0801.1404/ms.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: %%% PASJ LaTeX template for draft(body)<2007/01/19>
3: %%% 
4: %%% IMPORTANT NOTICE FOR AUTHORS
5: %%% 1. ``\draft'' creates single column and double spaces format.
6: %%% 2. If you comment out ``\draft'', the output will be double column
7: %%%    and single space.
8: %%% 3. For cross-references, the use of \label/\ref/\cite and the 
9: %%%    thebibliography environment is strongly recommended
10: %%% 4. Do NOT use \def/\renewcommand.
11: %%% 5. Do NOT redifine commands provided by PASJ00.cls.
12: %%% 
13: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
14: \documentclass{pasj00}
15: \draft
16: 
17: 
18: \begin{document}
19: \SetRunningHead{Okuda \& Iguchi}{Performance Measurements of 8-Gsps 1-bit ADCs.}
20: \Received{2007/08/21}%{yyyy/mm/dd}
21: \Accepted{2007/10/23}%{yyyy/mm/dd}
22: 
23: \title{Performance Measurements of 8-Gsps 1-bit ADCs 
24: Developed for Wideband Radio Astronomical Observations}
25: 
26: %%% begin:list of authors
27: % Do NOT capitalize all letters in "textsc".
28:  \author{%
29:    Takeshi \textsc{Okuda}\altaffilmark{1}
30:    and  
31:    Satoru \textsc{Iguchi}\altaffilmark{1}
32:  }
33:  \altaffiltext{1}{National Astronomical Observatory of Japan, 2-21-1 Osawa, 
34:  Mitaka, Tokyo 181-8588, Japan}
35:  \email{takeshi.okuda@nao.ac.jp}
36:  \email{s.iguchi@nao.ac.jp}
37: %%% end:list of authors
38: 
39: %% `\KeyWords{}' always has to be placed before `\maketitle'.
40: \KeyWords{techniques: spectroscopic--- instrumentation: interferometers--- instrumentation: 
41: spectrographs--- radio continuum: general--- radio lines: general}
42: %Do NOT move this preamble from here!
43: 
44: \maketitle
45: 
46: \begin{abstract}
47: 8-Gsps 1-bit Analog-to-Digital Converters (ADCs) were newly developed toward the 
48: realization of the wideband observation. 
49: The development of the wideband ADCs is one of the most essential developments 
50: for the radio interferometer. 
51: To evaluate its performance in interferometric observations, the time (phase) 
52: stability and frequency response were measured with a noise source and a signal 
53: generator. 
54: The results of these measurements show that the developed ADCs can achieve the 
55: jitter time less than 0.05 psec at the time interval of 1 sec, the passband 
56: frequency response with the slope less than 0.73 dB/GHz and the ripple less 
57: than 1.8 dB, and the aperture time less than 20 psec. 
58: The details of the developed ADC design, the measurement methods, and the 
59: results of these measurements are presented in this paper.
60: \end{abstract}
61: 
62: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
63: \section{Introduction}
64: 
65: A high-speed Analog-to-Digital Converter (ADC) can greatly contribute to the 
66: frontier of radio astronomy. 
67: The performance of ADC is very important because it decides the sensitivity of 
68: radio astronomical interferometric observations and the accuracy of astrometry 
69: observations and geodetic experiments. 
70: To realize the wideband observation for the Atacama Large 
71: Millimeter/Submillimeter Array (ALMA), the 2-bit ADC operating at the 4-GHz 
72: sampling rate have been developed \citep{oki02}. A fringe spectrum that contains 
73: 20 line features as well as continuum emission at $\sim$86.2 GHz was obtained 
74: though an interferometric observation of the Orion-KL regions using these ADCs 
75: connected to the Nobeyama Millimeter Array (NMA). 
76: This result shows that the realization of high frequency-resolution over a wide 
77: bandwidth is technically feasible \citep{oku02}. 
78: This is the first time to obtain such a wideband astronomical fringe from a 
79: \emph{single} correlation product of digitized signals; 20 line features can be 
80: found by a careful inspection of the spectrum.
81: 
82: In this paper, the development of newly-developed 8-Gsps 1-bit ADC is described 
83: in Section 2, the measurement methods and the performance of the ADC in 
84: Section 3, and the conclusion in Section 4. 
85: 
86: \section{The development of 8-Gsps 1-bit ADC}
87: The 8-Gbps 1-bit ADC was developed with commercially available DEcision Circuit 
88: (DEC) and 1:16 DeMUltipleXer (DMUX) which are currently used as optical 
89: communication GaAs IC chips at a bit rate of 10 Gbps. 
90: These chips, DEC (OKI KGL4215) and DMUX (OKI K422C), have been developed for 
91: high-speed optical communication systems. 
92: Fig.~\ref{fig:01} and Fig.~\ref{fig:02} shows the developed ADC board that has 
93: two 8-Gsps 1-bit ADC units. 
94: Fig.~\ref{fig:03} shows the functional system block diagram of the ADC board. 
95: Fig.~\ref{fig:04} shows the inner structure of the ADC unit. 
96: 
97: The input signal power level is -20 dBm at the ADC board end. 
98: In this ADC board, the input analog signals with a power of -20 dBm are 
99: amplified with AMP (Quinstar Technology QLJ-04082530-JO), and then the power of 
100: the input signals to ADC units becomes about $0$ dBm. After being amplified, 
101: the analog signals are sample-held and quantized at a sampling rate of 8192/4096 
102: MHz with GaAs DEC chips, and then the quantized waveforms are digitized and 
103: demultiplexed with 1:16 DMUX chips in the ADC units. 
104: Since the GaAs DEC chips have a response speed of 10 Gsps, the newly-developed 
105: ADC will be capable of covering the frequency range from 0 to 10 GHz. 
106: From the ADC units, the digital signals paralleled to 16 bits by 1:16 DMUX chips 
107: are output to the digital processing board (Fig. \ref{fig:04}). 
108: On the digital processing board, the digital signal level is first converted 
109: from ECL (Emitter Coupled Logic) to PECL (Positive Referenced Emitter Coupled 
110: Logic). 
111: In a FPGA (Field Programmable Gate Array), the data stream is adjusted and 
112: unified to be 64 (32 $\times$ 2) bits $\times$ 128 Mbps at both of the 4-and 
113: 8-GHz sampling rates by switching the 1:2 and 1:4 demultiplexers, and by 
114: selecting the appropriate signal paths (see Fig.~\ref{fig:05}). 
115: After being selected, the digital signals are multiplexed by 2:1 multiplexer in 
116: ``Output Control" in FPGA as shown in Fig.~\ref{fig:03}, and then the output 
117: data stream from the ADC board is 32 bits $\times$ 256 Mbps as shown in 
118: Fig.~\ref{fig:05}. 
119: The relationship between the sampling rate, the decimation factor and output 
120: data lines is listed in Table \ref{tab:sampling}.
121: 
122: A reference clock signal (input level of $+10$ dBm) at 4 or 8 GHz is divided 
123: into two ADC units, and then the clocks are provided to the DEC and DMUX chips 
124: (see Fig. \ref{fig:03}). 
125: In this case, there is a delay caused between the DEC and DMUX chips because the 
126: transmission path length is different (see Fig.~\ref{fig:04}). 
127: The delay is adjusted by the phase shifter (Anritsu Corp. A5N1102) in the ADC 
128: board. 
129: The ADC board also outputs the reference clocks of IF1 and IF2 which are 
130: synchronized with digital signals. 
131: Although these clocks are generated from the output signals of 1:16 DMUX chip, 
132: the phase of IF1 is not coherent with that of IF2 because they are transmitted 
133: via different communication paths. 
134: In this ADC board, this phase skew is absorbed in ``Output Control'' by selecting 
135: either of the reference clocks in ``Clock, Timing Control" (Fig.~\ref{fig:03}). 
136: Also, this internal reference clock can be generated and synchronized with an 
137: external clock by selecting it in ``Clock, Timing Control".
138: 
139: Bit distributions in the Analog-to-digital conversion can be estimated on the 
140: FPGAs. 
141: In the ADC board, the bit distributions are obtained on each of the 16 parallel 
142: lines (bits) or on the sum of all bits. 
143: The results of the bit distribution measurement can be read by accessing an 
144: internal Monitor $\&$ Control unit which is integrated in the ADC board. 
145: Also, to adjust the threshold level, the analog offset of the ADC board can be 
146: changed by a command issued from the Monitor $\&$ Control computer to the DEC 
147: chips via the Monitor $\&$ Control unit (Fig.~\ref{fig:03}).
148: 
149: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
150: \section{Measurement of ADC Performance}
151: Phase stability of the ADC was evaluated from derivation of the Allan variance 
152: by measuring a phase variation between the stable reference signals and the 
153: digital-to-analog converted signals \citep{mat95}. 
154: However, since the measured results included the effects of the time skew in 
155: digital signal processing and the stability of the DAC (Digital-to-Analog 
156: Converter), it was difficult to directly evaluate the phase stability or jitter 
157: time of the ADC.  
158: 
159: The measurement methods and results of the jitter time, the passband frequency 
160: response and the aperture time of thedeveloped ADCs are presented in this section. 
161: 
162: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
163: \subsection{ADC Measurement System}
164: 
165: Digital backend system of the wideband IF system has been developed for the 
166: Atacama Submillimeter Telescope Experiment (ASTE) that is a project to install 
167: and operate a 10-m submillimeter telescope at a high altitude site (4,800 m) in 
168: the Atacama desert, Northern Chile \citep{eza04}. 
169: This system is composed of a reference clock generator, high-speed ADCs, a 
170: 60-Gbps Data Transmission System (DTS), and a digital correlator that can 
171: operate as an 8-Gsps 3-bits correlator between four input signals. 
172: The system block diagram of the ASTE wideband digital backend system is shown in 
173: Fig.~\ref{fig:06}. 
174: 
175: Two types of reference clocks (8192/4096 MHz and 128 MHz), which are provided by 
176: a high-stability reference clock generator, lock the phase between two ADC units. 
177: DTS Transmitter (DTS-T) module including the developed ADC boards converts the 
178: astronomical analog signals into the optical transmission signals of 10.24 Gsps 
179: $\times$ 3 lines $\times$ 2 IFs. 
180: In this digital backend system, the WDM (Wavelength-Division Multiplexing) 
181: modules with 16 wavelengths is used for transmission of these optical signals. 
182: Of 16 wavelengths, only six wavelengths are used for the DTS system. 
183: As shown in Fig. 6, the six optical signals are output from the DTS-T modules 
184: through the WDM modules to the DTS Receiver (DTS-R) boards. 
185: The WDM modules are connected to each other with only one optical fiber cable. 
186: The WDM module on the digital correlator side is directly connected to the DTS-R 
187: boards with 6 optical fiber cables.
188: 
189: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
190: \subsection{Allan Variance Measurement}
191: 
192: The fraction frequency fluctuation averaged over the time interval $\tau$ is 
193: \begin{equation}
194: \bar{y}_k = \frac{\phi(t_k+\tau)-\phi(t_k)}{2\pi \nu_0 \tau}, 
195: \end{equation}
196: where $\phi(t)$ is a phase difference between the two ADCs, $\nu_0$ is a frequency 
197: of an oscillator. 
198: The Allan variance is defined as 
199: \begin{eqnarray}
200: \sigma_y^2(\tau)
201: &=& \frac{\left< \left( \bar{y}_{k+1} - \bar{y}_k \right)^2 \right>}{2} \nonumber \\ 
202: &=& \frac{\left< \left[ \phi(t+2\tau)-2\phi(t+\tau)+\phi(t)  \right]^2\right>}{8 \pi^2 \nu_0^2 \tau^2}.  
203: \end{eqnarray}
204: 
205: The instrument setup for measuring the Allan variance between the two 
206: developed ADCs is shown in Fig.~\ref{fig:07}. 
207: The noise signals generated from a noise source (Agilent Technologies 346B) are 
208: cut with a analog bandpass filter of 4 to 8 GHz. 
209: The limited-bandwidth noise signals are input to two ADCs. 
210: The phase difference between two ADC units is measured with the digital correlator. 
211: The obtained auto-correlations and cross-correlations were corrected from the 
212: Van Vleck relationship for 1-bit (two-level) quantization \citep{van66} as follows, 
213: \begin{equation}
214: \rho^\ast = \frac{2}{\pi} \sin^{-1} \rho, 
215: \label{eq:van-vleck}
216: \end{equation}
217: where $\rho$ is real and $\rho^\ast$ are measured correlation coefficients,
218: respectively. 
219: The measurement results of the Allan standard deviation are shown in 
220: Fig.~\ref{fig:08}.
221: 
222: The Allan variance can be written as 
223: \begin{equation} 
224: \sigma_y^2(\tau) =  \sum_{i=0}^{N_{A}} A_{i}^2 \tau^{-2} 
225: + \sum_{i=0}^{N_{B}}B_{i}^2 \tau^{-1} 
226: + \sum_{i=0}^{N_C}C_i^2 + \sum_{i=0}^{N_D}D_i^2 \tau, 
227: \label{eq:allan}
228: \end{equation} 
229: where $A_i$, $B_i$, $C_i$ and $D_i$ are the Allan variance in 1 second. 
230: $A_i$ is due to the white-phase noise and the flicker phase noise, 
231: $B_i$ is due to the random-walk phase noise and the white-frequency noise, 
232: $C_i$ is due to the flicker frequency noise, 
233: and $D_i$ is due to the random-walk frequency noise. 
234: The slope of measured Allan variance (see Fig.~\ref{fig:08}) indicates that the 
235: time stability of the developed ADC is dominated by the white-phase noise 
236: throughout the entire time scale. 
237: Then Allan variance can be considered as 
238: \begin{equation}
239: \sigma_y^2(\tau) =  A_0^2 \tau^{-2} + B_0^2 \tau^{-1} + C_0^2,   
240: \label{eq:allan-m}
241: \end{equation}
242: and these coefficients are derived from a least square fit of these curves. 
243: The time range for the fit is from 1 sec to $9.0\times10^2$ sec and from 
244: $9.0\times10^2$ sec to $1.5\times 10^3$ sec for AD1$\ast$AD2 and from 1 sec to 
245: $1.3\times10^3$ sec and from $1.3\times10^3$ sec to $2.3\times 10^3$ sec for 
246: AD3$\ast$AD4, but the fit is not performed in more than $\sim 1.5\times 10^3$ sec for 
247: AD1$\ast$AD2 and more than $\sim 2.3\times 10^3$ sec for AD3$\ast$AD4, 
248: because the curves of the derived Allan standard deviation turns a steep slope. 
249: The results of the least squares estimation are shown in Fig.~\ref{fig:09}, 
250: and then these coefficients are derived as listed in Table \ref{tab:allan}. 
251: %The ADC boards seem to be stable on a timescale of $10^4$ sec. 
252: 
253: \citet{rogers} defined the coherence function as 
254: \begin{equation}
255: C(T) = \left| \frac{1}{T}\int^T_0 e^{j\phi(t)}dt \right|, 
256: \end{equation}
257: where $T$ is a coherent integration time. When $\phi$ is a Gaussian random variable, 
258: the mean-square value of $C$ is defined as 
259: \begin{equation}
260: \left< C^2(T) \right> = \frac{2}{T}\int^T_0 \left( 1-\frac{\tau}{T} \right) \exp \left[ -2 \pi^2 \tau^2 \nu_0^2 I^2(\tau) \right] d\tau,
261: \label{eq:cohe}
262: \end{equation}
263: and the relation between the Allan variance and the true variance becomes 
264: \begin{equation}
265: \sigma^2_y(\tau) = 2\left[ I^2(\tau) - I^2(2\tau) \right],  
266: \label{eq:relation}
267: \end{equation}
268: where $\tau$ is the time interval, and $\nu_0$ is the arbitrary frequency before 
269: sampling [see Eqs. (5), (9) and (11) in \cite{rogers}]. 
270: When $\left< C^2(T) \right>$ is equal to 1, there is no loss of coherence. 
271: From the results of the phase stability measurement (Table \ref{tab:allan}) 
272: and the relation between the true variance and the Allan variance (Appendix A), 
273: the rms value of the coherence function, Eq.(\ref{eq:cohe}), can be represented as 
274: \begin{eqnarray}
275: \left< C^2(T) \right> = \frac{2}{T} \int^T_0 \left( 1-\frac{\tau}{T} \right) 
276: \exp \biggl[ -2 \pi^2 \tau^2 \nu_0^2  \biggl\{ \frac{2}{3}A_0^2\tau^{-2} \nonumber \\
277:  +B_0^2 \tau^{-1} + \frac{C_0^2}{2\log 2} \log \left( \frac{T}{\tau} \right) 
278: \biggl\} \biggl] d\tau
279: \label{eq:coherence}
280: \end{eqnarray}
281: Thus, the analysis results from the measurement of the Allan variance are of 
282: great help to estimate the coherence loss and the coherent integration time for 
283: evaluation of the ADC performance. 
284: Finally, the values of coherence function $\left< C^2(T) \right>$ of the developed 
285: ADCs are more than 0.9998 for $T=1$ sec, 0.99998 for $T=10$ sec, 0.99999 for $T=10^2$ 
286: sec and $T=10^3$ sec, which are derived from Eq.(\ref{eq:coherence}) by using the 
287: measured coefficients listed in Table \ref{tab:allan}. 
288: 
289: The derived Allan standard deviation shows the time fluctuation of a difference 
290: between two ADCs, but is not jitter time of one ADC. 
291: However, the Allan standard deviation can be represented with the root of the 
292: product of one ADC jitter time and another one. 
293: If two ADCs have the identical performance, the jitter time $\sigma_t$ is given by: 
294: \begin{equation}
295: \sigma_t = \frac{\tau}{\sqrt{2}} \sigma_y. 
296: \end{equation}
297: Since the white phase noise and the flicker phase noise are dominated in 1 
298: second as shown in Fig.~\ref{fig:09}, the jitter time is less than 0.05 psec 
299: from the Allan standard deviation at an averaging time of 1 second.
300: 
301: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
302: \subsection{Passband Response Measurement}
303: 
304: The sensitive and accurate astronomical observations need an analog IF system 
305: that has flat spectrum from the receiver to the ADCs. If there are a slope and 
306: a ripple, they have adverse effect on the observation. 
307: Several different frequency responses are caused throughout the entire analog 
308: processing and in ADC, and then final slope and ripple in the spectrum after 
309: correlation are caused by mixing these heterogeneous responses. 
310: For example, the slope and ripple affect the profiles of emission lines from 
311: astronomical objects. 
312: 
313: The instrument setup for measuring passband responses of the ADCs is shown in 
314: Fig.~\ref{fig:10}. 
315: Input signals are made with a noise source (Agilent Technologies 346B) and a 
316: high-stability signal generator (Agilent E8241A) that outputs a CW signal of 
317: $\pm$ 1.0 dBm with accuracy. 
318: Power of a CW signal is as much as that of the noise source at the frequency 
319: channel of the correlator output. 
320: Frequency of the CW signal is continuously changed within the frequency range 
321: of the correlator outputs from 4096 MHz to 8192 MHz. 
322: Then, the measured spectrum can be considered as 
323: \begin{equation}
324: B_{\rm ON}(\nu) -B_{\rm OFF}(\nu) 
325: = B_{\rm ADB}(\nu) \times B_{\rm A}(\nu) \times B_{\rm CW}(\nu)
326: \label{eq:passband}   
327: \end{equation}
328: where $B_{\rm ON}(\nu)$ and $B_{\rm OFF}(\nu)$ are correlator outputs, and ON/OFF 
329: indicates whether the CW signal is transmitted or not. 
330: $B_{\rm ADB}(\nu)$, $B_{\rm A}(\nu)$, and $B_{\rm CW}(\nu)$ are passband responses of 
331: the ADC boards, the analog IF system, and the signal generator, respectively.  
332: 
333: The passband response of the analog processing, which is obtained from the 
334: transmission processing of the signal between the signal generator and the ADC 
335: boards, can be derived with a network analyzer (Agilent E8362B) and its frequency 
336: range is from 3000 MHz to 9000 MHz with a frequency resolution of 1 MHz. 
337: With respect to $B_{\rm A}(\nu)$, since the relative accuracy is less than $0.1$ 
338: dB at each frequency, the synthesizer is assumed to have a flat spectrum. 
339: These passband responses are shown in Fig.~\ref{fig:11}. 
340: The measurement results are summarized in Table \ref{tab:passband}. 
341: In this paper, the values of ripple (p-p) is defined by subtracting the slopes 
342: from the passband responses. There are slopes of less than 0.73dB / GHz, which 
343: is similar to the specification of analog amplifiers (Quinstar Technology 
344: QLJ-04082530-JO) in the ADC boards. 
345: The peak-to-peak values of the ripple are less than 1.8 dB. However, the 
346: spectrum of AD4 seems to have another ripple that is less than $\pm 0.2$ dB. 
347: Since this ripple becomes small when an attenuator is inserted 
348: between the input port of the ADC boards and 
349: the cable connector end in the measurement system, 
350: it is assumed that the root cause of 
351: the ripple is an impedance mismatch between the ADC boards 
352: and this measurement system. 
353: 
354: Note that these results include the passband responses of the entire ADC boards 
355: that are composed of the amplifiers, the attenuators, semi-rigid cables, 
356: and DEC chips. 
357: 
358: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
359: \subsection{Aperture Time Measurement}
360: 
361: In the ADCs, the input signals are sampled during the aperture time and quantized 
362: for conversion to the digital signal. 
363: The 1-bit digitalized CW signal is expressed as 
364: \begin{equation}
365: f(t) = \sum_n \delta (t-n\frac{1}{\nu_0}) \ast g(t),  
366: \label{eq:f(t)}
367: \end{equation}
368: where $\nu_0$ is the CW frequency and $g(t)$ is a response of the ADC in the 
369: time domain. 
370: The Fourier transform of Eq.(\ref{eq:f(t)}) is 
371: \begin{equation}
372: F(\nu) = \nu_0 \sum_m \delta(\nu - m \nu_0) \cdot G(\nu) 
373: = \nu_0 \sum_m G(m\nu_0). 
374: \end{equation}
375: By measuring the transfer function, $\tilde{G}(\nu)$, a response of the ADC can 
376: be obtained in the time domain. 
377: In addition, this response provides the aperture time of the 1-bit ADC. 
378: 
379: The instrument setup for measuring the aperture time of the ADCs is shown in 
380: Fig.~\ref{fig:12}.
381: Input signals are generated with the noise source and the high-stability signal 
382: generator which are identical with those used for the measurement of passband 
383: response. 
384: Power of the CW signal is greater than $+$40 dB, and is as much as that of the 
385: noise source with the 1 MHz resolution of the correlator outputs. 
386: The frequencies of the CW signal are set to 5.0 GHz, 6.0 GHz, and 7.0 GHz. 
387: 
388: Figure.~\ref{fig:13} shows a comparison between the ideal and measured power 
389: spectra of the digitized CW signal ($\nu_0 = 5.0, \ 6.0, \ {\rm and} \ 7.0$ GHz) 
390: in a sampling frequency of 8192 MHz.  
391: Except a CW input frequency and aliasing 
392: effects, the real frequencies of these spectrum line before being folded in 
393: sampling become $\nu = 2(n+1)\nu_0$ ($n=$1, 2, 3, ...). The transfer functions 
394: (see Fig.~\ref{fig:14}), $\tilde{G}(\nu)$, are derived from the power spectra. 
395: The result shows that the measured values of the real transfer functions become 
396: smaller than the ideal transfer functions when the frequency ($\nu$) is more 
397: than $\sim 40$ GHz.
398: 
399: The frequency responses of the ADCs in the time domain are derived from the 
400: Fourier transform of the transfer functions below $\nu \sim 200$ GHz, and are 
401: shown in Fig.~\ref{fig:15}. 
402: Thus, the aperture time of the ADCs is less than 20 psec.
403: 
404: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
405: \section{Conclusion}
406: This paper has described the details of the developed 8-Gsps 1-bit ADCs and 
407: presented the measurement methods and the performances of the ADCs. Summary of 
408: this paper is as follows.
409: \begin{itemize}
410: \item
411: The ADCs were developed with the commercially 
412: available DEC and DMUX at a bit rate of 10 Gbps. 
413: Input signals with power of $-20$ dBm are quantized 
414: with a sampling frequency of 8192 MHz or 4096 MHz. 
415: \item
416: The Allan standard deviation of the ADC boards indicates that their performance 
417: is stable in the timescale of $\sim10^4$ sec. 
418: The jitter time in one second is estimated to be less than 0.05 psec.   
419: \item
420: There are slopes of $\leq0.73$ dB/GHz in the passband responses of 
421: the developed ADC boards. 
422: The peak-to-peak values of the ripple are less than $1.8$ dB. 
423: \item
424: The aperture time of the developed ADCs was derived from power spectra when a 
425: strong CW signal is input into the developed ADC boards. 
426: The aperture time of the ADCs is estimated to be less than 20 psec. 
427: \end{itemize}
428: The performance of the developed ADCs is summarized in Table \ref{tab:performance}. 
429: 
430: \section*{Acknowledgements}
431: We would like to acknowledge H.~Kiuchi for the useful comments on the 
432: calculations of the Allan variance. 
433: We also acknowledge S.~Asayamam, T.~Fujii, M.~Inada, and H.~Iwashita. 
434: We wish to thank Niizeki, Y., Fujii, Y., Takei, K., Saito, K., Ozeki, K., Kawakami, K.,  
435: and Onuki, H. for helpful advices about the measurements. T.~O.~ was financially 
436: supported by the Japan Society for the Promotion of Science (JSPS) for Young 
437: Scientists. 
438: This research was partly supported by the Ministry of Education, Culture, 
439: Sports, Science, and Technology, Grant-in-Aid for Young Scientists (B), 
440: No.~17740114, 2005 and by Grant-in-Aid for Scientific Research, No.~15037212.
441: 
442: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
443: % See the manual for the detail.
444: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
445: \begin{thebibliography}{}
446: \bibitem[Okiura et al.(2002)]{oki02} 
447: Okiura, M., Iguchi, S., Okumura, S.~K., Momose, M., Matsumoto, K., 
448: and Kawaguchi, N.\ 2002, Proc.~IEEE MTT-S Internatioanl Microwave Symposium
449: (Seattle, U.S.A.) 485
450: 
451: \bibitem[Okumura et al.(2002)]{oku02}
452: Okumura, S.~K., Iguchi, S., Chikada, Y., Momose, M., and Okiura, M.\ 2002, 
453: \pasj, 54, L59
454: 
455: \bibitem[Matsumoto et al.(1995)]{mat95}
456: Matsumoto, K., Kawaguchi, N., and Horiuchi, S.\ 1995,  
457: IEICE Trans.~B, J78-B-II, 553, in Japanese
458: 
459: \bibitem[Ezawa et al.(2004)]{eza04}
460: Ezawa, H., Kawabe, R., Kohno, K., and Yamamoto, S.\ 2004,  
461: SPIE, 5489, 763
462: 
463: \bibitem[Van Vleck \& Middleton(1966)]{van66}
464: Van Vleck, J.~H. and Middleton, D.\ 1996,  
465: Proc.~IEEE, 54, 2
466: 
467: \bibitem[Rogers \& Moran(1981)]{rogers} 
468: Rogers, A.~E.~E  and Moran, J.~M.\ 1981,  
469: IEEE Trans. Instrumentation and Measurment, 30, 283 
470: 
471: \bibitem[Thompson et al.(2001)]{thompson}  
472: Thompson, A.~R., Moran, J.~M., and Swenson, G.~W.,~Jr.\ 2001, 
473: Interferometry and Synthesis in Radio Astronomy 2nd Edition(John Wiley $\&$ Sons, New York, USA)
474: 
475: \end{thebibliography}
476: 
477: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
478: \appendix
479: \section{The relation between the Allan variance and the true 
480: variance, and the coherence loss.}
481: 
482: It is well known that the white-phase noise has an Allan variance of 
483: \begin{equation}
484: \sigma^2_y(\tau) = A_0^2 \tau^{-2}, 
485: \end{equation}
486: and from Eq.(\ref{eq:relation}) the true variance of 
487: \begin{equation}
488: I^2(\tau) = \frac{2}{3} A_0^2 \tau^{-2}, 
489: \label{eq:wp}
490: \end{equation}
491: where $A_0$ is the Allan variance at an averaging time of 1 second. 
492: And by substituting Eq.(\ref{eq:wp}) into Eq.(\ref{eq:cohe}), 
493: the coherence loss in the case of white-phase noise is written as
494: \begin{equation}
495: \left< C^2(T) \right> = \exp \left( \frac{-4 \pi^2 \nu_0^2 A_0^2}{3} \right). 
496: \label{eq:wp-loss}
497: \end{equation}
498: 
499: Also, it is well understood that the white-frequency noise and the 
500: random-walk-of-phase noise have an Allan variance of 
501: \begin{equation}
502: \sigma^2_y(\tau) = B_0^2 \tau^{-1}, 
503: \end{equation}
504: and from Eq.(\ref{eq:relation}) the true variance of 
505: \begin{equation}
506: I^2(\tau) = B_0^2 \tau^{-1}, 
507: \label{eq:rp}
508: \end{equation}
509: where $B_0$ is the Allan variance at an averaging time of 1 second. 
510: And by substituting Eq.(\ref{eq:rp}) into Eq.(\ref{eq:cohe}), 
511: the coherence loss in the case of white-frequency noise and the 
512: random-walk-of-phase noise is written as
513: \begin{equation}
514: \left< C^2(T) \right> = \frac{2 ( e^{-a T} +a T -1)}{a^2 T^2}, 
515: \label{eq:rp-loss}
516: \end{equation}
517: where $a=2 \pi^2 \nu_0^2 B_0^2$.
518: 
519: The flicker-frequency noise has a constant Allan variance of  
520: \begin{equation}
521: \sigma^2_y(\tau) = C_0^2, 
522: \end{equation}
523: but the relation between Allan variance and the ture variance 
524: has been directly solved and expressed yet \cite{thompson,rogers}. 
525: However, by defining the finite limited time in Eq.(\ref{eq:relation}), 
526: the true variance can be derived as 
527: \begin{equation}
528: I^2(\tau) = \frac{C_0^2}{2 \log 2}\log\left( \frac{\tau_{max}}{\tau} \right), 
529: \end{equation}
530: where $\tau_{max}$ is a constant of an arbitrary value to be fixed by 
531: introducing some constraints. 
532: For instance, in the atmospheric phase instability, the Allan variance 
533: profile changes from the flicker-frequency noise to the white-phase 
534: noise and the flicker-phase noise in the time range of 10 to 100 seconds. 
535: Also, the Allan variance profile changes from the flicker-frequency noise 
536: to the random-walk-of-frequency noise in the atomic frequency standard. 
537: Thus, it can be defined that $\tau_{max}$ is the turnover time of Allan 
538: variance profile. 
539: If the coherent time of $T$ is more than $\tau_{max}$, the ture variance 
540: becomes 
541: \begin{equation}
542: I^2(\tau) = 0, 
543: \end{equation}
544: while if $T$ is less than $\tau_{max}$,  
545: \begin{equation}
546: I^2(\tau) = \frac{C_0^2}{2 \log 2}\log\left( \frac{T}{\tau} \right). 
547: \label{eq:fp}
548: \end{equation}
549: And by substituting Eq.(\ref{eq:fp}) into Eq.(\ref{eq:cohe}), 
550: the coherence loss in the case of white-phase noise and flicker-phase noise 
551: is written as
552: \begin{equation}
553: \left< C^2(T) \right> = \frac{2}{T} \int^T_0 \left( 1-\frac{\tau}{T} \right) 
554: \exp \biggl[ -\frac{\pi^2 \tau^2 \nu_0^2 C^2_0}{\log 2} 
555: \log \left( \frac{T}{\tau} \right) \biggl]  d\tau, 
556: \label{eq:fp-loss}
557: \end{equation}
558: 
559: The random-walk-of-frequency noise has an Allan variance of 
560: \begin{equation}
561: \sigma^2_y(\tau) = D_0^2 \tau, 
562: \end{equation}
563: where $D$ is the Allan variance at an averaging time of 1 second. 
564: By using the same derivation as that of the flicker-frequency noise, 
565: it is also defined $\tau_{end}$ is a finite integration time.
566: When the coherent time of $T$ is more than $\tau_{end}$, 
567: the ture variance becomes 
568: \begin{equation}
569: I^2(\tau) = 0, 
570: \end{equation}
571: while when $T$ is less than $\tau_{end}$,  
572: \begin{equation}
573: I^2(\tau) = \frac{D_0^2}{2} (T - \tau). 
574: \label{eq:rf}
575: \end{equation}
576: And by substituting Eq.(\ref{eq:rf}) into Eq.(\ref{eq:cohe}). 
577: the coherence loss in the case of random-walk-of-frequency noies is 
578: written as
579: \begin{equation}
580: \left< C^2(T) \right> =  \frac{2}{T} \int^T_0 \left( 1-\frac{\tau}{T} \right) 
581: \exp \biggl[ -\pi^2 \nu_0^2 D^2_0 \tau^2 (T-\tau) \biggl] d\tau
582: \label{eq:rf-loss}
583: \end{equation}
584: 
585: \newpage
586: 
587: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
588: % Figure 
589: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
590: \begin{figure}
591:  \begin{center}
592:    \FigureFile(80mm,80mm){fig01.eps}
593:  \end{center}
594:  \caption{Top view of the 8-Gsps 1-bit ADC board. The top board, middle board, 
595:  and bottom board are a bias circuit board, a digital processing board, and an 
596:  ADC extended board, respectively.}
597:  \label{fig:01}
598: \end{figure}%
599: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
600: \begin{figure}
601:  \begin{center}
602:    \FigureFile(80mm,80mm){fig02.eps}
603:  \end{center}
604:  \caption{Side view of the 8-Gsps 1-bit ADC board. 
605:  This is top view of the ADC extended board (see Fig.~\ref{fig:01}. 
606:  ``A/D-1'' and ``A/D-2'' are the ADC units that have a built-in DEC and a DMUX.}
607:  \label{fig:02}
608: \end{figure}%
609: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
610: \begin{figure}
611:  \begin{center}
612:    \FigureFile(160mm,160mm){fig03.eps}
613:  \end{center}
614:  \caption{Functional system block diagram of the 8-Gsps 1-bit ADC board. 
615:  The solid, dashed, and gray lines indicate IF signals, clock signals, and 
616:  ``Monitor \& Control" buses, respectively. }
617:  \label{fig:03}
618: \end{figure}%
619: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
620: \begin{figure}
621:  \begin{center}
622:    \FigureFile(80mm,80mm){fig04.eps}
623:  \end{center}
624:  \caption{Inner photograph of the 8-Gsps 1-bit ADC unit that is in the ADC 
625:  extended board (see Fig.~\ref{fig:02}). 
626:  The left and right chips are a DMUX and DEC, respectively.}
627:  \label{fig:04}
628: \end{figure}%
629: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
630: \begin{figure}
631:  \begin{center}
632:    \FigureFile(143mm,160mm){fig05.eps}
633:  \end{center}
634:  \caption{Data formats of the developed ADC board according to 
635:  sampling rates. } 
636:  \label{fig:05}
637: \end{figure}% 
638: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
639: \begin{figure}
640:  \begin{center}
641:   \FigureFile(160mm,160mm){fig06.eps}
642:  \end{center}
643:  \caption{The system block diagram of the ASTE wideband digital backend system.} 
644:  \label{fig:06}
645: \end{figure}%
646: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
647: \begin{figure}
648:  \begin{center}
649:   \FigureFile(80mm,80mm){fig07.eps}
650:  \end{center}
651:  \caption{Block diagram of the instrument setup for Allan variance measurements. 
652:  NS is the noise source (Agilent Technologies 346B) and BPF is the 4-8GHz 
653:  bandpass filter.}
654:  \label{fig:07}
655: \end{figure}%
656: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
657: \begin{figure}
658:  \begin{center}
659:   \FigureFile(160mm,160mm){fig08.eps}
660:  \end{center}
661:  \caption{Allan standard deviation between samples on the second time scale. 
662:  Figures show Allan standard deviations of AD1$\ast$AD2 and those of AD3$\ast$AD4. 
663:  Solid lines, dashed lines, and dot lines indicate the results of an IF frequency 
664:  of 5 GHz, 6 GHz, and 7 GHz, respectively. The gray lines indicate the coherence 
665:  $\left< C^2(T) \right>$ of 0.99999 in the case of white-phase noise.}
666:  \label{fig:08}
667: \end{figure}%
668: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
669: \begin{figure}
670:  \begin{center}
671:   \FigureFile(160mm,160mm){fig09.eps}
672:  \end{center}
673:  \caption{Results of  least squares fitting to the derived Allan standard 
674:  deviations. 
675:  The top panels show the Allan standard deviations (gray) and fitting curves. 
676:  The solid lines indicate the sum total of the components, $\tau^{-1}$(dashed), 
677:  $\tau^{-1/2}$(dot-dashed), and $\tau^0$(3dots-dashed). 
678:  The bottom panels show values of relative residual 
679:  ($=(\sigma_y-\sigma^\ast)/\sigma_y$, where $\sigma^\ast$ is a derived fitting 
680:  curve).}
681:  \label{fig:09}
682: \end{figure}%
683: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
684: \begin{figure}
685:  \begin{center}
686:   \FigureFile(80mm,80mm){fig10.eps}
687:  \end{center}
688:  \caption{Block diagram of the instrument setup for measurements of passband 
689:  responses. SG is a signal generator(Agilent Technologies E8241A).}
690:  \label{fig:10}
691: \end{figure}%
692: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
693: \begin{figure}
694:  \begin{center}
695:   \FigureFile(80mm,80mm){fig11.eps}
696:  \end{center}
697:  \caption{Passband responses of the ADCs with the frequency resolution of 1 MHz. 
698:  The passband responses are normalized by 
699:  $\int^{\nu_2}_{\nu_1} B_{\rm ADB}(\nu) d\nu = 1$, 
700:  where $\nu_1$ and $\nu_2$ are $4097$ MHz and $8192$ MHz, respectively. }
701:  \label{fig:11}
702: \end{figure}%
703: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
704: \begin{figure}
705:  \begin{center}
706:   \FigureFile(80mm,80mm){fig12.eps}
707:  \end{center}
708:  \caption{Block diagram of the instrument setup for the measurements of the 
709:  aperture  times.}
710:  \label{fig:12}
711: \end{figure}%
712: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
713: \begin{figure}
714:  \begin{center}
715:   \FigureFile(160mm,160mm){fig13.eps}
716:  \end{center}
717:  \caption{Power spectra of CW signals, 5.0 GHz, 6.0 GHz, and 7.0 GHz with a 
718:  sampling frequency of 8192 MHz. 
719:  Amplitude is normalized by values of the amplitude of each CW signal. 
720:  The upper figures and the lower figures indicate the ideal spectra and the 
721:  measured spectra, respectively. }
722:  \label{fig:13}
723: \end{figure}%
724: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
725: \begin{figure}
726:  \begin{center}
727:   \FigureFile(80mm,80mm){fig14.eps}
728:  \end{center}
729:  \caption{Transfer functions for the CW frequencies of the AD1. Ratios are 
730:  normalized by a spectral-line power at the first-ordered frequency (5.0 GHz, 
731:  6.0 GHz, and 7.0 GHz).The dashed line indicates an ideal transfer function for 
732:  the CW frequency of 6.0 GHz. }
733:  \label{fig:14}
734: \end{figure}%
735: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
736: \begin{figure}
737:  \begin{center}
738:   \FigureFile(160mm,160mm){fig15.eps}
739:  \end{center}
740:  \caption{Aperture time of the ADCs. 
741:  The aperture time of the ADC is less than 20 psec.}
742:  \label{fig:15}
743: \end{figure}%
744: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
745: 
746: \newpage
747: 
748: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
749: % Table 
750: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
751: \begin{table}
752:  \begin{center} 
753:   \caption{The composition of 1:2(4) DMUX and the observation data according to 
754:  the sampling rates.}
755:   \begin{tabular}{ccc}
756:    \hline\hline
757: Sampling rate & Decimation & Description \\ 
758: \hline 
759: 8Gsps & 1:4 & Either of IF1 or IF2 data \\
760: 4Gsps & 1:2 & 64 bits each for IF1 and IF2 \\
761: \hline
762:   \end{tabular}
763:  \end{center}
764:  \label{tab:sampling}
765: \end{table}%
766: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
767: \begin{table}
768:  \begin{center} 
769:   \caption{ The Allan variances in 1 second for various types of noise. 
770:    Time rage indicate fitting ranges.} 
771:   \begin{tabular}{cccccc}
772:    \hline\hline
773:    Correlation & Freq [GHz] & $A_0$ & $B_0$ & $C_0$ & Time range [sec] \\
774:    \hline
775:    AD1$\ast$AD2 & $5.0$ & $6.3\times10^{-14}$ & $2.2\times10^{-15}$ & $6.5\times10^{-17}$ & $1.0 \le \tau \le 9.0\times10^2$ \\
776:                 &       & $1.1\times10^{-13}$ & ---                 & ---                 & $9.0\times10^2 < \tau  \le 1.5\times10^3$ \\
777:                 & $6.0$ & $5.4\times10^{-14}$ & $2.0\times10^{-15}$ & $2.8\times10^{-17}$ & $1.0 \le \tau \le 9.0\times10^2$ \\
778:                 &       & $8.3\times10^{-14}$ & ---                 & ---                 & $9.0\times10^2 < \tau  \le 1.5\times10^3$ \\
779:                 & $7.0$ & $5.0\times10^{-14}$ & $1.4\times10^{-15}$ & $5.6\times10^{-17}$ & $1.0 \le \tau \le 9.0\times10^2$ \\
780:                 &       & $8.0\times10^{-14}$ & ---                 & ---                 & $9.0\times10^2 < \tau  \le 1.5\times10^3$ \\
781:    \hline 
782:    AD3$\ast$AD4 & $5.0$ & $5.7\times10^{-14}$ & $3.2\times10^{-15}$ & $1.5\times10^{-16}$ & $1.0 \le \tau \le 1.3\times10^3$ \\
783:                 &       & $2.3\times10^{-13}$ & ---                 & ---                 & $1.3\times10^3 < \tau  \le 2.3\times10^3$ \\
784:                 & $6.0$ & $5.0\times10^{-14}$ & $3.1\times10^{-15}$ & $1.4\times10^{-16}$ & $1.0 \le \tau \le 1.3\times10^3$ \\
785:                 &       & $2.2\times10^{-13}$ & ---                 & ---                 & $1.3\times10^3 < \tau  \le 2.3\times10^3$ \\
786:                 & $7.0$ & $6.3\times10^{-14}$ & $4.0\times10^{-15}$ & $1.8\times10^{-17}$ & $1.0 \le \tau \le 1.3\times10^3$ \\
787:                 &       & $2.8\times10^{-13}$ & ---                 & ---                 & $1.3\times10^3 < \tau  \le 2.3\times10^3$ \\
788:    \hline
789:    \end{tabular}
790:  \label{tab:allan}
791:  \end{center}
792: \end{table}%
793: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
794: \begin{table}
795:  \begin{center}
796:   \caption{ Passband response for ADC boards.} 
797:   \begin{tabular}{ccc}
798:    \hline\hline
799:    ADC bord & slope [dB/GHz] & ripple(p-p) [dB]\\
800:    \hline
801:    AD1 & 0.58 & 1.6 \\
802:    AD2 & 0.63 & 1.4 \\
803:    AD3 & 0.73 & 1.8 \\
804:    AD4 & 0.64 & 1.5 \\
805:    \hline
806:    \end{tabular}
807:  \label{tab:passband}
808:  \end{center}
809: \end{table}%
810: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
811: \begin{table}
812:  \begin{center}
813:   \caption{ Performance of the developed ADCs.} 
814:   \begin{tabular}{ll}
815:    \hline\hline
816:    Sampling Frequency       & 4096 MHz or 8192 MHz\\
817:    Input level              & $-20$ dBm \\
818:    Number of bit            & 1 \\
819:    Jitter time in 1 second  & $<$0.05 psec\\
820:    Passband response        &  \\
821:    \ \ slope                & $\leq 0.73$ dB/GHz \\
822:    \ \ ripple(p-p)          & $\leq 1.8$ dB\\
823:    Aperture time            & $<$ 20 psec \\
824:    \hline
825:    \end{tabular}
826:  \label{tab:performance}
827:  \end{center}
828: \end{table}%
829: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
830: \end{document}
831: