Design and development of detector signal conditioning electronics for SST-1Thomson scattering systemAruna Thakar, Ajai Kumar, Jinto Thomas, and Chhaya Chavda Citation: Review of Scientific Instruments 79, 093505 (2008); doi: 10.1063/1.2972149 View online: http://dx.doi.org/10.1063/1.2972149 View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/79/9?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Demonstration of improvement in the signal-to-noise ratio of Thomson scattering signal obtained by using amulti-pass optical cavity on the Tokyo Spherical Tokamak-2a) Rev. Sci. Instrum. 85, 11D846 (2014); 10.1063/1.4891707 The circuit of polychromator for Experimental Advanced Superconducting Tokamak edge Thomson scatteringdiagnostic Rev. Sci. Instrum. 84, 093504 (2013); 10.1063/1.4820561 Synchronized operation by field programmable gate array based signal controller for the Thomson scatteringdiagnostic system in KSTAR Rev. Sci. Instrum. 83, 093505 (2012); 10.1063/1.4752408 Design of multipulse Thomson scattering diagnostic for SST-1 tokamak Rev. Sci. Instrum. 78, 043507 (2007); 10.1063/1.2724775 APD detector electronics for the NSTX Thomson scattering system Rev. Sci. Instrum. 72, 1129 (2001); 10.1063/1.1321751

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

128.42.202.150 On: Sun, 23 Nov 2014 17:03:42

Design and development of detector signal conditioning electronicsfor SST-1 Thomson scattering system

Aruna Thakar, Ajai Kumar,a� Jinto Thomas, and Chhaya ChavdaInstitute for Plasma Research, Near Indira Bridge, Bhat, Ghandhinagar 382 428, India

�Received 10 January 2008; accepted 25 July 2008; published online 10 September 2008�

An IR enhanced thermoelectrically cooled Si-avalanche photodiode �Si-APD� module is used fordetection of scattered photons from plasma electrons. Present design of signal conditioningelectronics for the APD has fast �50 MHz� and slow �500 kHz� channels to measure scattered andplasma background light, respectively. We report design analysis for different stages and theirperformance. The performance of fast channel is analyzed for two different group delays, speed,linearity, and its cross-talk with slow channel. Temperature dependence of APD’s responsivity isstudied in the wavelength range of 900–1060 nm. A minimum detection of �25 photoelectrons�with S /N=1� in the range of 5 to 25 °C is achieved at an APD gain of 75 in the present design.© 2008 American Institute of Physics. �DOI: 10.1063/1.2972149�

I. INTRODUCTION

Si-avalanche photodiodes �Si-APDs� are widely used fordetecting low yield light signals from visible to near IR re-gion �400–1000 nm� and have wide ranging applications inlight scattering experiments such as Thomson scatteringfrom plasma electrons.1–8 For SST-1 tokamak,9 Thomsonscattering diagnostics system10 uses fundamental wavelengthof Nd:YAG �yttrum aluminum gurney� laser �1064 nm�. Thescattered light from the plasma electrons is spectrally dis-persed and detected using an IR enhanced Si-APD and ex-pected number of scattered photons reaching the detector isin the range of 102–105 during 10 ns �full width at halfmaximum of laser pulse�.

A signal conditioning electronics �SCE� is designed anddeveloped for detecting Thomson scattered and plasma back-ground light with maximum rejection of noise and a stablegain for nearly eight hours of operation. It consists of �a� fastchannel to detect scattered photons, �b� a slow channel formeasurement of plasma background light �for error analysispurpose� and for calibration of filter polychromator, �c� acrowbar circuit to protect the detector against accidental ex-posure to a high photon flux, �d� a controller for thermoelec-tric cooler �TEC� of APD, and �e� a PC controlled dc-dcconverter to bias APD independently for its optimum perfor-mance.

Many methods have been reported for maintaining astable detector gain against thermal variation, e.g., keepingAPD in a separate chamber away from electronics,11 biasingAPD voltage with voltage feedback configuration at pre-defined gain,12 and switching on APD bias voltage only dur-ing plasma discharge. In the present design, the variation ofgain with temperature �2% / °C� is taken care by thermoelec-trically cooling the detector.

This article describes design and performance test for

different subassemblies, cross-talk between the slow and fastchannels and temperature dependence of APD’s responsivityfor different wavelengths.

II. SIGNAL CONDITIONING ELECTRONICSOF APD MODULE

Figure 1 shows the schematic diagram of SCE circuit forAPD detector. It consists of an APD module �Perkin-Elmer;CD2720�, a fast channel, a slow channel, a temperature con-troller for APD TEC, and a crowbar protection circuit forAPD.

A. APD module

APD module contains an IR enhanced Si-APD �Perkin-Elmer C30956E� having response from 400 to 1100 nm. Ithas an active area of 7 mm2 and quantum efficiency �QE� of85% at 900 nm, which reduces to 40% at the laser wave-length �1064 nm�. The detector noise is 1.1 pA /�Hz at22 °C. This module has an inbuilt transimpedance amplifier�gain: 13�103 V / I, bandwidth: 80 MHz� and a TEC. Itsoutput impedance is 13 �; and requires a load resistance of500 � �ac�. Module’s output-offset voltage varies from mod-ule to module.13 The expected output signal from the moduleis �624 �V �detector gain 75, TAPD=22 °C� for 100 pho-tons during laser pulse at 1064 nm.

B. Fast channel

It measures the scattered photons while rejecting slowlyvarying plasma background. This channel is designed to de-tect scattered photons without any pulse conditioning �actualpulse width� to minimize errors contributed by plasma back-ground light. This requires minimum bandwidth of the chan-nel to be 50 MHz �flattop�. The design consists of an externalvoltage amplifier, a Bessel filter �BF� �gain=1�, and a differ-ence amplifier �DA� �gain=1�, as shown in Fig. 1. The ex-ternal noninverting voltage amplifier �OPA 603, bandwidth:a�Electronic mail: ajai@ipr.res.in.

REVIEW OF SCIENTIFIC INSTRUMENTS 79, 093505 �2008�

0034-6748/2008/79�9�/093505/6/$23.00 © 2008 American Institute of Physics79, 093505-1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

128.42.202.150 On: Sun, 23 Nov 2014 17:03:42

100 MHz; gain: 1–10� amplifies output signal from the APDmodule by four times. Transfer function for the fast channelis given by

Vf = �NsQEMeGtGe�/t ,

where Vf is the output voltage, Ns is the number of scatteredphotons, QE is quantum efficiency, M is the detector gain�75�, e is the electronic charge, Gt is transimpedance ampli-fier gain �13�103�, Ge is external amplifier gain �4�, and t ispulse duration �10 ns�. For a dynamic range of 1000 �i.e., for102–105 photons�, the expected signal from fast channel is inthe range of 2.5 mV–2.5 V.

Output signal from APD module is ac coupled �R=500 �, C=500 pf, Fig. 1� to the external amplifier for re-moving the dc offset without affecting the shape of signal. accoupling also attenuates the background signal �3 dB attenu-ation at 636 KHz�. However, the 20 dB/decade rolloff of rcfilter does not fully remove plasma background signal atleast for frequencies below 500 kHz. Although the residualsignal is small in amplitude, it cannot be accepted as thechannel is designed for detecting as low as �100 photons. ABF and a DA are, therefore used for removing the residualslow signal.11

The BF �IC AD8047, two pole, low pass with a cutoff at2.7 MHz� delays the signal from external amplifier by100 ns. The DA �IC AD8047, gain�bandwidth: 130 MHz�subtracts the delayed signals from the nondelayed signals inorder to remove the low frequency background signal.11,12

Integrated chips used for this design have noise equivalentpower of �12 fW��Hz compared to 44 fW��Hz �at1060 nm� of detector. Noise from integrated fast channel anddetector is measured �with Maximum hold on� using spec-trum analyzer �Agilent, model: E4401�. Figure 2 �traces �a�and �b�, respectively� shows the noise power spectrum atroom temperature associated with detector �C30956E-TC, bi-

asing voltage: 300 V, under dark condition� and fast channel�with detector bias off�. The spectrum is shown up to 5 MHzas the maximum observed noise lies below 5 MHz and be-yond that it is negligible. Care is taken during design andassembly to reduce noise of SCE below the detector limit toimprove the detection limit.

To optimize the value of group delay, the designed BFcircuit14 is analyzed by observing the effect of two groupdelays �50 and 100 ns� on scattered signal and backgroundsignal through fast channel. To measure low frequency rejec-tion of the fast channel, signals up to 500 KHz are applied tothe external amplifier �before C1, Fig. 1�. Figure 3 showstransmission of slow components through this channel forboth the delays; low frequency rejection is better for 50 nsdelay. Similarly, for scattered signal, a 20 ns pulsed signal�from function generator� is applied to the external amplifier�Fig. 4, trace �a��. The measured output from the amplifier�trace �d��, BF �trace �c��, and DA �trace �b�� for both thedelays are shown in fig. 4. The integrated charge �30 ns in-tegration duration�, i.e., the areas under the curve are 2440and 2990 pV s for 50 and 100 ns delays, which are lower for50 ns delay. For the present design 100 ns delay is opted as itpreserves the shape of input pulse �Fig. 4, trace �b�� and totalcharge, which is close to expected 3000 pV s.

Low frequency rejection of fast channel is tested by themeasurement of detector noise. Figure 5 �traces �b� and �a�,respectively� is a detector �bias: 300 V, under dark condi-tion� noise spectrum before DA �at o / p of external amplifier,Fig. 1� and after DA. The noise up to 500 KHz is completelyremoved and peak noise at 1.11 MHz is reduced by 60%.

The response of the fast channel to a 10 ns pulsed lightfrom a laser diode �Hamamatsu, C6582, 873 nm� is shown inFig. 6, where channel 1 is the transistor-transistor logic�TTL� trigger output from the laser module and channel 2 isthe detected laser pulse. It is observed that the designed

FIG. 1. �Color online� Schematic ofSCE circuit where EA is external volt-age amplifier, BF is Bessel filter, andDA is difference amplifier.

093505-2 Thakar et al. Rev. Sci. Instrum. 79, 093505 �2008�

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

128.42.202.150 On: Sun, 23 Nov 2014 17:03:42

channel is fast enough to detect a 10 ns pulsed light withoutput having same rise and fall time as of the pulsed light.So, the analog to digital converter �ADC� will integrate thecharge only for laser pulse duration, which will result inimprovement of S/N ratio.

The signal from the DA is digitized using an in-housedeveloped charge-integrating ADC �C-ADC�.15 This C-ADCconsists of charge to time converter �MQT� and time to digi-tal converter �TDC�. The design of C-ADC permits placingof MQT along with the fast channel. This reduces the highfrequency pickup while transmitting the signal for digitiza-tion. The differential ECL output from MQT is fed to theTDC mounted in data acquisition rack. For the presentmeasurements, the gate interval of C-ADC is fixed to 30 nsalthough it has variable gate interval of 20–500 ns. The ex-pected signal from the fast channel is in the range of

2.5 mV–2.5 V or 125 fC–125 pC of charge, which is fed toC-ADC having input impedance of 200 �. The digitizeddata are acquired using a personal computer.

The APD module, the conditioning electronics and MQTare housed in an aluminum enclosure of outer dimensions of55�100�135 mm3 �Fig. 7�. The APD is directly solderedon the fast channel circuit board. It is inserted into a pre-cisely drilled hole on a 6.35 mm thick aluminum end plateand aligned with the help of three locating pins. The endplate acts as a heat sink for the module. The detectorassembly is locked to the xyz translator of the filterpolychromator10 for precise positioning of APD at the focalplane of the focusing lens. An optical fiber is placed in frontof the APD module at a different elevation as an alternateillumination port for performing interchannel calibration ofthe detector. The optical fiber is placed along an axis suffi-ciently skewed so that the fiber does not obscure the trans-mitted light through the interference filter.

As the detector assembly has to be operated in presence

FIG. 2. �Color online� Noise spectrumof detector under dark condition �trace�a�� and SCE �trace �b�, when detectoris biased off�.

FIG. 3. �Color online� Gain vs frequency �low� plot for scattered channel at50 and 100 ns group delay.

FIG. 4. The measured output from fast channel at DA �trace �b��, BF �trace�c��, external amplifier �trace �d�� for 50 and 100 ns delays with same inputsignal �trace �a��.

093505-3 APD signal conditioning electronics Rev. Sci. Instrum. 79, 093505 �2008�

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

128.42.202.150 On: Sun, 23 Nov 2014 17:03:42

of electromagnetic interference �EMI� radiation of laser andothers sources, care has been taken in terms of grounding,shielding of cables, layout of circuit boards, and soldering ofelectrostatic discharge sensitive components for noise pickupand oscillation-free performance. The detector assembly istested with Rayleigh scattered signal on test stand in pres-ence of Nd: YAG laser for EMI test. No EMI pickup problemwas noticed during test.

C. Slow channel

Slow channel is designed for the measurement of plasmabackground light �for error analysis purpose� and to performcalibrations associated with filter polychromator. In general,the plasma background signal is lower in amplitude com-pared to the scattered signal. Hence, a higher gain �program-

mable� is provided to measure it with higher accuracy. Thischannel �Fig. 1� has a buffer followed by a two-stage ampli-fier �with 3 dB bandwidth at 500 KHz�. The buffer protectsAPD module from loading due to its dc-offset voltage, whichis compensated by the first stage �G=5� of the amplifier.Programmable gains of 2, 4, 6, and 8 are provided to thesecond stage. Value of the programed gain during plasmadischarge is time stamped for data analysis purpose. Digiti-zation of background signal is performed in the same fashionas that of the scattered signal.

This channel is tested for programmability of gain, band-width, linearity against input signal, and associated noise.Although the bandwidth is restricted up to 500 KHz, its re-sponse to fast pulse is tested to avoid error due to scatteredsignal on plasma background measurement. For this test, a20 ns pulse �from function generator� is applied at the inputof buffer �Fig. 1�. Outputs from both the channels are mea-sured simultaneously. The slow channel output �Fig. 8� is abroader signal with small amplitude �channel 2, 27 mV� aftera �propagation� delay of 100 ns as compared to the fast chan-

FIG. 5. �Color online� Detector noise�under dark condition� measured be-fore and after DA.

FIG. 6. �Color online� Response to 50 ns pulsed light from laser diode asdetected by fast channel. FIG. 7. APD detector assembly with SCE and MQT.

093505-4 Thakar et al. Rev. Sci. Instrum. 79, 093505 �2008�

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

128.42.202.150 On: Sun, 23 Nov 2014 17:03:42

nel signal �channel 1, 150 mV�. This information is useful indeciding delay between time of integration for fast and slowsignal. Synchronized integration by both C-ADCs corre-sponding to fast and slow channels is found to be optimum.

D. Crowbar protection circuit

Computer controlled, high voltage modular power sup-plies �AD 521� are used to bias the APD detectors indepen-dently to maintain uniform gains. The biasing voltage isramped up or down �50 V /s� by software control. Accidentalexcess illumination or rise in its body temperature results inbreakdown or permanent damage to the APD. Therefore, acrowbar circuit is designed to protect APD by cutting downbiasing voltage when detector draws current beyond a certainlimit.

The output voltage from the first stage of the slow am-plifier is compared to a reference �ref� voltage by an analogcomparator LM 311 �Fig. 1�. The ref voltage is set to 3.2 V,which is equivalent to 50 �A of detector current. When de-tector current exceeds 50 �A ��650 mV from APD module�for more than 2 �s duration, comparator generates a TTLpulse, which triggers the microcontroller to cut down detec-tor biasing voltage. Crowbar’s response to a fast pulse �APDmodule signal equivalent to 100 �A, 100 ns detector currentwhich is more than expected maximum scattered current� isalso tested. The resulting TTL pulse from comparator is low,which keeps the detector bias ON.

E. Conditioning of APD detector

Thermoelectrically cooled APD module is opted to avoidthermal drift in gain over period of its operation. A tempera-ture controller circuit with accuracy better than �0.1 °C isdeveloped, which regulates the temperature of APD by con-trolling current through the cooler using thermister �Rth� as afeedback element.16 For CD 2720 module,13 Rth ��� as afunction of temperature T �K� is given by

Rth = 104 exp�3940 1

T−

1

298� .

The designed temperature controller stabilizes APD tempera-ture to the set value within few seconds. Continuous powerdissipation in transimpedance amplifier of APD module in-creases its body temperature continuously. Hence a suitableheat sink is provided to the module.

It is also observed that APD without transimpedance am-plifier �C30956E-TC� requires air purging to avoid conden-sation of moisture on the glass windows at low temperatures.However, this problem does not arise for APD module CD2720 �used in the present design� due to power dissipation inthe internal circuit.

III. PERFORMANCE TEST OF THE DETECTORSIGNAL CONDITIONING ELECTRONICS

Performance of SCE is tested for its gain, bandwidth,linearity against input signal, noise associated, temperaturedependence of responsivity/QE and dark current of detector,minimum detection limit, and variance �rms variation� in sig-nal during its operational period. The gain, bandwidth, andlinearity against input signal over the dynamic range of 1000are observed to be very stable.

The responsivity of APD against temperature is studiedin the wavelength range of 900–1050 nm. Light from stabi-lized tungsten lamp, chopped at 300 Hz and dispersed by aspectrometer, is coupled via optical fiber to APD operating atpreset temperature and biased to 300 V. Result shows thatthe responsivity at constant biasing voltage is higher at lowertemperatures for the wavelengths of interest. These resultsare in agreement with detector’s specifications. The lowerrate of increase in signal at higher wavelengths is due topositive temperature coefficient of QE at these wavelengths.

To measure the variance of the signal in terms of photo-electrons, experiments are carried out using techniques de-scribed in Ref. 11. A 10 ns pulsed laser �Hamamatsu, C6582,873 nm� is used as light source. A set of neutral densityfilters in the range of 0.1–4 optical densities are used to varythe incident photon flux so as to cover the dynamic range ofthe SCE. For a chosen photon flux, nearly 500 samples areacquired to estimate the average number of photoelectronsand their rms variation. Figure 9 shows the rms variation ofphotoelectrons at APD temperatures of 5 and 25 °C �at con-stant gain of 75�. The response of detector is found to belinear �better than 1%�. No significant change in rms varia-tion of photoelectrons with temperature is observed. Thelower limit for detection of photoelectrons by the presentSCE is found to be �25.

No considerable change in variance for a constant pho-ton flux �1064 nm� and gain is observed when temperature isincreased from 5 to 25 °C although the QE increasesslightly �5%�.17 Slow channel shows considerable reductionin dark current from 95 to 10 nA when APD is cooled downfrom 25 to 5 °C. Based on these observations, it is decidedto operate the APD at 10 °C to take advantage of low darkcurrent and to avoid condensation on the windows.

FIG. 8. �Color online� Signal output from fast �channel 1� and slow channel�channel 2� in response to 20 ns pulse applied at the input of buffer.

093505-5 APD signal conditioning electronics Rev. Sci. Instrum. 79, 093505 �2008�

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

128.42.202.150 On: Sun, 23 Nov 2014 17:03:42

IV. SUMMARY

SCE for APD is designed and tested for the SST-1Thomson scattering system. In the present design, the tem-perature of the APD is stabilized within �0.1 °C to maintainits responsivity constant. APDs are biased individually byprogrammable power supplies and bias voltage is ramped upor down �50 V /s� with software control. A crowbar circuit isdesigned to protect APD from accidental exposure to excesslight. The designed SCE has a fast and a slow channel. Thefast channel uses BF and DA for rejecting plasma back-

ground signal. Performance of SCE is analyzed for two dif-ferent �BF� delays. Frequency spectra of noise associatedwith detector and SCE are measured. The cross-talk betweenfast and slow channels is analyzed to find delay time forgating on both the channels. Effect of cooling on APD isstudied. With the optimized parameters, the SCE is found tobe capable to detect a minimum of 25 photoelectrons.

ACKNOWLEDGEMENTS

The authors would like to thank Rajwinder Kaur, VishnuChaudhary, Kiran Patel, Ranjeet Singh, N. C. Patel, andKaushal Pandya for their technical assistance during devel-opment and testing.

1 T. N. Carlstrom et al., Rev. Sci. Instrum. 63, 4901 �1992�.2 R. Pasqualotto, Rev. Sci. Instrum. 75, 3891 �2004�.3 K. Narihara et al., Rev. Sci. Instrum. 66, 4607 �1995�.4 D. Johnson et al., Rev. Sci. Instrum. 70, 776 �1999�.5 R. Rajesh et al., Pramana 55, 733 �2000�.6 R. Behm et al., Rev. Sci. Instrum. 70, 768 �1999�.7 H. Murmann and M. Huang, Plasma Phys. Controlled Fusion 27, 103�1985�.

8 F. Orsitto, A. Brusadin, and E. Giovannozzi, Rev. Sci. Instrum. 68, 1201�1997�.

9 Y. C. Saxena and SST Team, Nucl. Fusion 40, 1069 �2000�.10 A. Kumar et al., Rev. Sci. Instrum. 78, 043507 �2007�.11 C. L. Hsieh et al., Rev. Sci. Instrum. 61, 2855 �1990�.12 D. W. Johnson, B. P. LeBlanc, and D. L. Long, Rev. Sci. Instrum. 72, 1129

�2001�.13 Perkin Elmer data sheet for APD CD 2720.14 M. E. Van Valkenburg, Analog Filter Design, �Holt, Rinehart and Win-

ston, New York, 1982�.15 C. Chavada et al., Rev. Sci. Instrum. �unpublished�.16 T. F. Refaat, S. William, Jr., and J. Russell, NASA Tech. Memo. �,

209689 �1999�.17 P. P. Webb, R. J. McIntyre, and J. Conradi, RCA Rev., 35, 234 �1974�.

FIG. 9. rms variation of measured photoelectrons of fast channel at 5 and25 °C. The lower detection limit of photoelectrons at 5 °C is shown in theinset.

093505-6 Thakar et al. Rev. Sci. Instrum. 79, 093505 �2008�

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP:

128.42.202.150 On: Sun, 23 Nov 2014 17:03:42