150 GS/s Real-Time Oscilloscope Using a Photonic Front End Jason Chou, Josh Conway, George Sefler and George Valley

Bahram Jalali Department of Electrical Engineering University of California Los Angeles, USA

Electronics and Photonics Laboratory The Aerospace Corporation El Segundo, USA [email protected]

hold circuits to collect data at a much higher aggregate rate [3]. In this system each individual time sample must be interleaved in the time domain with data from a different ADC, and this puts tremendous strain on the back-end processing, calibration and equalization. As discussed in Ref [4], temporal interleaving also presents stiff requirements for the RF pre-amplifier that must provide a flat gain profile over the entire bandwidth with enough amplification and fidelity to drive all of the ADCs after a splitter. The second electronic approach uses frequency-domain interleaving [5]. This technique uses band-pass filters to demultiplex a wideband signal followed by downcoversion and digitization. One strong advantage of frequency-domain interleaving is that it reduces the constraints on both the ADC and on the RF amplifier, as front-end amplification is implemented after the demultiplexing. However, this technique is also highly demanding of the post-processing in the reconstruction of the time-domain waveform and in that broadband signals that overlap two frequency bins are difficult to reconstruct. In contrast to electronic approaches, a photonic technique has emerged which achieves time-domain bandwidth compression [6]. In this scheme, the RF signal to be digitized is modulated onto the intensity of a chirped optical carrier. Dispersion of the signal stretches the optical signal in the time-domain, thereby compressing the RF envelope. The key advantage is that commercial wavelength-division multiplexers (WDMs) can be used for the interleaving to route spectral bands to an array of photo-detectors and on to individual ADCs. This generates digitized time segments at each ADC, rather than single samples, greatly simplifying the complexity and reducing the number of time-domain interleaving steps to be stitched together by the back-end processing.

Abstract— We demonstrate an optical front end technology that multiplies the sampling rate of a real-time oscilloscope by a factor of three. Our approach uses an optical pre-processor to compress the signal bandwidth of continuous-time high speed RF waveforms. To operate in continuous-time mode, the optical signal, which carries the RF, must be segmented and demultiplexed into an array of N parallel channels. In prior work, large spectral overlap between channels was needed for calibration and this limited the multiplication factor, M, to values far below the maximum value of N, which is limited by the number of back-end digitizers. In this paper, we demonstrate a novel technique using temporal overlap between channels and achieve higher multiplication. The sampling rate of a four-channel 50 GS/s real-time oscilloscope is increased by a factor of 3, enabling us to digitize a 47 GHz tone at 150 GS/s. To our knowledge, this is a record in continuous time RF digitization.

I.

INTRODUCTION

The demand for high fidelity, high speed continuous-time digitizers ranges from research, to military technologies, to industrial applications. Examples of such needs include detection of transient responses of semiconductor devices, wideband radar, software defined radio and optical communications. While needs for such digitizers are growing rapidly, conventional analog-to-digital converters (ADCs) are straining to keep pace [1]. The device limitations of the sample, hold and quantization circuits limit the sampling rate and resolution through thermal, quantization and aperture jitter noise [2]. The effect of these degradations grows more significant with increased sampling rate, empirically degrading by 1 effective bit for every 2.3 dBS/s [2]. At extremely high frequencies greater than fT the ADC is limited by the fundamental device response of the transistors themselves.

The transient photonic time-stretch system is illustrated in Figure 1(a). The process begins with a short, broadband optical pulse. It is critical that the optical bandwidth of the input pulse be much larger than the bandwidth of the unstretched RF signal. The pulse is then passed through a dispersive medium, such as a dispersion compensating fiber or a chirped fiber Bragg grating [7], with dispersion D1 (seconds/Hz or ps/nm). Figure 1(a) illustrates the resulting

Two electronic approaches have been developed to interleave multiple lower rate ADCs. The first is temporal interleaving in which an array of ADCs is operated with subsampling period offsets. This enables the skewed sample-andThis work was supported under The Aerospace Corporation's Independent Research and Development Program

978-1-4244-2169-5/08/$25.00 ©2008 IEEE.

35

Fig. 2. Implementation of continuous-time 4-channel photonic bandwidth compressor.

Fig. 1. a) Illustrates the basic system principles of a transient photonic timestretch system. b) Depicts the shadow-casting analog to the time-stretch system in free space.

per segment are interleaved, rather than interleaving individual samples. As shown in Figure 2, stretching the input pulse to the inter-pulse period (the inverse of the pulse repetition rate of the laser) allows the continuous RF stream to be modulated onto a constant intensity repetitively chirped optical signal. Four adjacent channels are shown, limiting the maximum stretch factor (M) to 4 without signal ambiguity due to overlap within each channel. Practical considerations typically reduce the stretch factor to account for guard bands and channel overlap. Figure 2 also illustrates the reduction in interleaving accomplished through the photonic time-stretch method.

linearly chirped pulse with the long wavelengths travelling faster than shorter wavelengths. The RF signal is then applied to the chirped optical pulse through an intensity modulator. A second dispersive medium with dispersion D2 further stretches the pulse and the RF modulation commensurately. The optical signal is then square-law detected with a photo-detector, yielding an RF signal stretched in time by the factor M = (D1+D2)/D1. The frequency components are therefore reduced by the same factor, increasing the effective sampling rate of a given ADC. Kolner [8] has elucidated the mathematical duality between diffractive free-space optical propagation and pulse propagation through dispersive media. This equivalence allows us to illustrate our time-stretch system in its free-space shadow casting analogue, Figure 1(b), for clarity of concept. Note that the stretch factors are identical in both systems.

In our previous work we reported a four-channel continuous time photonic bandwidth compressor achieving 77 GS/s [4]. This demonstrated the use of an out-of-band pilot tone for inter-segment interleaving. The use of wavelength overlap for calibration, however, proved spectrally inefficient and a stretch-factor of only 1.55 was achieved. In the present work, the same 4-channel oscilloscope (Tektronix) was used to make a 150 GS/s scope employing photonic bandwidth compression. The stretch factor is increased to 3, thus achieving the highest sampling rate, real-time oscilloscope to our knowledge.

The transient time-stretch system using a single ADC at the back-end has been the subject of many successful demonstrations [6], including sampling rates up to 10 TS/s [9]. While ground-breaking, in these experiments the timesegments to be sampled are limited by the duration of the stretched pulse at the modulator, which scales as D1 times the optical bandwidth. Increasing D1 to provide longer sample durations leads to several issues, including frequencydependent signal fading known as the dispersion penalty [6] and increased optical loss in the proportionally larger D2.

II.

TEMPORAL VS. SPECTRAL OVERLAP

An important step in increasing the stretch factor is finding a spectrally efficient overlap technique. Redundancy in signal content at the edges of each segment is necessary to correct for static and slowly-varying mismatches between channels. In our previous work [4], custom trapezoidal filters were constructed with filter profiles similar to those shown in Figure 3(a). Unlike conventional WDM filters, this introduces intentional crosstalk amongst adjacent channels. In the construction of a photonic bandwidth compression system with spectral overlap, such trapezoidal filters are the most spectrally efficient method of overlap. This is due to limits of energy conservation and the fundamental behavior of the optical amplifiers. There is a finite amount of power spectral density (PSD) which must be shared in the overlap region. Unfortunately, this limits the stretch factor of systems employing spectral overlap. High M calls for a fast roll-off. This means that the wavelengths at which one channel has a high PSD, the adjacent channel is limited to low PSD. This

While transient digitizers have great scientific interest, extending the photonic time-stretch preprocessor to continuous time operation would be of great practical interest. This can be done by stretching the chirped pulse to a continuous time optical signal at the modulator and using conventional WDMs to spectrally sort the temporally overlapping optical signals after D2 and to route the signal to a parallel bank of ADCs (Figure 2). Continuous time operation using this process has several distinct advantages over the all electronic methods. First, unlike frequency interleaving in the electronic domain, no RF downconversion step is necessary and the strain on the processing is minimized. Second, in contrast to electronic time-domain interleaving, the time segments that include multiple samples

36

Fig. 3. a)Illustration of the spectral overlap technique in the frequency domain. b) Temporal overlap technique plotted in the time domain with spectrally diverse segments.

Fig. 4. Experimental setup

added to the optical modulation. This second, out-of-band tone is used for calibration and segment interleaving [4].

large power imbalance reduces the effective overlap region. Further amplifying this problem are the ADC quantization and noise sources that make the regions of low PSD unusable and forces a gradual filter roll-off.

The modulated signal is then dispersed a second time through DCF with dispersion parameter D2 = -3135 ps/nm. This combination of D1 and D2 yields a stretch factor of 2.99. A final WDM filter distributes the stretched signal to four parallel digitization channels. This analog-to-digital conversion is accomplished by a Tektronix DSA72004 realtime 50 GS/s oscilloscope. The continuous-time optical front end has increased the sampling rate to 150 GS/s and the analog bandwidth to 48 GHz.

The solution to these problems is to change the overlap technology. Instead of spectral overlap, we have developed a technique to overlap spectrally diverse segments in time. This temporal overlap, illustrated in Figure 3(b), offers several distinct advantages over the spectral means described above. Instead of relying on WDM filters with custom cross-talk profiles, conventional WDMs with guard bands can be used. Conservation of energy no longer limits the spectral roll-off in the overlap region, allowing for steep, brick-wall thin film filters making full use of the overlap. The SNR and quantization levels are also kept constant across the entire channel, making best use of the power in the channel. All of these features combine to allow a much higher stretch factor. III.

Fig. 5 shows single-shot real-time data of a 35 GHz tone and 47 GHz tone digitized at 150 GS/s. Figure 6(a) illustrates the implementation of a pilot calibration technique used to stitch four adjacent channels into a continuous time waveform [4]. Adjacent channels were concatenated by adjusting the gain, offset, and clock skew of each channel with the aid of the 6 GHz pilot tone. Figure 6(b) shows the FFT of the 6 GHz pilot tone and a 35 GHz signal tone for a single channel and a four channel output. The pilot tone provided real-time calibration data that was used to mitigate errors found between

4-CHANNEL BANDWIDTH COMPRESSION EXPERIMENT

A functional schematic of the 150 GS/s oscilloscope with a photonic bandwidth-compression front end is shown in Figure 4. The optical source of the system is a 1.5 ps modelocked fiber laser (Precision Photonics) that emits a pulse centered at 1550 nm. A repetition rate multiplier is used to increase the pulse repetition frequency to 74 MHz. This reduces the dispersion necessary in the first stage to achieve a continuous time stream of chirped pulses with a fixed optical bandwidth. These pulses are then dispersed in commercial dispersion compensating fiber (DCF) with dispersion parameter D1 equal to -1574 ps/nm. After the initial dispersion stage, the chirped pulses proceed to the temporal overlap filters. This configuration of brick-wall thin-film filters (Bookham) and precision delay lines takes four non-overlapping 2.5nm spectral channels and overlays them in time. The fiber delay lines have been fixed to achieve 15% overlap. These WDM filters also serve as implicit bandpass filters, removing extraneous frequencies to increase the system efficiency of the fiber amplifier that follows the WDM. A zero-chirp Mach-Zehnder (MZ) modulator then applies the RF tone to the chirped optical pulses through direct amplitude modulation. In addition to the 47 GHz tone to be digitized, a second 6 GHz pilot tone is

Fig. 5. Real-time continuous data of 35 GHz and 47 GHz tones.

37

maximize the system's spectral efficiency. In doing so we have increased the effective analog input bandwidth of a 16 GHz scope to 48 GHz. REFERENCES [1] [2]

[3] [4]

[5] [6]

Fig. 6. a) Plot of real time 150 GSa/s digitized data of a 35 GHz tone from four adjacent channels. b) Fast Fourier transforms (FFT) of both single channel and stitched digitized data.

[7]

adjacent segments. IV.

CONCLUSION

[8]

In summary, we have demonstrated the first 150 GS/s continuous-time oscilloscope implementing the photonic bandwidth compression technique. Such results have not been possible using electronic interleaving techniques to date. We have experimentally demonstrated a 4-channel system that multiplies the sampling rate of the state-of-the-art 50 GSa/s digitizer by a factor of 3. A novel technique of temporal overlap has been developed to increase the stretch factor and

[9]

38

R.H. Walden, "Analog-to-digital conversion in the early 21st century," MTT-S, Honolulu, HI, June 2007. B. Le, T.W. Rondeau, J.H. Reed and C.W. Bostian, "Analog-to-digital converters," IEEE Signal Processing Magazine, pp 69-77 (November 2005). J. Pickerd, “DSP in high performance oscilloscopes,” TektronixTM White Paper, March 2005. J. Chou, G.A. Sefler, J.A. Conway, G.C. Valley and B. Jalali, "4channel continuous-time 77 GSa/s ADC using photonic bandwidth compression," MWP, pp. 54-57 (2007). P. Pupalaikis, “Digital bandwidth interleaving,” LeCroyTM Technical Brief, March 2005. Y. Han and B. Jalali, “Photonic time-stretched analog-to-digital converter: Fundamental concepts and practical considerations,” Journal of Lightwave Technology, Vol. 21, No. 12, pp. 3085-3103, December 2003. J.A. Conway, G.A. Sefler, J. Chou and G.C. Valley, "Phase ripple correction: theory and application," Optics Letters, vol. 33, no. 10, pp. 1108-1110 (2008). B. Kolner, "Space-time duality and the theory of temporal imaging," IEEE Journal of Quantum Electronics, vol. 30, no. 8, pp. 1951-1963, August 1994. J. Chou, O. Boyraz, D. Solli and B. Jalali, "Femtosecond real-time single-shot digitizer," Applied Physics Letters, vol. 91, art. 161105 (2007).