Optical clock injection

Optics has several potential advantages for the precise clocking of electronic logic systems. Fundamentally, optics can generate and propagate short pulses, even over long distances, and with precise timing of the arrival of the pulses, but electrical wiring can do neither. So optics can be used for very precise timing injection, even to sub-picosecond levels, and for running even large systems synchronously.

David Miller has pointed out the basic advantages of optics for clocking [1], including the possibility of using optics to allow large synchronous systems. With his colleagues, he has analyzed the possibilities for optical clocking of chips [2][3], has demonstrated the necessary fast rise times in even silicon photodetectors [4][5], and has shown specifically low capacitance integrated silicon photodetectors for such applications [5]

Key conclusions from this work are that, though it would be difficult to provide enough optical power economically to allow clocking at all the points that need to be clocked on a chip, optics could provide very precise and predictable clocks over larger areas and systems. The use of “receiverless” approaches [2] are viable with integrated photodetectors and can also eliminate voltage amplification stages that can add to clock jitter.

Use of short optical pulses to read out optical modulator outputs can actually directly remove jitter and skew from the logical signals [6], and signaling with short optical pulses can also reduce latency. [7].

See this paper [8] for recent larger discussion of optical interconnects, including timing benefits.

See also the discussion of optically assisted analog to digital conversion. There, a key limitation in the digital precision of high-speed analog to digital converters is the precision with which the sampling timing can be set. Optics offers serious solutions to such problems.

Timing problems of wires

The problems in distributing precise timing in electrical wiring over any substantial distance come mostly from the resistance of the wires. As a result, electronic wiring cannot easily propagate short pulses or even sharp rising edges over long distances [1], and cannot guarantee the overall delay time of such timing signals over wires of any substantial length. The pulses or edges broaden because of the intrinsic resistance of wires – a problem for both electrical lines both in their “RC” (resistive-capacitance) mode (which dominates in all short lines, such as most of those on chips), and in their “RLC” (resistive-inductive-capacitive) mode (which dominates in most longer distances beyond chip size scales, especially at high speed. These resistance problems become even worse at high speeds or for fast transients because of the skin effect, which concentrates conduction near the surface of the wiring at high speeds. Furthermore, the effective arrival time of electrical signals is effectively temperature-dependent because of the temperature-dependence of the resistance of copper. Taken together, these problems prevent the precise electrical distribution of high-speed clock signals, especially over longer distances. Hence, it is essentially impossible to run large wired electrical systems synchronously, and clock timing can only be guaranteed in relatively small local domains.

Additionally, running high-speed clock signals down electrical wires can lead to inductive voltage drops on the wires, including on power supply lines, and hence to interference with logical signals that use the same lines for reference voltages.

Physical benefits of optics in distributing clock signals

Optical signals in free space have extremely well defined propagation times – the velocity of light is a constant – and these times do not change even as we go to very high bandwidth signals (or, equivalently, short pulses). The propagation times then depend only on distance. Even a change of a few millimeters in distance only leads to delay changes of 10 ps or so. (In free space, light propagates about 300 microns in 1 ps.) Even within optical fibers, the change in refractive index with temperature is very small, so there is very little change in the propagation time of signals on optical fibers within, say, the size of electronic systems even with large temperature changes. For example [1], in an optical fiber, the change of refractive index with temperature, which is ~ 10^-5 / ºC. For a 10 m optical fiber cable, the corresponding change in delay over a 100 ºC temperature range is only about 30 ps.

There is also very little dispersion of optical pulses (effectively, very little “spreading out” in time) over the size scales of meters, even inside optical fibers, so very precise clocking “edges” or pulses can be distributed over large distances.

[1] D. A. B. Miller, “Physical Reasons for Optical Interconnection,” Special Issue on Smart Pixels, Int. J. Optoelectronics 11, 155‑168 (1997)

[2] C. Debaes, A. Bhatnagar, D. Agarwal, R. Chen, G. A. Keeler, N. C. Helman, H. Thienpont, and D. A. B. Miller, “Receiver-less Optical Clock Injection for Clock Distribution Networks,” IEEE J. Sel. Top. Quantum Electron. 9, 400-409 (2003)

[3] D. A. B. Miller, A. Bhatnagar, S. Palermo, A. Emami-Neyestanak, and M. A. Horowitz, “Opportunities for Optics in Integrated Circuits Applications,” International Solid State Circuits Conference, 2005, Digest of Technical Papers, IEEE 2005, Paper 4.6, Pages 86-87

[4] A. Bhatnagar, S. Latif, C. Debaes, and D. A. B. Miller, “Pump-probe measurements of CMOS detector rise time in the blue,”  J. Lightwave Technol. 22, 2213‑2217 (2004)

[5] S. Latif, S. E. Kocabas, L. Tang, C. Debaes & D. A. B. Miller, “Low capacitance CMOS silicon photodetectors for optical clock injection”, Appl. Phys. A – Materials Science and Processing 95, 1129-1135 (2009)

[6] G. A. Keeler, B. E. Nelson, D. Agarwal, C. Debaes, N. C. Helman, A. Bhatnagar, and D. A. B. Miller, “The Benefits of Ultrashort Optical Pulses in Optically-Interconnected Systems,” IEEE J. Sel. Top. Quantum Electron. 9, 477-485 (2003)

[7] D. Agarwal, G. A. Keeler, C. Debaes, B. E. Nelson, N. C. Helman, and D. A. B. Miller, “Latency Reduction in Optical Interconnects Using Short Optical Pulses,” IEEE J. Sel. Top. Quantum Electron. 9, 410-418 (2003)

[8] D. A. B. Miller, “Attojoule Optoelectronics for Low-Energy Information Processing and Communications: a Tutorial Review,” IEEE/OSA J. Lightwave Technology 35 (3), 343-393 (2017)