ࡱ> FHEM Tbjbj== xWWF7_lhhhh\\\4, 4G<`7+7+7+:::::::$3= S?:\7+*^7+7+7+:1hh<1117+ph\:17+:1x18>:b\:T I"d .V::4<0G<^:,@0@:1hhhh Timing-Error Minimization of Optoelectronic Pulse Signals Bo~idar Vojnovi Laboratory for Stochastic Signals and Processes Research, Division of Electronics, Rudjer Boakovi Institute, Bijeni ka 54, 10000 Zagreb, Croatia e-mail:vojnovic@irb.hr Abstract In optical distance measurements, using time-of-flight techniques, associated time interval is defined by start and stop pulse signals. Measurement error depends strongly on timing-error of stop optoelectronic pulse signal. The pulse processing and pulse timing circuits, based both on charge-controlled properties of step-recovery diode properties were optimised in a way to achieve minimum timing-error. The system is very suitable to achieve picosecond resolution in timing pulses having longer rise-times. 1. INTRODUCTION In optical (laser) distance measurements, using time-of-flight techniques, associated time interval is defined by start and stop pulse signals. In the analysis it was supposed that the start signal is fully deterministic, very sharp and precisely defined in time. The stop signal in general case is stochastic, described by parameters; amplitude, rise-time and associated additive noise. The problem of high-resolution timing of stochastic pulses has been treated extensively in many scientific and technical areas: nuclear spectroscopy, radar ranging, optical (laser) distance/displacement measurements, laser altimetry, optical communications etc. [1,2,3,4,5]. These analysis were focused mostly on two main separate topics: (1) influence of the stochastic pulse parameters on pulse timing-error, assuming the ideal timing discriminator and (2) development of new timing-discriminator circuits and their optimisation depending on the application and desired time resolution. In this paper an attempt was made to optimise previously described method [6], which functionally integrate detecting and processing circuits with timing discrimination circuit. The analysis was made under the following presumptions: (1) Optical pulse signal is stochastic, but its arrival-time is considered jitter free; (2) Gaussian, white noise characterizes optical and electronic noise sources; (3) Signal processing circuits belong to the class of linear time-invariant filters; (4) System output pulse has to be bipolar, and time discrimination occurs at signal zero-crossing point; (5) Timing discriminator does not belong to the class of regenerative circuits, thus avoiding the effect of triggering charge (energy) sensitivity. 2. PULSE PROCESSING-DISCRIMINATION SYSTEM ANALYSIS 2.1. Criteria for timing-error minimization Pulse signal timing is a part of wider area of stochastic pulse arrival-time (time-delay) measurements, which belong to the class of stochastic signal parameters estimation problems. Different criteria were considered here to study estimation error minimization. (1) Maximum likelihood (ML) criterion was widely applied in stochastic signal amplitude and time-delay estimation [4, 8]. The idea was to use ML criterion to find out the impulse-response function h() of the optimum filter (optimum estimator), which minimizes pulse timing-error. The problem was analysed for the case of signal accompanied with white, Gaussian noise, as well as for the case where the signal is non-stationary Poisson process. In both examples it was assumed the pulse width is mach shorter than the observation time. The minimum error was obtained at the time point, which maximizes signal output from the optimum filter. This method thus is not very applicable for nanoseconds pulse widths and required sub-nanosecond time resolution. (2) Minimum error-rate approach attempts to find out the filter that) minimizes the mean-number of zero-crossings of the noise component of the stochastic signal. This method could be interesting in the case when the spike noise is predominant. Theoretically white (current) noise contains the series of random positive and negative pulses qi(t1), where (t) is Dirac delta-function, qi and ti are stochastic variables (charge content and arrival-time of each pulse). The mean number of the noise zero-crossings is [11] is defined as:  EMBED Equation.3  (1) where: c is a constant, h() and h((() are the impulse response function of the noise filter and its second derivative, Ryy() and Ryy((() are the auto-correlation function of the noise at the filter output and its second derivative respectively. Minimum of the functional F[h()] gives us h() that minimizes  EMBED Equation.3 , which is well known  cusp shaped function. Because h() is unipolar function, as it is also pulse signal, zero-crossing timing is not possible and timing is amplitude sensitive. The optimal timing point is not explicitly defined and depends on the point where the noise-to-signal slope ratio has a minimum. (3) Minimum square-error criterion [10] leads to maximization of signal-to-noise ratio, which in the timing case means minimization of signal variation (noise)-to-signal slope ratio. This criterion is more appropriate approach if the signal variations (noise) are much slower than the signal rise-time. The analysis results in optimum (matched) filter that minimizes pulse timing-error against signal variations for assumed ideal timing discriminator circuit. To find optimum impulse response function h0() the following expression should be minimized  EMBED Equation.3  (2) where Tm is a timing point and s2 is the input signal variance which includes effects of statistical processes of optical signal generation, propagation and detection. The denominator in (2) represents squared average signal slope. From expression (2) a functional G=G[h()] could be made and minimized, which gives us h0() of the optimum timing filter. The general solution, which is of our interest, has the form  EMBED Equation.3  (3) where  is a factor indicating relative noise at the system input. 2.2. Theory of system operation The system will be analysed under the condition (S(Tm-), which is valid for larger optical distance measurements. In this case response of the filter has the form h0() = S((Tm-). The same result [5] was obtained replacing signal variance by noise variance in equation (2). The timing measurement chain contains the following essential parts: (1) optical signal detector (sensor); (2) electronic pulse preamplifier; (3) signal pre-processing circuit and (4) filter-discriminator circuit. Optical detector and preamplifier have to convert optical signal into electronic one with minimal response time and timing-jitter as well as with maximum signal-to-noise ratio. As the preamplifier noise is usually coloured, the pre-processing filter is aimed to whiten this noise, making simpler synthesis of filter-discriminator circuit to achieve minimum timing-error of the output pulse. Impulse response function of the whole optimum filter is thus convolution of impulse response functions of all parts in the chain. Detector signal, containing charge Q, was considered as current impulse, i(t) = Q(t). The input noise has two components: voltage and current white noises, having power spectral-densities Ne/2 and Ni/2 respectively. We consider detector-preamplifier part as a low-pass, single-pole circuit, thus the whitening filter is pole-zero cancellation circuit. Output of the filter could be approximated by the signal v(t)=Q/CTe-t/(c and white voltage noise with Ne/2, c2=CT2Ne/Ni, where c is the  noise-corner time constant, CT is the total input capacitance of the detector-preamplifier circuit. The impulse response function of the system (optimum filter) is first derivative of  cusp shaped function, having time constant  and shifted in time by Tm [5,9]. The impulse response function hs() of the proposed system (Fig.1) with step-recovery diode represents, under certain, realizable conditions, good linear approximation of the h0() of the theoretical (unrealisable) optimum timing filter. Step-recovery diode circuit is based on switching properties of step-recovery diode, a P-I-N junction structure, that acts as a charge-controlled switch. Diode forward current, static or transient, injects the charge into the diode. When this charge is being removed by the reverse pulse current, the diode continues conducting (low impedance state) until all the charge is removed. At the point when the total amount of charge in the diode becomes zero, it stops conducting (high impedance state) very abruptly. In our system the input current signal injects the charge into unbiased diode. The same pulse inverted and amplified by a factor k, starts after well-defined delay-time td to remove the stored charge. At the moment Tm when the total charge becomes zero, we get the sharp edge pulse having rise-time less than 100 ps, which gives us the time reference relying very strongly on pulse arrival-time. In the analysis it is assumed that the diode is ideal current integrator and charge zero-crossing discriminator, because the following requirements are met: (1) Injected charge is completely removed during reverse recovery phase; (2) The leakage charge due to reverse diode capacitance is negligible; (3) The minority carrier life-time could be chosen long compared to diode conducting time; (4) The diode does not exhibit charge (energy) triggering sensitivity because it does not belong to the class of regenerative circuits.  Figure 1. Impulse response function of the system Figure 2. Impulse response function of DL-SRD circuit That is important especially for slower pulse discrimination, because there is no time-walk due to different signal slope at the point of discrimination; (5) The diode is true time zero-crossing discriminator, thus avoiding the problem of initial discrimination level adjustment and stability. The best ultra-fast, low noise preamplifiers have rise-time in the order of nanosecond, integrating time-constant could be some milliseconds and c is usually in the range 10-100 nanoseconds. Under these conditions, the proposed system could be realized simply by combination of pulse amplifiers, delay lines (DL) and step-recovery diode (SRD) with minority carrier life-time constant mc greater than 100 ns. As the diode acts as an integrator, the DL-SRD circuit has impulse response function as on Fig.2. If the relation cTm is valid, the input of DL-SRD circuit sees the step voltage of the amplitude Vs= Q/CT and white voltage noise with Ne/2. The timing-error variance is, consequently  EMBED Equation.3  (4) 2.3. Discussion of results Calculation of timing-error with some realistic example of signal and circuits parameters (w=20ns,  EMBED Equation.3 =1nV/ EMBED Equation.3 , Q=10pC, CT=100pF, k=5), gives us T= 0.5ps. This result was obtained under the assumption of idealized circuits, as indicated before and without calculation of SRD noise contribution to timing error. Nevertheless, we can say that the proposed method is capable to achieve picosecond timing-resolution against the noise influence. Regarding the amplitude and rise-time variation as well as triggering charge sensitivity, we can conclude the following;(1) the method is insensitive to pulse amplitude variations;(2) For linear signal-rise at the DL-SRD timing circuit input, system is fairly insensitive to rise-time variations;(3) Timing discriminator does not exhibit triggering charge sensitivity, because SRD is non-regenerative discriminator circuit. The simplified version of DL-SRD circuit was tested in two types of measurements;(1) static, measuring time-walk T against pulse amplitude and rise-time, and (2) dynamic, measuring statistical distribution of output pulse time-delay. Obtained results gave us T =04ns, with amplitude and rise-time dynamics of 10:1, (20-200ns). Standard deviation of timing-error was calculated from measured time spectra (pdf of delay-times), and we got 120ps, 220ps and 300ps in three measurements with input pulse rise-times of 100ns, 220ns and 500ns respectively. The pulse amplitude dynamics was 10:1. 3. CONCLUSION The proposed DL-SRD timing circuit (charge balanced discriminator) is very suitable for applications in longer range optical distance measurements. Theoretical analysis and experimental results have shown that the method assures picosecond time-resolution even with wider pulse signals of several hundreds of nanoseconds, exhibiting significant insensitivity against signal amplitude and rise-time variations. REFERENCES E. Gatti, V. Svelto, Optimum Filter for Timing Scintillation Pulses, Nuclear Instruments and Methods, Vol. 39, pp. 309-313, 1966. J. Llacer, Optimum Filter for Determination of the Position of an Arbitrary Waveform in the Presence of Noise, IEEE Transactions On Nuclear Science, Vol. NS-28, No.1, pp. 630-633, 1981. E. Gatti, S. Donati, Optimum Signal Processing for Distance Measurements with Laser, Applied Optics, Vol. 10, No.11, pp. 2446-2451, 1971. G. Lee, G. Schroeder, Optical Pulse Timing Resolution, IEEE Transactions On Information Theory, pp. 114-118, 1976. B. Vojnovi,  Resolution Improvement of Stochastic Pulse Arrival-Time Determination , Ph.D. Dissertation, Faculty of Electrical Engineering, University of Zagreb, Zagreb 1973. B. Vojnovi,  A Subnanosecond Timing Circuit Using Snap-off Diode , Compte rendus du Colloque sur l'Electronique nucleaire, pp. 59-1 to 59-7, Versailles, 10-13 Sept. 1968., Paris B. Vojnovi, Private communication R.N. Donough, A.D. Wahlen,  Detection of Signals in Noise , 2nd Edition, Academic Press, New York 1995. M. Konrad,  Detector Pulse Shaping for High Resolution Nuclear Spectroscopy, IEEE Transactions on Nuclear Science, NS-15, No.1, pp- 268, 1968. V. Radeka, N. Karlovac, Least-Squared Error Amplitude Measurements of Pulse Signals in Presence of Noise, Nuclear Instruments and Methods, Vol.52, pp.86-92, 1967. J.S. Bendat, A.G. Piersol, Random Data, Analysis and Maesurements Procedures, 2nd Edition, John Wiley & Sons, New York, 1986. ( 1/(1 hs(() k-1 (1 (2 (1 (2 (w hD(() Tm (1 (1 -1 1 (w (1 ( xB%dfxXZfh:<bdfh(*P㥪j5EHU\&j(D 5CJUVmHnHsHtHj5U\ 5H*\ jH*&j5CJU\mHnHsHtHu jEHU#jI(D CJUVmHnHsHtH jUH*5\mH sH  CJmH sH  CJmHsHCJ5/vxB &Z j(t"# ^`$a$dSTPRTV!!""""""##&$($0$2$%%"%$%(%*%.+0+@+B+,,,---B-D-N-P-R-V-X-Z-\-^-b-d-v-x---////0ָ֧̗֟ jt5H*\ j5H*\ ja5\j 5EHU\&j^)D 5CJUVmHnHsHtH 5H*\ 5H*\5\j5U\j5EHU\&jZ(D 5CJUVmHnHsHtH;###[7\7]7^7a7b7c7d7e7f7g7h7i7j7k7l7m7n7o77777777 ^`00g4h444\6]6[7^7`7o77777$;&;= =>>">$>>>>>??n?p???????@@@@@@@@@@AAAA,A.Aj5EHU\&j)D 5CJUVmHnHsHtHj5EHU\&jd)D 5CJUVmHnHsHtHj] 5EHU\&j(D 5CJUVmHnHsHtHj5U\jCJUmHnHu5\ 5H*\57777777777777777779?n???HHJJ ^`.A^A`AHHJK8KKKgLwLLLLLLNNOVPP QQQQQQQ/Nقz  , xڥV]HQݙ\m] _|·, te^"| [a!!zʕL0{g9{{.{!P tdYBaAu FŻD@KU L42!3~g;[; hȷqE0p0s Zɳ_p QAS&ƅ ^uRh!F߶Bw7I%|/՘Ft#x5-Z{5q}mцTZ vigmΛZp )\RhHr~iQyrfɰs ]㫪UU#Œ~iG]e4iG/=|Yo}vY*<]kvյ{F"֌y _rRfr|Ryשʁ5][duFEG_^:euGT~#Y@煮,YOi S|B`z&@q?lP\cu3>?b `³f2WՐjMvrCwkES>"NQ^ϯ.9*ṉR柟krhOO[/(Scn̉xq.>7lSnc@nR'O,Dq7`7НP\e^[})laCR vYx0zξ}f_"I!NXthخ { DdTJ  C A? "2D.$!LjYO8 L`!.$!LjYO8f HXJxcdd``$d@9`,&FF(`TI A?d-Aǀ7$# ( XXbS% d1D YB2sSRsW|)bF:`00d*.b;OF&&\ J$]0;]`q]"7EDd`IPJ  C A? "2+l;k].`!U+l;k` X #xڭTkA~n3ACDA<JCE ¹V"i,ED-%ZX;!r^c̛]c Dz}vo">U*4Mmu0M ᤮RCouz33g+d 0KF{H';=#JWk&u5±JܳA>fG4\ΡL^뷑+q81xx`ݕ't˸߸bcxJXTDd\ 4\UJ  C A? "2R;KM `!R;Kj` "!pXxڭUMhA~vM6lD<,DJ(Ŋ&E0U`XP$7zD,)"Ԋz\zb ].φ;o{a0^ ])qg111}_F> yytRrQD>'q;ee.ќd?oId* z t-Zut-brQvmf׹06f Al3[|zpG׹;ޣx1G8#g}ƌv*I <ɟ{-?O1d~"1tW?c [NWWx[  D%_ =?7FF}FqWX|Ž^>_*YTɥRl0%z':x;MO2`CX:?j3Dd`LMJ  C A? "2@I ?j^pQq `!i@I ?j^pQZ#7xڥTkA3dѤIMEsRZoKQē(҃PaAX! o z-^ zxx(ؓBynXe&~y7;<dꀏ+GD QI 9uc֯HT Bg( -uk'Gn7|?M R+wޝc6{XDMM =?0KWn,Γge4Ӿѹl|okzuP}p} syS2% vBN=D\xX$8dR bDdJ  C A?  !"#$%&'()*+,-./0123456789:;<>?@ABCDXGJKLMNOPQRTSUWVZ[Y\^_`abcdefghijklmnopqrstuvwxyz{|}Root Entry( F ~W"IData =WordDocument'xObjectPool* Ъ/" ~W"_1143536457FЪ/"Ъ/"Ole CompObjfObjInfo  #&'(),156789:;=>?@AC FMicrosoft Equation 3.0 DS Equation Equation.39qP Fh()[]=N() 2 ="1 2 R yy2 2 0()R yy 0()Equation Native _1143536670 F8"8"Ole  CompObj f=C"h() 0" +" h 2 2 ()dh 2 ()d 0" +" FMicrosoft Equation 3.0 DS Equation Equation.39qObjInfo Equation Native 0_1143536730Fpf="pf="Ole  N FMicrosoft Equation 3.0 DS Equation Equation.39q   t2 = s2 t()ddtSt()[] t=CompObjfObjInfoEquation Native _1143543390Fs@"s@"T m 2 FMicrosoft Equation 3.0 DS Equation Equation.39qd h 0 ()=dSt"()d"1St"()Ole CompObjfObjInfoEquation Native +[] t=T m  FMicrosoft Equation 3.0 DS Equation Equation.39q  t2 =12k"1()"N e C T2 Q 2  w_1143537570"FPA"PA"Ole !CompObj"fObjInfo$Equation Native %:_1143537764F.E".E"Ole *CompObj +f =N e V s2 " w 2k"1() FMicrosoft Equation 3.0 DS Equation Equation.39q"PT  N eObjInfo!-Equation Native .>_1143537810$F@F"@F"Ole /CompObj#%0fObjInfo&2Equation Native 361Table]@ FMicrosoft Equation 3.0 DS Equation Equation.39q  Hz Oh+'0(8 DP l x  :Timing-Error Minimization of Optoelectro "2}m1:GWY,4Y`!Qm1:GWY,4@ xڝQN1 "NC 0vR1) Z#t nq@l'=X9f 2%!&BPt)׬y-ѱ`^^BыDipr=#0U~R!cޢvmkp1qt:j}hc͑IfE6D'}FP nJ7=7rx gKvTaCIw?={[*|ث^u7HwWJi^Ȣ~(vJ DdTJ  C A? "2muT _X I`!AuT _X ` :XJxcdd``~ @c112BYL%bpuQSummaryInformation()4DocumentSummaryInformation8<hCompObjBjnic Pulse SignalseimiBranka Medved RoginatioranranNormalM...3.Microsoft Word 9.0a@e@r"@w+"@۴"-՜.+,08 hp  Ruder Boskovic Institutema7 :Timing-Error Minimization of Optoelectronic Pulse Signals Title  FMicrosoft Word Document MSWordDocWord.Document.89q i8@8 NormalCJ_HaJmHsHtH<A@< Default Paragraph FontJYJ Document Map-D M OJQJ^JBOB odi normal$a$CJaJmH sH tH :O: odi title$dha$5CJ$4O"4 odi abstract5CJHO2H odi title1$Vxx^`V5>OB> odi reference7^7`2B@R2 Body Text$a$5\ !%&*159=ADHIJKLMNOPQRSTUVW[^7WuVrQP|poMfecbX      !%&*159=ADHIJKLMNOPQRSTUVW[^a      !$h%7x)xVx;<N & - <] 3<\$$$$$$$$$$$$$$$$$$$%%% %!%"%#%$%%%&%'%(%)%*%+%,%-%.%/%0%1%g%h%i%&)G)i))^/m/U1`112,33Q45'5566F7H7I7N7O7U7V7Z7[7^7_7b7c7f7g7j7k7l7o7p7v7w7z7{7~7777777777777777777777777777770@0@0@0@0@0@0@0@0@0@0@00@0@0@0@0@0000@0@0000@0@000`0000@0@0000000000@0@00@000@0000000000@000000@00@`0000@00@0@000@0 0 0 0 0 0 0 0 0 0 0  0 0000000000000000000000000000000000000000000000000000P0.AST0357:#7JSST14689;T2<PR !#G)[)]))))***7:::::::8@30 (  HB y C D \  %)C1  #" T  # +,!. &t {.Z1 #  #" .C1` | C |v>/Z1# :h /,51 ~ #" {.v0f r S r/,51' TB } C D:0V0f p S  p %+).&  ` e C e 5",$." ` f C  f ,.$  fB g s *LD {%0`B h c $D +'+"`B i c $DjJ *(*`B j c $DjJ */`B k c $DjJ /h!/`B l c $DjJ ]!*]!/`B m c $DjJ S!*%*`B n c $DjJ %*%+f o S  o  % '  lB s 0Dg  ,,fB t s *D q!,%,` u C u r)<m+ lB v 0Dg  (z.S!z.fB wB s *D u/u/fB x s *D %+%0"f  S " m&)(Y, " \ $.1  #" ` V C V z$' ` M C  M  )Z+  ` P C P  . 0 ` Q C Q  . 0 fB R s *LD W>%X'0`B S c $D c+H+`B T c $D 5 +< H0fB U s *D P+U/f W S W ).+ f X S X )+ lB Y 0Dg  W/ /fB Z s *D i /`/`B [ c $DjJ c) +`B \ c $DjJ  )& .`B ] c $DjJ 1 +l.lB ^ 0Dg   /< /`B _B c $D  ) S0!fB ` s *D c) )fB aB s *D 6.< .f b S b (h*! f c S c 6-D/& Z  3 !.O1 !B S  ?]$$7yAAta$ t t,<CDL*2 wz68hjr"t"$$$%1%g%!'#'(((())-+7+..R0\0c1h1m1s111222233T4\4a4n44444444444444455,535:5@555"6(6-656666667F7I7K7L7M7O7Q7V7Z7\7]7`7a7d7e7h7j7m7n7p7r7w7y7|7}777777777777 w{^cqs38AChk& _ $%1%f%& &((//44F7I7M7O7T7V7Y7\7^7`7b7d7f7h7j7m7o7p7u7w7y7|7~777777777777333333333333333333333<]G)i)F7I7M7O7T7V7Y7[7]7_7a7c7e7g7i7l7n7p7u7w7y7{7}777777777777Branka Medved Rogina(D:\BV\ODIMAP_IV\ODIMAPIV-Paper-2_BMR.docBranka Medved Rogina(D:\BV\ODIMAP_IV\ODIMAPIV-Paper-2_BMR.docBranka Medved Rogina(D:\BV\ODIMAP_IV\ODIMAPIV-Paper-2_BMR.docBranka Medved Rogina(D:\BV\ODIMAP_IV\ODIMAPIV-Paper-2_BMR.docBranka Medved Rogina(D:\BV\ODIMAP_IV\ODIMAPIV-Paper-2_BMR.docBranka Medved Rogina(D:\BV\ODIMAP_IV\ODIMAPIV-Paper-2_BMR.doc...xC:\Documents and Settings\vojnovic\Local Settings\Temporary Internet Files\Content.IE5\MVEHWHAP\ODIMAPIV-Paper-2_BMR.doc...dC:\SAV95TMP\My Documents\MZT\Zavod za Elektroniku\VOJNOVIc\B.V.-DOCUMENTS\ODIMAPIV- Final Paper.doc...qC:\Documents and Settings\vojnovic\Application Data\Microsoft\Word\AutoRecovery save of ODIMAPIV- Final Paper.asd...dC:\SAV95TMP\My Documents\MZT\Zavod za Elektroniku\VOJNOVIc\B.V.-DOCUMENTS\ODIMAPIV- Final Paper.docTH6Zz8^`o([.]^`.pLp^p`L.@ @ ^@ `.^`.L^`L.^`.^`.PLP^P`L.TH6FJ$7F77777@//y// yyyyyyyy||||&*,.3457PP @PP$@PPP8@PP P"P$PL@P*P,P0Pd@P:P@PDP@PNPPP@UnknownGz Times New Roman5Symbol3& z Arial5& z!Tahoma"q{{{ -a!P207F7 2Q9Timing-Error Minimization of Optoelectronic Pulse SignalsBranka Medved Rogina...