Kari Määttä, Juha Kostamovaara, Risto Myllylä1kari/Papers/appliedoptics.pdf · Kari Määttä...

1

Kari Määttä Applied Optics

Profiling of hot surfaces by pulsed time-of-flight laser rangefinder techniques

Kari Määttä, Juha Kostamovaara, Risto Myllylä1

Electronics Laboratory, Department of Electrical Engineering, University of OuluSF-90570 Oulu, Finland

1Technical Research Centre of Finland, Optoelectronics LaboratoryP.O. Box 202, SF-90570 Oulu, Finland

ABSTRACT

The possibilities for using the pulsed time-of-flight laser radar technique for hotrefractory lining measurements are examined and formulae are presented forcalculating the background radiation collected, the achievable signal to noise ratio andthe measurement resolution. An experimental laser radar device is presented based onthe use of a laser diode as a transmitter. Results obtained under real industrialconditions show that a SNR of 10 can be achieved at measurement distances of up to15 - 20 m if the temperature of the converter is 1400°C and the peak power of thelaser diode used is 10 W. The single shot resolution is about 60 mm (sigma value), butit can be improved to mm range by averaging techniques over a measurement time of0.5 s. A commercial laser radar profiler based on the experimental laser radar device isalso presented, and results obtained with it in real measurement situations are shown.These measurements indicate that it is possible to use the pulsed TOF laser radartechnique in demanding measurement applications of this kind in order to obtainreliable data on the lining wear rate of a hot converter in a steel works. Key words:Range finders, laser optics, thickness measurement, radar, optical.

2


I. INTRODUCTION

Pulsed time-of-flight (TOF) laser distance measurement techniques based onsemiconductor laser diodes have been developed extensively for industrial inspectionpurposes in recent years /1/ /2/. The fact that the measurement point is known is animportant advantage in profiler applications, as the transmitter beam can be focusedon a small spot by means of a single lens. The size of the optics can be reduced due tothe shorter wavelength of the light compared with microwaves, and this enables small,even handheld laser systems to be constructed. The TOF technique makes it possibleto achieve a larger linear dynamic measurement range with smaller constructiondimensions than by the triangulation method /3/. The radiation source in the pulsedTOF technique is usually a semiconductor laser diode, which is stable, long-lived andallows a pulse repetition rate of several kiloherz which means reliable and fastermeasurement than with devices using the other TOF principle, modulation of acontinuous wave (the CW technique). The pulsed TOF technique always gives anexact result for an object at an unknown distance, while the CW technique usuallyrequires the distance to be measured at at least two modulation frequencies to producean exact result /4/. The transmitted laser pulses used in the pulsed TOF technique areusually narrow, which means that a high peak power of up to tens of watts can beachieved while the average optical power level remains low. This and the well definedbeam are important points if measurements are to be carried out in environmentswhere a large number of workmen are present. If the received signal-to-noise-ratio(SNR) is small, so that the single shot resolution is not good enough, resolution can beimproved as far as the millimetre range by using an averaging technique without themeasurement time becoming unnecessarily long.

One important field of application for this technique concerns profiling measurementsin industry e.g. measurement of the shape of huge steel blocks in shipyards /5/, whichcan be done to an accuracy of a few millimetres before assembly. This means savingsin assembly time and costs. Another application in which benefits can be achieved,especially in terms of a faster measurement time, is described in this article. This isconcerned with checking the thickness of hot refractory linings in steel works /6/, /7/,/8/.

In the modern LD-KG (Linz Donawitz - Kawasaki Gas) steelmaking process adoptedat the Raahe Steel Works in Finland in 1984, the LD converter in which the steel isproduced is a large, cylindrical steel vessel lined with heat-resistant bricks. Thediameter of the converter body at the Raahe Steel Works is 5.4 m and its depth 7.1 m,while the lining thickness varies between 60 cm and 100 cm, being thickest on thebottom of the converter. A typical campaign time for the converter is about1400 - 1600 heats, which takes about 50 days, after which it must be relined, at a costof about $250,000. To achieve maximum savings, the converter should be used for aslong as possible without the risk of burn-out becoming too high. These requirementsneed an effective, reliable means of checking the lining thickness, which it should bepossible to measure between tapping and recharging to prevent unnecessary costs andtime-consuming delays in production. The total profiling time should also be as shortas possible, preferably less than 10 minutes, to prevent the converter from cooling

3


down too much. This means that the temperature of the lining during measurementwill be high, usually between 1100°C - 1400°C, which induces severe backgroundradiation and noise problems.

The possibilities for applying TOF techniques to the profiling of hot surfaces arediscussed here. First, the theory of the measurement method is examined as far as theamount of background radiation, noise and achievable signal to noise ratio areconcerned, and then the basic construction of the experimental laser radar device,containing an optomechanical measuring head and distance measurement electronics,is presented. After this, basic measurement results obtained in the laboratory and inreal industrial environments are shown, and finally a commercial laser profiler,LR-2000, based on the prototype described in this article is presented and some realmeasurement results obtained in hot converters are quoted.

II. SIGNAL AND NOISE

A. Background radiation

Many problems arise when one is carrying out TOF laser profiling measurements in ahot converter. Firstly, the high temperature introduces background radiation whichdetracts from the SNR and the single shot resolution of the measurement or increasesthe measurement time. Secondly, the speed of light is dependent on the refractiveindex of the air, which is a function of temperature /9/, and this variation in therefractive index may alter the transit time of the laser pulses and cause instability asthe temperature of the air in front of the target varies. Possible temperature gradientsalong the path of the measurement beam can also cause local variations in therefractive index, which may disturb the beam and cause instability in themeasurements. Thirdly, the temperature of the air decreases further away from the hotlining of the converter and if the surface is measured at an angle, the laser beam canbe refracted because the refractive index changes. This may cause non-linearity in theresults at different measurement angles. Also a huge amount of dust and smog existsin front of the converter immediately after tapping, and this can attenuate the receivedsignal and scatter some of it back to the receiver, which can take the form of anunwanted stop pulse prior to the real one.

The radiance of an ideal black body can be calculated by integrating Planck's law ofblack body radiation over a desired wavelength interval. The radiance of a real targetis obtained by multiplying the spectral radiance of the black body by the emissivity εof the surface. ε is usually a function of temperature and wavelength /10/, but in thisarticle it is assumed to be a constant (0.8), as no exact estimate of its variation as afunction of temperature is available. An example of the radiance LB of a grey bodyand the wavelength λP at which LB is at its maximum are presented in Table 1,calculated around the wavelength of a GaAs single heterojunction laser diode,906 nm.

To achieve an estimate of the background radiation collected from a hot target, aschematic diagram of the typical receiver optics of a laser profiler constructed withone positive lens is presented in Figure 1. The optics consist of a detector of area AD

4


at a distance s1 from the lens. The signal reflected from the target is collected on thedetector by the receiver lens, with an area of AR and focal length fR. The backgroundradiation is reduced by a narrowband optical interference filter located either in frontof the receiving lens or between the lens and the detector. The receiving optics areadjusted in a such way that the lens forms an image of the detector of area AI at adistance s1' from itself. This image can be either real or virtual, which means that thefocus distance can be from minus infinity to plus infinity. The distances between thedetector and lens and the image and lens can be calculated according to the Gaussianlens equation.

Calculation of the background radiation PB is based on the formula:

P L A L AA

sL AB R B R I R B R

IR B I R= = =τ τ τΩ Ω

’,

12 (1)

where τR is the transmission of the receiver optics. Eq. (1) can easily be understood tobe true if the hot target is at the distance of the image plane of the receiver lens, but itis in fact valid at all target distances. If the position of the target changes, the detectorand its image remain fixed, and the detector can be thought to be replaced at theposition of its image. Since the solid angle which collects radiation ΩR = AR/s1'2 andthe area of the detector image AI are independent of the position of the target, thebackground power collected remains constant even if the distance between theradiating target and the image of the detector changes. The situation is the same as ifthe detector is looking at the radiating target through a hole of distance s1' and areaAR. The assumption that the background power received is independent of the targetdistance as long as the focus distance remains constant was confirmed by measuringthe optical radiation received from a hot oven at a temperature of about 800°C withthe optics shown later. The results are given in Figure 2.

To reduce the background radiation by optical design methods, the product ARΩIshould be kept as small as possible. This product can be expressed as a function of thearea and focal length of the receiver lens, the detector diameter and the focus distancewith the formula

.’

1’

)’(

’

2

1222

1

21

21

−=−==Ω

s

f

f

AA

fs

fsAA

s

AAA R

R

DR

R

RDRIRIR (2)

Eq. 2 implies that a change in the focus distance from 1 m to infinity, for example,will increase the product AR⋅ΩI only by a factor of about 1.3 if the focal length of thereceiver lens is 120 mm. Thus the focus distance of the receiver optics can be freelyadjusted without having much effect on the received background radiation.

5


B. Total noise of the system and its contribution to measurement resolution

The resolution of a TOF measurement is determined by the signal to noise ratio of itsdetection. A simplified formula by which the distance resolution σRCL of a single shotmeasurement can be estimated is /11/

σ σRCL

nc

du dt

c

B SNR= ≈

2

0 35

2

., (3)

where c is the speed of light in a vacuum, σn the root-mean-square value (RMS) of thenoise, du/dt the slope of the timing pulse at the moment of timing and B thebandwidth of the timing pulse. The slope of the timing pulse in Eq. (3) isapproximated by the peak value of the signal us divided by the risetime tr of thesignal, and the risetime by the formula tr ≈ 0.35/B. SNR is the signal to noise ratiouS/σn. If the number of measurements to be averaged is N, then the resolution in Eq.

(3) is improved by a factor of 1/N .

Eq. (3) can be applied directly only to constant level detection and it assumes that thenoise in the reference level is small compared with that in the signal. Timingdiscrimination in pulsed TOF laser radar devices is determined by a constant fractiontiming discrimination (CFD) principle of some kind /12/, in which the timing isperformed by dividing the incoming pulse into two parts, delaying one part in a suchway that the timing moment occurs when the falling edge of the undelayed pulse andthe rising edge of the delayed pulse cross, at a moment when the amplitude of thesepulses is about a half of the maximum value. In this detection principle the rms valueof the noise in the rising edge is about the same as that in the falling edge at themoment of timing. If the noise levels in the leading and trailing edges areuncorrelated, the distance resolution of the CFD σRCF can be estimated to be

( ),2

22

dtdudtdu

c

TL

TLRCF +

+≈

σσσ (4)

where σL, σT, duL/dt and duT/dt are the RMS noises of the leading and trailing edgesof the timing pulses and the slopes of these edges, respectively. If σL = σT = σn andthe slopes of the leading and trailing edges are equal, then the distance resolution ofthe CFD detection is σRCL/ 2 .

Eq. (4) should also be applied to the detection of the start signal. If the noises in thestart and stop pulses can be assumed to be uncorrelated and if the timing moment inthe start channel is also determined on the CFD principle described here, the totalresolution σR can be approximated by summing the distance resolution of the start andstop channels as random variables

σRSTART STOP

c

B SNR SNR= +0 35

2 2

1 12 2

., (5)

6


where the signal bandwidths in the start and stop channels are assumed to be equal.Any other jitter sources which reduce the resolution, such as the single shot resolutionof the time interval measurement must also be summed as a random variable in Eq.(5) to obtain the total resolution.

TOF rangefinding systems involve three main noise sources which reduce the SNR:the signal itself, the noise in the electronics (preamplifier and AP diode) andbackground-induced noise. The summary of each noise source is shown in Table 2,reduced to current noise in the input of the preamplifier. Typical numerical values foreach noise component are shown in Table 3.

The signal level at the timing moment is scaled from the peak value of PS using thefactor kTP in the calculation of the signal noise. If the timing point is at a half of thepeak amplitude of the pulse, a first approximation to the value of kTP is 0.5 if thebandwidth of the receiver channel is about the same as that of the optical pulse orhigher. The primary response Ro of the silicon AP diode is 0.44 A/W at a wavelengthof 900 nm, if the quantum efficiency is 60%. The term F(M) is the excess noise factorof the AP diode, which is a function of the APD gain M. There are several formulaefor the excess noise factor F(M) in the literature, of which that given by themanufacturer of the diode /13/ is used in these calculations:

F MM

M( ) . ( ) . .= − +0 98 21

0 02 (6)

The value of the signal noise in Table 3 was calculated with two values of PS,corresponding to ratios of the peak signal current to the rms value of the electronicsnoise current of 10 and 100. The terms k, q and Bn in Table 2 corresponds toBolzmann's constant, electron charge and the noise bandwidth of the receiverelectronics, respectively.

Noise in the AP diode is caused by the detector dark current, which contains twoparts: Ids which is not multiplied, and Idb, the multiplied part /14/. The amount ofnoise contributed by the dark current depends on the active area of the diode, so thatdiodes with a small area (0.2 mm2), like C30902E, have noise which is about onetenth of that of the preamplifier and can thus be neglected.

The noise contribution of the transimpedance-type preamplifier arises from foursources: the base current IB, the base spreading resistor and collector current IC of thefirst transistor stage and the feedback resistor RF of the preamplifier itself. If the totalcapacitance in the input of the amplifier Cin is small, the dominant noise sources inthe preamplifier will be IB and RF /15/.

The reflected current noise in/ Bn in the input of the preamplifier is thus

i

Bq k P P R M F M in

nTP S B o na= + +2 2( ) ( ) . (7)

7


When the target temperature increases, noise from the background radiation becomesthe dominant noise component, whereas below about 760°C the noise contribution of

the electronics is the dominant one, with a value of about 6.6 pA/Hz in thisexperimental system.

C. Received signal as a function of distance

The optical signal PS received at the detector is usually approximated as a function ofdistance by the radar equation, which gives a good approximation over long distances.In this case the whole signal which reaches the receiver lens goes on to the detector,reduced only by the transmission of the receiver optics. Over short distances only afraction of the signal collected by the receiver lens goes on to the detector, because theimage of the illuminated target does not usually fit entirely on the detector surface.The signal received at the detector surface, PS(x), can be approximated at all distancesx by the modified radar equation:

P xP A

xT x eS

T T R R x( ) ( ) ,= −ρ τ τπ

γ2

2 (8)

where ρ is the reflectivity of a diffusely reflecting target in the direction of the normalof the surface, PT the optical power of the laser diode and τT the transmission of thetransmitter optics. The exponent term represents a coefficient by which the signal isattenuated in the round trip for atmospheric reasons /16/. The constantγ, an "extinction coefficient", represents the attenuation factor, which is usuallyassumed to be zero except when measuring distances over hundreds of metres or ifatmospheric conditions are very severe due to dust, smog or smoke, for example. Theterm T(x) in Eq. (8) represents the effect of the fixed optics on the signal level, andcan be assumed to have a value of 1 only if the detector sees the whole areailluminated by the transmitter. As explained above, this is usually the case with longdistances, and also with short distances if the transmitter and receiver optics can befocused and their beams can be kept overlapping at the measurement distance.

The value of T(x) is dependent on the distance, usually in a highly complicatedmanner depending on the construction and adjustments of the optics. This isimpossible to evaluate in symbolic form because the transmitter beam profile is notuniform at out-of-focus distances and the receiver usually sees only a fraction of it atshort distances if paraxial optics are concerned. To calculate T(x), the radiance causedby the transmitter and the solid angle at which the receiver is seen must be determinedand the product of these two must be integrated over the whole area from which thereceiver optics collect the reflected signal. Details of this calculation, which is bestcarried out in numerical form by means of a computer program, are presented in /17/.As an example, three simulated curves (γ = 0) with different optical adjustments,together with the 1/x2 curve obtained by the radar Eq. (8) when T(x) = 1, are shown asa function of distance in Figure 3. The scale is set to 1 at a distance of 8.5 m, wherethe curve A reaches its maximum, and the other two signal curves are scaled withrespect to this value. The adjustment of the optics clearly has a great effect on the

8


amount and dynamics of the received signal in the measurement area of 3 m - 30 m,for example.

The signal current iS in the input of the preamplifier can be obtained from Eq. (8) bymultiplying it by the response of the AP diode:

i x P x M RS S o( ) ( ) .= (9)

D. Estimation of the signal-to-noise ratio

The signal to noise ratio at the detection level can be expressed by combining Eqs. (7)and (9):

SNR xi x

i

M R P x

q k P x P R M F M B i B

S

n

o S

TP S B o n na n

( )( ) ( )

( ( ) ) ( ).= =

+ +2 2 2(10)

In real measurement situations in iron works, the dominant noise component when thetemperature is above 760°C is the background-induced noise. It can be seen from Eqs.

(1), (8) and (10) that in this case the SNR is proportional to AR/ΩI . To increase theSNR, the transmitter spot should be made as small as possible, so that receiver beam,and thus ΩI, can be made small. This can be done by increasing the focal length of thetransmitter lens, but at the same time its diameter must be increased to avoid NAlosses in the transmitter optics. The diameter of the active area of the radiating sourcecan be reduced, of course, but the output power usually decreases, too. On the receiverside, any increase in the diameter of the receiver optics will increase the SNR. Thenumerical aperture of the detector sets the upper limit for the diameter of the receiverlens unless its focal length is increased, which will then reduce the receiver beamdiameter.

III. CONSTRUCTION OF AN EXPERIMENTAL LASER RADAR

A. Construction of the distance measurement electronics

The main parts of the distance measurement electronics are a laser transmitter, tworeceiver channels for start and stop pulses, a time-to-digital converter (TDC) and aµP-based control unit. A simplified schematic diagram of the electronics is shown inFiqure 4, a more detailed description is presented in reference /6/ and typical electricalparameters are summarised briefly in Table 4.

The transmitter consists of a pigtailed semiconductor laser diode and pulsingelectronics based on avalanche breakdown of a specially selected transistor. The laseris inserted inside a box, the temperature of which is stabilized by a Peltier element.This improves the stability of the distance measurement with respect to temperature.

9


The start channel consists of a pigtailed PIN photodiode, transimpedance preamplifier,two post-amplifiers and a CFD to convert analogue pulses to accurate ECL-level(emitter-coupled logic) start pulses. Optical start pulses are generated in the measuringhead and converted to current pulses by the PIN photodiode. Τhe signal bandwidth ofthe receiver channel is matched to that of the optical laser pulse and the amplificationof the start channel is adjusted so that the signal level in the CFD is at its optimumvalue.

The stop channel includes a pigtailed silicon avalanche photodiode, preamplifier, gaincontrol electronics, two post-amplifiers and the CFD. An AP diode is used becausethe signal level received from the converter is too low for a PIN photodiode. Gaincontrol is performed electrically and has a dynamic range of 25 dB, which is usuallysufficient when making measurements in actual factory environments. If moreattenuation is needed, the processor lowers the gain of the AP diode by reducing thebias voltage.

The timing moment in the start and stop channel CFD is based on crossing of theleading and trailing edges of the delayed and direct timing pulses, respectively, so thattiming does not take place at the peak signal level but at the half-amplitude point. Ashort ECL-timing pulse of about 10 ns produced by the CFD is transmitted to the TDCat the timing moment as a start or stop pulse. The typical walk error of this kind ofCFD is within +/-3 mm in a dynamic range of 1:4. Distance measurement is possibleif the ratio of the peak value of the signal to the rms value of the noise from theelectronics is greater than about 10. With a lower ratio there are too many unwantedstop pulses in the receiver channel caused by noise, which slows down reliabledistance measurement too much.

The time interval between the start and stop pulses is measured with the TDC, whichis a fast, accurate, stable time interval measuring device employing a digital countingtechnique together with an analogue interpolation method /18/.

B. Construction of the optical measurement head

The optomechanical measuring head of the experimental laser radar consists ofparaxial optics with a separate transmitter and receiver channel. Paraxial optics werechosen because they have a better transmittance than coaxial optics, thus improvingthe inherently low SNR of measurements involving hot refractory linings. Theschematic diagram of the optics is shown in Figure 5, a detailed description of themeasurement head can be found in reference /6/ and typical optical parameters aresummarized in Table 5.

The pulses from the laser diode are guided to the measuring head by an optical fibre,and the light emerging from the fibre is collimated and focused by a high qualityachromatic lens. A fraction of the transmitter light pulse is picked up as a start pulse.This optically generated start pulse improves the stability and resolution of the systemrelative to an electrically generated start pulse. An optical interference filter is placedin front of the transmitter fibre to reduce the heating effect of a hot target. Thediameter and focal length of the receiver lens are the same as those of the transmitterlens. The stop fibre feeds the light to the AP diode of the stop channel receiver, and in

10


front of it there is a narrowband interference filter which deflects most of thebackground radiation away. A heat-resistant glass shield is placed in front of thetransmitter and receiver lenses to protect the optics from dust and other mechanicalhazards present in industrial environments.

The total losses in the transmitter optics consist of NA mismatch losses between thetransmitter fibre and lens and transmission losses in the interference filters, glassshield and air/glass surfaces of the transmitter lens, and those in the receiver optics oflosses in the glass shield, air/glass surfaces of the receiver lens and interference filterand connection losses between the receiver fibre and detector. The detector is placedinside a receptacle to which the receiver fibre is connected with a SMA-type fibreconnector. The fibre/diode transmission τFD is calculated from the core diameter ofthe fibre, the diameter of the active area of the detector (0.5 mm), the distancebetween the fibre and detector (1.3 mm) and the NA of the receiver fibre (0.3, 1/e2 -intensity level).

The diameter of the measurement beam varies as a function of the measurementdistance, being a minimum of 25 mm at the focus distance of the transmitter, 10 m.Variations in the reflectivity properties and undulations of the target surface inside themeasurement beam area can cause error in distance measurement of the order of sometens of millimetres. The finite size of the measurement spot rounds off any sharpstepwise variations in the surface and smooths out any undulations of the surface. Theeffect of these variations of the surface can be estimated if the beam intensity profile,the shape of the laser pulse in the time domain and the principle of the timingdiscrimination method are known.

IV. MEASUREMENT RESULTS UNDER LABORATORY CONDITIONS

Results regarding the resolution, linearity and stability of the experimental distancemeasurement electronics are presented here with respect to temperature and time asmeasured under stable laboratory conditions.

The measured and calculated single shot resolutions of the electronics as a function ofthe SNR in the stop channel are shown in Figure 6. The SNR is expressed as the ratioof the peak value of the signal current to the rms value of the noise current of theelectronics. The single shot resolution of the TDC, 6 mm (sigma value) /18/, is addedto the calculated resolution. The value of the SNR was altered during measurement byusing neutral density filters in front of the transmitter optics. The calculated andmeasured resolutions are in good agreement and the resolution at high SNR values islimited to about 10 mm. The ratio of the peak value of the signal current to the rmsvalue of the noise current of the electronics in the start channel was about 90.

The effect of averaging on the resolution can be seen in Figure 7, where the measuredresolution is plotted at two SNR values. The measured value corresponds to this, as Nvaries from 1 to 4000. A small deviation from the 1/N curve can be seen in bothcurves at large values of N. The resolution can be improved to a mm range, however,if the SNR is more than10 and if N is about 1000.

11


The non-linearity of the device as a function of distance is presented in Figure 8. Themeasurement was performed on a black lining brick with a reflectivity correspondingto that of the cool converter lining, 0.06. The brick was moved along a scale in 0.5 msteps and was capable of being positioned with an estimated accuracy of +/-1 mm.The maximum signal was collected at a distance of 10 m and the variation in signallevel was about 23 dB in the area measured. The gain control electronics had aninternal delay which depended on the attenuation. This delay was partly compensatedfor by the processor in Figure 8.

The stability of the electronics with respect to ambient temperature is shown in Figure9. The temperature of the laser was stabilized to +20°C by the Peltier element. Thedrift of the electronics without temperature compensation is linear and about-1.1 mm/°C. If the temperature of the laser had not been stabilized, the drift wouldhave been non-linear, as presented in /6/, due to the variation in the shape of theoptical pulse in the time domain, for which the CFD cannot compensate. Thetemperature stability of the TDC of this device is better than -0.2 mm/°C, whichmeans that one probable reason for the temperature drift is non-equal drift between thestart and stop channels. The drift was found to be independent of the measurementdistance and the amplification of the receiver channel, which means that it can easilybe compensated for. Figure 9 also presents the drift of the device when thetemperature drift is compensated for by the processor according to a preprogrammedtemperature/drift table for the electronics. The stability of the temperature-compensated electronics is about +/- 4 mm over the whole temperature range-20°C - +40°C.

The stabilization time of the electronics after power on is shown in Figure 10. Themeasured distance is stabilized to an accuracy of +/- 5 mm within 5 min and to anaccuracy of +/- 1 mm within 20 min. The temperature of the electronics rises about4°C during stabilization. The total drift during stabilization is -20 mm, of which theTDC accounts for 3.5 mm. The reason for the stabilization drift lies in the two-channel electronics. The start and stop channels have non-equal drift during thermalstabilization of the electronics.

The long-term stability of the electronics has been measured over a time period ofabout 60 hours, and the drift in the measured distance was found to remain within+/- 3 mm. The temperature of the electronics varied periodically by about +/-3°Cduring measurement.

V. MEASUREMENT RESULTS IN REAL ENVIRONMENTS

The measurements described in this chapter were made with the experimental laserradar device in an actual industrial environment, the Raahe Steel Works ofRautaruukki Ltd, during normal operation.

A. Background radiation

The background radiation collected by the optics was measured with theoptomechanical head described earlier and an optical power meter. The target was a

12


hot converter and distance to the bottom was 10 m. Measurements were made at aright- angle to the bottom lining of the converter. The background radiation was fed tothe power meter via an optical fibre of the same type as used in the stop channel. Thefocus distance of the receiver optics was 24 m. The temperature of the convertersurface was measured with a pyrometer.

The measured and calculated background radiations in the receiver fibre are shown inFigure 11 as functions of the temperature of the converter. The background power wascalculated using Eq. (1). The transmission of the receiver optics from the shield glassto the output of the receiver fibre, without the connection coefficient between the fibreand detector, was 0.67.

The measured and calculated values are not exactly the same if the emissivity is 0.8,as was assumed. Instead, an emissivity value of 0.7 gives results within +/- 10% ofmeasured background radiation. The fact that the measured background radiationdiffers only slightly from the calculated value nevertheless proves that the design ofthe optics is correct, for the optics collect radiation only within the solid angledetermined by the diameter of the receiver fibre and the focal length of the receiverlens. There is no excess radiation reaching the measurement head from outside thesolid angle and transmitted to the detector by internal reflections. The interferencefilter effectively blocks the background radiation, which lies beyond the passband.

B. Gain of the AP diode

To measure the dependence of the gain of the AP photodiode as a function of biascurrent, the receiver fibre was installed in the transmitter optics during the previousmeasurement and the transmitter optics focused at the same distance as the receiveroptics, with their axes overlapping over a measurement distance of 10 m. Thus theoptical power meter and AP diode were looking at the same point in the converterthrough identical optics. The dependence of the gain of the AP diode with respect toambient temperature was corrected by increasing the bias voltage by a factor of+0.7 V/°C .

The gain of the AP diode M was calculated as a function of the measured bias currentIB by the equation

M =I

P R B

FD B oτ. (11)

The calculated and measured bias currents of the AP diode are shown as a function ofthe temperature of the converter in Figure 12 and the gain of the AP photodiode as afunction of the bias current in Figure13.

The measured gain of the AP diode at a current level of 10 µA is only slightly morethan the minimum value promised by the manufacturer on the diode datasheet. Thecalculated gain would be higher if the transmission of the fibre-diode connection waslower, but as mentioned earlier, the given value was calculated from the geometry ofthe connection. The decrease in the gain of the AP diode when the dc-bias current is

13


increased from 10 µA to 80 µA is 27%. A decrease in gain of this kind wasascertained later in laboratory measurements carried out on a number of AP diodes.According to the manufacturer, the response of the AP diodes is linear over a wideinput power level, 0.01 pW - 10 nW, depending on the bandwidth and quantumefficiency /19/. Gain saturation can occur due to a number of effects, however,principally voltage drop across the load resistance or bias circuitry, heating effects andspace-charge effects. All three reasons seem to be probable in this application.

The 80 µA bias current level corresponds to a converter temperature of 1405°C, whichis achieved immediately after the molten steel is poured out. The maximum allowablecontinuous dc reverse current of the AP diode is 200 µA /13/, which can be estimatedto be achieved if the temperature of the converter is about 1530°C and no bias currentlimitation is included in the bias circuitry.

C. Background radiation-induced noise

The background radiation-induced noise was also measured at the same time as thegain of the AP diode. The radiation collected by the optics from the hot converter wasguided to the AP diode via the stop fibre and the noise level was measured at theoutput of the stop channel amplifiers with a wideband rf millivolt meter. Themeasured and calculated noise curves, reduced to the input of the preamplifier, arepresented in Figure 14 as a function of the temperature of the converter.

Three noise curves are shown the first obtained from the calculated backgroundradiation, Eq. (1), after which the noise was calculated by Eq. (7), the secondcalculated from the measured background radiation, as shown in Figure 11, alsoemploying Eq. (7), and the third calculated from the measured electrical noise, scaledto the input of the preamplifier by dividing it by the bandwidth and gain of thepreamplifier and postamplifiers. In all these calculations the gain and excess noisefactor of the AP diode are corrected with respect to the bias current shown earlier.

All three noise curves are almost linear and identical in slope, having only differentamounts of offset. If the emissivity were expected to have a value of 0.7, the first andsecond curves would be identical. It can be deduced by extrapolation from all thecurves that the background radiation-induced noise is equal to the electronics noise6.6 pA/ Hz at temperatures of about 800°C.

D. Signal-to-noise ratio

The SNR was measured with a laser transmitter of pulse power about 10 W. Thetransmitter and receiver optics were focused and their axes set to cross at 24 m and10 m, respectively. The measurement distance was 10 m. The signal and noise levelsin the output of the stop channel amplifiers (input of the CFD) were measured with awideband oscilloscope and a rf millivoltmeter, respectively.

Measurement was begun immediately after tapping of the converter, at a temperatureof about 1400°C. The received signal increased about 28% during measurement,partly due to the increase in the gain of the AP diode, while the signal level varied

14


randomly by about +/-10% with a frequency of about 1 - 2 Hz, probably on account ofthe smoke and dust issuing from the hot converter.

The measured and calculated SNR values are presented in Figure 15. The signal levelwas calculated by the radar equation (8) at a distance of 10 m, in which caseT(x) = 0.6 because the focus distance of the transmitter and receiver optics was 24 mrather than 10 m.

The SNR was about 15 immediately after commencing measurement, after which itincreased to about 55 by the end of measurement about 15 min later. The calculatedSNR curve has values about four times larger than the measured ones over the entiretemperature range, and since the measured and calculated noise levels are about thesame, the reason must lie in the intensity of the received signal. The extinctioncoefficient γ is assumed in these calculations to be zero, whereas if it is assumed to beof 0.069 m-1 an attenuation of 1/4 will be obtained in the signal. This attenuationcould be caused by smoke or dust in the hot converter. On the other hand, theattenuation would probably not be constant during measurement because the dust andsmoke would disperse with time. The most probable reason is thus that the reflectioncoefficient of the hot target may vary depending on whether the measurement point isthe fire bridge itself or a coating of slag, other iron oxides or even pure iron on itssurface. Test measurements under laboratory conditions and the modified radar Eq.(8) give equal results when carrying out measurements on a target of knownreflectivity.

E. Stability of the distance measurement

A stability measurement was performed to assess the dependence of the measureddistance on the temperature of the converter. The speed of light is dependent on therefractive index of the air, which depends on the air pressure, temperature andhumidity /9/. The flow of hot air out of the converter can cause the measurement beamto refract, which can be seen as a variation in the measured distance.

The stability, signal level and single shot resolution of measurements of a hotconverter at a distance of 14 m are shown in Figures 16 and 17. The time 0corresponds to the point at which the converter was turned just after tapping so thatmeasurement became possible. The temperature of the converter was about 1400°Cinitially and about 1050°C by the end of the measurement. Reliable distance resultswere obtained 1.5 min after beginning the measurement because of the huge amountof smoke and dust rising in front of the converter, which caused a proportion of thesignal to scatter back, to be detected in the oscilloscope in the form of an additionalwide pulse of randomly varying position and amplitude in the receiver channel.

The measured distance drifted -5 mm during the first 5 minutes, after which the rate ofdrift decreased. This drift may be caused by variation in the refractive index of the airas the converter cooled down. All the results remained within +/-6 mm during the22 min measurement period, but the signal level measured by the system processordoubled during this time, partly as a result of an increase in the gain of the AP diodebecause the bias current decreased as the converter cooled down, and partly becausethe smoke and dust dispersed, reducing the extinction coefficient and increasing the

15


signal. The single shot resolution measured by the processor was about 50 mm(sigma-value) at first, which corresponds to the SNR value of 10 obtained from Figure6. By the end of the measurement the resolution had improved to 20 mm,corresponding to a SNR value of 50.

VI. SOME REAL LINING WEAR MEASUREMENTS

After developing the experimental version of the laser radar device, the authorsdeveloped two new laser radar instruments together with two Finnish firms, NoptelLtd. and Rautaruukki New Technology Ltd., based on more advanced electronics andoptics. This has resulted in the production of two commercial laser radar devices, theolder LR-1500 for point measurement in hot converters and the more recent LR-2000for automatic profiling of the bottom region of a hot converter.

A. Construction of the laser profiler LR-2000

The laser profiler consists of five units: an optomechanical measurement head, stepmotors, distance measurement electronics, power supply and a PC/AT-compatiblecontrol computer. A drawing of the laser profiler is provided in Figure 18.

The distance measurement electronics contain all the electronics needed to measurethe transit time of a very short laser pulse to a target and back. It has a uP controllerwhich communicates with the PC via a RS-232C serial bus. Signals between thedistance measurement electronics and the optical head are transmitted by opticalfibres. The optical head contains transmitter optics to collimate the laser pulse to thetarget and receiver optics to collect part of the reflected signal in the receiver fibre.The measurement head can be scanned with good accuracy and high speed by meansof two step motors which contain an integrated package of a servo motor, gear box,angle encoder and driver electronics. The scanning motors communicate with the PCvia a RS-232C interface. The power supply provides DC voltages for the other units,and also includes batteries, which make it possible to use the system for about 4 hourswithout a mains supply. The control computer includes an electroluminescent EGA-compatible flat display, keyboard, disc drive and program package to handle all themeasurement and analysis tasks. The whole measurement system is installed in anequipment cart which is easy for one man to move to the measurement site. Heatshield elements for protection against thermal radiation are also included, and theweight of the whole system is about 120 kg.

The measurement sequence is as follows. The first step is to fix the coordinate systemof the measuring equipment to the object coordinate system, after which automaticmeasurement can be performed according to a preprogrammed sequence. A manualmeasurement mode with manual scanning and single point measurement is alsopossible. The results are displayed on a screen and stored for post-processing andanalysis. The wear rate of the lining is obtained by comparing the results with thoseobtained upon first measuring the relined converter.

16


B. Measurement results obtained in normal use of the LR-1500 and LR-2000

The LR-1500 has been in use for several years for performing lining wearmeasurements at the Raahe Steel Works of Rautaruukki Ltd. /8/ /20/. It does notsupport computer-controlled scanning and must be aimed manually at the desiredpoint. It is designed for point measurements to obtain lining wear data on the tuyereregions of converters. These are the most critical areas because the rate of wear ishighest on account of the N2/Ar gas blown into the molten steel during the process.Two typical lining wear rates during campaign times with two types of lining brick arepresented in Figure 19. The typical measurement distance was about 10 m. The wholemeasuring process for six tuyeres lasts about 20 min, including all preparations,calibration and data processing.

The more advanced model of the device, LR-2000, with computer-controlledscanning, has been in use continuously for over a year at the Raahe Steel Works. Theprofiler has been used to measure lining wear in the bottoms of the converters and incritical regions of the sides. Typical lining wear results are shown in Figure 20. Thesolid line shows the new lining profile of the converter and the dashed line the liningwear rate after 849 tappings. The upper picture shows a cross-section about 2.5 mfrom the bottom of the converter seen from in front and the lower picture is the cross-section in the middle of the converter, left side up. Due to the shape of the converter,it is impossible to measure the whole of the bottom and both sides in a single profilingperiod. Both pictures in Figure 20 are combinations of results obtained in twomeasurement sequences measured from different places in front of the converter.

The number of measurement points was about 100 per measurement, the total timetaken being about 10 min, including fixing of the coordinates (1 - 2 min) and changingthe position of the profiler. The time per measurement point is about 0.5 second. Thisconsists of the time required to measure the distance, which is approximately 310 msbecause 500 successive results were averaged, reading and storing the result andturning the measurement head, which together take about 200 ms. The savings inmeasurement time compared with a laser profiler based on the modulation of acontinuous wave is obvious. The measurement speed of most other devices istypically between 3 - 10 second/point depending on the temperature and reflectiveproperties of the lining /7/ /21/.

VII. DISCUSSION

The possibilities for measuring hot refractory lining wear rate by pulsedTOF techniques are examined here, including considerations of the collection ofbackground radiation, the signal-to-noise ratio and its effect on resolution. Anexperimental laser radar device has been constructed to verify the calculations. Thetest system consists of distance measurement electronics and a separateoptomechanical measurement head connected to the electronics by optical fibres. Thedistance measurement electronics consist of a semiconductor laser diode, two channelreceivers for the start and stop pulses, gain control electronics and a novel timeinterval measuring system, the TDC. The optomechanical measurement head isconstructed with paraxial optics and separate transmitter and receiver channels toachieve maximum efficiency. The narrowband optical interference filter in the

17


receiver channel reduces the high background radiation from the hot converter andthus improves the inherently low SNR.

Test measurements in laboratory and industrial environments show that the achievabledistance measurement resolution of a pulsed TOF laser radar device can be estimatedto an accuracy better than a few percent using the formulae presented here. Thebackground radiation collected is independent of the measurement distance anddepends only on the dimensioning and adjustment of the receiver optics. The formulaeestimate this to an accuracy of +/-10%, but the accuracy to which the noise in thereceiver channel can be estimated is somewhat lower, +/-15%, due to variation insome parameters, such as emissivity of the hot target.

The pulsed TOF laser radar device is capable of measuring distances reliably if theratio of the peak signal to the noise of the electronics in the input to the constant-fraction-timing discriminator is greater than about 10. This signal to noise ratio can beachieved at measurement distances at up to 15 - 20 m if the temperature of the targetis 1400°C and the optical peak power of the laser at least 10 W. The single shotdistance resolution is about 60 mm (sigma-value) if the SNR is 10, but resolution canbe improved to the mm level by averaging, the measurement time still remainingshort, about 0.5 s. The measurement time is about one tenth of the time required bythe CW technique. The total accuracy of this experimental device is limited to+/-10 mm due to non-linearities caused by the electronic gain control and temperaturedrift related to the electronics with separate receiver channels for start and stop pulses.

A new commercial laser profiler, LR-2000, has recently been developed based on theexperimental version described above. This again uses the pulsed TOF principle, themain advantage of which as compared with the CW technique is its shortermeasurement time. The LR-2000 has been used successfully at Raahe Steel Works forone year for measuring the lining wear rates of three converters. No converterbreakouts have occurred during this time. Exact knowledge of the wear rate in eachcampaign has provided good data for selecting the optimal brick material, therebyincreasing converter productivity. The pulsed TOF laser radar device or modificationsof it may prove applicable in future to other demanding industrial measurementsituations requiring cm-class accuracy.

ACKNOWLEDGEMENTS

This project was carried out in co-operation with two Finnish firms, Noptel Ltd. andRautaruukki New Technology Ltd. The authors wish to thank them for their financialand technical support. Thanks are also due to the Raahe Steel Works of RautaruukkiLtd. and Heikki Tuomikoski for the opportunity to carry out measurements in realindustrial environments. Hannu Jokinen and Jukka Kauniskangas helped the authorsduring the measurements in Raahe and prepared the actual lining wear measurementresults, for which particular thanks are due.

18


REFERENCES

/1/ M. Manninen Task-Oriented Approach to Interactive Control of Heavy-DutyManipulators Based on Coarse Scene Description. Doctoral thesis. ActaPolytechnica Scandinavia, Mathematics and Computer Science Series No. 42,Technical Research Centre of Finland, Electronics Laboratory, 81 pp, 1984.

/2/ J. Kostamovaara, K. Määttä, M. Koskinen, R. Myllylä: Pulsed laser radars withhigh-modulation-frequency in industrial applications, SPIE Proceedings Volume1633 Laser Radar VII: Advanced Technology for Applications, pp. 114 - 127,Los Angeles, California 1992.

/3/ T. A. Clarke, K. T. V. Grattan, N. E. Lindsey: Laser-based triangulationtechniques in optical inspection of industrial stuctures, SPIE proceedingsVolume 1332 Optical Testing and Metrology III: Recent Advanced in IndustrialOptical Inspection, pp. 474 - 485, 1990.

/4/ D. E. Smith: Electronic Distance Measurement for Industrial and ScientificApplications, Hewlett-Packard Journal, Number 6, June 1980, Volume 31, pp.3 - 10.

/5/ M. Manninen, J. Jaatinen: Productive Method and System to ControlDimensional Uncertainties at Final Assembling Stages in Ship Production,Proceedings of the 1991 Ship Production Symposium by The Society of NavalArchitects and Marine Engineers, 7p., San Diego 1991.

/6/ K. Määttä, J. Kostamovaara, R. Myllylä: On the Measurement of Hot Surfacesby Pulsed Time-of-flight Laser Radar Techniques, SPIE Proceedings Volume1265 Industrial Inspection II, pp. 179 - 191, The Hague1990.

/7/ L. Green: Condition monitoring - its impact on plant performance at BritishSteel, Scunthorpe Works, Ironmaking and Steelmaking , Volume 17 no. 5. 1990,pp. 355 - 362.

/8/ K. Määttä, J. Kostamovaara, R. Myllylä: A Laser Rangefinder for hot Surfaceprofiling measurements, SPIE Proceedings Volume 952 Laser Technologies inIndustry, pp. 356 - 364, Porto 1988.

/9/ J. F. Ready: Industrial Applications of Lasers, Academic Press Inc., New York,1978.

/10/ R. Siegel, J. R. Howell: Thermal Radiation Heat Transfer, HemispherePublishing Corporation, New York ,1981.

/11/ G. Bertolini: Pulse shape and time resolution. In: Bertolini, G. & Coche, A.(eds.) Semiconductor detectors, North-Holland Publishing Co., Amsterdam,1968, pp. 243 - 276.

/12/ J. Kostamovaara: Techniques and devices for positron lifetime measurementand time-of-flight laser rangefinding. Acta Univ. Oul. C37, Electr. 9, 1986.

19


/13/ C30902E, C30902S, C30921E, C30921S Data Sheet, RCA Inc., Electro Optics

/14/ H. Kressel: Topics in Applied physics: Semiconductor Devices for OpticalCommunication. Springer-Verlag Berlin Heidelberg New York 1980

/15/ R. G. Meyer, R. A. Blauschild: A wide-band low-noise monolithictransimpedance amplifier. IEEE Journal of Solid-State Circuits, vol. SC-21, no.4, August 1986, pp. 530 - 533.

/16/ RCA Corporation: RCA-electro-optics handbook, Technical Series EOH-11,USA 1974.

/17/ J. Wang, K. Määttä, J. Kostamovaara: Signal Power Estimation in Short RangeLaser Radars, to be published in the proceedings of the ICALEO´91 Conference,Optical sensing and measurement Symposium, 10p., San Jose 1991.

/18/ K. Määttä, J. Kostamovaara, R. Myllylä: Time-to-digital converter for fast,accurate laser rangefinding, SPIE Proceedings Volume 1010 IndustrialInspection, pp. 60 - 67, 1988 Hamburg.

/19/ P. P. Webb, R. J. McIntyre, J. Conradi: Properties of avalanche photodiodes.RCA Review, RCA Inc., Electro Optics, June 1974.

/20/ M. Karppinen, J. Kostamovaara, R. Myllylä, M. Seppänen, K. Määttä: A NovelLaser Rangefinder System for The Profile Measurements of Refractory Linings,Proceedings of the UNITECR'89 Meeting, pp. 1340 - 1353, November 1 - 4Anaheim California USA 1989.

/21/ G. Klages: Measurement of refractory lining in metallurgical vessels.Metallurgical Plant and Technology, vol. 12, no. 6, 1989, pp. 32 - 41.

20


TABLES:

TABLE 1. Radiance of the grey body LB at different temperatures (λ = 906 nm,∆λ = 886 nm ... 926 nm = 40 nm and ε = 0.8).

t [°C] 800 900 1000 1100 1200 1300 1400 1500 1600

λP [µm] 2.70 2.47 2.28 2.11 1.97 1.84 1.73 1.63 1.55

LB [W sr-1 m-2] 2.3 8.1 23.4 58.3 127.9 254.1 465.1 795.1 1283.7

TABLE 2. Noise sources of the TOF-laser radar

Noise component Noise value

Signal - induced noise ins2 2 2q B P k R M F Mn S TP o ( )

AP diode noise ind2 2 2q B I I M F Mn ds db( ( ))+

Preamplifier noise ina2 2

4q B I

k T B

Rn Bn

F+

Background radiation - induced noiseinb2 2 2q B P R M F Mn B o ( )

TABLE 3. Typical noise levels [pA/Hz ]

ins ind ina inb

PS = 14.7 nW PS = 147 nW C30902 Cin = 0 pF Cin = 2 pF T = 760 °C T = 1400 °C

7.4 23.5 0.23 2.3 3.4 6.6 88

21


TABLE 4. Typical values of the electronics parameters

Optical power of the laser PT 10 - 20WFWHM of the optical pulse 10 nsRise time of the optical pulse 3.5 nsWavelength of the laser 906 nmFWHM of the laser wavelength 2 nmTypical pulse repetition rate of the laser 4 kHzElectrical bandwidth of the receiver B 100 MHzNoise bandwidth of the receiver Bn 122 MHzInput noise of the preamplifier ina 6.6 pA/ Hz Dynamic range of the electronic gain control 25 dBWalk error (1:4 dynamic range) < ±3 mmPrimary response of the AP diode Ro 0.44A/WGain of the AP diode M (PB = 0 W) 113Excess noise factor of the AP diode F(M) 4.2

TABLE 5. Parameters of the optical measurement head

TRANSMITTER:Focal length of the transmitter lens fT 120 mmDiameter of the transmitter lens 50 mmFocus distance of the transmitter 10 mDiameter of the transmitter fibre 0.3 mm

RECEIVER:Focal length of the receiver lens fR 120 mmDiameter of the receiver lens 50 mmFocus distance of the receiver 24 mDiameter of the receiver fibre 0.4 mmBeam overlapping distance 12 m

INTERFERENCE FILTER:Wavelength 906 nmbandwidth (FWHM) 40 nmbandwidth (10 % transmission) 80 nm

TRANSMISSION OF THE OPTICS:Total transmission of the transmitter optics τT 0.5Total transmission of the receiver optics τR 0.32Transmission of the shield glass 0.86Transmission of the air/glass surface 0.96Transmission of the fibre/diode connection τFD 0.50Transmission of the interference filter 0.78

22


s’1

AD

Image of thedetector

Receiver lens

Detector

ΩI

ΩI

ΩR

A R

s1

A I

Fig. 1. Schematic diagram of the receiver optics.

23


DISTANCE [m]

BA

CK

GR

OU

ND

RA

DIA

TIO

N[u

W]

0.4

0.6

0.8

1.0

1.2

1.4

0 2 4 6 8 10 12 14 16

Fig. 2. Measured background radiation in a hot oven as a function of distance at focusdistances of the receiver optics of 6 m ( ), 8 m ( ), 10 m ( ) and 12 m( ).

24


DISTANCE [m]

RE

LAT

IVE

SIG

NA

L

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 5 10 15 20 25 30

A

B

C

D

Fig. 3. Calculated beam overlap function with three adjustments of the transmitterand receiver optics.A: Transmitter and receiver focus distance is 10 m and optical axes cross at 10 m.B: Transmitter and receiver focus distance is 24 m and optical axes cross at 10 m.C: Transmitter and receiver focus distance is 24 m and optical axes cross at 24 m.D: Received signal observes the 1/x2 rule when it is fitted to curve A at 10 m or to

curve C at 24 m.

25


PRE-AMPLIFIER

PIGTAILED (400 um, 4m)SILICON PHOTODIODE

POST-AMPLIFIERS

CONSTANTFRACTION

TIMINGDISCRIMINATOR

(CFD)

PRE-AMPLIFIER

POST-AMPLIFIERS

CONSTANTFRACTION

TIMINGDISCRIMINATOR

(CFD)

PIGTAILED (400 um, 5 m)AVALANCHE PHOTODIODE

ELECTRONICGAIN

CONTROL

AMPLITUDEMEASUREMENT

TIME-TO-DIGITAL-

CONVERTER(TDC)

LASERTRANSMITTER

uP-BASEDCON-

TROLLER

RS-232C

START CHANNEL

STOP CHANNEL

ECL-START

ECL-STOP

PIGTAILED (300 um, 4 m)SEMICONDUCTORLASER DIODE

AMPLITUDE

DISTANCELASER CONTROL

GAIN

Fig. 4. Block diagram of the distance measurement electronics.

26


TRANSMITTERFIBRE 300um

STOP RECEIVERFIBRE 400um

START RECEIVERFIBRE 400um

SEMI-TRANSPARENT(93 %) MIRROR

BOTH LENSES:DIAM: 50 mmF: 120 mmHIGH QUALITYACHROMATS

DIFFUSE REFLECTING SURFACE

OPTICALINTERFERENCE FILTER

ROBAX SHIELD GLASSOPTICALINTERFERENCE FILTER

Fig. 5. Schematic diagram of the optomechanical measuring head.

27


SIGNAL-TO-NOISE RATIO IN STOP CHANNEL

RE

SO

LUT

ION

[sig

ma-

valu

e in

mm

]

0

10

20

30

40

50

60

70

80

1 10 100 1000 10000

Fig. 6. Calculated ( ) and measured ( ) single shot resolution of the electronics asa function of SNR (ratio of the peak value of the signal to the rms value of theelectronics noise).

28


NUMBER OF MEASUREMENT RESULTS AVERAGED

RE

SO

LUT

ION

[sig

ma-

valu

e in

mm

]

0

1

10

100

0.1 1 10 100 1000 10000

Fig. 7. Resolution of the electronics as a function of the number of measurementsaveraged, with SNR values of 10 ( ) and 270 ( ).

29


DISTANCE [m]

NO

N-L

INE

AR

ITY

[mm

]

-30

-20

-10

0

10

20

30

0 5 10 15 20 25 30

Fig. 8. Non-linearity of the electronics.

30


AMBIENT TEMPERATURE [°C]

DR

IFT

[mm

]

-40

-30

-20

-10

0

10

20

30

40

-30 -20 -10 0 10 20 30 40 50

Fig. 9. Temperature stability of the electronics with temperature compensation ( )and no temperature compensation ( ).

31


TIME [min]

DR

IFT

[mm

]

-30

-20

-10

0

10

20

30

0 5 10 15 20 25 30 35 40 45

20

21

22

23

24

25

26

ELE

CT

RO

NIC

ST

EM

PE

RA

TU

RE

[°C

]

Drift

Temperature

Fig. 10. Stabilization of the electronics after power on. The temperature drift of thedistance is compensated for by the processor.

32


TEMPERATURE OF A CONVERTER [°C]

RA

DIA

TIO

N [u

W]

0

1

2

3

4

5

6

1000 1100 1200 1300 1400 1500

Fig. 11. Measured ( ) and calculated background radiation in the hot converter. Theemissivity of the surface is 0.6 ( ), 0.7 ( ) and 0.8 ( ).

33



AP

D B

IAS

CU

RR

EN

T[u

A]

1

10

100

1000

1000 1100 1200 1300 1400 1500 1600

Fig. 12. Measured ( ) and calculated ( ) bias current of the AP photodiode as afunction of the temperature of the converter.

34


BIAS CURRENT [uA]

GA

IN

0

20

40

60

80

100

120

140

0 10 20 30 40 50 60 70 80 90

Fig. 13. Gain of the AP photodiode as a function of its bias current.

35



NO

ISE

CU

RR

EN

T [p

A/s

qrt(

Hz)

]

0

20

40

60

80

100

1000 1100 1200 1300 1400 1500

Fig. 14. Measured ( ) and calculated noise scaled to the input of the preamplifier asa function of target temperature. Calculations are based on the measured backgroundradiation ( ) or the calculated background radiation ( ).

36



SIG

NA

L-T

O-N

OIS

E R

AT

IO

0

50

100

150

200

250

1000 1100 1200 1300 1400 1500

Fig. 15. Measured ( ) and calculated ( ) signal-to-noise ratio as a function of thetemperature of the converter.

37


MEASUREMENT TIME [min]

DR

IFT

[mm

]

-20

-15

-10

-5

0

5

10

15

20

0 5 10 15 20 25

Fig. 16. Stability of actual measurements of a hot converter. Each measurement is anaverage of 1000 successive single-shot results.

38


MEASUREMENT TIME [min]

SIG

NA

L [m

V]

0

200

400

600

800

1000

0 5 10 15 20 25

0

10

20

30

40

50

RE

SO

LUT

ION

[mm

sig

ma-

valu

e]

Signal

Resolution

Fig. 17. Measured signal and single shot resolution of measurements of a hotconverter.

39


Fig. 18. LR-2000 laser profiler. Its height, width and depth are 160 cm, 55 cm and73 cm. The weight is about 120 kg. Thermal shields are not shown.

40


Fig. 19. Tuyere brick lengths during campaigns with two brick types.

41


Fig. 20. Typical converter profiling results. The grid scale in both pictures is 500millimetres.

Date post:	23-Jul-2020
Category:	Documents
Upload:	others
View:	3 times
Download:	0 times

Kari Määttä, Juha Kostamovaara, Risto Myllylä1kari/Papers/appliedoptics.pdf · Kari Määttä...

Documents