GB2540744A - Photonic Digital-to-Analogue Converter - Google Patents

Abstract

A photonic digital-to-analogue converter (DAC), and its method of operation are disclosed. The photonic DAC comprises an array of semiconductor laser diodes (408, 410, 412), each of which is modulated with different relatively low frequency (<2GHz), low power (<-10 dBm), Large bit count (>14) source digital signals at least partially representative of an original RF signal. The optical outputs of the laser diodes have varying wavelengths and are fed into a wavelength division multiplexer 414 which combines them into ideally one multiplexed optical output which is then pulse modulated by a Mach-Zehnder modulator 416 with a pulse train 418 having a sample rate relatively much higher than the frequency of the source modulating digital signal to provide pulsed optical waveforms. The pulsed optical waveforms are then passed through a dispersive medium 422, such as an optical fibre, which stretches the pulses 424 before they fall on a photodetector 426 which provides a corresponding electrical analogue output 428. The invention is characterised in that each of the semiconductor laser diodes is directly modulated, and in that the response time of said photodetector is less than the inverse of the sampling rate of the pulse modulation means such that the analogue signal resulting from the photodetector resembles the original RF signal. This arrangement provides wide bandwidth and low jitter and finds use in communication systems, radar, TV and radio broadcast services.

Description

'vvf S’hi:

The present invention relates to a photonic Digital-to-Analogue Converter (DAC).

There are almost innumerable applications for Radio Frequency (RF) or Microwave/miliimetre wavelength signals, and in the vast majority of modern applications, there is commonly a requirement for digitally-encoded (arbitrary) signals to be converted to their analogue equivalents, most commonly a waveform of some description. DACs which employ solely electronic components are inherently disadvantaged on account the well-known issue of timing jitter, and equally well-known bandwidth limitations, increasingly, there is a general requirement to produce high-fidelity RF waveforms up to tens of GHz, and yet such frequencies are arguably beyond the theoretical limits of electronic DACs. Indeed, producing such high frequency waveforms is currently impossible using Silicon or Gallium Arsenide technologies and although signals up to 5GHz have been produced, it Is considered highly improbable, or at least unlikely, that true RF arbitrary waveforms above this figure will be realised using these technologies. in an effort to address these issues, photonic technologies have more recently been proposed, and in some circumstances, the disadvantages of purely electronic DACs have been mitigated to some extent. Photonic DACs have inherently wide bandwidth, low-jitter high-speed sampling, and reduced sensitivity to electro-magnetic interference.

In terms of known prior art, US7889107 describes a photonic DAC which includes multiple electro-optical converters (most commonly incorporating, or being embodied In, semiconductor laser diodes) to generate multiple first optical signals in response to multiple input signals, multiple optical attenuators to attenuate intensifies of the first optical signals and to generate multiple second optical signals, an optical coupler to combine the second optical signals and to generate a third optical signal, and a photodetector to convert the third optical signal into an electrical analog signal, in one embodiment the multiple electro-optical converters generate multiple first optical signals of multiple different wavelengths in response to multiple input signals, and after attenuation, said optical signals are combined in an array waveguide grating {AWC}, the unified optical signal output of which is then received by said photodetector which produces the analogue signal However, this photonic DAC relies fundamentally on attenuating the intensities of the first optical signals, and therefore the DAC proposed requires separate attenuation of each and every optical signal produced by the electro-optical converters. Furthermore, and perhaps more importantly as far as the present invention is concerned, the bandwidth of the resulting signal is the same as that of the input signal, For example, the arrangement disclosed might receive five 10GHz 1-bit streams and output a single 5-bit, 1QGHz stream. Practically, this arrangement Is limited by the laser modulation speed/bandwidth. in BDigital-to~Analog Conversion Using Electro-optic Modulators” IEEE Photonics Technology Letters 15, p.117-119, 2003, A, Yacoubian et al, the use of N external electro-optic modulators is suggested, as opposed to utilising directly modulated lasers, to generate N different levels of signal and combine them together again to generate signal which has 2N levels. Low-pass-filtering Is used to smooth the signal, thus generating analogue signal. However, in this arrangement, the bandwidth of the resulting signal Is the same as the Input signal which is applied to each external modulator, and therefore, as in the arrangement of US7889107, there is no possibility for Increasing the effective bandwidth. Also, external modulators have lower modulation efficiency than that of lasers in general. Additionally, and more importantly, such external modulators may have lower modulation linearity, are often expensive, and commonly introduce loss into the system.

In "High speed integrated InP photonic digital-to-anaiog converter,” 2006 international Conference on Indium Phosphide and Related Materials Conference, paper MA1.1, by A. Leven et al„ a photonic DAC is proposed consisting of 4 phase modulators integrated together to provide a 4~bit D/A conversion. The manner in which optical sampling and pulse-shaping is achieved is however considered deficient because external optical pulses are required for operation, and there is little facility for increasing the effective bandwidth of output signal. in "Ail-optical digital-to-analog conversion using pulse pattern recognition based on optical correlation processing,” Optics Express 13, p.10310-10313, ZOOS, T, Nishitani et at, a N bits DAC is proposed requiring a 2M~1 photonic DAC module which uses pulse pattern recognition based on optical correlation processing to achieve the conversion. The device proposed herein requires external optical pulses for operation and signal bandwidth is dependent on that optical source. Again, the device does not provide any increase in the effective bandwidth of output signal. it is an object of the present invention to overcome the existing disadvantages and deficiencies of existing prior art photonic DACs.

According to the present invention there is provided a photonic DAC comprising an array of semiconductor laser diodes, each of which is modulated with different relatively low frequency, low power, large bit count source digital signals at least partially representative of an original RF signal and which provide first optical outputs of varying wavelengths, a wavelength division multiplexer for combining said first optical outputs into comparatively fewer second optical outputs, pulse modulation means whereby said second optical outputs are further modulated by a pulse train having a sample rate relatively much higher than the frequency of the source modulating digital signal to provide pulsed optical waveforms corresponding In number to the incoming second optical outputs, a dispersive medium through which said pulsed optical waveforms are constrained to pass and are thus at least partially dispersed, and a photodetector which receives the incident dispersed pulsed optical waveforms and provides a corresponding electrical analogue output, characterised in that each of the semiconductor laser diodes is directly modulated, and in that the response time of said photodetector is less than the inverse of the sampling rate of the pulse modulation means such that the analogue signal resulting from the photodetector resembles the original RE signal.

Preferably the wavelength division multiplexer combines all the incoming first optical outputs into a single second optical output.

Preferably* the pulse modulation means is one of: a Mach-Zehnder modulator, an electro-absorption modulator, a nonlinear sampling gate.

In a particularly preferred arrangement, particularly if a larger on/off contrast is required (typical modulators may provide a typical on/off contrast of 20-30 dB, whereas a most preferred Level of on/off contrast is greater than this level, most preferably of the order of SO dB}* then the pulse modulation means may comprise 2 or more modulators arranged in series.

Preferably, any modulator forming part of, or constituting said pulse modulation means may be actively configured to enhance a base level on/off contrast value thereof.

Most preferably, the source digital signal is of a frequency (or speed) of the order of 1-10GHZ, more preferably <SGHz, most preferably 2-4GHZ, ideally 2GHz, and has a power of preferably <0 cBm, more preferably <~5dBm, and most preferably -10 dBm,

Preferably the bit count of the source digital signal is >8, preferably >»12, further preferably >“14, and most preferably >14.

In a second aspect of the present invention, there is provided a photonic method of digitai-lo-analogue conversion comprising the steps of: - Individually modulating a plurality of semiconductor diodes within an array with a relatively low frequency, low power source digital signal at least partially representative of an original RF signal to provide a plurality of different first optical outputs of varying wavelengths, > wavelength division multiplexing said first optical outputs into comparatively fewer second optical outputs, - pulse modulating said second optical outputs with a pulse train having a sample rate relatively much higher than the frequency of the source modulating digital signal to provide pulsed optical waveforms corresponding in number to the incoming second optical outputs, - subjecting said pulsed optical waveforms to dispersion, - and converting the energy of dispersed pulsed optical waveforms from optical to electrical by photodetection thereof, characterised in that' each of said semiconductor diodes Is directly modulated with a different source of digital signals, and further characterised in that the response time of said photodetector is less than the inverse of the sampling rate of the pulse tram.

The present invention finds advantage over the prior art because each of the semiconductor lasers in the array is directly modulated with different digital information, and therefore the optical output of each laser carries different information of far greater bandwidth than that available in corresponding electrical DACs, Furthermore, the optical sampling and time stretch techniques followed by a fast photodetector allows for high-fidelity (low noise, high linearity) combination of the individual analogue optical signals generated by the directly-modulated lasers in the optical domain and their conversion back to the RF via the photo-detection, it is therefore poss ble to create arbitrary analogue waveforms from a digital input signal with a high number of resolution bits. Furthermore, whereas the arrangement of US7889107 was limited by the laser modulation speed/bandwidth, the arrangement of the present Invention is not so limited, or at least limited to a far lesser extent A specific embodiment of the invention is now described by way of example and with reference to the accompanying drawings wherein.

Figure 1 provides a schematic depiction of a prior art time-stretch analog-to-digitai converter (ADC) including an optical front- end, providing a stretch factor of 4,

Figure 2 schematically depicts one possible prior art configuration for the optuai front-end of Figure 1,

Figure 3 schematically depicts a prior art photonic DAC configuration,

Figure 4 schematically depicts a photonic DAC configuration according co one embodiment of the present invention,

Figure 5 provides a graphical representation of the phenomenon of optical dispersion,

Figure 6 pfpides a graphical representation of the pulse-shaping which may be achieved by the photodetector utilised in one embodiment of the present invention, and

Figure 7 graphically presents data resulting from experiments carried out on a standard directly modulated semiconductor laser, {Model Specification: Iblana Photonics, model number is EP-1SS0-NLW), and in particular provides an indication of the spurious-free dynamic range (SFDR) {the available maximum ratio in d8 between the powers of fundamental or main signal and the greatest undesired spur, mist commonly identified as third-order intermodulation distortion 0MD-3) noise),

Oeteifed. Description

Referring firstly to Figure 1, an original analog RF signal 100 having a relatively wide bandwidth 102 (i.e, the RF signal consists of a relatively large number of different frequencies) is time-stretched and wavelength division multiplexed (WDM) with the help of a time-stretch preprocessor or optical frontend 104. The resulting RF signal segments 106, 108,110,112 are “slowed down" relative to the original signal 100 in that their frequency bandwidth is, in this case, quartered as shown at 114, and this facilitates captured and conversion of such signals by conventional electronic ADCs, as indicated at 116,118,120,122. Subsequently, digitized samples can be rearranged in the digital domain to obtain a digital representation of the original signal

An example optical front-end is depicted in Figure 2. An original analog signal 200 is modulated by an intensity modulator 202 over a chirped optical pulse 20¾ ip essence, this is an arrangement for indirect modulation, as opposed to one for direct modulation wherein the input signal or rather an electric analogue of it, is fed directly to the semiconductor laser s© that the optical output thereof is immediately modulated by, in effect, the incoming electrical signal the shading 206 across the breadth of the schematically depicted pulse 204 is indicative Of the different wavelengths within the pulse which are obtained by dispersing an ultra short supercontinuum pulse output from a chirped pulse source 208, such as a mode-locked laser (MIL). Intensity (or amplitude) modulator 202, of which one example is a Maeh-Zehnder modulator, produces the signal shown at 210, essentially the original RF signal 200 modulated with the chirped optical pulse 206, A second dispersive medium 212 stretches the optical pulse further, as schematically depicted at 214, which is direcied onto a photodetector (PD) 216, the output of which provides a stretched replica 218 of original signal

The reader will note that Figures 1 &amp; 2 relate to an ADC, as opposed to the present invention, which is a DAC. in the interests of completeness, it should be mentioned that motivation for the present invention was provided at least In part by work done on photonic ADCs, In particular, the technical publication A.O.j. Wiberg, 2. Tong, Liu, L., Ponsetto, J.L, Ataie, V., Myslivets, L, Aiic, N., Radic, $., /’Demonstration of 40 GHz analog-to-digital conversion using copy-and-sample-all parametric processing/’ Optical Fiber Communication Conference and Exposition (OFC/NFOEC), 2012 and the Rational Fiber Optic Engineers Conference , vol, no., pp.1-3,4-8 March 2012, OW3C.2 discloses use of a signal copied on many optical carriers with different wavelengths in conjunction with dispersion to achieve broadband photonic-assisted analog-to-digital conversion (ADC).

Figure 3 shows one embodiment of a prior art photonic DAC from US7889107. DAC 300 may Include a multiple number of electro-optical converters 310-1, 310-2,310-3,.,., and 310-N that generate a multiple number of first optical signals, a multiple number of optical attenuators 320-1, 320-2,320-3,..., and 320-N that attenuate intensities of the first optical signals to generate a multiple number of second optical signal, an optical coupler 330 that combines the second optical signals to generate a third optical signal and a photodetector 340 tip converts the third optical signal into an electrical analog signal ASC the structure of the DAC 300 is substantially the same as the structure of the DAC 100 except that an Nx1 arriyed waveguide gr ating (AWG) is used as the optical coupler 330, and as such effectively multiplexes the various different second optical signals together. The ©tectro-optical converters 310-1,310-2, 310-3,,,., and 3ΊΡ-Ν may generate a multiple number of flit optical signals PSC1-1, PSC1-2, PSC1-3,.... and PSC1-N having various different wavelengths H Ag, %... Aw. The electro-optical converter 310-1, 310-2,310-3,..., or 310-N may Include a driver 360-1, 360-2, 360-3,.,..,. or 360-N and a laser diode 370-1, 370-2, 370-3,,.,, or 370-N. Optical coupler 330 combines second optical signals PSC2-1, P5C2-2, PSC2-3,and PSC2-N having various wavelengths to generate a third optical signal PSC3,

In accordance with the invention, and referring to Figure 4, there Is shown schematically a photonic DAC configuration 400 in which multiple conventional electronic DACs 402, 404, „.406 (series references being 1, „„ I, ,N) are provided and directly modulate multiple semiconductor lasers 408, 410,412 (1,.... i,.... N) each outputting light of different wavelengths λ-η h,..., An- DACs 402, 404, 406 modulate their respective semiconductor lasers with an analogue signal of relatively low frequency (e,g, 1-2GH?.) compared to the originating RF signal, such as that referenced at 100,200 in Figs, 1 and 2 respectively, Thus, there are in total N laser outputs, all modulated with analogue signals emanating from N respective DACs 402,404,.... 406 , and these are then combined in an optical coupling device 414, such as a wavelength division multiplexer (WDM), possibly In the form of an arrayed waveguide grating (AWG) is optically sampled by using a Mach-Zehnder modulator (MZM) driven by extremely short (typically of the order of tens of picoseconds) electrical pulses 420. it should be mentioned that a 10 ps pulse has already bandwidth of the order of 60-100 GHt and Applicants do not currently envisage any applications requiring a greater bandwidth. Indeed for most current commercial application requirements, a 25 ps pulse (for 40 GS/s sampling) is envisaged to be more than sufficient. However, to scale the technique up, should such be required in future, it is certainly possible, and in some cases preferable, to use an optical sampling gate. Such an arrangement most simply comprises an optical modulator that Is driven by an optical signal (as compared to a MI modulator, that controls optical signal with an electric signal). Typically, such an optical sampling gate may be based on a non-linear optical material and the gating signal would be a pulse from a mode-locked laser (MIL), which can be of the order of 0.1-100 ps in duration. In any event, the pulses must be short enough to have duration shorter than any instantaneous period of change of original RF analogue signal. MZM 416 and pulse train 418 together have the effect of providing a pulse train 420 which is modulated by the light, and all its constituent wavelengths, emanating from the output of the WDM 414, and this composite pulse train, after passing through a dispersive medium 422 and thus being subjected to dispersion therein, is effectively stretched as illustrated at 424.

Finally, the optically processed light falls on a photodetector, which converts the incident light back to the desired analogue signal 428. The dispersive medium 422 referred to herein Is to be understood as encompassing any device winch gives rise to, or causes dispersion, such obviously included optical fibre, or any other device which is capable of inducing dispersion In, or causing dispersion of electromagnetic radiation (e.g., reflection-operated fibre Bragg gratings - e.g,, as discussed in http://terexion.com/en/cs-dcmli

Figure 5 demonstrates the effect of dispersion one a single one 500 of the modulated pulses shown at 420 in Fig. 4. Here as an example, we assume that four lasers with different wavelength fill to X4) are used.

As can be seen from the Figure, the phenomenon of dispersion effectively causes the different wavelengths of light within the single pulse 500 to propagate at different velocities through the dispersive medium, in essence, the fundamental relationships which underpin dispersion are that: - the phase velocity of a wave depends on Its frequency, and - the product of the phase velocity of a wave in a uniform medium and the refractive index of that medium is a constant (c, the speed of light in a vacuum); expressed in terms of wavelength, the refractive index of a medium varies with the wavelength of the light travelling through it

Accordingly, for any medium except free space (having a refractive Index >1), the phase velocity will be reduced by different amounts depending on the different wavelengths, and therefore different wavelength radiations propagate more or less quickly through the medium, and the effect shown In Figure 5 is realised. With a dispersive material such as optical fibre, each wavelength thus experiences a time delay, and as a result, the pulses (of different wavelengths) are stretched in time domain, essentially filling the “gap” Ts in 418, which although extremely short (< 1 ns, and possibly as low as a pic©- or femtosecond), would be sufficiently well resolved by the photodetector such that the required analog RF signal would be either impossible or extremely difficult to recreate, However, the evolution and extension of pubes within and between successive (optical) sampling times (e.g, 7s and Ts±i) as the light is propagating through the dispersive material, ideally so that each of the separated pulses is equally spread within the time Interval defined therebetween as shown at 502, permits a photodetector 428, preferably having a response time longer than each pulse duration 7s to produce the correct and required output electrical signal as seen in Fig, 6. In essence, the photo-detector does the bw-pass-filtering so only the envelope of the signal can be detected, which generates the required analogue RF signal

As the linearity of directly modulated semiconductor lasers is one of most critical properties for the present invention and that which ultimately would determine the overall performance of the proposed photonic DAC, the inventors herefer undertook some validation studies. These are represented in Figure 7, Specifically, they conducted a preliminary study on the linearity of a modern semiconductor laser (Model # XXXXXX manufacture by YYYYYYY). Figure 7 shows the third-order Intermodulation distortion {IMD-3}, being the level of unwanted signal generated by the nonlinearity of the semiconductor laser, and the fundamental signal The difference between the fundamental signal and IMP-3 is required to be as large as possible. The laser under test was measured to have a 83 dBc (for 1 Hz resolution) spurious-free dynamic range (5FDR), being the available maximum ratio in dB between the powers of fundamental signal and the greatest undesired spur, as limited by the background noise of used photo-detector. Although the results dearly demonstrate that, the semiconductor laser under test is more than capable (i.e. of sufficient linearity) of performing satisfactorily In the proposed photonic DAC, it is to be mentioned that the SFDR may be improved even further by using low noise photo-detectors., and we confirmed that direct modulation of laser can provide enough linearity for our invention.

The application of photonics DACs, such as described, will extend into any technological and scientific domain where agile, precise, low noise, phase-stable wide bandwidth RF signals are required. Such applications, to name but a few, indude test equipment, Communication systems, radar and radar test equipment, bio medical analysis, TV and radio broadcast services.

Claims (16)

Ckiims

1. A photonic DAC comprising an array of semiconductor laser diodes, each of which is modulated with different relatively low frequency, low power, large bit count source digital signals at least partially representative of an original RF signal and which provide first optical outputs of varying wavelengths, a wavelength division multiplexer for combining said first optical outputs into comparatively fewer second optical outputs, pulse modulation means whereby said second optical outputs are further modulated by a pulse train having a sample rate relatively much higher than the frequency of the source modulating digital signal to provide pulsed optical waveforms corresponding in number to the incoming second opticaloutputs, a dispersive medium through which said pulsed optical waveforms are constrained to pass and are thus at least partially dispersed, and a photodetector which receives the incident dispersed pulsed optical waveforms and provides a corresponding electrical analogue output, characterised in that each of the semiconductor laser diodes Is directly modulated, and in that the response time of said photodetector is less than the inverse of the sampling rate of the pulse modulation means such that the analogue signal resulting from the photodetector resembles the original RF signal.

2. A photonic DAC according to claim 1 wherein the wavelength division multiplexer combines all the incoming first optical outputs into a single second optical output,

3. A photonic DAC according to any preceding claim wherein the pulse modulation means is one of: a Mach-Zehnder modulator, an electro-absorption modulator, a nonlinear sampling gate.

4. A photonic DAC according to any preceding claim wherein tie pulse modulation means is actively configured to enhance a base level on/off contrast value thereof.

5. A photonic DAC according to any preceding claim wherein the pulse modulation means possess an on/off contrast of the order of SOdB.

6. A photonic DAG according to any preceding claim wherein the source digital signal has one of the following characteristics: - a frequency (or speed) between 1-20GHZ, - a frequency (or speed) between 2--40¾ • a frequency of 2GHa.

7. A photonic DAC according to any preceding claim wherein the source digital signal has one of the following characteristics: - a power of less than 0 dBm, - a power of less than 5dBm a power of -10 dBm,

8. A photonic. DAC according to any preceding claim wherein the source digital signal has one of the following characteristics: - a bit count greater than 8, a bit count greater than or equal to 12, - a bit count greater than or equal to 14, - a bit count greater than 14.

9. A photonic method of digitai-to-anaiogue conversion comprising the steps of: - Individually modulating a plurality of semiconductor diodes within an array with a relatively low frequency, low power source digital signal at least partially representative of an original RF signal to provide a plurality of different first optical outputs of varying wavelengths, -- wavelength division multiplexing said first optical outputs into comparatively fewer second optical outputs, pulse modulating said second optical outputs with a pulse train having a sample rate relatively much higher than the frequency of the source modulating digital signal to provide pulsed optical waveforms corresponding in number to the incoming second optical outputs, - subjecting said pulsed optical waveforms to dispersion, - and converting the energy of dispersed pulsed optical waveforms from optical to electrical by photodetection thereof, characterised in that each of said semiconductor diodes is directly modulated with a different, source digital signal and further characterised in that the response time of said photodetector is less than the inverse of the sampling rate of the pulse train,

10. A method according to claim 9 wherein the wavelength division multiplexer combines all the incoming first optical outputs into a single second optical output

11. A method according to either of claims 9 or 10 wherein the pulse modulation Is achieved using one of: a Mach»Zehnder modulator, an electro-absorption modulator, a non-linear sampling gate.

12. A method according to any of claims 9-11 wherein the pulse modulation means is actively configured to enhance a base level on/off contrast value thereof.

13. A method according to any of claims 9-12 wherein the pulse modulation means possess an on/off contrast of the order of 50dB.

14. A method according to any of claims 9-13 wherein the source digital signal has one of the following characteristics: - a frequency (or speed) between 1-2GCHZ, - a frequency (or speed) between 2-4GHz, - a frequency of 2GHz,

15. A method according to any of claims 3-14 wherein the source digital signal has one of 11½ following characteristics: > a power of less than 0 dBm, - a power of less than -SdBm, - a power of 10 dBm,

16. A method according to any of claims 9-IS wherein the source digital signal has one of the following characteristics: a bit count greater than 8. - a bit count greater than or equal to 12, - a bit count greater than or equal to 14, - a bit. count, greater than 14,