GB2102567A - Photometer - Google Patents

Abstract

A double-beam photometer includes a single photometric detector which is subject to inertia. A beam switching device causes sample and reference light beams to be alternately directed upon the detector with intervening dark intervals. The theoretical square waveform is distorted as the result of the inertia of the detector to give the illustrated waveform. Thus, during each reference and sample interval the detector signal 48, 49 approaches exponentially a final value 40, 50 respectively and during the subsequent dark interval decreases again exponentially along the line 44, 45 towards the theoretical dark value 46. In processing these signals the value of the signal during each reference and sample interval has substracted from it the value of the signal from the subsequent dark interval, the signals being measured at a specific moment in each interval, i.e. tv2, t'v2, tp2 and t'p2 respectively. It is shown mathematically that, in this way, an exact compensation of the dark current is achieved in spite of the inertia of the detector. <IMAGE>

Description

SPECIFICATION Photometer This invention relates to double-beam photometers of the type having components defining sample and reference paths of rays and including a single photometric detector and a beam switching device by means of which a light beam from the sample path of rays and a light beam from the reference path of rays are directed alternately upon the detector.

In photometers operating in the infra-red range, especially in spectral photometers, photometric detectors are used which are subject to inertia. A sudden increase of the light flux impinging upon the detector is not followed by a sudden increase of the output signal, but by an output signal which approaches a new output signal exponentially and asymptotically and with a time constant. If the light flux impinging upon the detector suddenly decreases, this does not lead to correspondingly sudden decrease of the output signal but again to an exponential decrease approaching a new stationary value with a time constant. Such detectors moreover show a dark current. In other words, they supply a finite output signal even if no radiation from the sample or the reference light beam impinges upon the detector.

To obtain a "base line" for the signal amplitude during signal processing, when using a detector subject to inertia, the sample and the reference paths of rays are interrupted by the beam switching device during first dark intervals between the sample intervals and the reference intervals, each of these first dark intervals therefore following a sample interval. In the same way, the sample and the reference paths of rays are interrupted by means of the beam switching device during second dark intervals between the reference intervals and the sample intervals, each of which second dark intervals therefore follows a reference interval.

Usually, the signals of the detector are integrated with respect to time over the different intervals. To eliminate the dark current, the signals obtained in the dark intervals are subtracted from the "measured signals" obtained in the sample and the reference paths of rays. As the dark current may be subjected to variations an average value is formed over a number of dark intervals. This average value represents the "base line" for the measured signals, with respect to which base line the amplitudes of the measured signals are measured for the signal processing.

Due to the inertia of the detector, the signal amplitude of the detector depends not only on the radiation flux impinging upon the detector during the corresponding sample, reference or dark interval, but theoretically also on the radiation which has impinged upon the detector during all previous intervals and the signal components of which decay exponentially. In practice, however, only those signals have to be taken into account during each sample interval, reference interval or dark interval, which have been obtained during the preceding interval. It may be estimated that the influence of a dark interval on the following sample interval or reference interval is negligible. However, the influence of the radiation flux in the sample or reference intervals on the base line determined by the dark intervals is by no means negligible.If this base line or reference depends, for example, on the transmission of the sample and varies therewith, this results in an error of the measured differences of signal amplitudes and reference.

It is the object of the invention, to design the signal processing circuit in a photometer of the type referred to in such a way that an exact compensation of the dark current independent of the radiation flux of the sample and the reference light beams is achieved.

According to the invention, this object is achieved by the inclusion of a signal processing circuit connected to the photometric detector and which comprises signal detecting means for detecting the detector signals during each sample interval, reference interval and first and second dark intervals and difference forming means for forming the difference of the signals generated by means of the signal detecting means during each sample interval and during the first dark interval following each sample interval and for forming the difference of the signals generated by means of the signal detecting means during each reference interval and during the second dark interval following each reference interval.

The invention is based on an appreciation of the fact that for compensation of the dark current, a distinction must be made between the first dark intervals following each sample interval and the second dark intervals following each reference interval. The differences between the signals generated thus need to be formed from signals from those dark intervals which follow the respective sample intervals and reference intervals concerned. It can be shown that, in this manner, an exact compensation of the dark currents may be achieved which is independent of the radiation fluxes.

If an average over several dark intervals is to be formed, this may be achieved if the signal processing circuit comprises means for forming a first average of the signal supplied by the signal detecting means over a number of first dark intervals and means for forming a second average of the signals supplied by the signal detecting means over a number of second dark intervals, and the difference forming means form the difference of the signals supplied by the signal detecting means during each sample interval and the first average and the difference of the signals supplied by the signal detecting means during each reference interval and the second average.

The signal detecting means may comprise means for detecting the instantaneous value of the detector signal at a predetermined moment of the sample interval, the reference interval and the first and second dark intervals respectively. They may, however, alternatively comprise RC-networks for pseudo-integration of the detector signal over predetermined integration intervals within the sample interval, the reference interval and the first and second dark intervals respectively.

A further possibility consists in the signal detecting means comprising integrators for integrating the detector signal over predetermined integration intervals within the sample interval, the reference interval and the first and second dark intervals respectively.

An example in accordance with the invention will now be described in greater detail with reference to the accompanying drawings, in which: Figure 1 shows a chopper disc for a beam switching device; Figure2 shows a corresponding signal processing circuit; Figure 3 shows a waveform of the radiation flux effective at the detector; and Figure 4shows the corresponding signal waveform.

The chopper disc 10 shown in Figure 1 comprises a reflecting sector 12, a transparent or cut-out sector 14 and, in between, two blackened sectors 16 and 18. A sheet metal projection 20 is provided on the chopper disc 10 in an angular area corresponding to, but somewhat smaller than that of the reflecting sector 12. Two light barriers 22 and 24 are disposed on diametrically opposite sides of the chopper disc 10, which light barriers are alternately interrupted by the sheet metal projection during rotation of the chopper disc 10.

The sheet metal projection 20 lies in the plane of the paper in Figure 1. In a second plane lying therebelow, a second sheet metal projection 21 is disposed in an angular area corresponding to, but somewhat smaller than that of the blackened sector 16. The sheet metal projection 21 interrupts two further light barriers 23 and 25 respectively, lying in the second plane below the light barriers 22 and 24.

The chopper disc 10 is arranged as a beam switching device in a known manner such that it alternately reflects a light beam 26 by the reflecting sector 12 into a sample path of rays or lets it pass through a transparent or cut out sector 14 into a reference path of rays. Between them, the light beam 26 is completely interrupted by each of the blackened sectors. The sample light beam directed along the sample path of rays and the reference light beam directed along the reference path of rays are super-imposed again by means of a beam combining means in known manner and alternately impinge upon a common photometric detector 28 which may be a lead sulphide detector.Assuming that the sample light beam passes through an absorbing sample and thereby is attenuated as compared to the reference light beam, then the waveform illustrated in Figure 3 results for the radiation flux B occurring at the detector 28.

During a sample interval 30, the radiation flux Bp from the sample path of rays impinges upon the detector 28. During a reference interval 32, the radiation flux Brfromthe reference path of rays impinges upon the detector 28. A first dark interval 34 lies between each sample interval 30 and reference interval 32, which dark interval thus follows sample interval 30 in time and is caused by the blackened sector 16 of the chopper disc.

A second dark interval 36 lies between each reference interval 32 and the subsequent sample interval 30, which second dark interval thus follows the reference interval 32 in time.

As can be seen from Figure 3, the reference interval 32 lasts from tvo to t'vo. The subsequent second dark interval 36 lasts from t'vo to tpo. The sample interval 30 lasts from tpo to t'po and the first dark interval 34 following sample interval 30 lasts from tipo to tvo of the next period. The sheet metal projection 20 interrupts the light barrier 22 during an interval portion from tp, to tp2 lying entirely within the sample interval 30. It interrupts the light barrier 24 during an interval portion from tV1 to tV2 lying entirely within the reference interval 32.The sheet metal projection 21 interrupts the light barrier 23 during an interval portion from t'p1 to t'p2 lying entirely within the first dark interval, following the sample interval 30. The sheet metal projection 21 interrupts the light barrier 25 during an interval portion from t'vi to t'V2 lying entirely within the second dark interval 36 following the reference interval 32.

The signal waveforms at the output 38 of the detector 28 do not correspond to the square wave signal illustrated in Figure 3 because of inertia of the detector 28. A signal is rather obtained as illustrated in Figure 4. At the end of the reference interval 32 and at the end of the sample interval, the detector signal approaches its respective final stationary value. In Figure 4, this final value is designated by the numeral 40 for the reference interval 32. At the beginning of the dark interval 36, the signal produced by the radiation flux B decays exponentially, as illustrated by the waveform portion 44 in Figure 4, towards the dark current i, illustrated by the horizontal dashed line 46. At the beginning of the sample interval 30, the radiation flux from the sample path of rays becomes effective at the detector 28.After the waveform portion 48 the signal increases asymptotically from the value at the beginning of the sample interval 30 to a final value illustrated by the dashed line 50 in Figure 4. In practice, the time constants Tf and Tr of the decrease of waveform portion 44 and of the increase of waveform portion 48 are different. It can be shown that the influence on the signal in the sample interval 30 of the preceding dark interval 36 and of the preceding reference interval 32 is negligible. The influence of the reference interval is not negligible, however. This influence is compensated for by the processing circuit controlled by the light barriers 22, 23, 24 and 25.

Figure 2 shows schematically one example ofthe signal processing circuit.

A detector signal is present at an output 52 of the detector 28. Let it be assumed that the detector current is transformed into a voltage proportional thereto in conventional manner. This output 52 is connected to a plate of a capacitor 56 through a resistor 54. The other plate of the capacitor 56 is, on the one hand, connected to the input 58 of an amplifier 60 and, on the other hand, to ground through a switch 62. The output of the amplifier 60 is applied to the input 66 of an amplifier 68 through a switch 64. Furthermore, the input 66 is connected to ground through a capacitor 70.

In the same way, the output 52 of the detector 28 is connected to a plate of a capacitor 74 through a resistor 72. The other plate of the capacitor 74 is, on the one hand, connected to the input 76 of an amplifier 78 and, on the other hand, to ground through a switch 80. The output of the amplifier 78 is applied to the input 84 of an amplifier 82 through a switch 82. Furthermore, the input 84 is connected to ground through a capacitor 88.

The outputs 90 and 92 of the amplifiers 68 and 86 respectively, are supplied to a quotient former 94 providing a transmission signal at an output 96.

The switches 62, 64 and 80, 82 are controlled by the light barriers 22, 23, 24 and 25 (Figure 1).

The control of the switches 62,64,80 and 82 by the light barriers 22,23,24 and 25 may be achieved in different manners as specified in the following tables.

Version 1

switch \ 62 64 80 82 interv; 62 64 80 82 tvi to tv2 closed open open open t'vi to tiv2 open closed open open tpi to tp2 open open closed open t'pitOt'pz open open open closed Version 2

switch \ 62 64 80 82 interval 62 64 80 82 tvi to tv2 open closed open open t'vi to tlv2 closed open open open tpi to tp2 open open open closed t'pi to t'p2 open open closed open The operation of the arrangement described will be explained below by means of theoretical considerations.

Figure 3 shows the waveform of the radiation B with respect to time, Figure 4 the associated waveform of the detector current i. In each interval it approaches the dashed equilibrium value in accordance with an exponential function with the time constants Tr during increase and Tf during decrease (tf < Tor). It is assumed that Bv and Bp are constant over several periods, a condition which in general is fulfilled at least reasonably well.

The reference interval starts at tVo and ends at t'vo. Accordingly, the beginning and end of the following dark interval are at t'vo and tpo, those of the sample interval at t'po and tVo.

Within each interval, two moments having the subscripts 1 and 2 are marked. The first moment having the subscript 1 is not considered at first. It is only taken into account for an integration or the use of an electric filter having its own time constant. At first it is assumed that the time constant of the signal processing is negligibly small as compared to the detector time constants Tr and Tf.

Each of the signals is detected at the second moment, i.e. the sample signal at tp2 and the subsequent dark signal att'p2.

In general, the detector current i in its equilibrium state (i.e. after a sufficiently long irradiation as compared to time constants Tr and If, respectively of the detector) is: (1) i=io+EB/ i, being the dark current and (2) #=#i #B being the sensitivity of the detector.

If, at a moment to, the radiation suddenly increases to a then constant value, 1a (initial value) is the detector current flowing at a moment to, and le (final value) is the stationary detector current, which would be flowing after a very long period of radiation, then the waveform of the detector current i with respect to time is given by (3) i=ie-(ie-ia)e-t-to/t This general equation is applied to each of the intervals of Figure 4.

In the reference interval, the detector current which is variable with respect to time is referred to as lv(t) - in contrast to the final value iv which is not achieved - and, correspondingly in the dark interval after the reference interval iv(t'), in the sample interval ip(t) and in the subsequent dark interval ip(t'). At the end of each interval the final value is reached except for a small amount Ai, which is always positive: thus at the end of an increase phase i (t) is smaller by Ai, at the end of a decrease phase, i (t) is larger by Al than the final value.

In general, Ai has different values in the four intervals. Ai at the end of the reference interval is referred to as Alv, at the end of the following dark interval as #i'v, at the end of the sample interval Aip, and at the end of the subsequent dark interval as Ai'p. These four values of A may be exactly calculated under the condition of a stationary equilibrium, i.e. when several cycles follow each other, always having the same radiation Bv in the reference intervals and the same radiation Bp in the sample intervals. For the following computations only the fact that A is small as compared to ie -- 1a is required.

Starting with the reference phase, the initial value of the detector current (at the moment tvo) is () 'a = e + the final value (not reached) is (5) 'a V The application of equation (3) now yields (6) iv(t)=iv-(iv-io-#i'p)e-t-tvo #r According to equation (1 ) a value proportional to the radiation is obtained, if the dark current ie is subtracted from the detector current (7) i-i0-B.

If accordingly the equilibrium value i, (not reached) of the dark current is subtracted in equation (6), then for t = tv2, at the moment at which the signal is processed: -tv2-tvo -tv2-tvo (8) iv(tv2)-io=iv-io-(iv-io)e + #i'pe #r #r is obtained.

This signal difference iv(tv2)-io is made up of three components, namely the ideal value iv-io, a first error component (i-iO)e - tv2 tvo Tr substantially indicating the amount by which the signal is still different from the final value, and a second error component tv2 - tvo #i'pe- , #r being a correction of the first error component.

In general, tV2-tVo > Tr Thus e-tv2-tvo #r becomes small as compared to 1, and the first error component becomes small as compared to the ideal value iv-io. Furthermore, Ai' is small as compared to iv-io. Thus, the second error component becomes small as compared to the first error component; it is a small fraction of a small error component and thus negligible. This has been tested quantitatively. A practical example having an experimentally measured time constant Tr of the detector and having a usual chopper frequency showed that the second error component was smaller by several orders of magnitude than the maximum error admitted of the instrument (and corresponding to the usual specifications).

The first error component, however, is not a true error falsifying the measuring result, as it is proportional to the ideal value. If the second error component is neglected, (8) may be written: (9) iv(tv2)-io = 1-e-tV2-tvo (iv-io).

#r The difference between the measured signal iv(tv2) and the ideal value i, is proportional to the ideal signal difference. As tV2-tVo and Tr are predetermined characteristics of the instrument, this can be written using a constant proportionality factor Kv: tv2-tv0 (10) iv(tv2)io=Kv(iv-io),Kv=1-e- #r .

Kv is somewhat smaller than 1 and is standardised for the measuring value output.

Thus the reference signal measured is sufficiently exact and provides a result which is sufficiently proportional to the ideal value, if the ideal dark current i, is subtracted. In an analogue manner, this is also valid for the sample signal. In reality, however, the dark current is also measured incorrectly.

The dark current is calculated in the dark interval following the reference interval. in analogy to (6), this now gives: t'-t'vo (11) i'v(t')=io + (iv-#iv-io)e- #r the dark signal is detected at the momentt'2. If t = t'v2 in (11), and if the right hand side is again split into three components, this yields (12) i'v(t'v2)=io+(iv-io)e-t'v2-t'vo - #ive-t'v2-t'vo #f #f This again gives a first component having the ideal value io, a first error component (iv-iO)e - t V2 - ttVo If and a second error component #ive-t'v2-t'vo .

#f Again, the second error component is small as compared to the first error component, but the first error component is not necessarily small as compared to the first ideal component io, as, in general, i,-i, is large as compared to o The estimation of these error components may be made only on the basis of the difference between reference signal and dark signal. If, on the left hand side of (8) the dark current i'v(t'v2) measured according to (12) is inserted instead of the ideal dark current io, this yields (13) iv(tv2)-i'v(t'v2)=iv-io-(iv-io)e-tv2-tvo #r t'v2-t'vo tv2-tvo t'v2-t'vo -(iv-io)e- +#i'pe- +#ive- .

#f #r #f The right hand side of this equation consists of five components; the ideal value iv-io, two error components proportional to the ideal value, and two correcting terms. The two error components are small as compared to the ideal value and the two correcting terms are small as compared to the error components and thus are negligible. This yields (14) iv(tv2)-i'v(t'v2)= 1 -e- tv2tvo ~e~t v2-tVo (iv-io) Tr #f or, if a constant characteristic of the instrument is introduced: (15) iv(tv2)-i'v(t'v2) = K'v(ivlo).

In analogue manner, (16) ip(tp2)i'p(t'p2) = K'p(ipi0) is obtained for the sample signal.

In general, tV2-tv2 = tp2-tpo, ttV2-tRvo = t'p2t'po, and thus K v = K'p.

This however, is not necessarily required and even if it is aimed at, small differences due to adjustment errors in the timing may result. In the ratio forming for deriving the sample transmission, any differences may be compensated for by a scaling factor.

In each of equations (15) and (16), a measured value i-i' is obtained which is proportional to the respective ideal value iv-io and ip-iO, respectively, if the measured dark value is subtracted from the respective measuring value (reference signal or sample signal) during the subsequent respective dark interval.

The essential idea of the invention consists in subtracting from each measured value the subsequently measured dark value.

In contrast to this, conventional electrical circuits suggest subtracting from each measuring value the previously measured dark value in the usual manner; the dark value is measured and electrically fixed.

Afterwards, only that part of the following measuring value is evaluated which exceeds the fixed value.

In applying such a circuit to the invention, this procedure may be reversed; the measured value is fixed and the subsequent dark value is evaluated proceeding from this fixed value. Then, indeed, a sign reversal (dark value minus measuring value instead of measuring value minus dark value) is obtained, which, however, is not relevant for the application.

It is not necessary to subtract the real dark signal immediately following each measured signal, though this would be the exact way of operation. It is also within the invention to subtract from the measured signal the dark signal following another but similar measured signal, especially the dark signal following the preceding similar measuring signal. This presupposes that, at the end of an entire period, the measuring signal is not substantially changed, this being approximately the case in general. Thus, the dark signal following a measured signal is fixed and retained until the next similar measured signal begins. From this measured signal, only that part is evaluated which exceeds the fixed dark value. Then the sign reversal mentioned above does not occur.

If, in contrast to the invention, the respective dark current, measured immediately previously is subtracted, distinct errors result therefrom. To show this, equation (14) relating to the method of the invention (subtraction of the following dark signal) may be written in a somewhat different form: tv2-tvo t'v2-t'vo (17) iv(tv2)-ilv2)= 1-e- (iv-io)-e- (iv-io) #r #f or (18) iv(tv2)i'v(t'v2) = Kv(iv-io)-K"v(iv-io) and accordingly (19) ip(tp2)i'p(t'p2) = Kp (ipio) K"p(iplo).

If the previously measured dark current is subtracted in the usual way, then these become: (20) iv(tv2)l'p(t'p2) = Kv(iv-io)-K"p(ip-io).

(21) ip(tp2)-i'v(t'v2) = Kp(ip-io)-K"v(iv-io).

The reference value is now falsified and inverted by a component proportional to the sample value.

It is also usual to average the dark value over several dark intervals. The measured dark value then contains a component proportional to (iv-io) and a component proportional to (ip-iO). According to (20) and (21), a cross correlation also results in this case as the respective "wrong" error component is not proportional to the measured value.

The invention has been described for the case that the detector current is detected at a fixed moment at the end of the respective interval, for example the reference current at the moment tvo. To reduce the noise level, however, it is more favourable to evaluate during each phase the waveform with respect to time of the detector current during a longer period of time by means of integration, for example. In that case, for example, the reference current is integrated from tvi tot,2, the following dark current from t'vi tot',2, and so on (see Figure 4).

A quasi-integration may also be carried out in having the signal pass an electric filter which has its own time constant, which preferably is larger than Tr and Tf. In such a quasi-integration, it is advantageous to average over several similar intervals, for example over several reference intervals.

In all these cases, too, the invention may be applied. Substantially, again the equations (18) and (19) with the methods according to the invention and the equations (20) and (21) with the methods usual up to now are obtained. This can be shown mathematically.

Firstly, the integration is carried out. To this end, proceeding from the general equation (3) the detector current is integrated over an interval from t1 to t2. This yields: t2tc tito (22) t idt = ie(t2-t1)+(ieia)# e- -e- # According to (3), the current i itself comprises a component proportional to 1a and a component proportional to (ie-ia) at a moment t. Integration over an interval t1 -t2 gives the same result. As all previous considerations and conclusions were based on such proportionalities having 1a and ie-ia, they may also be applied to integrated measuring values in the same way.

Dealing now with the quasi-integration and proceeding from the detector current according to equation (3) but designating the time constant of the detector by T, gives (23) i = ie-(ie-ia)e - ~~~~~~~ #1 If this current is passed through a filter having the time constant T2, then beyond the filter: ie-ia t-to t-tc (24) i=ie- #2e- -#1e- #2-#1 #2 #2 For T2, this equation provides the indeterminate expression 1a - 0/0.The the following equation has to be applied with 11 = (25) i = ie(ieia) 1 + t to e - t to I I Here also there is always (in (24) and (25)) a component proportional to (ie-ia). Thus this again leads to the same conclusions as when detecting the measured values at a predetermined moment without quasi-integration.

Even with the quasi-integration, the measured value is detected at a determined moment (namely t in (24) and (25) respectively). But the measured value there detected is obtained by an integration from to tot, the value to be integrated within this interval being weighted dependent on time according to an exponential function.

The essential statements are also maintained if the quasi-integration is extended over several similar phases. To this end, the measured value obtained beyond the filter having a time constant 12 at the end of such an interval is sampled and maintained until the beginning of the following similar phase. The effects depend on whether this value is fixed immediately beyond the filter - as happens in the example to the signal to be subtracted - or only beyond another amplifier followed by a holding device which has a negligibly small time constant - as happens in the example to the difference signal.

In the first case, equation (24) - it is also possible to proceed from equation (25) - has to be provided with an additional term. It is necessary to add: t-to (26) (if-ia) e - #2 where it is the sampled and held final value. This results from (24) if therein t is inserted for the moment at which sampling and holding begins. Thus if - ia is only proportional to ie - ia. Thus the expression (24) additionally provided with (26) again comprises a component proportional to ie and a component proportional to ieia The second case is now very easily dealt with; the value held beyond the additional amplifier is without importance as the next initial value.As a negligibly small time constant is effective in this case, the initial value before the amplifier (thus immediately beyond the filter) becomes practically fully effective at once. In the method according to the invention, in any case, this initial value consists of a component proportional to ie and a component porportional to 1a - The four light barriers of Figure 1 control the four switches 62,64,80,82 of Figure 2 in the following manner.

During the reference interval (here and subsequently reference is always made to the somewhat shorter measuring time t1 .... t2 within the interval concerned) the switch 62 is closed and the switch 64 is opened.

The right hand side of the capacitor 56 is connected to ground. The left hand side of this capacitor is charged via a resistor 54.

The chain 54, 56 has a time constant. A quasi-integration is obtained: the waveform of the voltage (subsequently called the detector voltage for simplicity) following an exponential function with the detector time constant Ir is weighted with the time constant of the filter 54, 56. The left hand side of capacitor 56 is charged up to a potential u (t) given by (24), if i is replaced there by u and 11 = Ir and 12 = X (the left hand side of (24) is u (t)).

The switch 62 is opened at the moment tV2. From then on, the voltage difference applied to capacitor 56 between the plates is maintained, even if the detector voltage u varies: the average capacitor potential only varies in accordance with the detector signal. The potential on the right hand side of the capacitor is then equal to the difference between the potential just applied to the left hand side and that potential which was applied to the left hand side at the moment of opening of switch 62.

The potential applied to the left hand side of capacitor 56 at the opening of 62, however, was the quasi-integrated reference signal. During the following dark interval, the waveform with respect to time of the dark signal (also quasi-integrated) is obtained there. Thus, on the right hand side of capacitor 56 and thus at the amplifier input 58 the waveform with respect to time of the quasi-integrated dark signal minus the sampled and held and thus constant quasi-integrated reference signal is obtained.

Switch 64 is closed during this dark interval. The quasi-integrated signal difference is supplied to the amplifier 68. The switch 64 is re-opened at the end of this dark interval. The capacitor 70 constantly maintains the signal applied to the input 66 of the amplifier 68. Switch 62 remains open. The voltage difference at capacitor 56 is still maintained. This condition is also maintained during the following sample interval and the subsequent dark interval. Only when the next reference interval starts (at the moment tv) switch 62 is re-closed (64 remains open) and the process described is repeated. Thus a quasi-integration is obtained from several periods, as treated mathematically above.

The switches 80 and 82 for forming the sample signal are correspondingly controlled in the same way, i.e.

with a phase shift of 180 with respect to the control of switches 62 and 64.

In version 2, the switches are controlled in such a way that the dark signal is subtracted from the measuring signal.

At the beginning of the dark phase following the reference signal, switch 62 is closed. Switch 64 is open. At the end of this dark interval, switch 62 is opened: the dark signal is sampled and held across the capacitor 56.

Switch 64 remains open. Both switches also remain open during the following sample interval and the subsequent dark interval. During this time, the dark signal remains held by the capacitor 56, and the amplifier input 66 constantly remains at the potential generated during the preceding cycle.

Switch 64 is closed only at the beginning of the reference interval, whereas switch 62 remains open. The reference signal exceeding the dark value is now supplied to the amplifier input 66.

At the end of the reference interval, switch 64 is opened. The capacitor 70 maintains the signal applied to the amplifier input 66 constant until the beginning of the next reference phase.

With the beginning of the following dark phase, the described process is repeated.

Claims (5)

1. A double-beam photometer having components defining sample and reference paths of rays and comprising a single photometric detector which is subjected to inertia such that it responds with delay to variations of the radiation flux falling thereupon and supplies a dark current in the absence of radiation flux, a beam switching device by means of which a sample light beam from the sample path of rays and a reference light beam from the reference path of rays are alternately directed upon the detector during respective sample intervals and reference intervals and the sample and reference beams are simultaneously interrupted during first dark intervals between the sample intervals and the reference intervals and during second dark intervals between the reference intervals and the sample intervals and a signal processing circuit connected to the photometric detector and which comprises signal detecting means for detecting the detector signals during each sample interval, reference interval and first and second dark intervals and difference forming means for forming the difference of the signals generated by means of the signal detecting means during each sample interval and during the first dark interval following each sample interval, and means for forming the difference of the signals generated by means of the signal detecting means during each reference interval and during the second dark interval following each reference interval.

2. A photometer according to claim 1 in which the signal processing circuit comprises means for forming a first average of the signals supplied by the signal detecting means over a number of first dark intervals and means for forming a second average of the signals supplied by the signal detecting means over a number of second dark intervals, and the difference forming means form the difference of the signals supplied by the signal detecting means during each sample interval and the first average and the difference of the signals supplied by the signal detecting means during each reference interval and the second average.

3. A photometer according to claim 1 or claim 2 in which the signal detecting means comprise means for detecting the instantaneous value of the detector signal at a predetermined moment of the sample interval, the reference interval and the first and second dark intervals respectively.

4. A photometer according to claim 1 or claim 2 in which the signal detecting means comprise RC-networks for a pseudo-integration of the detector signal over given integration intervals within the sample intervals, the reference interval and the first and second dark intervals respectively.

5. A photometer according to claim 1 or claim 2 in which the signal detecting means comprise integrators for integrating the detector signal over predetermined integration intervals within the sample interval, the reference interval and the first and second dark intervals respectively.