the other side of the coin path and opposed to the optical sensor array, and wherein the length of the optical sensor array and the position of the sensor array relative to the coin path is such that for at least the largest diameter coins of said predetermined range the coins will obscure the entire sensor array and some will brush thereagainst.
The largest coins will thereby provide a cleaning action over the full length of the sensor array.
The coin conveying means may be an active or a passive coin conveyor.
A passive coin conveyor may simply be a chute defined by opposed walls through which the coin passes under gravity. Many coin discriminators are, however, designed to be capable of handling a large number of coins
in a rapid manner, in which case the coins are driven along the coin path, either by a drive belt or held by- a rotating coin conveying disc, for example as described in patent specifications WO 99/33030 and WO 2007/031770 of Scan Coin.
According to a second aspect of the invention we provide a coin discriminator adapted to measure the diameters of a predetermined range of coins comprising a rotatable coin gripping means which grips one edge of a coin and which carries gripped coins in single file between a linear optical sensor array and an opposing illumination means, the sensor array extending radially with respect to the rotational axis of the coin gripping means, processing means responsive to the output of the sensor array and configured to measure that angle through which the coin gripping means has turned during which the coin blocks at least one pixel of the sensor array, and which measures that further angle through which the coin gripping means turns during which the coin completely covers the sensor, and which calculates the coin diameter from the first and further angles.
According to a third aspect of the invention we provide a coin discriminator adapted to measure the diameters of a predetermined range of coins and comprising a rotatable coin gripping means which grips one edge of a coin and which carries gripped coins in single file between a linear optical sensor array and an opposing illumination means, the sensor array extending substantially radially with respect to the rotational axis of the coin gripping means, processing means responsive to the output of the sensor array and configured to measure the angle through which the coin gripping means has turned during which the coin blocks at least one pixel of the sensor array, and which measures the radial location of the pixels that are first covered and/or uncovered by the coin, and which computes the diameter of the coin from those measurements.
Some embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings in which:
Figure 1 shows a prior art arrangement in which coins pass along a datum edge, but in which the largest diameter coins do not completely cover the sensor array;
Figure 2a shows the sensor re-positioned in accordance with the invention so as to be covered entirely by the larger coins;
Figure 2b shows the coin in the two positions in which it first affects the sensor, and when it last affects the sensor;
Figure 3a shows three successive serial outputs of the sensor, in terms of pixels obscured, as determined by a coin moving past the sensor array;
Figure 3b shows three successive serial outputs to those of Figure 3a, for a coin of the same diameter as in Figure 3a, but which has a slightly different timing relative to the timing of the serial outputs of the sensor array;
Figure 4 shows how two coins of different diameter, but positioned differently with respect to an arcuate datum, could give rise to sensor outputs indicating the same time between first encounter and leaving the sensor;
Figure 5 shows the angle A over which the coins affects at least one pixel of the sensor, and angle B over which all of the sensor pixels are covered;
Figures 6 to 10 show geometric parameters used to calculate the coin diameter;
Figures 11a and Hb show plots of the sensor output for a Danish IKr and 2Kr coins respectively;
Figure 12 is a schematic partial perspective view sectioned on a radius of a coin conveying disc assembly showing a coin gripped by the disc margin; and
Figure 13 is a schematic radial cross-section on an enlarged scale showing a coin being carried through a coin discriminator in accordance with the invention, fitted to the assembly of Figure 12.
With reference to Figure 2, the method described here uses the same type of optical sensor as used in the prior art of Figure 1 , but placed closer to the datum edge ie in this arrangement the larger coins of the set of predetermined coins to be discriminated completely cover the sensor sweeping it clear of dirt and coin dust. However, this creates the obvious problem that we do not see the top of the coin. To overcome the problem, we need to have information about how the coin is moving.
If the movement information were very high resolution, determining coin diameter would be trivial. The diameter is the amount of movement between positions 13 and 14 of the coin. In practice the movement data does not usually have the required resolution. There are two reasons for this:
1. The optical sensor does not have infinite bandwidth. In a digital sensor, this limit is the time taken to serially clock the light levels for each pixel out of the sensor. This limits the measurement rate
to a few thousand per second. Of course, if the coin were moving slow enough, this would give the required resolution. However commercial coin counting machines require a higher through-put than that.
2. In a coin counting machine an electric motor transports the coins via a belt or disk system. The amount of movement is determined by counting pulses from an encoder, or the number of steps, if a stepper motor were used. In both cases, practical limits restrict the resolution that can be obtained for a given through put of coins per minute.
In an optimised system the above two limits are combined so that for each pulse from the encoder, one measurement is clocked from the optical sensor. The clock rate must be near the maximum in order to extract the reading before the next encoder pulse.
With a digital sensor, the movement resolution can be increased by derivation from measurements of the lengths of chords on the coin. For example, with reference to Figure 3a the first non-zero set of pixel readings shown to the left has 12 black pixels followed by 18 in the second sample corresponding to two chords on the coin. These measurements can be taken on the leading (or trailing) portion of the coin where the chord is less than a diameter, and can be measured by a sensor array that is shorter than the full diameter of the coin.
Compare this to the case in Figure 3b where the first non-zero sample has 4 black pixels followed by 16 in the second sample.
By doing some mathematics, it is possible to use these numbers to give an estimate of coin diameter that has a higher resolution than the horizontal spacing between measurements.
For example, in Figure 3a, it is possible to calculate the position/timing of the leading edge Z of the coin, by knowing the lengths of the two lines of obscured pixels that correspond to two chords of the coin and the speed of the coin. These two chords mathematically define a unique position for point Z. This process of calculating the position/ timing of point Z can be considered to be a derivation or extrapolation from the measurements represented by the two chords.
We now need to distinguish between coin sorters using belts and those using disks. In a belt sorter, a belt moves the coin through the sensor in a straight line. In a disk sorter, such as that of previously mentioned WO 99/33030 the coin is held near the edge of a circular disk. Thus the coin moves through the sensor on a circumference of a circle.
In a belt sorter, the diameter measurement method described above not only solves the dirt problem; it also gives the correct answer if the coin is off the datum edge. A coin 0.5mm away from the datum does not appear 0.5mm bigger. Additionally, once the diameter is measured and knowing which pixels were blocked first, or last, the distance off the datum edge can be calculated. The above description provides all the principles needed to implement this idea on a belt sorter.
On a disk sorter, such as that shown in Figure 12 and described hereafter, the problem is more difficult; the datum edge is usually part of a circle. What the encoder pulses measure is not the distance moved by the coin, but the angle rotated by the disk. Because of this, the method described does not automatically solve the coin off datum problem.
The sketch of Figure 4 shows why. The datum edge, 20, is part of a circle. The coin blocks some of the pixels while the disk D rotates over the angle A. This can occur either with a small coin, 21, on the datum or with a larger coin, 22, that is off the datum.
With a digital sensor, the simplest solution to this problem is to determine which pixels were blocked first or unblocked last. In Figure 4 the first pixels blocked by coin 21 are closer to the centre of the disk than those blocked by coin 22. Knowing angle A and the position of the first/last pixels covered, it is possible to find the coin diameter and the distance from the datum by solving two simultaneous non-linear equations.
An alternative way of finding the coin diameter and distance from the datum edge is by measuring two angles, as shown in Figure 5.
Angle, A, is the angle for which any of the pixels are covered. Angle, B, is the angle for which all the pixels are covered.
Knowing the two angles A and B and the position of the top of the optical sensor from the centre of the disk, it is possible to calculate the coin diameter and distance from the datum.
PREFERRED EMBODIMENT
The preferred implementation is the two-angle method. This is because of practical problems with digital linear sensors and the light sources used to illuminate them. It is difficult to create a light source that is the same brightness over the entire length of the sensor and some pixels are more sensitive to light than others. This sensitivity problem becomes worse as the sensor becomes covered in dirt and dust. The result is that
the pixels which "go black" first are the least sensitive ones rather than the ones most covered by the coin. This can produce significant errors in determining which pixel was blocked first and last.
The mathematics used with the two-angle method to calculate the diameter is as follows:
With reference to Figure 6, the angle, A, is the angle rotated by the disk between covering the first pixel and uncovering the last pixel. Using the sine formula for a right-angled triangle gives:
Sin{AI2) = r k + e + r
Rearranging gives:
r k + e + r = (1)
Sin(A/2)
Where r = the radius of the coin ie half the diameter
k - Distance from the centre of the disk to the datum edge
e = Distance off the datum edge (should be zero)
With reference to Figure 7, the angle, B, is when all pixels are covered. Using the COSINE rule we have:
{2yf = 2(k + g + s) 2 (\ - Co S (B))
hence
Where: 5 = Length of the sensor
g = Gap between datum and the start of the sensor.
With reference to Figure 8, the half the angle of B gives:
C / -os / (-B D 1 / I 2 \ ) = k + e + r +x k + g + s
Rearranging gives
(k + g +s)Cos(B /2) = k + e + r + x
Using equation (1) to substitute for k + e + r gives
(k + g + s )Cos{BI2) = —-—- + x
Sm(A 12)
and hence
x 2
(3)
From Pythagoras r 2 = x 2 + y 2
Using equations (2) and (3) to substitute for x 2 and j> 2 gives
{k + g (1 - Cos(B))
(4)
To make this equation easier to handle, let us substitute:
C = (k + g +s)Cos(B/2)
E = Sin(A 12)
Substituting C, D and E into equation (4) gives
Multiplying out and rearranging gives
C' -^r + ^ + D -r 2 = 0 E E 2
and hence the quadratic
This can be solved using the quadratic formula:
l - E 2
Where a r —
E
c = D + C 2
The quadratic formula gives two results for the value of coin radius, r, and hence the two diameters. This is because there are two coin diameters that will produce the angle A and block the top pixel of the optical sensor, as shown in Figure 9. In practice it is always the smaller diameter that we want. Which is given by the formula:
Diameter
Having now found the value of r (half the coin diameter) , we can now find the value of e (the distance off the datum edge) . Rearranging equation (1) we have:
e k — r
Sin(A/2)
Note it is possible for e to be slightly negative. This happens with deeply knurled coins when two tips of the knurl are on the datum.
Small Diameter Coins
With reference to Figure 10, when the coin's diameter is small enough, less than 22mm in one embodiment, the coin will not completely cover the optical sensor. This makes the angle, B, zero and the above formula
will not work. In this case we need to return to the original idea and determine how much of the optical sensor is in the shadow of the coin. If we combine this with the movement information, we can calculate the distance off the datum and the true diameter.
As before
Sin(A/2) = (5) k + e + r
and
e + 2r = g + x (6)
Where r = the radius of the coin ie half the diameter
k = Distance from the centre of the disk to the datum edge
e = Distance off the datum edge (should be zero)
g = Gap between datum and the start of the sensor.
x = The maximum length of the sensor covered by the coin.
From equation (5)
e = r -k (7)
Sin{AI2)
From equation (6)
e = g + x - 2r (8)
Combining equations (7) and (8) and adding 2r+ k
r
+ r - g + x +k
Sin(A /2)
Hence
or
Diameter + x + k)
Given the diameter then from equation (8) the distance off the edge is given by
e = g + x — Diameter
Calibration
Some of the variables in the above calculations are measured for each coin of the predetermined set of coins, such as the angles and the distance along the sensor for a small coin. Other variables such as, k, the radius of the datum edge and g, the distance from the datum edge to the bottom of the sensor are held as constants. These constants need to be known with great precision in order to measure the diameter accurately in millimetres.
Rather then trying to manufacture the coin machine with the value of say, g, fixed to better than 0.1mm it is easier to manufacture the machine
with normal engineering tolerances and then calibrate it using close tolerance test coins.
We need at least as many different diameter calibration coins as we have unknown constants in the above equations. The accurately known diameters of the calibration coins are used with the above formulas to find the values of k, g and the other constants
Coins with holes
Some countries have coins with holes in the middle, for example Denmark. Figures 11a and lib show plots of the sensor output for Danish 1 and 2 Kr coins passing over an optical sensor.
The coin of Figure 11a is the Danish IKr. It is 20.2mm diameter and does not completely cover the sensor. The diameter of this coin is converted into millimetres using the small diameter method on the previous page. The coin of Figure l ib is the 2Kr. Its diameter of 24.4mm does completely cover the sensor. The diameter of this coin is calculated using the two-angle method described above.
The coins do not appear round in Figures 11a and l ib. This is because the vertical and horizontal resolutions are different. The vertical resolution is determined by the pixel spacing on the optical sensor. Whereas the horizontal spacing is determined by how often the encoder takes a reading from the optical sensor.
The sizes of the holes is found by counting the maximum number of white pixels in the middle of the coin and multiplying this count by the pixel spacing. This gives the hole size in millimetres and can be used to check that the hole is present and the correct size.
Disc sorter example
Figure 12 is provided to show one prior art example only of a disc sorter to which the invention may be applied. The Figure corresponds to Figure 4 of Patent Specification No. WO 99/33030 (PCT/SE98/02406) . In that sorter a single file of coins 11 is created, and the coins 11 are gripped at one edge 33 between the radially outer margin of a rotating main horizontal disc 30 and an annular resilient strip 31 carried by an annular strip carrier 32 that is fixed relative to disc 30, to be carried on an arcuate coin path 34 about the disc axis. In one arcuate portion of the path of the disc periphery the gripped coins are taken past a stationary coin discriminator, not shown.
Prior to the coins reaching the angular position of the coins 11 shown in Figure 12, the coins have been urged against an arcuate datum edge to position the coin edge 33 radially of the disc axis, as described in WO 99/33030.
Figure 13 shows the sensor 12 of the invention positioned below the arcuate path of coins 11 and shows the light source 35. Some coins will brush over the upper face of the sensor 12. The light source 35 is arranged to produce a collimated beam of light, for example by using an LED source and suitable lens.
The sensor array 12 outputs signals to the processing means 36 configured to perform the algorithms previously discussed.
Next Patent: EXTRACTION AND PURIFICATION OF BIOLOGICAL CELLS USING ULTRASOUND