Low noise optical receiver

Abstract

A low noise optical receiver includes an amplifier in a feedback network, which allows the value of a feedback resistor to be increased. The magnitude of the gain of the feedback amplifier is greater than one. By increasing the value of the feedback resistance, the effective noise of the receiver is lower.

Description

BACKGROUND OF THE INVENTION

This application claims the benefit of U.S. Provisional Application No. 60/601,018, filed Aug. 12, 2004, entitled OPTICAL RECEIVERS AND AMPLIFIERS FOR LINEAR BROADBAND DISTRIBUTION SYSTEMS, the disclosure of which is herein incorporated by reference.

1. Field of the Invention

The present invention relates generally to circuits for optical receivers, and more particularly to a design for a low noise optical receiver.

2. Description of the Related Art

The delivery of video services over communication systems such as Hybrid-Fiber-Coax (HFC), Fiber-To-The-Curb (FTTC), and Fiber-To-The-Home (FTTH) often necessitates the use of high dynamic range technologies to support legacy analog NTSC signal formats. These video systems all use amplitude modulated (AM) optical carriers and require an optical transmitter to modulate the information onto the light. They also require an optical receiver to demodulate and amplify the signal for use by customer premise terminals, such as set top boxes or NTSC television sets.

A basic optical link used in an analog RF video delivery system is shown in FIG. 1. In this case, an FTTP analog RF system is shown, but the principles are equally applicable to HFC or FTTC systems. An optical transmitter takes in a multi-channel signal and amplitude modulates (AM) a light source in a linear fashion. The content can be standard NTSC analog TV channels or digitally modulated carriers such as used in cable modem systems. This system is primarily designed to carry video services, but quite often is used to carry advanced digital services such as high-speed data and telephony. The output of the optical transmitter provides an input to an Erbium Doped Fiber Amplifier (EDFA), which greatly increases optical power levels without adding significant noise or distortion. The EDFA high power output is then fanned out by means of an optical splitter to provide signal to a number of customers. Typically the fan-out or split ratio is 1:32 or 1:64. The amplitude-modulated optical signal is then demodulated into an electrical signal by a photo-detector, which functions as an envelope detector on the incoming light. A trans-impedance amplifier provides electrical gain such that the resulting signal is suitable for distribution to customer premise equipment, or to further coaxial distribution systems.

In these systems, the optical dynamic range at the receiver input is the difference between the maximum optical input level before the onset of distortion, less the minimum optical input level before noise degrades signal quality. In the case of analog RF video signals, either excessive distortion or excessive noise will degrade customers' viewing experience. Consequently, video system architects spend considerable time optimizing their systems around distortion and noise performance, and an optimum system design will carefully balance the distortion and noise performance against cost.

Because of the spatial diversity of customers and the variable nature of optical link budgets in typical deployments, optical path losses can widely vary. For instance, fiber runs will be longer in rural areas than in urban environments. Depending on the specific optical plant deployed, the number and locations of loss elements such as patch panels and splices will vary. To make wide-scale deployments over a large range of optical plants easier it is very desirable to have an optical receiver able to operate over a wide optical dynamic range. For instance, in some three-wavelength FTTP systems now in the early stages of deployment, the desired optical loss budget is between 10 to 28 dB. Unfortunately, optical receivers for the 1550 nm wavelength video portions of the FTTP system only support about 7 or 8 dB of dynamic range. The small optical dynamic range of video optical receivers can make FTTP deployments more difficult since more effort must be expended to meet the relatively narrow optical input window. A wider 1550 nm wavelength video receiver dynamic range will make FTTP deployments easier.

Thus, one of the key goals presented to the designer of analog RF optical receivers is to increase the usable optical dynamic range. As stated, this involves two elements. While the noise and distortion must be within acceptable limits over the entire specified optical input range, generally two corner conditions form the basis for the design. First, the receiver must not cause significant distortion when the input optical condition is large. Minimizing distortions in any amplifier can be accomplished by a number of means such as increasing the size of active transistor devices inside the amplifier. Unfortunately larger transistor active area leads to increased power consumption and cost. Another technique to minimize distortions is to apply multi-device amplifier topologies that can have inherently lower distortion. The familiar cascode topology is commonly used for this purpose and has two transistors. Second, the receiver must not contribute significant noise when the input condition is low. Minimizing noise likewise involves a careful selection of circuit topology and bias conditions. Minimizing noise is often done by maximizing the value of key resistors in the circuit such as the primary shunt feedback resistors used in broadband circuits.

It is important to note with regard to the design that distortion and noise are different concepts. That is, a design specifically optimized for good distortion performance will have degraded noise performance, compared with a design which targets low noise. Similarly, a design specifically optimized for low noise performance will have comparatively worse distortion than a design optimized for distortion. In most cases, the principle task of the design is to carefully balance the noise and distortion of the receiver while holding costs to a minimum.

It is also worth mentioning that poor distortion and noise performance affect systems differently depending on the type of content transmitted. For example, a system carrying QAM modulated digital information will be quite sensitive to distortion effects such as clipping, but less sensitive to noise effects when compared with an analog NTSC signal. Noise and distortion are not the same, but rather must be carefully balanced in the design.

FIG. 2 illustrates the tradeoff between noise and distortion. FIG. 2 shows a generalized trans-impedance amplifier design which includes a high gain Amplifier A having voltage gain from Vin to Vout of A. It's assumed that the Amplifier A has near infinite input impedance. A shunt feedback resistor, Rfb, regulates the output voltage of the trans-impedance amplifier by feeding back a portion of the output signal to the input. It can be shown that the trans-impedance gain, Ztia, of the circuit is:
Ztia=Vout/Iin=Rfb*A/(1−A)=˜−Rfb (large A)
The quantity, Iin, is the input current provided by a photo-detector when it is illuminated. The value of Iin is determined by the input optical power and the responsivity of the photo-diode. The range of Iin the circuit experiences is then a direct result of the optical dynamic range. The output voltage, Vout, is significant in that the Amplifier A must provide reasonable linearity up to the Vout level indicated by:
Vout=Iin*Rfb*A/(1−A)
Vout=˜−Iin*Rfb (large A)
Vout(max)=˜−Iin(max)*Rfb (large A)
For a given range of input optical powers, the maximum Vout is then directly set by the value of Rfb. The amount of distortion generated in the circuit will depend on the non-linear characteristics of Ztia with respect to Iin. The non-linear relationship between Vout and Iin can be described as a power series:
Vout(Iin)=m1*Iin+m2*(Iin)ˆ2+m3*(Iin)ˆ3+higher order terms
Here m1 and m2 are the standard power series coefficients for the 1^st, 2^nd, and 3^rd order responses, respectively, of the complete trans-impedance amplifier in FIG. 2.

The equivalent input noise of a trans-impedance amplifier is the sum of all noise sources within the trans-impedance amplifier lumped into a single equivalent noise current source, Ieqt, placed at the input in parallel with the photo-detector. Although photo-detector impedances can influence Ieqt, no photo-detector noise sources (such as shot noise) are included in Ieqt. Assuming that Amplifier A is noise-less, the only noise source contributing to the equivalent input noise is that of Rfb. For amplifiers fabricated from field-effect devices (FET), this is a useful approximation due to the high input impedance and very low noise performance FET devices offer. It is not a good approximation for amplifiers fabricated from bipolar junction devices (BJT) due to the comparatively high base current and correspondingly high shot noise. Assuming photo-detector impedance is infinite, Ieqt of the circuit in FIG. 2 is the thermal noise associated with Rfb:
(Ieqt)ˆ2=4kTB/Rfb
For example, a feedback resistor of 1000 ohms will generate 4 pA/rtHz of equivalent input noise. Thus, we would like to increase Rfb to achieve the lowest noise performance. However, as previously stated, a larger Rfb implies that a larger output voltage Vout must be supported with good distortion characteristics by our Amplifier A. When Vout increases, so does the distortion generated in Amplifier A. This leads to a direct trade-off between noise and distortion performance in the circuit of FIG. 2.

One of the primary methods for improving this tradeoff involves a push-pull topology in which two separate amplifiers are operated 180 degrees out of phase with respect to each other (FIG. 3). Outputs from these separate amplifiers are added together in a push-pull signal combiner such as a transformer or balun shown in FIG. 3. This approach is described in R. B Childs, T. A. Tatlock, and V. A. O'Byrne; “AM Video Distribution System with a 64-Way Passive Optical Splitting”, IEEE Photonics Technology Letters, Vol. 4, No. 1, January 1992. In this design, a photo-detector's two outputs are used to drive two separate amplifiers, whose outputs are combined with a transformer.

Much the same technique is also covered in detail in Little, Jr, et al, U.S. Pat. No. 5,239,402, as well as follows on works by Skrobko, U.S. Pat. No. 5,347,389 and U.S. Pat. No. 6,674,967. The basic elements of these approaches all include a photo-detector, two separate amplifiers, and a means for coupling the amplifier outputs in a push pull fashion.

The advantages of this push-pull approach are twofold. First, because thermal noise contributions of each feedback resistor Rfb and Amplifier are independent from one another, noise power from these sources will be additive at the output. In addition, the push-pull operation of the circuit insures that the desired signal's output voltage will be additive through the output transformer. It can be shown that the net effect of this is to reduce the Ieqt of the push-pull implementation to be sqrt(2) of that from each half. For example, a pair of 1000 ohm feedback resistors will generate 2.82 pA/rtHz of equivalent input noise in a push-pull design. The second advantage of the push-pull approach is that 2^nd order distortion terms can be made to cancel provided the circuit in FIG. 3 maintains complete balance. Any imbalance of current flow into the separate amplifiers, or imbalance in the power series characteristics of the amplifiers, or imbalance in the characteristics of the push-pull combiner, will lead to a direct loss of 2^nd order cancellation. Should this imbalance become too large, the noise reduction properties will also degrade.

While the circuit in FIG. 3 improves noise and 2^nd order distortion, it requires an output balun or transformer to combine outputs. Output transformers are typically wound using bifilar wire around appropriately sized ferrite cores to achieve the desired bandwidth. Winding of baluns and transformers is labor intensive and therefore expensive. The circuit in FIG. 4 overcomes much of this by combining signals in an active differential amplifier, as contained in Witkowicz, U.S. Pat. No. 4,139,767. In this implementation, the noise performance of the full receiver will be sqrt(2) lower than that caused by each separate amplifier, and the 2^nd order distortion produced in the 2 input amplifiers will cancel in the output differential amplifier. However, 2^nd order distortions emanating in the differential amplifier output stage will not cancel without an output transformer or balun device, as in U.S. Pat. No. 5,239,402. In this regard, the design in U.S. Pat. No. 4,139,767 is both distinctly different from, and is inferior to, U.S. Pat. No. 5,239,402. However, U.S. Pat. No. 4,139,767 is easier to implement in an integrated circuit (IC) since it does not require a ferrite wound balun or transformer, and therefore can be made with significant cost advantages over U.S. Pat. No. 5,239,402.

In summary, balancing noise, distortion, and cost are the primary challenges in the design of optical receivers. A push-pull technique has been useful in improving noise and distortion by adding a completely separate 2^nd amplifier.

SUMMARY OF THE INVENTION

A low noise optical receiver includes an amplifier in a feedback network, which allows the value of a feedback resistor to be increased. The gain of the feedback amplifier is greater than one. By increasing the value of the feedback resistance, the effective noise of the receiver is lower.

According to one embodiment of the present invention, an optical receiver comprises a photodetector having a cathode connected to a first node, a photodetector biasing network connected to the photodetector and the first node, a main amplifier having an input connected to the first node, a feedback amplifier connected to an output of the main amplifier, and a feedback resistor connected between the feedback amplifier and the first node.

The optical receiver may further comprise a biasing network having high impedance. The main amplifier may also have high input impedance. Additionally, the magnitude of the feedback amplifier's gain response is greater than one (1).

According to the method of the present invention, a method for reducing the noise of an optical receiver comprises detecting an optical signal with a photodetector, the photodetector biased with a high impedance network, applying an output current from the photodetector to a high impedance main amplifier, and feeding back an output signal from the main amplifier through a feedback network, the feedback network comprising a feedback amplifier and a feedback resistor, wherein the magnitude of the feedback amplifier's gain response is greater than one (1).

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will be readily understood by the following detailed description in conjunction with the accompanying drawings, wherein like reference numerals designate like structural elements, and in which:

FIG. 1 is a block diagram of a standard optical link system;

FIG. 2 is a schematic of a prior art trans-impedance amplifier design;

FIG. 3 is a schematic of a prior art a push-pull topology;

FIG. 4 is a schematic diagram of a further prior art design;

FIG. 5 is a block diagram of one embodiment of the present invention; and

FIG. 6 is a schematic of one circuit implementation of the embodiment shown in FIG. 5.

DETAILED DESCRIPTION OF THE INVENTION

The following description is provided to enable any person skilled in the art to make and use the invention and sets forth the best modes contemplated by the inventor for carrying out the invention. Various modifications, however, will remain readily apparent to those skilled in the art. Any and all such modifications, equivalents and alternatives are intended to fall within the spirit and scope of the present invention.

Receiver noise may be reduced according to one embodiment of the present invention shown in FIG. 5. A photo-detector 50 is biased using a high impedance network 52 connected between a supply voltage, Vcc, and the cathode of the photo-detector 50. This network 52 may be comprised of inductors, resistors, or active elements such that the cathode sees a very high impedance path for current flow. The photo-detector 50 demodulates the incoming AM light and drives a current, Iin, past a high input impedance main amplifier 54, having voltage gain of A, and into a feedback resistor Rfb. The output of the main amplifier 54 is then coupled to the input of a feedback amplifier 56 having voltage gain B, where the magnitude of B is greater than 1. The trans-impedance gain of the circuit in FIG. 5 is: $\begin{array}{c}\text{Trans-impedance}=\mathrm{Voutput}/\mathrm{Iin}\\ =\mathrm{Rfb}*A/\left(1-B*A\right)\\ ~-\mathrm{Rfb}/B\left(\mathrm{large}A\right)\end{array}$
Note that for large values of main amplifier gain A, the expression approximates to a quantity independent of A. Also note that if the value of B, the feedback gain, is set to 1, the expression for gain reduces to that of the prior art shown in FIG. 2, or −Rfb. It can be seen then, that the application of feedback gain into the receiver reduces the net trans-impedance of the receiver by an amount equal to the feedback gain B. This fact can be used to increase the value of Rfb such that its thermal noise into the receiver is reduced. Because of the gain in the feedback path, a higher value of feedback resistance can be selected without a change in the amount of current traveling though Rfb. Because the main amplifier 54 ideally presents a very high impedance at the input, the feedback current is essentially the input current Iin. Consequently, the feedback resistor can be made larger if there is gain in the feedback amplifier, and a larger feedback resistor will reduce the noise referred to the input. The present invention specifically makes use of feedback amplifier 56 voltage gain values with a magnitude greater than 1, since values for B which are equal to or less than 1 are of no value in reducing noise.

Since the circuit in FIG. 5 applies feedback, the phase responses of the gains on amplifiers A and B are important. The main amplifier 54 and feedback amplifier 56 may be either inverting or non-inverting provided they are not both either inverting or both non-inverting. Hence, it is possible to have an inverting main amplifier and an inverting feedback amplifier, or to have an inverting main amplifier and a non-inverting feedback amplifier.

The present invention is also beneficial for receivers using both polarities of the photo-detector, as shown in the embodiment of FIG. 6. The photo-detector and associated biasing circuitry operates as previously described. The main amplifier comprises Q1 and Q2 which provide voltage gain from their gates to their respective drains. The drains of Q1 and Q2 are the outputs of the receiver. The main amplifier provides inverting gain. The feedback amplifier is based upon Q3 and Q4, which form a differential pair and provide voltage gain from their gates to their drains. The input to the feedback amplifier is coupled through C5a and C5b into the gates of Q3 and Q4. The outputs of the feedback amplifier are cross-coupled drains of Q3 and Q4 such that the gain of the feedback amplifier is non-inverting. The amount of gain the feedback amplifier achieves depends on the size and biasing of Q3 and Q4 and the values of Re1, Re2, and Rgain. The values of R2a and R2b are generally quite high relative to Rgain and do not significantly impact the gain.

Biasing of Q1 and Q2 is set by the interaction of Rb1, and Rb2 with device parameters such as pinch-off voltage, Vp, which is the gate voltage needed to completely turn off the device, and saturated drain current, Idss, which is the current flow when Vgs=0. The DC current flow into the gates of Q1 and Q2 is extremely low, so the resistors R2a and R2b effectively provide a DC ground to the gates Q1 and Q2. Alternatively, resistors R2a and R2b may be tied from their respective gates to a common control voltage which is useful in adjusting the bias current of Q1 and Q2. The bias current occurring in Q1 and Q2 is the point when the equation Vgs1=−Ids1*Rb1 and Vgs2=−Ids*Rb2. For Q1 or Q2 currents on the order of 40 mA to 80 mA and a typical FET process, a correspondingly low value of Rb1 and Rb2 results, typically about 5 ohms. The low amount of voltage lost across Rb1 and Rb2 also serves to preserve voltage headroom in the circuit, which maximizes efficiency and linearity performance. Since power is consumed in the biasing resistors and because FET device linearity can be generally improved by increasing the biasing condition from drain to source, a small amount of voltage drop across Rb1 and Rb2 is desirable. Best efficiency and linearity results by minimizing the voltage at the sources of Q1 and Q2 respectively. With no major changes in performance, it is possible to tie the sources of Q1 and Q2 together and combine Rb1 with Rb2 into a resistor ½ their respective values, or about 2.5 ohms. The resulting circuit still operates as two amplifiers because the small value of resistance needed to properly set the drain currents together for best efficiency and linearity is too small to provide common-mode rejection, and the balanced input current coming from the photo-detector makes it unnecessary to have common-mode rejection performance in the main amplifier. Such common-mode rejection is best achieved with a very high impedance current source in place of Rb1 and Rb2, but at significant expense of the aforementioned voltage headroom.

In a preferred embodiment, the amount of feedback gain is just under 2 and the value of the feedback resistor is over 1800 ohms. Although the gain in the feedback path reduces the effective gain, the differential behavior of the invention brings the gain back to approximately 1800 ohms. The circuit in FIG. 6 is fabricated in 0.25 micron GaAs pHEMT in a single RFIC. Extensive computer modeling and simulation shows the equivalent input noise to be less than 2.2 pA/rtHz.

Distortion products generated in the feedback amplifier can impact the overall receiver linearity, and it is there that the voltage levels are highest. The present invention does have the benefit that the current required from the feedback amplifier can be much smaller than the current required from the main amplifier. The feedback amplifier has only to drive the R2a, Rgain, and Rfb. This impedance may be designed to be suitably large so that non-linearities in Q3 and Q4 can be minimized without excessive bias current. In other words, the feedback amplifier load conditions are very light, which provides very helpful design flexibility in making a feedback amplifier with low power consumption. In a preferred embodiment, the bias currents on Q3 and Q4 are set to 25 mA each, which is less than ½ the currents in Q1 and Q2 of the main amplifier.

The dual outputs of Q1 and Q2 in FIG. 6 are useful in a number of applications. They may serve as the two outputs for different RF distribution networks, which might be useful there be multiple devices connected to the RF distribution network. For example, subscribers often have more than one terminal device such as television set or set top box in their house, so having two outputs may save the cost of an RF power splitter in the RF distribution network. The two outputs may also be coupled into a differential amplifier as performed in Witkowicz, U.S. Pat. No. 4,139,767 shown in FIG. 4.

Those skilled in the art will appreciate that various adaptations and modifications of the just-described preferred embodiments can be configured without departing from the scope and spirit of the invention. Therefore, it is to be understood that, within the scope of the appended claims, the invention may be practiced other than as specifically described herein.

Claims

1. An optical receiver comprising:

a photodetector having a cathode connected to a first node;

a photodetector biasing network connected to the photodetector and the first node;

a main amplifier having an input connected to the first node;

a feedback amplifier connected to an output of the main amplifier; and

a feedback resistor connected between the feedback amplifier and the first node.

2. The optical receiver of claim 1, wherein the biasing network has high impedance.

3. The optical receiver of claim 2, wherein the main amplifier has high input impedance.

4. The optical receiver of claim 3, wherein the feedback amplifier has a feedback voltage gain magnitude greater than one.

5. The optical receiver of claim 4, wherein the feedback voltage gain is approximately two and the value of the feedback resistor is greater than approximately 1800 ohms.

6. A method for reducing the noise of an optical receiver, the method comprising:

detecting an optical signal with a photodetector, the photodetector biased with a high impedance network;

applying an output current from the photodetector to a high impedance main amplifier; and

feeding back an output signal from the main amplifier through a feedback network, the feedback network comprising a feedback amplifier and a feedback resistor;

wherein the feedback amplifier has a voltage gain magnitude greater than one.