A compact, laser diode based phase Doppler system

Experiments in Fluids

Experiments in Fluids 9, 153-158 (1990)

9 Springer-Verlag 1990

A compact, laser diode based phase Doppler system R. W. Sellens* Lehrstuhl ffir Str6mungsmechanik, Universit/it Erlangen-Nfirnberg, D-8520 Erlangen, F. R. Germany Abstract. Recent advances in laser diode technology provide the opportunity to reduce the size and cost of laser Doppler and phase Doppler anemometry systems. The application of laser diodes to phase Doppler systems is discussed, and the construction of a compact, laser diode based phase Doppler system is described. Test measurements in a monosized drop stream, a water spray and a fluidized bed are reported, demonstrating the functionality of such a system.

1 Introduction The measurement technique of phase Doppler anemometry (PDA) has developed rapidly in the last few years, becoming recognized as a flexible and robust approach to the measurement of the size and velocity of spherical particles. The improvements in instrumentation have also stimulated interest in spray flows, as it is now possible to quickly and accurately measure quantities that, until recently, were out of reach with practical techniques. Present PDA instruments are based on gas lasers and include relatively bulky optical and signal processing systems. While these systems work well in a laboratory environment, a number of factors make them ill suited for in situ measurements in industrial environments. - They are relatively large and heavy, making them difficult to install at odd angles and in cramped locations. - They are costly and, primarily due to the laser, fragile. - The danger of sparks from the high voltage supplies for the laser and photomultipliers prohibits their use in explosive environments. The same list of disadvantages applies to most laser Doppler anemometry (LDA) systems as well. However, recent developments based on solid state laser diodes (Dopheide et al. 1988; Durst et al. 1989) have shown that most of the functionality of larger, gas laser based LDA systems can be provided by compact laser diode systems. Such systems have been built, providing a full LDA system * Present address: Dept. of Mechanical Engineering, Queen's University, Kingston, Ontario, Canada K7L 3N6

of laser, optics and detector, in a housing the size of a flashlight. This makes it possible to measure in many environments where it was previously impossible, or where the measurement location could be reached only using fiber optic probes. All of the advantages that laser diodes bring to LDA, they also bring to PDA. The advantages of compact drop size and velocity instrumentation are obvious in applications such as gas turbine combustion, fuel injection in gasoline and diesel IC engines, fuel sprays in rocket engines and multitudinous spray processes in the chemical and petroleum industries. The present work addresses these needs, describing the construction and testing of a laser diode based phase Doppler system.

2 Basic principles of phase Doppler The potential for particle sizing based on a shift in the phase of Doppler signals was first recognized by Durst and Zar6 (1975). The technique has subsequently been applied by various workers (Bachalo 1980; Bauckhage and F16gel 1984; Saffman et al. 1984) to provide simultaneous measurements of drop size and velocity. The measurements of individual drops may be accumulated to provide drop size and velocity distributions the form in which such information is usually presented. Although the frequency of an LDA signal is invariant with the location of the detector, the phase of the signal is a function of location in space and the optical characteristics of the scattering particle. Thus, by measuring the phase difference between signals taken at two or more locations in space it is possible, assuming that the particle is spherical, to calculate the particle diameter. The basic principle involved is the fact that, if light is scattered through slightly different angles by a particle, the length of the optical path followed through the particle will depend on the scattering angle and the radius of curvature of the surfaces. This difference in the optical path length, on the order of the wavelength of the light, will result in a phase shift in the scattered light. This phase shift mixes down when two fields are combined, just

154 as the frequency difference between the fields mixes down to provide the Doppler difference frequency. Under appropriate optical conditions the phase shift is linearly proportional to the particle diameter. For detectors symmetrically arranged about the plane of symmetry of the sending system, the phase relationship may be reduced to a single equation based on geometry, laser wavelength and refractive indices (Bauckhage et al. 1986). For asymmetrically arranged detectors the same type of analysis produces a, necessarily more complex, set of relations based on the same parameters (Sellens 1990a). Although adequate for most phase Doppler situations, the geometric optics do not provide an exact solution to the light scattering equations. Simple geometric approaches fail when multiple scattering modes are significant, or when the particle size approaches the wavelength of the laser light. In these cases a more exact solution is required. Bachalo and Sankar (1988) have shown that the effects of multiple scattering modes may be adequately addressed using a more general geometric model. For small particles the exact Mie scattering equations provide good results, although care must be taken to mathematically integrate over the collection aperture, just as takes place in the actual instrument. Point calculations (A1 Chalabi et al. 1987) will show strong fluctuations in the phase/size relationship that are simply not present in the actual instrument. Figure 1 shows comparisons between the phase/size relations obtained from Mie scattering calculations and those obtained from simple geometric optics for several different configurations of the present system. Note that the magnitude of the fluctuations in the phase/size curves is strongly affected by relatively minor changes in the optical parameters. Such checks are absolutely essential to avoid major errors. For example, although reasonable measurements of water drops (refractive index 1.33) in air may be made at scattering angles around 150 ~ glass spheres (refractive index 1.50) in air do not provide a well behaved phase/size relationship at any of the backscatter angles.

3 Sending systems The requirements for a phase Doppler sending system are virtually identical to those for an LDA system. Both require a pair of well focussed beams, crossing to provide a control volume of sufficient intensity that particle signals may be detected at some remote location. The focussing of the beams must provide wavefronts that are sufficiently planar to provide flat, evenly spaced interference fringes in the control volume. For LDA the crossing angle is determined solely by considerations of spatial resolution and the frequency range of signals that can be dealt with by the signal processing system. Although spatial resolution is less of a consideration with PDA, due to the off axis collection angles, signal frequency is still a consideration and the beam crossing angle enters into the phase/size calculations as well. The basic

Experiments in Fluids 9 (1990) requirements of both LDA and PDA systems are well satisfied by gas laser systems using simple optical configurations for beam splitting and focussing. A laser diode by itself may not be directly substituted for a gas laser. While a gas laser provides a circular Gaussian beam of minimum divergence, diode lasers provide elliptical, astigmatic, highly divergent beams, due to the geometry of the resonant cavity in the semiconductor. In addition, while the wavelength of a gas laser is well defined, the wavelength of diode lasers varies with temperature, and from one diode to another nominally equivalent diode. However, with appropriate beam conditioning, a diode laser can provide an LDA/PDA control volume of a quality comparable to those provided by gas lasers. To provide such a control volume, certain difficulties must be overcome. The specific requirements of LDA systems have been addressed by Dopheide et al. (1988) and Durst et al. (1989), among others. The particular problems are collimation of the divergent beam and in some cases correction for astigmatism, stabilization of the laser wavelength and the focussing characteristics of a non-Gaussian beam. The highly divergent beam requires strong collimation elements very close to the face of the laser diode. In some instances even the glass window of the diode casing is an obstacle to proper collimation. Due to the strength of the lenses and the small distances involved, the collimation of diode beams requires alignment to within micrometers. Durst et al. (1989) have observed that even the thermal expansion of a collimator mounting system can have significant detrimental results. The wavelength of diode lasers is dependent on operating temperature, and is subject to jumps due to '~ hopping" as the laser cavity expands. Therefore, some form of temperature control is necessary to stabilize the wavelength and avoid the unstable regions where mode hopping may occur. Although it may appear similar, the beam from a diode laser does not have the same Gaussian profile as a typical gas laser beam. This results in different focussing characteristics, which may be significant for LDA and PDA, as described by Durst et al. (1989). The characteristics of the control volume may be adversely affected by strong focussing of the beam, among other things, therefore it is essential to verify the quality of the control volume, by scanning or some other means, when assembling a new instrument. The major portion of the problems mentioned above may be eliminated by employing a commercially packaged laser diode. A large number of manufacturers now supply laser diode packages with integral collimation and temperature control. With such a package the user need only verify that the system meets the necessary requirements, rather than solving all the problems from scratch. Such a packaged system is used in the present work. The diode laser used in the present system is provided by Melles Griot. The package includes a 30 roW, 830 nm diode laser with integrated collimation optics. The combination of

R. W. Sellens: A compact, laser diode based phase Doppler system 80~

60o.

P

z'O~

/

r

o

laser package. (Figure 2 shows a schematic of the sending optics.) The beam contractor, composed of an 80 mm positive lens and a 20 mm negative lens, provides a four to one beam contraction. The reduced beam then enters a constant path length beam splitter, and is focussed to a control volume through a 200 mm achromat lens. The entire system is housed in a rack built up from microbench components.

~

'

20o

4 Receiving s y s t e m

.1~i~ I,'' . ~ . ~ r

ftuidized bed

.20 ~

-z,O~

,

0

,

5

,

~

,

10 15 20 Porticte size(Fm)

25

30

Fig. l. Comparison of Mie scattering calculations with results of simple geometric optics for various instrument configurations (Table l); the curves for the water spray have been plotted in the negative direction solely to avoid overlap

Beom

controctor

L

o

I ,L

s

Beomsptitter e ~

I 2SOmm

2oomm

-,

Fig. 2, Schematic of sending optics Aval.onche photod[ode \

~ \

155

Troversemount tens

q

Fig. 3. Schematic of a single receiver laser and collimator is cooled by a thermoelectric element to stabilize the laser wavelength and avoid collimation errors at extreme temperatures. The beam is circularized by a pair of prisms to provide a collimated beam of 8 mm absolute diameter (approximately 5 mm l/e 2 diameter) at the exit of the laser housing. Immediate focussing of such a large beam would result in a very narrow, very intense control volume, especially for a P D A system. To increase the size of the control volume a beam contractor is placed immediately in front of the diode

In the present system separate receivers have been constructed, which may be shifted in position independently of one another. This brings both advantages and disadvantages. Figure 3 shows the basic configuration of a single receiver. The receivers are constructed using the same microbench components as the sending system. Elevation of the receivers above the plane of symmetry of the sending system is achieved by specially constructed wedges that maintain the alignment of the receivers with the control volume. As all components are mounted on a c o m m o n aluminum plate, the receivers must lie in the plane of the sending system or above it. Separate receivers have the disadvantage of more difficult alignment, and the potential errors that go with misalignment (Sellens 1990b). In addition, care must be taken to maintain the spatial relationship between the sending system and the separate receivers. Fortunately, in a compact system, the lighter weight of the components and the smaller dimensions make this less difficult. Also, if the receivers are separate and arbitrarily placed, the geometric dependence of the relationship between phase and size becomes somewhat more complex, compared to a symmetrically arranged system, however this problem has been dealt with by Sellens (1990a). The use of separate receivers provides the advantage that the vertical separation may be adjusted independently of the size of the collection apertures. This adjustability allows easy adaptation to a wide range of measurement conditions. In the present system the measurement range is roughly ] 5 jam for every millimeter of vertical separation (the dominant direction) between the receivers. Thus, typical vertical separations between the collection lens centres are 10 or 20 mm, for measurement ranges of about 150 and 300 lain, respectively. If round collection lenses were arranged symmetrically, the lens sizes would have to be severely restricted to avoid overlap. The present system uses 30 mm collection lenses, giving a collection aperture of 8.6 ~ Large collection apertures have the advantage of providing smoother instrument response, due to the averaging of small scale optical variations over a larger area. This is demonstrated very clearly by Bachalo and Sankar (1988). Figure 4 shows the arrangement of the two receivers in relation to the sending optics. The entire system is mounted on a specially constructed bench with rotational symmetry, making it easy to install components at particular scattering angles, and to fix components at a known distance from a

156

Experiments in Fluids 9 (1990) ratio of the in-phase component to the out-of-phase component provides the tangent of the phase angle. The signals are validated on three criteria; SNR must exceed a minimum value, the signals must have the same amplitude to within a factor of two and the calculated velocity must lie within the bounds of accuracy of the signal processing technique. The current arrangement can process up to 35 events per second. If a faster, but more noise sensitive, zero crossing technique is used the data rate approaches 60 events per second.

~2

6 Test measurements

Sending system

Fig. 4. Top view of optical system layout; in addition to the off axis scattering angles shown here, each of the receiving units is elevated out of the sending plane by way of a wedge block

Table 1. Optical and sampling arrangements for different measurements Monosized drops Detector 1 - Angle q51 [deg] Elevation ~91 [deg] Detector 2 - Angle ~92 [deg] - Elevation qJz [deg] Particle material Refractive index [~] Sample interval [ns] Size range [gm] Velocity range [m/s]

Swirl Jet nozzle spray

Recirculating fluidized bed

50 40 40 5.71 2.86 2.86 70 60 60 0 5.71 5.71 Propanol Water Glass 1.37 1.33 1.46 50 200 1000 up to 139 up to 379 up to 364 7.8-83 1.9-20.8 0.4-4.2

central point, namely the control volume. Avalanche photo diodes (APDs) are used as photodetectors. They provide especially good performance with near infrared diode lasers, as the peak quantum efficiency of the APDs corresponds to the laser wavelength.

5 Data acquisition and signal processing Signals from the APDs are band pass filtered and passed to a two channel, 8 bit transient recorder with a digitization rate of up to 20 MHz. A level trigger on the filtered signal is used to detect bursts. The digitized signals are transferred to a high performance personal computer (20 M H z 80386 and 80387), where all signal processing for phase and frequency is carried out in software. The signals are processed using the cross spectral density technique first applied to P D A by Domnick et al, (1988). This technique operates in the frequency domain, separating the in-phase and out-of-phase components into the real and imaginary parts of the spectrum, respectively. The frequency is determined from the peak in the power spectrum, while the

A series of test measurements was carried out to verify the functioning of the system. First, the control volume characteristics were measured to insure that an adequate focus had been achieved. This was followed by measurements in a monosize drop stream to determine the resolution and accuracy of the instrument. Two practical tests were then undertaken to evaluate the operation of the instrument in a normal flow. Table 1 specifies the optical configurations used in the different test measurements. 6.1 Control volume characteristics

The sending system was partially assembled and then adjusted to provide a parallel beam at the exit of the beam contractor. The beam splitter and focussing lens were then added to provide two crossed beams. The control volume dimensions were measured using a BeamScan Model 1080 with a 4 gm slit. The lie z diameter was measured to be 180 gm, with a variation of ___6gm when one of the beams was blocked, indicating good alignment and uniformity between the beams. The local characteristics of the control volume were measured by scanning a 5 Jam pinhole, mounted on rotating disk, across the control volume at constant speed. The position of the pinhole carrier was adjusted by an electronic micrometer stage with an accuracy of 1 lam. The resulting signals were collected on an APD, stored in a digital storage scope, and transferred to a personal computer, where the frequency was determined from the power spectral density of the signal. Figure 5 shows a burst recorded using this technique. Figure 6 shows the frequency variation along the control volume axis and across the control volume width. The consistency of the frequency is very good within the useful limits of the control volume. From this one may conclude that the packaged laser and focussing optics combine to produce a control volume of sufficiently high quality. 6.2 Monosized drop measurements

A stream of monosized propanol drops with a diameter of 44 lam was produced using a TSI Aerosol Generator. When this stream was passed through the control volume a continuous train of signals was produced by the detectors. Figure 7 shows typical signal trains, after passing through the band pass filter. Due to the signal density, it was c o m m o n for one

R. W. Sellens: A compact, laser diode based phase Doppler system

! 57 51 Variation of measured drop size with sampling interval

49

-6

=L

::k

v

~q

~47

"4

ca.

L. "0 t,t3

IE 45

-2 .....

Fig. 5. Signal scan; this signal is typical of those taken in the central region of the control volume; as signals are collected in side scatter, only the central region is imaged on the detectors

-1200 1.10

Longitudinal distance(#m) -400 0 400 800

-800 i

I

I

1200

~1.05

g & 1.00

~ 0.95 cr

I

-80

0

s'o

160

2+0

35o

0

Sampling interval (ns)

Fig. 8. Variation of measured mean and RMS drop size with sample rate

I

I

Longitudinal scans Transverse scans

0.90 -120

43

RMS drop size

!

4'0

- 413 0 80 Transverse distance (~Lm)

120

Fig. 6. Variation of the Doppler frequency with position in the control volume; variation in both the transverse and longitudinal directions is less than 1% over the useful region of the control volume

signal processing event to include scattered light signals from more than one drop, thus increasing the error in size detection due to the phase noise at the transition between particles, This effect was minimized by measuring at a high sample rate and increased trigger level, so that only three or four periods near the centre of the burst were taken. At these settings the mean drop size was measured accurately, while the R M S d r o p size was found to be 1.5 pm approximately 1% of the total d r o p size range. Although this accuracy is sufficient for most spray measurement requirements, this dense particle stream represents a less than optimal measurement situation and better values should be expected for the standard error in measurements of less dense systems where bursts will be separated by a significant time. Figure 8 shows the effect on the measured mean and R M S drop sizes with decreasing sample rate and thus increasing overlap between bursts. Note that while errors increase substantially, there is no catastrophic failure due to the overlap.

6.3 Swirl jet nozzle i

o

lb

2'o

3'o

4'o

50

Time (~ts)

Fig. 7. Typical signal trains from monoslzed drop stream

Measurements were made in the region near the exit of a swirl jet atomizer nozzle, a flow qualitatively similar to one previously investigated by Sellens (1988) using a gas laser based P D A system. The results, shown in Fig. 9, are similar to those expected, based on the previous work. N o difficulties with signal quality were encountered, even in this relatively dense region of the spray. Measurements of the m a x i m u m burst length as a function of d r o p size indicate that there is little bias against smaller drops, as seen from the corrected size distribution in the figure.

6.4 Recirculating fluidized bed As a further test, measurements were made in a recirculating fluidized bed at the Lehrstuhl ffir Mechanische Verfahrens-

158

Experiments in Fluids 9 (1990) c.-===

S w i r l jet nozzte

Number

::::~

12.

:.:.::..= =====

8 Conclusions

distribution Corrected

l e o n velocity RMS velocity

A diode laser phase Doppler system has been constructed for use in spray and particle now measurements. The system performance has been measured under different test conditions and found to be similar to that of gas laser based PDA systems. The present system is compact enough that all components can be carried at once by one man. In addition, the entire system, including laser, optics, APDs, transient recorder and computer for signal processing, has been assembled for less than US$ 20,000. This represents a new level for phase Doppler systems in both compactness and price. This development opens new doors for in situ measurements for research, as well as process monitoring and control.

n

gS'

e,

gt,

References I

I

i

i

i

I

i

i

I

100

I

i

i

i

i

[

i

i

I

200 Port~cte size ( ptm 1

I

i

i

i

i

l

I

[

300

Fig. 9. Measurement results: swirl jet nozzle technik in Erlangen. This system is composed of a vertical duct with an air flow carrying glass spheres vertically upwards in the core, but with significant down flow near the walls of the duct. This flow presented several potential difficulties. Firstly, the measurements were made through windows (glass on the sending side and plexiglas on the receiving side) which could not be kept clear of particle accumulation on the inside of the duct. Next, the particle density was very high, so that an observer could not see 20 cm through the duct. Lastly, the glass particles had been in use in this system for three years, so that their surfaces had been substantially roughened by abrasion. In spite of these difficulties good signals were obtained and validation rates were around 80%. The measured size distribution was in good agreement with the known size characteristics of the original glass spheres.

7 Planned future developments The present system is limited by its lack of frequency shift in the sending system. A rotating diffraction grating is being added to the system to provide both beam splitting and frequency shift. This addition will permit measurement in flows with recirculation or high RMS velocities. Data rate is currently limited by both the data acquisition system and the signal processing software. While the data rate is sufficient for many measurement tasks, an increase in speed is almost always an improvement. Purpose built signal processing systems provide much higher data rates, but are in opposition to the cost and size objectives for the present system. The use of faster data acquisition cards and signal processing hardware to improve the data rate is being investigated.

Al-Chalabi, S. A. M.; Hardalupas, Y.; Jones, A. R.; Taylor, A. M. K. P. 1987: Calculation of the calibration curves for the phase Doppler technique: comparison between Mie theory and geometrical optics. In: Int. Syrup. Optical Particle Sizing: Theory and Practice. Mont Saint-Aignan, France Bachalo, W. D. 1980: Method for measuring the size and velocity of spheres by dual beam light-scatter interferometry. Appl. Opt. 19, 363 370 Bachalo, W. D.; Sankar, S. V. 1988: Analysis of the light scattering interferometry for spheres larger than the wavelength of light. In: 4th Int. Symp. Appl. Laser Anemometry Fluid Mech. Lisbon, Portugal Bauckhage, K.; Fl6gel, H. H. 1984: Simultaneous measurement of droplet size and velocity in nozzle sprays. In: 2nd Int. Syrup. Appl. Laser Anemometry Fluid Mech. Lisbon, Portugal Bauckhage, K.; Fl6gel, H. H.; Fritsching, U.; Hiller, R.; Sch6ne, F. 1986: Analysis of particle sizes with the aid of laser-Doppleranemometry: the influence of specific geometrical and optical properties of particles. In: 3rd Int. Symp Appl. Laser Anemometry Fluid Mech. Lisbon, Portugal Domnick, J.; Ertl, H.; Tropea, C. 1988: Processing of phase-Doppler signals using the cross-spectral density function. In: 4th Int. Syrup. Appl. Laser Anemometry Fluid Mech. Lisbon, Portugal Dopheide, D.; Faber, M.; Reim, G.; Taux, G. 1988: Laser and avalanche diodes for velocity measurement by laser Doppler anemometry. Exp. Fluids 6, 289 297 Durst, F.; Miiller, R.; Naqwi, A. 1989: A semiconductor laserDoppler-anemometer for applications in aerodynamic research. In: 13th Int. Congr. Instrument. Aerosp. Facilties. G6ttingen, FRG Durst, F.; Zare, M. 1975: Laser Doppler measurement in two phase flows. Proc. LDA Syrup. pp. 403 429. Copenhagen, Denmark Saffman, M.; Buchave, P.; Tanger, H. 1984: Simultaneous measurement of size, concentration and velocity of spherical particles by a laser Doppler method. In: 2nd Int. Symp. Appl. Laser Anemometry Fluid Mech. Lisbon, Portugal Sellens, R. W. 1988: Phase Doppler measurements near the nozzle in a low pressure water spray. In: 2nd Symp. Liquid Part. Size Measurement Tech. ASTM, Atlanta/GA Sellens, R. W. 1990a: A derivation of the phase Doppler measurement relations for an arbitrary geometry. Exp. Fluids 8, 165-168 Sellens, R. W. 1990b: Alignment errors in phase Doppler receiving optics. Part. Part. Syst. Characterization (in press) Received October 16, 1989