Noise Measurement in Amplifying System for Radiation Detectors

L. Lombigit*, M. A. Khalid Malaysian Nuclear Agency

43000 Kajang, Selangor, Malaysia *lojius@nuclearmalaysia.gov.my, ashhar@nuclearmalaysia.gov.my

M. N. Hamidon Functional Device Laboratory of Institute of Advance

Technology. University of Putra Malaysia,

43400 Serdang, Malaysia. mnh@eng.upm.edu.my

N. Sulaiman Department of Electrical and Electronic Engineering,

University of Putra Malaysia 43400 Serdang, Malaysia nasri@eng.upm.edu.my

Abstract— Electronic noise is a critical parameter especially for electronic devices which deal with low level input signals such as preamplifier-shaper chains in nuclear instrumentation. It degrades the capability of the amplifying system as a radiation measurement device in terms of detection level, energy resolution and event measurement. In this paper, we propose an experimental approach to noise measurement in amplifying systems for radiation detectors. The experimental results are verified with theoretical computation and validated with an established experimental method. The results are in good agreement with theoretical and established methods.

Index Terms—Noise, amplifying system, radiation detectors

I. INTRODUCTION Radiation detectors convert the radiation energy deposited

in its volume into electrical signals. Those signals are usually small, and contaminated with interference noises. Therefore it requires an amplifying system normally in the form of a preamplifier-shaper chain which amplifies and shapes the signal to obtain the best possible information of the deposited radiation energy. These noises will determine the minimum detectable radiation events, reduce the energy resolution and may alter the time dependence of the signal in time measurement [1].

The noise in a radiation detection system is originated from the detectors and amplifying system, but this paper will focus on the noise generated by the analog amplifying system or more specifically the preamplifier-shaper chain. In radiation detection, the amplifying system noise is usually expressed as equivalent noise charge (ENC) so that it can be directly compared to the input charge. Noise analysis and the optimum filter design in a preamplifier-shaper chain system is an established topic, compressive researches and discussions are found elsewhere in [2-4]. Fig.1 shows noise generators in preamplifier-shaper associated with radiation detectors. The

input amplifier (CSP) might be a transistor, an integrated circuit, or an operational amplifier (Op-amp). CT is the total input capacitance; T(s) is the transfer function of the shaper, SV and SI are the series and parallel input noise generators, the noise power spectral densities of which are [4] | | 1 | |, 2

where a and b are the white noise components due to thermal voltage and current noises. The af and bf denote the 1/f noise coefficients. The total ENC in a preamplifier-shaper system is the integration of all noise generators over the transfer function of the shaper [5], and the squared equivalent noise charge (ENC) is given by the following expression: ENC CT 2 CT 2 3

This ENC is expressed either in unit charge (Coulomb) or number of electron (e-). CT is the total input capacitance in the system which includes stray capacitance, CTest (used to converts voltage to charge pulse), detector capacitance (Cd) and preamplifier feedback capacitor (CF). A1, A2, A3 and τ are the shape factors and time parameter which depend on shaper transfer function. Eq. (3) shows that, the ENC is dependent on two parameters i.e. input capacitance, CT and time parameters,τ.

Fig. 1. Noise generators in amplifying system associate with

radiation detectors (excluding detectors noise source).

218978-1-4673-3119-7/12/$31.00 ©2012 IEEE

In measurement terms, the input-referred noise i.e. ENC is obtained by first measuring the output noise then referred to input by dividing it with the system gain. The output noise can be measured either with or without the presence of input signal. With input signal connected and assisted by a discriminator, the output noise is evaluated based on the relationship between noise count and discriminator threshold levels. Finally, the output noise is extracted using the S-Curve method [6-7] or method described in [8-9]. Other method to measure the output noise without connecting the input signal is by direct measurement using a special measuring instrument [10-12]. In this paper, we propose an experimental-based method to evaluate the ENC in an amplifying system associated with radiation detectors. We estimate the ENC in our amplifying system with theoretical computation using a noise model as in eq. (3) and validate the method with an established experimental method.

The remainder of this work is organized as follows. In Section II, the noise analysis of the preamplifier-shaper is discussed. In Section III, we illustrate the proposed procedure for the proposed ENC measurement. In Section IV, we report and discuss the experimental results in comparison with the computed and established method. Finally, in Section VI, we present our conclusions

II. NOISE ANALYSIS In this section, we briefly summarize the noise analysis in

the preamplifier-shaper. The test subject for this study is the previously developed UPM-MNA-RENA board [13]. The board is a charge amplifier comprised of eight channel charge sensitive preamplifier and 5th order semi-Gaussian shaper. It was developed entirely with Op-amp and single channel diagram of this board is shown in Fig. 2. The ENC is computed under the following assumption;

(a) The stray capacitance is unknown and omitted from the calculation, therefore CT = CTest (1pF) + CF (3.3pF) + Cin.

(b) The white noise components are dominant and 1/f components are negligible since, from the device datasheet, its contribution is only significant at a frequency below 1 kHz [14].

The CSP section of the UPM-MNA-RENA board is an Op-

amp (OPA657) with input voltage noise, en = 4.8nV/√Hz and current noise, in = 1.2fA/√Hz [14]. The ENC in the

preamplifier-shaper is computed using eq. (3) with τ = 300ns and semi-Gaussian (n=4) shape factors [4]. The series and parallel white noise spectral power density are calculated as follows: , , 4KTRF RF . Where: ieq = Equivalent input noise current spectral density 4KT = 1.66E-21J at 300K RF = 75kΩ

III. PROPOSED EXPERIMENTAL METHOD The instruments setup for this method is shown in Fig. 3.

The input signal is a rectangular voltage pulse generated by the nuclear pulse generator. The pulse rate and counting period are set around 1 kHz and 10s respectively. The voltage input pulses are converted into charge pulses by injecting the signal into a 1 pF capacitor (CTest). An external capacitor (Cin) is added parallel to the CTest to simulate the effect of input capacitance variation.

Fig. 3. The instruments setup for the noise measurement.

Fig. 4. A Graphical illustration describing the noise sampling.

Fig. 2. A Single channel diagram of the UMP-MNA-RENA board.

219

The electronic noises at the top of the output signal (Fig. 4) are sampled and counted at different amplitudes using an integral discriminator (ORTEC SCA 550A). With this setting, all pulses with amplitudes greater than the lower level threshold (LLT) will be counted. A graphical illustration describing the sampling method is shown in Fig. 4. We start the measurement by setting the LLT just around the top of the output signal; this is the reference point for measurement (signal side). This reference point is verified by taking a 10s counting and with an input signals around 1 kHz. The pulse count at this point is around 10000 counts. The LLT and pulse count at this point is recorded as , . . The LLT is increased at equally spaced interval until the discriminator pulse count is zero. For this study, we set the LLT interval around 2 mV because of the discriminator LLT resolution limitation. It is worth noting that a smaller interval will increase accuracy. The noise amplitude and their frequency are computed as follows: With n = 1, 2… n. k =1, 2….k and equally spaced threshold ∆

. Noise amplitudes , , … … , , , … … , , 4 Frequencies , , … … , . . , … . , . . . 5 The noise amplitudes are plotted with respect to its

frequencies; this plot represents the noise distribution. Assuming the white noise components are dominant, this plot shall follow a Gaussian distribution and therefore the sigma (σ) is equal to the rms value of the electronic noise. The sigma (σ) of distribution is extracted using a nonlinear least squares curve fitting with standard Gauss fit function. In order to get enough data for the dependence of ENC on input capacitance variation, we perform the measurement with different Cin values (Cin = 5.6, 22, 33, 47, 56, 68 & 82 pF). The ENC referred to input is computed by dividing the measured rms noise with the preamplifier-shaper gain. The obtained ENCs are then plotted against Cin and the actual ENC of the preamplifier-shaper chain is extracted by extrapolating the plot to Cin = 0pF.

For validation, we apply the ENC measurement according to the method described in [7] as reference. The reference and proposed methods will be referred to as established (EM) and proposed method (PM) for the remainder of this manuscript.

IV. RESULTS AND DISCUSSION As shown in Fig. 5, the measured ENCs are in agreement

with computed in terms of dependence of ENC on input capacitance. The measured ENC is higher by at least one order of magnitude. This is identified as due stray capacitance effect, which is omitted from computation in Section II. This stray capacitance is inherited from wiring and poor PCB layout. The 8-channels preamplifier-shaper chain in the UPM-MNA-RENA board channels are arranged in parallel configuration with CH4 and CH5 located in the middle. Test shows significant signal crosstalk between the adjacent channels due to capacitive coupling. The PCB design is double-sided where signal tracks and components are placed at top side and ground plane in the

bottom. This PCB configuration increased stray capacitance between signal tracks or component pads and the ground plane. This effect could be minimized by removing ground plane underneath the signal tracks and component pads. Due to this effect, we found that the individual channel output gain suffers from systematic drop (outer to middle channel). The CH4 and CH5 suffers most from this stray capacitance effect, therefore it can be seen from Fig. 5 that they have higher ENC. After recalculation, this stray capacitance is estimated around 30pF.

As described previously in Section III, the RMS noise for

different Cin is equivalent to sigma of the distribution. We use the area version of Gauss function and error function (Illario Defendi-Detectors & Electronics Group, FRMII Germany- Private communication) as fitting functions for the PM and EM respectively. Sample of plot and extracted sigma for both methods is shown in Fig. 6. The measured rms noise at different Cin for channel 1 is tabulated in Table I. Our measurement error (fitting error) is small, average at around 2%. This indicates that the measurements data are highly accurate and consistent. The RMS noise obtained with PM only deviates at average around 3% from the EM.

The rms noises obtained with this procedure are output referred noise. Therefore, the input referred noise in ENC form is obtained by dividing this rms noise with the system gain. The ENCs is plotted with respect to Cin and the actual ENC of the UPM-MNA RENA board is obtained by extrapolating the plot to Cin = 0pF. A sample of this plot for channel 1(CH1) is shown in Fig. 7. We also perform ENC measurement for the other channels and the results are shown in Table II. The result shows that the average deviation of the PM from the EM is small at around 1%. Results from the rms noise and ENC measurements show that the PM and EM are comparable. This validates our proposed method. Proving which method is superior in terms of accuracy, however, is beyond the scope of this work. One advantage of the PM is that the fitting function, i.e. Gauss function, is readily available in most scientific analysis software, whereas the error function must be user defined.

Fig. 5. Computed and measured ENC as a function of input capacitance.

220

TABLE I. EXTRACTED RMS NOISE, FITTING ERROR AND DEVIATION OF PM FROM EM NOISE FOR CHANNEL 1.

External Capacitor (pF)

rms noise PM & EM (mV), Fitting error (%) and Deviation of PM noise from EM noise (%)

Cin PM Fitting Error EM Deviation

5.6 2.45 1.3 2.44 0.4

22 2.81 1.4 2.70 4.1

33 3.18 2.5 3.38 5.9

47 3.66 0.3 3.71 1.3

56 4.03 3.4 3.91 3.1

68 4.41 0.5 4.27 3.3

82 5.07 1.5 5.04 0.6

Average - 1.6 - 2.7

TABLE II. ENC VALUES FOR ALL CHANNEL ACQUIRED WITH PM AND EM.

Channel

No. ENC (e_) and deviation of PM ENC from EM ENC (%)

PM EM Deviation

CH1 1803 1808 0.3

CH2 2071 2047 1.2

CH3 2262 2308 1.9

CH4 2264 2282 0.8

CH5 2340 2351 0.5

CH6 2220 2224 0.2

CH7 2067 2017 2.5

CH8 1863 1853 0.5

Average - - 1

V. CONCLUSION We have shown an experimental procedure for

measurement of ENC in an amplifying system for radiation detectors. This method allows extraction of electronic noise/ENC using instruments commonly available in radiation detection research. We demonstrate the application of this method by measuring the electronic noise/ENC in our newly developed preamplifier-shaper system. The method is verified with theoretical computation and validated using an established experimental method. Results are in good agreement with theoretical and established measurement methods.

ACKNOWLEDGMENT We acknowledge Dr. Ilario Defendi, Dr. Karl Zeitelhack

and Dr. Marcus Petertill from Detector & Electronic Group, FRM II Germany for the development of the error function. This works is supported by the Ministry of Science, Technology and Innovation Malaysia (MOSTI).

REFERENCES [1] P. W. Nicholson, Nuclear Electronics. John Wiley & Sons, 1974. [2] H. Spieler, “Low noise electronics in practical applications,”

Nucl. Instrum. Meth. A, vol. 636, pp.S149-S154, 2011. [3] E. Gatti and P. Manfredi, “Processing the signals from solid-state

detectors in elementary-particle physics,” La Rivista del Nuovo Cimento (1978-1999), vol. 9, pp.1-146, 1986. V. Radeka, "Low-Noise Techniques in Detectors," Ann. Rev. Nucl. Part. Sci., vol. 38, pp. 217-277, 1988.

[4] E. Gatti, M. Sampietro and P. F. Manfredi, "Optimum filters for detector charge measurements in presence of 1/f noise," Nucl. Instrum. Meth. A, vol. 287, pp. 513-520, 1990.

[5] G. Bertuccio and A. Pullia, "A method for the determination of the noise parameters in preamplifying systems for semiconductor radiation detectors," Rev. Sci. Instrum., vol. 64, pp. 3294-3298, 1993.

[6] R. Dinapoli, "A radiation tolerant pixel detector pixel detector system for the ALICE and LHCB experiments at CERN,"CERN-THESIS-2004-006, CERN, Geneva, 2004.

Fig. 6. A Sample of plots and fitted rms noise (sigma) for the PM and EM.

Fig. 7. A sample of plot and curve fitting (PM & EM) for the extraction of actual ENC of the UPM-MNA-RENA board.

221

[7] J. Y. Yeom, I. Defendi, H. Takahashi, K. Zeitelhack, M. Nakazawa and H. Murayama, "A 12-Channel CMOS Preamplifier-Shaper-Discriminator ASIC for APD and Gas Counters," IEEE Trans. Nucl. Sci., vol. 53, pp. 2204-2208, 2006.

[8] A. Kashchuk, "Method for test and diagnostics of the on-detector front-end electronics for the LHCb Muon system," LHCb-2002-013, CERN, Geneva, 2002.

[9] A. P. Kashchuk, N. A. Kuchinskii, O. V. Levitskaya, and S. A. Movchan, "Equivalent noise charge for different impedance-matching schemes of the drift straw-tube and the amplifier," Phys. Part. Nucl. Lett., vol. 8, no. 1, pp-40-45, 2011.

[10] "IEEE Standard Test Procedures for Amplifiers and Preamplifiers Used With Detectors of Ionizing Radiation," IEEE Std 301-1988, p. 0_1, 1989.

[11] T. Zimmerman. (1990, 5/8/2012). A Simplified, Improved Method for Making Amplifier Equivalent Noise Charge Measurements using a New Generation Digitizing Oscilloscope. Available: http://lss.fnal.gov/archive/tm/TM-1699.pdf

[12] J. Pahor and D. Ponikvar, "RMS noise meter," Nucl. Instrum. Meth. A, vol. 496, pp. 461-464, 2003.

[13] L. Lombigit, M.N. Hamidon, M. A. Khalid and N.Sulaiman, "Low cost front-end readout electronic for instrumentation used in neutron experiments," Int. J. Phys. Sci., vol. 7, pp. 2812-2817, 2012.

[14] OPA657. (2001, 7/8/2012). (Device datasheets)1.6GHz, Low-Noise, FET-Input Operational Amplifier. SBOS197E. Available: http://www.ti.com/lit/ds/symlink/opa657.pdf

222