Use of predetection mixing of intensity modulatedoptical beams
Marion Scott and Phil Garcia
Predetection mixing of the intensity modulation on optical beams is shown to have performance advantages in
certain laser radar applications. The method allows detection of signals with high frequency intensitymodulation using a low frequency sensor. The method can lead to a signal-to-noise ratio improvementcompared to a system using a high frequency sensor. The technique is demonstrated by using a sensor with a5-kHz cutoff frequency to receive a laser beam intensity modulated at 4 MHz.
I. Introduction
A standard radar technique for measuring the rangeto a target is to transmit a sinusoidally modulatedelectromagnetic signal and measure the phase shift ofthe return.' This technique has been used in laserradar by intensity modulating the laser output.2 Thebest range resolution is obtained from these radarswhen the modulation frequency and the signal-to-noise ratio are both as high as possible. But as modu-lation frequency is increased, the electronics in thereceiver circuit must pass higher frequencies and willadd more thermal noise. The decrease in signal-to-noise ratio at the higher modulation frequency ad-versely affects the range resolution and limits the oper-ating range of the radar.
An electronic mixing technique has been reported 3' 4
which translates the high modulation frequency of thereceived signal to a lower frequency for subsequentamplification. The receiver in this system consists ofan avalanche photodiode with a modulated bias volt-age. The received signal is electronically mixed withthe photodiode bias modulation frequency to producea difference frequency output from the photodiode.The low difference frequency signal is then amplifiedby low frequency, low noise electronics.
The signal-to-noise ratio of this receiver has beencalculated for the circuit noise-limited case and is pre-dicted to be superior to a receiver without electronicmixing. 3 However, the electronic mixing receiver re-
The authors are with Sandia National Laboratories, P.O. Box5800, Albuquerque, New Mexico 87185.
quires a high frequency local oscillator circuit to injectthe local oscillator signal into the detector. There willbe a noise contribution from this high frequency circuitwhich may prevent the predicted signal-to-noise im-provement from being realized in practice.
There is an alternate mixing technique in which themixing takes place in the optical beam, prior to detec-tion. In this system, the transmitted laser beam isintensity modulated with a sinusoid of frequency ft.The return beam is intensity modulated at a frequencyfr. The intensity of the optical beam will now containa difference frequency component ft - fr. This alter-nate mixing method has the advantage of allowing alow frequency detector and processing electronics tobe used to detect an optical beam with high frequencyintensity modulation. As will be discussed later, thislow frequency sensor can achieve a higher signal-to-noise ratio than the corresponding high frequency sys-tem. This technique has the disadvantage of requir-ing a high speed, high efficiency modulator. But highspeed modulators have been reported and their avail-ability may make this technique more attractive.
The next section discusses the advantages of thepredetection mixing technique for a laser radar appli-cation. The results of a proof-of-principle experimentare then given. The technique is not limited to laserradar applications and can be used to improve signal-to-noise ratios on other circuit noise-limited systemsdesigned to detect intensity modulated signals.
II. Background and Theory
The phase shift technique for measuring range to atarget consists of transmitting a sinusoidally modulat-ed signal and measuring the phase shift of the return.For laser radar applications, the transmitted signal isan intensity modulated laser beam. The range mea-surement is ambiguous, with the ambiguity distanceequal to one-half of the modulation wavelength. For a
sufficiently high signal-to-noise ratio, the range reso-lution of this type of system is given by5
AR = c/(4irf mI7N), (1)
where f is the modulation frequency, S/N is the powersignal-to-noise ratio, and c is the speed of light. Thetwo methods for improving range resolution that areevident from Eq. (1) are increasing signal-to-noise ra-tio or increasing modulation frequency.
To see the effect of increasing modulation frequencyon signal-to-noise ratio, consider the laser radar receiv-er circuit shown in Fig. 1. (The transimpedance am-plifier is shown because it is frequently used for laserradar applications and has advantages when used in anintegrated array of detectors and amplifiers.6 Thedesign of the preamplifier has been discussed previ-ously.
7 Only one aspect of this topic will be discussedhere to illustrate a concept.) When the operationalamplifier has sufficiently high gain and the detectorhas sufficiently high impedance over the frequencyband of interest, the output voltage (V) is related to theinput signal current (Is) by
V = -zf4s (2)
where Zf is the parallel combination of the feedbackresistor Rf and the capacitance Cf of the feedback path.The cutoff frequency of the amplifier will be
f = 1/(27rRC). (3)
Once the stray capacitance Cf has been minimized, ahigher bandwidth is obtained by reducing Rf. Whenreferenced to the amplifier input, the mean-squarenoise current due to R is
Z = (4kTB)/Rf, (4)
where h is Boltzmann's constant, T is the temperatureof the feedback resistor, and B is the noise bandwidth.The noise bandwidth may be set by filtering the signalat some subsequent stage of the processing. But theneed to preserve a high cutoff frequency in the pream-plifier leads to a decrease in Rf [as shown in Eq. (3)] anda corresponding increase in the noise per unit band-width [as shown in Eq. (4)]. An increase in the modu-lation frequency of the laser transmitter to achievebetter range resolution therefore results in a decreasein the signal-to-noise ratio.
When detecting low amplitude signals, the domi-nant noise source in a laser radar receiver is often thethermal noise term given in Eq. (4). For this situation,and assuming that the cutoff frequency in Eq. (3) isequal to the modulation frequency, it can be shownthat Eq. (1) predicts that the range resolution willimprove in proportion to the square root of the modu-lation frequency.
The proposed alternate receiver architecture con-sists of a high speed modulator followed by a lowfrequency detector and preamplifier. The transmit-ted signal has the form
Pt = P0 [1 + cos(2rmt)], (5)
where P0 is the average power of the laser, and a 100%
VBIAS
Fig. 1. Typical laser radar receiver circuit, the transimpedanceamplifier. Signal current in the photodiode is converted to an
output voltage V.
depth of modulation has been assumed. The powerincident on the detector is modulated a second time.If this modulation is sinusoidal and also has a 100%depth of modulation, the power incident on the detec-tor has the form
Pi = (P/2)[1 + cos(27rfmt)][1 + cos(27rfrt)], (6)
where Pr is the average power collected at the receiveraperture. The difference frequency component of Piis
Pd = (Pr/4) cos[2r(fm - fr) t]. (7)
Equation (7) shows that the return laser beam nowhas an intensity modulation component at the differ-ence frequency. The magnitude of this component isone-fourth the magnitude of the signal componentthat would have been received by a high frequencydetector with no mixing. However, the reduction insignal power can be more than offset by the noisereduction in the detector circuit. If the mixing processhas shifted the signal frequency from, for example, 10MHz to 1 kHz, the mean-square noise power from thefeedback resistor can also be decreased by 104, as pre-dicted by Eqs. (3) and (4). At the lower frequency thedetector may become dark current noise-limited, pre-venting realization of the full 104 noise reduction.However, a substantial noise reduction can be ob-tained.
This technique makes the signal-to-noise ratio inde-pendent of the modulation frequency. The range res-olution now improves in proportion to the modulationfrequency, in contrast to the square root dependencefor the high frequency detector system.
Ill. Experimental Results
An optimum experimental implementation of thereceiver described above requires a very high speedmodulator with high throughput. Although such de-vices have been reported, 8' 9 the only modulator avail-able to the authors was an acoustooptic cell with a lowdiffraction efficiency. Because of the low diffractionefficiency, it was not possible to implement an opti-mum system for comparison with a standard high fre-quency receiver. The experiment described here wasdesigned to characterize the use of a low frequency
(b)Fig. 2. Spectrum analyzer display of the two high frequency com-ponents of the doubly intensity modulated optical beam. The de-tector circuit has its 3-dB cutoff frequency at 5 kHz. The spectrumanalyzer horizontal sensitivity is 0.2 kHz/div, vertical sensitivity is10 dB/div, and reference level is -30 dBm. (a) Laser drive currentmodulated at 5 kHz and acoustooptic cell modulated at 4 kHz. (b)Laser drive current modulated at 50 kHz and acoustooptic cell
modulated at 49 kHz.
receiver to detect optical beams with high frequencyintensity modulation.
The experimental apparatus consisted of a GaAslaser source, a silicon PIN photodiode with a transim-pedance amplifier, and an acoustooptic modulator.The output of the laser is first modulated by directlyvarying its drive current. The acoustooptic cell pro-vides another high-speed modulation of the light inci-dent on the detector. For this experiment, these twomodulation frequencies were separated by 1 kHz. The3-dB cutoff frequency of the detector-amplifier com-bination was -5 kHz.
Initially, the laser current was modulated at a fre-quency of 5 kHz and the acoustooptic cell was modu-lated at 4 kHz. The laser modulation frequency wasthen increased to 50 kHz and the acoustooptic cellmodulation frequency to 49 kHz. The high frequencycomponents of the sensor output for these two cases asrecorded on a spectrum analyzer are shown in Fig. 2.The output for the 49- and 50-kHz frequencies is -15dBm below the output for the 4- and 5-kHz cases.This is because the 50-kHz signal is well beyond thecutoff frequency of the sensor. The sensor output atthe difference frequency is shown in Fig. 3. The dif-ference frequency output is the same amplitude, -80dBm, for both cases. This indicates that the differ-ence frequency output is independent of laser modula-
Fig. 3. Spectrum analyzer display of the difference frequency com-ponents for the conditions described in Fig. 2. Spectrum analyzersensitivities are the same as in Fig. 2; (a) and (b) correspond to Figs.
2(a) and (b), respectively.
Fig. 4. Spectrum analyzer display of the difference frequency out-put when the laser current is modulated at 4.000 MHz and theacoustooptic cell at 3.999 MHz. Horizontal sensitivity is 0.2 kHz/div, vertical sensitivity is 10 dB/div, and the reference level is -20
dBm.
tion frequency up to 50 kHz, even though the highfrequency component is attenuated.
The laser drive frequency was then increased to 4MHz and the acoustooptic cell modulation frequencyto 3.999 MHz. This frequency was near the maximummodulation frequency of the cell. The output of thesensor at the high frequencies was below the noise floorof the spectrum analyzer and therefore could not bedetected. The output at the 1-kHz difference fre-quency is shown in Fig. 4. Note that the referencelevel for the spectrum analyzer output is at -20 dBm inFig. 4, so the difference frequency signal amplitude isstill at -80 dBm, as before. (The spectrum analyzerused for the measurement shown in Fig. 4 was a differ-ent instrument from the one used to generate the data
in Figs. 2 and 3. This is the reason for the higher noisefloor in Fig. 4. A different, higher frequency, spec-trum analyzer was used to verify that the higher fre-quency components of the signal were not present inthe sensor output.) This experiment clearly indicatesthat the low frequency detector can be used to detecthigh modulation frequency signals. The experimentwas limited to megahertz frequencies only by instru-mentation. The same principles should apply at evenhigher frequencies.
IV. Summary and Conclusions
A method of receiving electromagnetic signals withhigh frequency intensity modulation using a low fre-quency sensor has been described. A signal-to-noiseratio improvement can be obtained when applying thistechnique to circuit noise-limited systems. The use ofthe technique can be advantageous in laser radar sys-tems where the phase shift method of range measure-ment is employed. An experimental implementationwas demonstrated using a 5-kHz cutoff frequency sen-sor to receive a laser beam which was intensity modu-lated at 4 MHz. Although the example of laser radarwas used to illustrate the advantages, the method canbe applied to other communication links as well.
This work was performed at Sandia National Lab-oratories and supported by the U.S. Department ofEnergy under contract DE-AC04-76DP00789.
References
1. R. Wilson, "Phase Comparison Position-Determining Systems,"in Range Instrumentation, E. H. Ehling, Ed. (Prentice-Hall,Englewood Cliffs, NJ, 1967).
2. A. Sona, "Lasers in Metrology: Distance Measurements," inLaser Handbook, Vol. 2, F. T. Arecchi and E. 0. Schultz-Dubois,Eds. (Elsevier, New York, 1972).
3. W. K. Kulczyk and Q. V. Davis, "The Avalanche Photodiode as anElectronic Mixer in an Optical Receiver," IEEE Trans. ElectronDevices ED-19, 1181 (1972).
4. D. K. W. Lam and R. I. MacDonald, "GaAs Optoelectronic MixerOperation at 4.5 GHz," IEEE Trans. Electron Devices ED-31,1766 (1984).
5. J. Freedman, "Radar," in System Engineering Handbook, R. E.Machol, Ed. (McGraw-Hill, New York, 1965).
6. P. P. Webb and R. J. McIntyre, "Multi-Element ReachthroughAvalanche Photodiodes," IEEE Trans. Electron Devices ED-31,1206 (1984).
7. R. G. Smith and S. D. Personick, "Receiver Design for OpticalFiber Communication Systems," in Semiconductor Devices forOptical Communication, H. Kressel, Ed. (Springer-Verlag, NewYork, 1982).
8. A. Morimoto, T. Kobayashi, and T. Sueta, "Active Mode Lockingof Lasers Using a Fast Electrooptic Deflector," IEEE J. QuantumElectron. QE-24, 94 (1988).
9. R. G. Walker, "High-Speed Electrooptic Modulation in GaAs/GaAlAs Waveguide Devices," IEEE/OSA J. Lightwave Technol.LT-5, 1444 (1987).
0
Meetings continued from page 5050
1989February
27-1 Mar. Fiber Optic Communications course, Tempe Ctr. forProfessional Development, Coll. of Eng. & AppliedSciences, AZ State U., Tempe, AZ 85287
27-1 Mar. Microphysics of Surfaces, Beams & AdsorbatesTop. Mtg., Salt Lake City OSA Mtgs. Dept., 1816Jefferson Pl., NW, Wash., DC 20036
March
1-3 Photonic Switching II Top. Mtg., Salt Lake CityOSA Mtgs. Dept., 1816 Jefferson Pl., NW, Wash., DC20036
5-10 Imaging & Photographic Science course, RochesterRIT/T&E Sem. Ctr., One Lomb Memorial Dr., Roch-ester, NY 14623
6-8 Quantum Wells for Optics & Optoelectronics Top.Mtg., Salt Lake City OSA Mtgs. Sept., 1816 Jeffer-son Pl., NW, Wash., DC 20036
