Abstract
In most Terahertz time-domain spectrometer (THz-TDS) experiments, the lock-in amplifier works with trans-impedance pre-amplifier to amplitude the terahertz pulse accepted from detector. This paper discusses the development of data acquisition system for the transmission THz-TDS. In this system, the cross-correlation software algorithm in SR830 lock-in amplifier from Stanford Research Systems, that is usually used in THz-TDS, has been replaced by parallel hardware algorithm of Field Programmable Gate Array (FPGA) chip with the parallel processing ability. This chip has a faster processing speed and higher accuracy than others. A 24 bit Delta-Sigma Analog Digital (AD) was used in place of the 16 bit successive approximation ADC of SR830. The new AD convertor can reduce the complexity of trans-impedance pre-amplifier circuit and replace the SR555 current amplifiers that designed to work with SRS lock-in amplifiers. Besides trans-impedance pre-amplifier circuit, all function circuits, such as low-pass digital filter, phase-locked loop, Direct Digital Synthesis (DDS) reference source and the core algorithms, are integrated in a FPGA chip, which make the new designed lock-in amplifier with a small volume reduce a dozen times SR830 size.
Introduction
As an electromagnetic wave with wavelength between infrared and microwave, terahertz wave has characters with both electromagnetism and light electronics, which exists widely in nature. The Terahertz time-domain spectrometer (THz-TDS) is a coherent measurement technique with broadband terahert pulse and the spectral information can be achieved by Fourier transform. It is one of the latest methods of the spectrometric detection, which has important applications in many areas of basic research. It can be used to investigate material behavior with increasing interest owing to the potential of various applications in security (Robin et al. 2014), biomedicine (Fischer et al., 2005; Kawase et al., 2003; Siebert et al., 2002;), and non-destructive materials testing (Cristofani et al., 2014). The THz-TDS is an important way to obtain the pulse terahertz spectrum of the material. The most important factor in the whole system is the data acquisition module. Normally, the data acquisition system used in THz-TDS contained lock-in amplifier and trans-impedance pre-amplifier (Li et al., 2012). These two devices are very expensive and have bulky volume. It is difficult to integrate all devices into a system. However, THz-TDS have to shrink in size and price to become acceptable for real-word application.
The terahertz pulse generate the periodicity weak voltage signal, only of microvolt with large amplitude of noise at the same time. The noise has millivolt and bandwidth. To reduce bandwidth of noise and improve the signal-noise-ratio (SNR), the weak terahertz signal needs to be amplificated and denoised to obtain useful information of terahertz signal. There are two methods for terahertz signal acquisition in the laboratory: one is using the lock-in amplifier which specific detection of weak signal to amplify the terahertz pulse and reduce the noise; the second method uses current or voltage preamplifier to extract terahertz signal by multiple averaging. SR830 digital signal processing (DSP) lock-in amplifier was used in the THz-TDS at present in the laboratory for this research.
Lock-in amplifier technique is one of the important methods for weak signal detection (Meade, 1982). The lock-in amplifier uses a cross-correlation principle design of synchronous detection apparatus (Hu and Wen, 2003; Wu et al., 2003). Usually in the measurement, noise is a interference signal, it is restricted and impacts the sensitivity of a measuring instrument and is difficult to shield. In order to reduce the effect of noise on the useful signal and improve the SNR (Naftaly and Dudley, 2009), the normal procedure is using narrow band filter to remove the noise out of the band. Because of the problem of center frequency instability and Q value of general filter, it cannot meet the high requirements of filtering noise. The measured signal and referenced signal are synchronously measured from the lock-in amplifier. It does not have the frequency stability problem, and can be regarded as a tracking filter. Its equivalent Q value is decided by the time constant of low pass filter, so the stability of the elements and the environmental requirement is not high.
Recent surveys show that the lock-in amplifier worked with trans-impedance pre-amplifier could improve the SNR at least ten thousand times in more than 80 dB. Owing to its advantages, it is suitable for the THz-TDS to detect the weak signal.
From the perspective of the practicability of the THz-TDS, this paper has proposed a new program about the data acquirement of teraherz pulse signals instead of the former lock-in amplifier, which is not conducive to integration because of its large size. This paper mainly focuses on the design, development and test of data acquisition system for the transmission THz-TDS. We designed the trans-impedance pre-amplifier and lock-in amplifier with the Field Programmable Gate Array (FPGA) and Advanced RISC Machines (ARM) chip. A well-designed data acquisition system can make the THz-TDS spectrometer work much more effectively and reduce the expensive cost and volume of commercial lock-in and trans-impedance pre-amplifier. In addition to this, it is easy to integrate the data acquisition system with the whole THz-TDS, which makes the whole system small size and lightweight.
THz-TDS layout
The schematic diagram of the THz-TDS is show in Figure 1. The basic THz-TDS setup is given with an Yb-doped fiber laser source. A stable laser system that has low jitter is a necessity. In our work, a fiber laser whose gain medium is efficient for wavelength about 1560 nm is used to obtain an amplified laser that has stable operation and low jitter. The femtosecond fiber generates optical pulses with duration of 30 fs and a repetition rate of 80 MHz. The pump/probe measurement technique can be implemented in transmission THz-TDS. The laser beam is split by a beam splitter (BS) into a pump beam and a probe beam. In this system, photo conductive antenna is used for THz generation and detection. And the average powers of pump/probe beam need to be reduced to 30 mW before irradiated on the photoconductive antenna. Four off-axis parabolic mirrors are used in THz-TDS for focusing and collimating THz beam. The output signals from the detector antenna were amplified by a current preamplifier at first then processed into a lock-in amplifier. By scanning the delay line, we can obtain the terahertz electric field waveform.
The schematic diagram of the THz-TDS .
Design of the system
Based on the principle of lock-in amplifier (Alexander, 2009; Brisges 1988; Brown, 1983; Connor, 1982; Fish, 1993; Howard, 2002; Jiri, 2000; Mcdonough and Whalen, 1995; Mohanty, 1986; Ott, 1988; Van, 1987; Wilmshurst, 1985) and THz-TDS optical path, we propose the general design of the data acquisition system as below. Input is the signal to be measured through terahertz signal optical chopper after chopping, the reference signal output from chopper is served as the reference signal. As shown in Figure 2, the system consists of the low noise preamplifier circuit, AD/DA digital to analog conversion interface, cross-correlation digital detection unit, a current voltage input and output unit, and a communication interface of five parts.
General diagram of the trans-impedance pre-amplifier and lock-in amplifier.
Cross-correlation detection digital unit was completed in one FPGA chip, consisting of a digital phase locked loop (PLL), a reference signal generating unit, programming direct digital frequency synthesizer (DDS), four phase sensitive detectors (PSD), programming Finite Impulse Response low pass digital filter, two vector calculation units that can calculate the signal peak, amplitude value, and phase values at the same time.
In the current/voltage input and output unit, the input unit includes: input voltage A channel, B channel of voltage, differential signal channel A-B and external reference frequency signal input interface. Input reference signal relates to the use of external reference frequency, such as sine wave and square wave, and can also use internal reference. When using internal reference, it can achieve two kinds of frequency signal demodulation, and output synchronization of arbitrary waveform reference signal. The output unit includes analog output and digital output of the signal obtained by the simultaneous extraction. The ARM processor of 32 bit for the USB protocol driver and process management hsa been used in this system, USB as a bridge between the phase lock amplifier and the host computer.
First, the weak terahertz current pulse passed into Automatic Gain Control (AGC) circuit through the input port. AGC circuit is mainly composed of trans-impedance chip and an AGC chip. The trans-impedance chip in the circuit will be converted the weak current signal into voltage signal. The AGC chip will produce DC voltage changed with the amplitude of input level, and will sample the amplitude of input signal, to obtain a control voltage and adjust to the reversed gain of the amplifier. It will make the output terahertz signal remain constant or keep the change in a small range.
After the terahertz signal passed through the AGC circuit, the frequency trap and two harmonic trap filters were set to reduce the interference of power supply noise on the signal characteristics. In order to obtain the better discretization analog signal and prevent the high frequency in the terahertz signal to fold into the lower frequency, which will generate the false frequency component, the anti-alias filter has been used in this system to realize the clean signal and then transfer the real signal into the analog-to-digital conversion circuit in Analog Digital Circuit (ADC).
Digital signal after the ADC discretization is divided into four channels, and entered the four multipliers respectively, as a multiplicand of the multiplier. At the same time, the reference signal through the reference port input the shaping circuit, then entry to the PGA chip. The FPGA chip produced two groups’ orthogonal sine and cosine signal that four channels of orthogonal sine and cosine signals multiplied four channels of quantization signal processed by ADC. After that, we can obtain the sum frequency components and the difference frequency components of the reference signal and the input signal. The signal contained the sum frequency component and the difference frequency components by means of the Cascaded Integrator Comb (CIC) filter to achieve the conversion of sampling rate. And Finite Impulse Response (FIR) filter is mainly used to filter out the high frequency noise in the signal. When the signal passed through the combination filter of CIC and FIR filter, and then entered into the amplitude and phase calculator, eventually the amplitude and phase of real-time signal can be obtained accurately.
In order to facilitate better exchange information between users, we will manage the calculation of amplitude and phase of data with one piece of Main Control Unit (MCU) chip. The ARM is served as the system MCU, because of its advantages of signal processing. The user can carry on the instruction to read and write control through MCU communication interface such as USB, RS232, to achieve control of the lock-in amplifier. At the same time, if we need to obtain the analog quantity form of amplitude or phase information, the original data of digital quantity form can be converted into analog quantity form with Digital Analog Circuit (DAC) interface.
Test result
The performance test on amplitude discrimination, phase discrimination and signal detection under noise of different levels has been completed with standard and noise-mixed signal given by Multifunction Signal Generator.
The result shows that the system has good performance amplitude discrimination and phase discrimination. It can obtain millivolt level signals under strong background noise with good signal noise ratio.
Linearity is the corresponding relationship between the amplitude of output signal and input signal, is an important parameter to value the design of lock-in amplified current. The linearity typifies the capacity of the amplitude discrimination circuit of lock-in amplifier. The factors that affect lock-in amplifier are many-sided and complex, such as distortion of the preamplifier circuit response, bias voltage and retardation effects of capacitance and so on. We used the waveform generator to output the squared pulse with the 1 kHz frequency and 5 V amplitude as the reference signal for this lock-in amplifier. The input signal is the sine waveform with the same frequency and phase to reference signal. We changed the amplitude of input signal, and the lock-in amplifier was connected to the Mixed Signal Oscilloscope to monitor the output signal. Changing the input signal step by step, we obtained the relationship between the input voltage and output voltage, as shown in Figure 3. Using the data processing tool to analyze the detection data to build the regression equation between input and output. From the software, the result can give linear correlation coefficients between input and output above 0.99. The regression equation with significant correlation turns out that the linear relationship between the input and output signal is true. When the input signal is 0, the static output is -0.008 V, this influence can be diminished by the compensating circuits. The performance was repeated once, twice, and a third time at irregular voltage steps; the results show that the linear correlation coefficients of our lock-in amplifier remain basically unchanged: this amplifier has the stable linearity.
The linear relationship between input and output voltage signal.
The detection experiment was conducted under simulation condition. In this experiment, the sinusoidal signal with 100 mV amplitude and 1 Hz frequency generated by Multifunction Signal Generator was set as the reference signal. The chopped terahertz pulse was simulated with superimposition of a white noise of 2 V amplitude and 1 KHz frequency and this reference signal, as shown in Figure 4(a).
(a) The simulation of the chopped terahertz pules; (b) The detection signal from the developed data acquisition system.
After the signal and reference signal are processed by our data acquisition system, the result is shown in Figure 4(b). The detection signal is closed to direct current; the amplitude is about 50 mV. This result proved that our system contained the trans-impedance pre-amplifier and the lock-in amplifier can detect and extract the useful signal in high noise background.
We connected the data acquisition system in our THz-TDS to obtained the terahertz pulse. We measured six runs of the time traces; the mean time domain picosecond pulses and their corresponding Fourier transform spectra are shown in Figure 5. The bandwidth extends beyond 3 THz – an excellent value for this system.
THz pulses in the time domain and in the frequency domain. The time trace was recorded by our lock-in amplifier.
The dynamic ranger (DR) and SNR are key specifications of a THz-TDS. Whereas DR describes the maximum quantifiable signal change and determines the measurement bandwidth; SNR indicates the minimum detectable signal change and reflects the amplitude resolution or sensitivity. The first is that the data are acquired as time-domain traces, whereas measured optical parameters are derived from the Fourier-transform spectra. The dynamic range of spectral data is strongly frequency-dependent, and typically decreases steeply with frequency.
The dynamic range is defined as the ratio between the largest and smallest measurable signal. In practical terms, it is defined as (Naftaly et al., 2009; Tonouchi 2007)
where RMS of noise floor the root-mean-square (RMS) of the noise floor. It is clear from this expression that the dynamic range of terahertz system in the frequency domain will follow the spectral profile of its signal amplitude, as seen in Figure 6. It is a widely accepted practice to quote the maximum value with 1700 in this case as the dynamic range of terahertz time-domain system.
SNR of the amplitude spectrum, calculated as the ratio of the amplitude and its standard deviation. DR of the amplitude spectrum, calculated as the ratio of the amplitude and the RMS of noise floor.
The SNR indicates the minimum detectable signal change, and as such is a complementary system parameter to the dynamic. The SNR is defined as (Naftaly et al., 2009; Tonouchi 2007)
Similarly to the dynamic range, the SNR is different for the temporal and spectral data, and is frequency dependent. Figure 6 shows the SNR of the calculated spectrum for a typical time-domain trace averaged from six scans. The DR and SNR of the calculated spectrum in the frequency domain is shown in Figure 6. The SNR is then given by the ratio of the mean and standard deviation (SD) of FFT; while the DR is the ratio of the mean of FFT to the RMS of noise (here SNR∼50 and DR ∼1700).
It is a widely accepted custom to quote the maximum value (in this case ∼1700) as the DR of a THz TDS. Provided that the frequency dependence of the DR is borne in mind, this approach is justifiable to a degree, because the great majority of THz TDS produce similar spectral profiles.
Conclusion
In conclusion, we have developed the system containing the lock-in amplifier and trans-impedance pre-amplifier successfully and given the test results of the analog signals. In our system, parallel hardware algorithm of FPGA, a 24 bit Delta-Sigma AD chip, and trans-impedance pre-amplifier circuit are developed as the new approach for the data acquisition. It will reduce the size of the commercialization instruments and can be used more widely in the field. This data acquisition system based on ARM and FPGA is mainly used in weak signal detection under strong background noise.
It means this system can get weak signals under a noisy background easily. If the noise intensity is 1 to 100 million stronger than signal intensity (i.e. dynamic range of 120 dB), in this condition, our system can be used for extraction of weak signal, including the amplitude of the signal, modulus and phase value in real-time. It can be widely used in physics, astronomy, optical communication, radar, and biomedical engineering field. With the corresponding sensor, it can be used for detection of weak light, small displacement, micro vibration, tiny temperature deviation, small capacitance, weak magnetism, weak voice, micro electrical conductivity and micro current.
Footnotes
Acknowledgements
The authors are grateful to Maojiang Song for their valuable help. We want to thank Fei Yang and Feng Han for providing excellent advice on this work.
Declaration of Conflicting Interests
The authors declare that there is no conflict of interest.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors would like to acknowledge the financial support provided by Guizhou Science and Technology Department. This work was supported by Guizhou Science and Technology Department (NO.SY20143065, NO. J20142107) and General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China (NO. 2014QK063). This work is supported by National Natural Science Foundation of China (No. 21503045, No. 61540038).