NON-LINEAR FILTER FOR DML

Abstract

A circuit is disclosed having a component having repeatable distortion characteristics; and a drive circuit for providing a drive signal and comprising a non-linear filter for pre-compensating for distortion introduced by the component having repeatable distortion characteristics in response to the drive signal, the distortion having a non-linear response to the drive signal.

Description

This application claims the benefit of U.S. Provisional Patent Application No. 61/984,621, filed on Apr. 25, 2014, the entire contents of which are incorporated herein by reference.

FIELD OF THE INVENTION

The invention relates generally to communication and more particularly to optical communication.

BACKGROUND

Optical transmitters employing Directly Modulated Lasers (DML) such as Vertical Cavity Surface Emitting Lasers (VCSELs) are rated to operate up to a predetermined data rate. Problematically, when operating at higher data rates, distortion from the DML itself limits performance of the device and thus the data link. The DML transmits an optical signal that differs from the drive signal provided thereto such that signal reception is substantially affected beyond short transmission distances. Added jitter and vertical eye closure from distortion introduced by VCSEL can cause significant reduction in signal-to-noise ratio (SNR). These limitations on performance place a limit on the transmission distances for higher data rates.

Linear filters are used conventionally to partially compensate for the distortion due to the DML itself However, linear filters fail to achieve optimal compensation for the distortion. It would be advantageous to overcome some of the shortcomings of the prior art.

SUMMARY OF EMBODIMENTS OF THE INVENTION

In accordance with an aspect of at least one embodiment there is provided a component having repeatable distortion characteristics; and a drive circuit for providing a drive signal and comprising a non-linear filter for pre-compensating for distortion introduced by the component having repeatable distortion characteristics in response to the drive signal, the error having a non-linear response to the drive signal.

In accordance with an aspect of at least one embodiment there is provided a method comprising: providing a drive current for driving a Directly Modulated Laser (DML); filtering the drive current with a non-linear filter to provide pre-compensated drive current pre-compensated for errors in a signal resulting from driving the DML with the drive current, wherein an output signal from the DML in response to the pre-compensated drive current better approximates the drive current to incur reduced errors.

In accordance with an aspect of at least one embodiment of the invention there is provided a circuit comprising: an input port for receiving a first signal; a plurality of taps, each tap comprising an input port for receiving a tap input signal, a first input port for receiving a first weight, a second input port for receiving a second other weight, and a biasing circuit for biasing an applied weighting between the first weight and the second weight to bias the tap signal, the biased tap signal for modifying the first signal.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 illustrates a typical optical output signal amplitude of a DML (VCSEL) in response to a direct driving input signal.

FIG. 2 is a simplified block diagram of a linear finite impulse response (FIR) filter.

FIG. 3 is a logic diagram of non-linear FIR filter.

FIG. 4 is a diagram of a non-linear FIR filter implementation.

FIG. 5 is a diagram of another non-linear FIR filter implementation optimized for performance.

FIG. 6 is an eye diagram of an unfiltered drive signal alongside an eye diagram of an output signal corrected with a non-linear FIR filter such as that of FIG. 4 or FIG. 5.

FIG. 7 is a graphical representation of the transmit signal before and after filtering with a 4 tap non-linear FIR filter.

FIG. 8A shows a sample circuit for implementing a non-linear filter for pre-compensating a drive signal for driving a directly modulated laser (DML).

FIG. 8B shows another sample circuit for implementing a non-linear filter for pre-compensating a drive signal for driving a directly modulated laser (DML).

FIG. 8C shows another sample circuit for implementing a non-linear filter for pre-compensating a drive signal for driving a directly modulated laser (DML).

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

The following description is presented to enable a person skilled in the art to make and use the invention, and is provided in the context of a particular application and its requirements. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and applications without departing from the scope of the invention. Thus, the present invention is not intended to be limited to the embodiments disclosed, but is to be accorded the widest scope consistent with the principles and features disclosed herein.

Referring to FIG. 1, shown is a typical optical output signal amplitude of a Directly Modulated Laser (DML) in the form of a Vertical Cavity Surface Emitting Laser (VCSEL) in response to a direct driving input signal. As is evident, the optical signal generated (thin line) fails to follow accurately the signal provided (thick line). The resulting overshoots and undershoots add distortion to the signal. The distortion appears as both jitter affecting the width of the eye and amplitude variations affecting the opening of the eye.

Referring now to the eye diagram that is shown on the left-hand side of FIG. 6, the signals are distorted horizontally—i.e., increased jitter—and vertically—i.e., degraded SNR. As is evident, the inner eye opening is significantly smaller than it would be if the signal was undistorted. Correction of these distortion artifacts within the transmit signal are important to enable transmission of higher data rates over longer distances.

The distortion artifacts resulting from DML optical response are amplitude dependent and thus non-linear in nature. The rising edge and falling edge responses are different and they each need to be compensated differently. Further, compensating one edge response may adversely affect the other edge or may fail to achieve significant improvement without compensating for the other edge as well. Thus, conventional approaches using linear filters for compensating for the distortion from the DML response are not optimal.

A second problem is implementation efficiency. If the distortion is repeatable and calculable, it may be possible using a DSP to reduce the nonlinear distortion within the DML signal; that said, such an implementation would be costly and would not lend itself to inexpensive, low power and compact implementation. A more simple non-linear distortion reduction method would be preferred.

Referring now to FIG. 2, shown is a simplified block diagram of a typical linear finite impulse response (FIR) filter. A signal is provided to the filter and is then summed with a weighted delayed version of the signal, termed as a delayed tap, or a plurality of weighted sequentially delayed versions of the signal or delayed taps. Linear FIR filters are well known and well studied.

Because the distortion is non-linear in nature, a linear filter is not suitable to addressing the distortion concerns completely. In fact, such a linear filter, will fail to substantially correct the problems disclosed above, reducing distortion in one of the rising or falling edge response while compounding the distortion in the other.

Referring now to FIG. 3, shown is a non-linear FIR filter design for providing pre-compensation for some of the non-linearities shown in FIG. 1. Here, a signal is provided to the filter consisting of multiple delayed taps and each tap's contribution is weighted by two different factors dependent on input signal's instantaneous amplitude, resulting in an amplitude-dependent, non-linear filter response. The filter generates a different response to each of the rising and falling edges and approximately compensates for the non-linear response of the DML in response to a driver signal. The use of two weights per tap, combined with scaling the tap contributions with instantaneous input signal amplitude, allows for non-linear filter response. While the use of the FIR architecture supports compact and efficient implementation. For example, the non-linear FIR filter shown is implementable as an analogue circuit within a semiconductor, for example, without relying on complex processing circuitry such as a DSP.

Referring now to FIG. 4, shown is a simplified diagram of a practical implementation of non-linear FIR filter for Non-Return to Zero (NRZ) signaling. Again, a signal provided is delayed and tapped, and each tap contribution is weighted differently, depending on input signal level of one or zero, to provide level dependent non-linear operation. The tap contributions are added back into the signal to provide filtering thereof. Signal scalars are reduced to gates in the filter shown, as suits an integrated hardware implementation.

Referring now to FIG. 5, shown is a diagram of another non-linear FIR filter implementation optimized for performance in the present embodiment. Here again, each tap signal is acted on by two different weights. A multiplexer is used to select the weighting for multiplication. The input signal level is used to select one of the multiplexer's input weights, and the multiplier scales the tap signal as per selected weight for each stage. Since the weights for each tap switch as per the input signal's level, the filter output need not follow a linear contour. The resulting non-linear filtering pre-compensates for distortion in the DML.

Referring again to FIG. 6, shown is an eye diagram of an uncorrected output signal (left-hand side) alongside an eye diagram of an output signal corrected with a non-linear FIR filter such as that of FIG. 4 or FIG. 5 (right-hand side), sand according to the present embodiment. As is shown, the eye has opened up considerably with reduced jitter and improved SNR. An improved eye diagram is typically reflective of improved ability to transmit over greater distances and reduced error in signal reception.

FIG. 7 shows a graphical representation of the transmit signal before (left-hand side) and after filtering with a 4 tap non-linear FIR filter (right-hand side). Most noteworthy, signal distortion is greatly reduced after a short time reducing distortion central to the eye. At the rising edge and falling edge, distortion remains, but it is significantly reduced. Thus, the eye opening in an eye diagram is improved. Further, other frequency components resulting from the distortion are reduced with reduced distortion.

Just looking to the falling edge, it is seen that whereas without filtering, the signal bounces at the bottom down and up, with filtering the signal remains substantially in alignment with the desired signal contour. On the rising edge, two notable bounces are reduced to one smaller bounce, thereby limiting the effect of the bounce on the top of the eye.

FIGS. 8A-C shows three sample circuits for implementing a non-linear filter according to the embodiment. Each circuit has different drawbacks and advantages, but effectively, the filter design allows not only for analogue hardware implementation, but for varied implementation to take advantage of different power sources, power levels, and other design criteria. Architectures supporting implementation flexibility are typically desirable as they are useful in many different applications and well suited to implementation in many different devices.

As is seen in each of the circuit diagrams, two currents proportional to weights are shown designated with “w” (w_0 and w_1) being multiplexed into the scaling circuit for each tap determined by level of the input signal (Dp, Dn). Alternatively, the currents proportional to weights are applied to a scaling circuit such that they are first scaled by the input signal (Dp, Dn) followed by the tap signal (Tnp, Tnn). Further alternatively, currents proportional to the weights are applied to a scaling circuit where the signals being scaled are a logical combination of input signal (Dp, Dn) and the tap signal (Tnp, Tnn). The logical combinations include input signal (Dp, Dn) logically OR'd with tap signal (Tnp, Tnn) designated as “Dp+Tnp”; and input signal (Dp, Dn) logically AND'd with tap signal (Tnp, Tnn) designated as “Dp.Tnp”. The scaled version of these logically combined signals in current form is then summed through a wire OR to produce a single tap contribution that is dependent on the weights and the input signal amplitude. Multiple tap contributions are summed to generate a resulting signal that has an amplitude dependent non-linear characteristic.

Though FIGS. 8A-C show one tap for each architecture, it is understood by those of skill in the art that any number of taps is supported and selection of a number of taps is dependent upon the circuit design requirements. Further, though two weights are shown, the filter architecture described above may be implemented with additional weights to correct for more complex amplitude dependent non-linear effects requiring higher granularity or resolution in amplitude levels.

Though the above embodiments are directed to pre-compensating the drive current, filtering of received signals to improve data detection is also supported. The general architecture for non-linear filter as shown in FIG. 3 can be used for a received signal. In end-to-end fibre optic communications such as fibre optic cables for communicating, the transmitter and receiver pairing is known and the weights within the receiver are tuned for use with a specific receiver or are adjusted based on a transmitter from which a signal is received.

In use, a circuit is designed and manufactured. Once manufactured, the circuit is tested with a representative DML component and based on the combined circuit and DML transmit signal characteristics, the non-linear FIR filter weights are adjusted to pre-compensate the drive current for the DML. Thus, each product is compensated individually, accounting for known DML response issues as well as circuit specific response issues for a given DML. Once compensated, the circuit operates in compensated mode. Optionally, the circuit's operating parameters are readjusted to re-compute the weights for the non-linear FIR filter at intervals.

In another embodiment, the optical output signal is tapped and provided as feedback to the transmit circuit where the non-linear FIR filter is adjusted in response to changes in performance of the DML output signal. Further optionally, the circuit is designed and manufactured with fixed weighting for the non-linear FIR filter.

In another embodiment, the manufactured devices are tested, the non-linear FIR filter is tuned—weights are set—and the circuit is tested again. Based on its performance, the circuit is assigned a quality level. Thus, some manufactured drive circuits support 25 GHz while others support only 15 GHz—determined after tuning in the manufacturing stage. This allows for a more coarse tuning process with the performance assignment then dividing between circuits with best tuning and those with less effective tuning results.

Numerous other embodiments may be envisioned without departing from the scope of the invention.

Claims

1. A circuit comprising:

a component having repeatable distortion characteristics; and

a drive circuit for providing a drive signal and comprising a non-linear filter having at least a tap for pre-compensating for distortion introduced by the component having repeatable distortion characteristics in response to the drive signal, the distortion having a non-linear response to the drive signal.

2. A circuit according to claim 1 wherein the component comprises a Directly Modulated Laser (DML).

3. A circuit according to claim 2 wherein the DML comprises a Vertical Cavity Surface Emitting Laser (VCSEL).

4. A circuit according to claim 2 wherein the DML comprises a Distributed FeedBack (DFB) laser.

5. A circuit according to claim 3 wherein the DML is operated at at least 25 Gbps.

6. A circuit according to claim 4 wherein the non-linear filter comprises a non-linear Finite Impulse Response (FIR) filter having at least 2 weights for application at each delayed tap and supporting at least one delayed tap.

7. A circuit according to claim 4 wherein the non-linear filter comprises a non-linear Finite Impulse Response (FIR) filter having at least 2 weights for application at each delayed tap and supporting at least 3 delayed taps.

8. A circuit according to claim 1 wherein the non-linear filter comprises a non-linear Finite Impulse Response (FIR) filter having at least 2 weights for application at each delayed tap and supporting filtering of both a rising edge, low to high signal level response and a falling edge, high to low signal level response.

9. A circuit according to claim 8 comprising:

for each tap a first input port for receiving a first weight, a second input port for receiving a second other weight, a switch for switching between the first weight and the second weight, and a weighting circuit for weighting of a signal within the tap to produce a tap output, tap output signals from different taps combined to form the drive signal.

10. A circuit according to claim 8 comprising:

for each tap a first input port for receiving a first weight, a second input port for receiving a second other weight, a scaling circuit for scaling the first weight and the second weight, and a weighting circuit for weighting of a signal within the tap to produce a tap output, tap output signals from different taps combined to form the drive signal.

11. A circuit according to claim 1 wherein the non-linear filter comprises a non-linear Finite Impulse Response (FIR) filter having greater than 2 weights at each delayed tap supporting filtering of a complex amplitude dependent non-linear distortion for a signal with a modulation scheme having greater than 2 amplitude levels of consequence for a given data symbol, such as PAM4 or 4-Level Pulse Amplitude Modulation.

12. A circuit according to claim 11 consisting of an analogue filter circuit.

13. A circuit according to claim 12 wherein the circuit is implemented in an integrated semiconductor.

14. A circuit according to claim 11 comprising:

for each tap a first input port for receiving a first weight, a second input port for receiving a second other weight, a scaling circuit for scaling the first weight and the second weight, and a weighting circuit for weighting of a signal within the tap to produce a tap output, tap output signals from different taps combined to form the drive signal.

15. A method comprising:

providing a drive current for driving a component;

filtering the drive current with a non-linear filter to provide pre-compensated drive current pre-compensated for distortion in a signal resulting from driving the component with the drive current, wherein an output signal from the component in response to the pre-compensated drive current has reduced distortion and better approximates an ideal transmit signal for an intended modulation.

16. A method according to claim 15 wherein the component comprises a Directly Modulated Laser (DML).

17. A method according to claim 16 wherein the directly modulated laser comprises a Vertical Cavity Surface Emitting Laser (VCSEL).

18. A method according to claim 16 wherein the directly modulated laser comprises a Distributed FeedBack (DFB) laser.

19. A method according to claim 18 wherein filtering is performed with an analogue filter.

20. A method according to claim 15 wherein the analogue filter is implemented in semiconductor.

21. A method according to claim 15 wherein the non-linear filter comprises a non-linear FIR filter.

22. A method according to claim 15 wherein filtering corrects for both a rising edge, low to high signal level response, and a falling edge, high to low signal level response.

23. A circuit comprising:

an input port for receiving a first signal;

a plurality of taps, each tap comprising an input port for receiving a tap input signal, a first input port for receiving a first weight, a second input port for receiving a second other weight, and a scaling circuit for scaling an applied weighting based on the first weight and the second weight to scale the tap signal, the scaled tap signal for modifying the first signal.

24. A circuit according to claim 23 wherein the scaling circuit comprises a switching circuit for switching between the different weights to select one weight for application at a first time and another weight for application at another time within a same signal to be filtered.

25. A circuit according to claim 23 wherein the scaling circuit comprises a switching circuit for switching between the different weights to select one weight for application at a first time and another weight for application at another time in dependence upon a content of the signal to be filtered.

26. A circuit according to claim 23 comprising a summer for summing an output of each of the plurality of taps.

27. A circuit comprising:

an input port for receiving a first signal;

a plurality of taps, each tap comprising an input port for receiving a tap input signal, a first input port for receiving a first weight, a second input port for receiving a second other weight, and a scaling circuit for scaling an applied weighting between the first weight and the second weight to scale the tap signal, the scaled tap signal for modifying the first signal.

28. A circuit comprising:

an input port for receiving a first signal;

a plurality of taps, each tap comprising an input port for receiving a tap input signal, a plurality of input ports each for receiving a weight, and a scaling circuit for scaling an applied weighting based on the received weights to scale the tap signal, the scaled tap signal for modifying the first signal.

29. A circuit according to claim 28 wherein the scaling circuit comprises a switching circuit for switching between the different weights to select one weight for application at a first time and another weight for application at another time within a same signal to be filtered.

30. A circuit according to claim 28 wherein the scaling circuit comprises a switching circuit for switching between the different weights to select one weight for application at a first time and another weight for application at another time in dependence upon a content of the signal to be filtered.

31. A circuit according to claim 28 comprising a summer for summing an output of each of the plurality of taps.

32. A circuit comprising:

an input port for receiving a first signal, the first signal received at a receiver via a communication interface and from a remote location;

a plurality of taps, each tap comprising an input port for receiving a tap input signal, a plurality of weight input ports each for receiving a weight, and a scaling circuit for scaling an applied weighting based on the received weights to scale the tap signal, the scaled tap signal for modifying the first signal.

33. A circuit according to claim 32 wherein the scaling circuit comprises a switching circuit for switching between the different weights to select one weight for application at a first time and another weight for application at another time within a same signal to be filtered.

34. A circuit according to claim 32 wherein the scaling circuit comprises a switching circuit for switching between the different weights to select one weight for application at a first time and another weight for application at another time in dependence upon a content of the signal to be filtered.

35. A circuit according to claim 32 comprising a summer for summing an output of each of the plurality of taps.

36. A method comprising

providing a receiver for receiving a signal transmitted across an optical fibre and for providing an electrical first signal;

using a filter, filtering the first signal with a non-linear filter to provide compensation to the first signal for distortion in the signal when transmitted resulting from driving a transmitter at a transmit end, wherein an output signal from the filter better approximates an ideal transmit signal for an intended modulation.

37. A method comprising:

manufacturing a circuit comprising: an input port for receiving a first signal; a plurality of taps, each tap comprising an input port for receiving a tap input signal, a plurality of input ports each for receiving a weight, and a scaling circuit for scaling an applied weighting based on the received weights to scale the tap signal, the scaled tap signal for modifying the first signal; testing the circuit and determining each of the plurality of weights based on testing thereof; and setting each of the plurality of weights based on a result of the testing thereof and fixing each of the plurality of weights.

38. A circuit comprising:

a non-linear FIR filter comprising a plurality of taps, each tap having multiple weights and a scaling circuit for scaling the multiple weights to affect a signal propagating within the tap for nonlinear filtering of a first signal.

39. A circuit according to claim 38 wherein the non-linear filter is implemented as an analogue component within an integrated circuit.