GB2073985A - Improvements in or relating to laser doppler anemometers - Google Patents

Abstract

A laser Doppler anemometer for measuring fluid flow velocity includes a beam splitter for splitting an incident laser beam 2 into a reference beam 7 and a second beam 6 for transmission through said fluid, a diplex optical spectrometer 14 for diplexing the reference beam and light scattered from the second beam and dispersing the different wavelengths of each beam and producing demultiplexed beams, and a pair of photo-multipliers 18, 21 each arranged to receive one of the demultiplexed beams and detect photon bursts therein. The spectrometer may employ light polorization to diplex the reference and the second beam onto a single optical channel, and the phase difference between the signals produced by the photo-multipliers 18, 21, which will be indicative of the velocity to be measured, may be measured using digital or analogue techniques. <IMAGE>

Description

SPECIFICATION Improvements in or relating to laser doppler anemometers The present invention relates to laser Doppler anemometers and in particular relates to anemometers which measure Doppler shift by optical spectrometry.

Jackson and Paul described in J Phys E Sci Instrum 4173-7, 1971 a laser Doppler anemometer which although having a frequency resolution of 7.5 MHz, corresponding to a few metres per second, could only be used conveniently to measure velocities at least one or two orders of magnitude greater than this. The anemometer included a scanning confocal Fabry-Perot spectrometer and a single frequency laser source.

Light scattered from particles in the flow under investigation is combined with a weak reference beam from the laser source. The spectrometer produces at a photomultiplier input a time dependent light intensity which corresponds to the spectral nature of the combined beams. When the Doppler shift is too small it cannot be resolved and so low speed flow measurements cannot be made with such anemometers.

The present invention provides a laser Doppler anemometer which is capable of Doppler frequency shifts associated with low speed flow.

According to the present invention a laser Doppler anemometer for measuring fluid flow velocity includes means for splitting an incident laser beam into a reference beam and a second beam for transmission through said fluid, a diplex optical spectrometer for diplexing the reference beam and light scattered from the second beam and dispersing the different wavelengths of each beam and producing demultiplexed beams, and a pair of photomultipliers each arranged to receive one of the demultiplexed beams and detect photon bursts therein.

The optical spectrometer may comprise a spectrometer which employs light polarization to diplex signals in a single optical channel as disclosed in our copending Patent Application No. 8006889.

One embodiment of the invention further includes the analogue signal processing means to process output signals from the photomultipliers.

A second embodiment of the invention further includes digital signal processing means connected to receive signals from the photomultipliers and having an output responsive to the time delay between corresponding photon burst detected in the photomultiplier.

Embodiments of a laser Doppler anemometer in accordance with the invention will now be described by way of example only and with reference to the drawings of which: Figure l is a schematic plan view of a laser Doppler anemometer including an analogue signal processor in accordance with the invention, Figure 2 is a schematic circuit diagram of the analogue signal processor of Figure 1, Figures 3 and 4 are a series of graphs showing idealized waveforms in various parts of the analogue signal processor of Figure 1, Figure 5 is a schematic circuit diagram of a digital signal processor which is an alternative to the analogue signal processor of Figure 1, Figures 6 and 7 are a series of graphs showing idealized waveforms in various parts of the digital signal processor of Figure 5.

The laser Doppler anemometer shown in Figure 1 employs an argon-ion laser 1 operated with an intra-cavity etalon at a wavelength of 514.5 nm and a power level of 100 mW. A vertically polarized input beam 2 from the laser 1 passes via an aperture 3 in a container 4 to a partially silvered mirror 5 which separates the input beam into two component beams 6 and 7. Beam 7 is used as a reference beam whilst beam 6 provides excitation for flow-borne scatterers comprising atomized oil from an atomizer 30 which introduces a stream of oil droplets into the air-receiver 33 supplied from a compressed air-line 32. The seeded airflows in a direction indicated by arrow A from the reservoir through a 2.7 mm nozzle 11 set with its axis at an angle of 45" to the beam 7.The assembly 30, 33, 11 has x and y-axis positioning controls 31 which enable the nozzle 11 to be positioned, as required, relative to the intersection point P of the two laser beams 6 and 7. A pipe 8, suitably cut away to admit the beams 6 and 7 allows the seeded jet to flow unhindered away from the region P and be discharged safely.

A lens 34, of focal length 15 cm, focusses the beam 6 into a diffraction limited focal cylinder the diameter of which defines the scattering volume and hence spatial resolution of velocity measurement. The beam 7 is directed through a half wave plate 9 to rotate the vertical plane of polarization of the reference beam by 90" and through a Glan-Thompson prism 10 to ensure that the polarization is strictly planar. The beams 6 and 7 intersect at a measurement zone P in the pipe 8. Vertically polarized scattered light and horizontally polarized light from zone P is directed via a pair of lenses 12, 13 to a confocal 'Fabry-Perot' spectrometer 14which includes two spherical mirrors of radius of curvature 10 cm and which has an FSR (free spectral range) of 750 MHz.After analysis in the spectrometer 14 components of the beams 6, 7 are directed by a lens 15 onto a 'Wolf' polarizing prism 16 which separates the components into two beams 6' and 7' which correspond to light from beams 6 and 7. Beam 7' is directed via a pair of mirrors 22,23 collected by a lens 17 and focussed into an entrance pupil of a photomultipliertube 18 via an interference filter 19 which removes black body emission from the laser 1.

The scattered beam 6' is collected by a lens 20 which produces an image of the scattering volume on the entrance pupil of a second photomultipliertube 21 which is similar to the photomultiplier 18. By limiting the size of the image of the scattering volume, the aperture defines the size of the scattering volume. The aperture diameter is 500 Ftm and the lenses 12, 13, 15, 20 have a combined magnification of unity, and produces an effective scattering volume of the same length.

A signal processing system 35 receives signals from the reference photomultiplier 18 and the signal photomultiplier 21. Figure 3 shows typical scanning waveforms for the processing system 35. Scan amplitude and flybacktime are properly related to preserve timing information and are illustrated in Figure 3(a) which shows a waveform having a repetition frequency of about 15 Hz together with a detector signal from the reference channel only as shown in Figure 3(c). Detected output is regarded as a voltage proportional to light intensity. Three broad peaks P1, P2, P3 followed by three narrow ones P4, P5, P6 in Figure 3(c) indicate that the spectrometer scans over three FSR.For an arbitrary amplitude of scan, however, peaks which occur during the first forward sweep will not necessarily be correctly phased with respect to those generated on the second forward sweep. In order to remedy this, the three narrow peaks are removed as in Figure 3(d), by inhibiting the detector on flyback with a gating waveform shown in Figure 3(b). It is necessary to suppress the detector for longer than the flyback time so as to allow the spectrometer sufficient time to settle into the new forward scan. Then for a given ratio of upward and downward scan times, TUTTED say, the amplitude Vs of the scanning voltage is adjusted to generate a detector output which, apart from missing peaks, is entirely regular as shown in Figure 3(d).

In particular, if V0 is the voltage required to scan over one FSR, then the number of orders (not integral) scanned during the forward part of the waveform is x = Vs/V,. Similarly the number of orders that would be scanned during the time TD, were it not for the flyback is xTD/Tu. Thus, if the detector output is to be regularly phased over all time, the sum of these two quantities, x(1 + TD/TU) must be kept as an integer, N. The amplitude of the scanning voltage is therefore determined as, V NVo (1) 1 + TD/Tu as shown in Figure 3 with N = 4. For the waveform in Figure 3, the spectrometer scans effectively over four FSR in the scan cycle time Ts = 4T.

Once set up in this way, the phase uniformity of the detected peaks in Figure 3(d) is preserved, even if the laser frequency changes slightly. The effect is to shift the positions of the peaks but since only the forward part of the scan is observed all peaks are moved in the same sense and of course by the same amount, provided that the scan is linear. The same comments apply to the Doppler shifted light in the signal channel.

Similarly any change in the optical length of the spectrometer, due for instance to temperature variations, has the same effect, ie the output in Figure 3(d) is shifted, slightly along the time axis in one direction or the other, the regularity of the peaks remaining unaffected.

All anemometry measurements which involve a scanning spectrometer can be reduced to measurements of the time tD between the appearance of corresponding points on the electrical representation of the shifted and unshifted optical spectra. The only information required from the train of pulses in Figure 3(d) is the position of the peaks: the height and shape of the pulses can be neglected. Since the pulse train is entirely regular, all the required information is contained in the phase of the first Fourier component of the waveform. In practice this may be obtained simply by applying the voltage of the detector output in Figure 3(d) to a narrow band filter tuned to a frequency T-' = NTs-' where N is the integer in equation (1) and Ts is the scan cycle time which in practice is about 4 x 15 = 60 Hz.

The signal processing system 35 is shown in detail in Figure 2. Outputs from the photomultipliers 18 and 21 are carried on lines 37 and 38, respectively, and line 39 carries a flyback suppression logic signal. The signals on lines 37, 38 and 39 are gated to their respective channels 34,44.

First and second 60 Hz filters 81,82 respectively are included in the reference channel 84, and similar pairs of 60 Hz filters 41,42 are used in the signal channel 44. In each case, the second filters 82 and 42 form part of amplitude stabilising loops 83 and 43, respectively. The output waveform from the loop 83 is indicated in Figure 2 at A and corresponds to Figure 3(e), and is independent of the magnitude of the photomultiplier signal. Only the 60 Hz component of the photomultiplier output is required and so the 60 Hz filters are supplied directly with the individual impulses derived from a photon counting set.

Provided that the Value of the tuned filter is high enough, the missing peaks in Figure 3(d) do not affect its output and a good sine-wave is obtained, Figure 3(e). The last waveform clearly retains the timing information provided by the original photomultiplier signal, Figure 3(c). The phase of the sine-wave will depend on the precise nature of the phase-shifts in the electrical circuit but provided these remain constant, their magnitude need not be known. For convenience the sine-wave has been shown with its peaks corresponding to the peaks in the original waveform, Figure 3(c).

In order to extract the phase of the sine-wave in Figure 3(e), it is transformed into a square-wave, indicated at B which corresponds to 3(f) using a zero-crossing comparator 85. This square-wave is then used to generate a phase-shifted waveform by first producing a pulse of adjustable width, l, indicated at C which corresponds to 3(9), using a variable delay monostable circuit 86 triggered by the rising edges of the square-wave. To reconstitute a precisely symmetric square-wave, which is required at a later stage a bistable circuit 30 is triggered by the falling edges of the pulse train from the monostable circuit 86. The output waveform from the bistable circuit 87 preserves timing information from the original spectrum, see Figure 3(c), except that now an adjustable phase shift can be introduced.In Figure 3(h) the waveform has rising and falling edges which mark the positions of the peaks in the photomultiplier output. Referring to Figures 2 and 4, the square-wave outputs from bistable 87 and from a corresponding bistable 47 in the signal channel 44 are shown at D (see Figures 3(h) and 4(c)) and E (see Figure 4(c')). These square-wave outputs are gated in a phase comparing gate 51 and are input to a low pass filter 50 via buffer and offset circuits 52. The waveform of the input to the filter 50 is shown at F in Figure 2 which corresponds to Figure 4(d).

Figure 4 shows idealized waveforms present in the two channels 34,44 when the anemometer is operating. Figures 4(b,b') show the waveforms derived from the two photomultipliers whilst Figures 4(c,c') show corresponding square-waves containing the required timing information. By adjusting the time delay, t, in each channel appropriately, the leading and trailing edges of each square-wave can be made to coincide with the positions of the spectral peaks in their respective channels. By gating the two square-waves together, see Figure 4(d), a train of pulses of width tD is produced where tD is simply the required Doppler related time delay, #D tD = (FSR) . T (2) where T is the time taken to scan over one FSR (free spectral range) and v0 is the Doppler frequency shift.

A direct analogue measure of tD is produced by averaging the waveform of Figure 4(d) in a low pass filter 50 shown in Figure 3. If Vp is the pulse height then the output voltage from the low pass filter is, V VD=ItDI (3) The modulus sign indicates that pulses of negative width cannot be produced, and shows that the instrument in this form is not sensitive to changes in the direction of flow.

This difficulty is overcome by increasing the phase delay in the signal channel so that in the absence of a Doppler shift the square-waves, Figures 4(c,c') (dotted lines) are a quarter of a period out of phase. Equation (3) is then modified to, VD = 1 Vp tD T T 2 (4) where tD is in the range - 2 T < tD < 2 T. By inserting a voltage shift of magnitude - Vp/4, the DC output from the anemometer is directly proportional to velocity. Using equation (2) the sensitivity of the system to Doppler frequency shifts is, VD = t D Vp 2(FSR) and VD is in the range +2 FSR.

The magnitude of the Doppler shift is related to the velocity vector V of the flow as follows: 2vc= (ko-kL).V (5) where k0 and kL are the wave vectors of the scattered light and laser light respectively. In Figure 1, V is arranged to be parallel to the scattering vector K = ks - kL and so, 20 vDzsin (24) where U is the magnitude of the velocity, k = 2z/k is the wavelength of the laser and (3 = 90" is the scattering angle.

The velocity sensitivityof the anemometer is then, VD = U Vp - V2X(FSR) In practice the instrument was arranged to provide a direct voltage output in the range +10 V corresponding to Doppler shifts of +l(FSR) - j375 MHz, a velocity range of +136 my~1. By adopting a Fabry-Perot of different FSR or a different scattering angle, the range of the anemometer may be modified according to the velocities encountered.

The digital signal processor shown in Figure 5, which is an alternative form of processor to the analogue processor described above, has three input lines comprising a signal detector line 61, an UP/DOWN logic line 62, and a reference detector line 63. The relationship between the logical level on line 62 and the scanning voltage applied to the piezoelectric transducer of the spectrometer can be seen from the waveforms (a) and (b) of Figure 6.

The signals on line 61 are switched by a logic switch 65 to one of a pair of output lines 77 and 78 in accordance with the logic level on line 62. Similarly, signals on line 63 are switched by a switch 64 to line 75 or 76, Line 75 is connected to one input of an up/down counter 68 via a divide-by-two circuit circuit 66 and line 76 connected to a second input of the counter 68. Similarly, line 77 is connected to one input of a second up/down counter 69 via a divide-by-two circuit 67 and line 78 is is connected to a second input of counter 69.

The switches are arranged such that UP logic on line 62 switches the signals on lines 61 and 63 to lines 77 and 75 respectively. Signals from the counters 68 and 69 are input to a zero crossing latch 70, 71 respectively and the outputs from the latches 70,71 gated through an output gate 73 controlled by a clock oscillator 72. A multi-channel scaler and display 74 receives the output signal from the gate 73. Clock pulses from each scan cycle are stored in successive channels of the scaler.

In operation, in order to determine the timing of photon bursts in each channel individual photon pulses are transferred during a rising frequency sweep via the counters 68 and 69 operating in the count-up mode, see Figure 6, waveform (e). By way of illustration it is assumed that there are 40 photons in the burst shown on the left hand side of wavefrom (c), Figure 6. On the falling frequency sweep pulses from a second burst of photons are transferred directly to the counters 68,69 now operating in the count-down mode. If the number of photons in each burst is sufficiently large and noise photons are relatively few, the statistical fluctuations in the number of photons counted in the two peaks is smali and the counter reaches zero (point X in waveform (e), Figure 6) at a time which is very close to the peak in the photon rate.Under these conditions the logical level (see wavefrom (f) in Figure 6) generated by the up/down counter 68 contains all the timing information which is relevant to the reference channel. A similar, time shifted, logic level is generated in the signal channel. Figure 7 shows the relative timing of corresponding waveforms in the presence of a Doppler related time delay to between the optical channels. Waveforms c and c' in Figure 7 show the photon pulses from the two detectors and waveforms d and d' show the corresponding logic leveis generated by the up/down counters 68, 69. The signals from the counters 68, 69 are gated with the output from the stable clock oscillator 72. The number of clock pulses obtained during each scan cycle is a direct measure of the flow velocity.

In situations where reversals in flow velocity are encountered, negative Doppler shifts can be measured by introducing a constant time delay 't' into one of the channels (see dotted waveforms (d) and (e) in Figure 7).

The multichannel scaler and display 74 shown in Figure 5 displays a velocity profile 79 wherein the height of each point of which the profile is composed represents the number 'n' of clock pulses stored during each scan cycle, where na velocity. Since one measurement is made within every scanning cycle, this repetition time represents the limiting rate at which the anemometer can follow changes in velocity. The scan cycle time can be varied according to convenience provided it is not less than 20 ms.

At low light intensities velocity measurements will include an unacceptably high noise level. An averaging facility was provided whereby clock pulses from a number of successive scan cycles were stored in the same channel of the multichannel scaler.

Claims (5)

1. A laser Doppler anemometer for measuring fluid flow velocity including: means for splitting an incident laser beam into a reference beam and a second beam for transmission through said fluid, a diplex optical spectrometer for diplexing the reference beam and light scattered from the second beam and dispersing the different wavelengths of each beam and producing demultiplexed beams, and a pair of photomultipliers each arranged to receive one of the demultiplexed beams and detect photon bursts therein.

2. An anemometer as claimed in claim 1 wherein the optical spectrometer employs light polarization to diplex signals in a single optical channel.

3. An anemometer as claimed in claim 1 or 2 further including analogue signal processing means connected to receive output signals from the photomultipliers.

4. An anemometer as claimed in claim 1 or claim 2 further including digital signal processing means connected to receive signals from the photomultipliers and having an output responsive to the time delay between corresponding photon bursts detected in the photomultipliers.

5. An anemometer substantially as described herein with reference to the accompanying drawings.