Lab Exercise PN: Phase Noise Measurements Phase noise is a critical specification for oscillators used in applications such as Doppler radar and synchronous communications systems. It is tricky to measure because it involves measuring a very, very small signal (the noise) in the presence of a relatively large signal (the fundamental). Phase noise is specified by the noise power in a one Hz sideband relative to the power at the fundamental. It is measured at various offsets (f m ) from the fundamental. In the following, you will measure the phase noise of a high quality synthesizer, the HP 8648c, and a Syntonic D2000-1000 oscillator (see data sheet for specifications). You will make the measurements in two different ways, using a spectrum analyzer and using the delay line discriminator method. The measurements will be made at 1 GHz. Prelab There is no required prelab for this lab. 1 Spectrum Analyzer Connect the HP8648C synthesizer to the spectrum analyzer. Set the frequency to 1.0 GHz and the amplitude to 10 dbm. Set the spectrum analyzer s resolution bandwidth to 10 khz, center the display on 1 GHz. Measure the power in dbm at 1GHz + 100kHz. To get the per Hz power, subtract 10log(resolution bandwidth). The 10 khz resolution bandwidth includes noise power over a fairly broad range of the side band. Use the spectrum analyzer s averaging feature. Reduce the resolution bandwidth to 1kHz and remeasure. If the per Hz noise level changes, then you are at a frequency that is close enough to the fundamental that the side band noise spectrum is changing significantly within the resolution bandwidth. Try reducing the resolution bandwidth again. To obtain a more accurate measurement, it may help to reduce the span to 1 khz and set the center frequency to the measurement frequency. It may occur that the noise level you are trying to measure is below the noise level of the spectrum analyzer (It s just a receiver after all.). A test would be to increase the spectrum analyzer s attenuation by 5 db. Your noise measurement should not change. If it does change, you are at the noise floor of the instrument and all you can do is put a low noise pre-amplifier before the spectrum analyzer. It will also have its own noise and may generate intermodulation distortion. Subtract the measured power of the noise from the measured power of the fundamental and add 2.5 db to get L(f m ) (where f m = 100 khz in this case). The 2.5 db is to compensate for the fact that the network analyzer uses a peak detector and does not directly measure power. Measure L(f m ) for the frequencies shown in the table below. As you move down in frequency, the noise level will increase and so making a measurement above the noise floor will be easier. However, the closer the measurements get to the fundamental, the more important it will be to have a small resolution bandwidth. Otherwise as described above- the resolution bandwidth filter will allow some of the fundamental signal to add to the noise power you are trying to measure. You can see that if you get to the vicinity of Lab Exercise PN: Phase Noise Measurement - 1 -
f m = 1 khz, it will be difficult to get a good measurement even at the minimum resolution bandwidth setting. f m L(f m )synthesizer L(f m )VCO 100 Hz 200 Hz 500 Hz 700 Hz 1 khz 2 khz 7 khz 5 khz 10 khz 20 khz 50 khz 70 khz 100 khz 500 khz 1 MHz Table 5.1 Comparison of measured noise chatacteristics. Now measure the Syntonic oscillator. Add an attenuator to adjust the power to about 10 dbm. The output frequency will be in the vicinity of 1.0 GHz. It doesn t really matter that much exactly what the frequency is as far as noise characteristics are concerned. However, the frequency must not drift since this would make it difficult to measure using the spectrum analyzer. If you were trying to measure the noise at 10 khz (f m ) from a 1.0 GHz fundamental, the fundamental would have to drift only 0.001 % to drift right into the frequency where you are trying to make your measurement. Use your measured results to fill in the relevant parts of Table 5.1 as accurately as you can. Measurements may not be possible at low f m. Record the output power and frequency of Syntonic oscillator..2 Delay Line Discriminator Measurement As you have no doubt found from the previous measurements, it is quite difficult to measure the phase noise for small values of f m due to the presence of the relatively large fundamental frequency component at f o. The delay line discriminator, discussed in lecture, is one way around this problem. You will use a simplified delay line set-up, as shown in Figure 5.2, to make phase noise measurements. Lab Exercise PN: Phase Noise Measurement - 2 -
Figure 5.2 Setup for calibrating the delay line discriminator Component Mfg part number Comment Mixer Mini-Circuits ZFM 2000 Low pass filter SLP250 Dynamic signal analyzer HP 3561A Power divider Mini-Circuits ZSFC 2-4 Directional coupler KDI 20 db Coupler Low frequency amplifier MIL special 40 db, non inverting op-amp OP-27, 50 ohm input Delay line BNC connectorized coax 57, 72.5 ns, -10 db @ 1 GHz Phase Shifter Time delay adjustable 0.25 ns Table 5.2 Components for delay line discriminator measurement 5.2.1 Simplified Delay Line Discriminator Set up the measurement system as shown in Figure 5.2. Turn the audio source off for now and set the synthesizer to 1.0 GHz with an output power of about 10 dbm. Adjust the phase shifter such that the DC level as seen on the dynamic signal analyzer is a minimum. From lecture, you know that after this adjustment, the output voltage is related to the calibration spectrum by, 2 v d(rms) 2 = K 2 2 δf d 2 f rms m 2 f = K 2 2 d 2 f 2 m L( f m ) (5.1) m where K d is the discriminator constant in volts per Hz. You will determine K d via a calibration step before making any measurements. But first, calculate an estimate of what K d should be. An approximate expression for K d is K d A o 2 τ d G d G c G audio (5.2) where A o is the signal amplitude in volts at the input of the power divider; t d is the delay line time delay in seconds; G d, G c, are the transducer gains of the delay line and the mixer respectively; G audio is the voltage gain of the low frequency amplifier. Lab Exercise PN: Phase Noise Measurement - 3 -
.2.2 Calibration: Turn on the audio source and set the frequency to 5kHz. On the spectrum analyzer you should see small sidebands 5kHz away from the fundamental on both sides of the fundamental. Adjust the amplitude of the audio source such that the sidebands observed on the spectrum analyzer are about 15 db down from the fundamental. Record the relative level (db). Go to the dynamic signal analyzer and observe a spur at 5KHz. You should be on the dbv scale. As discussed in class, the discriminator constant is found from, 20 log(k d ) = K d [db] = V d [dbv] 3 db 20 log f m Λ(f m ) [db] (5.3) where K d has units of volts/hz. V d is the output of the discriminator measured by the dynamic signal analyzer. The dynamic signal analyzer measures dbv which is the rms output voltage normalized to 1 rms volt and written in db ( 20 log[v d(rms ] ). Λ(f m ) is the relative height of the sideband to the fundamental as observed by the spectrum analyzer. Change the audio frequency and repeat the procedure. Fill out table 5.2 for K d. K d will not change very much versus f m. f m V d [dbv ] 20 log f m Λ(f m ) [db] 1 khz 2 khz 5 khz 10 khz 20 khz 50 khz 70 khz Table 5.2 Calibration data when P o = dbm K d [db] 2.3 Measurement of Synthesizer Phase Noise. Turn off the audio calibration source. The spectrum analyzer should show no distinct spurs, only relatively smooth noise sidebands. The same should be true for the dynamic signal analyzer. Set the signal analyzer span to 100 khz, the display to dbv and the averaging to 20. The display is reading 20log(V rms ) or, equivalently, 10 log(v d(rms)2 ) where V d(rms)2 is the mean square noise voltage of noise with spectra in a range Δf centered at f m. Δf is the bin size set by the FFT in the signal analyzer. It s like the resolution bandwidth in the spectrum analyzer. It can be read off the middle of the screen just below the spectral display. To normalize V d(rms) 2 to a 1 Hz bandwidth, divide by Δf. In db this means 20log(V d(rms) ) 10 log Δf. Fill out table 5.3 by using the marker function on the signal analyzer to V d in dbv normalized to 1 Hz at the frequencies indicated. Equation 5.1 written in db can be used to get the Λ(f m ) [db] entries in the table. Lab Exercise PN: Phase Noise Measurement - 4 -
Λ(f m ) [db] = V d [dbv] K d [db] -3 db 20 log f m (5.4) where K d comes from table 5.2. Estimate K d for frequencies not in the table (f m below 1 khz). Compare Λ(f m ) from the discriminator measurements the spectrum analyzer measurements in Table 5.2. They should be in close agreement above 10 khz. 5.2.4 Measurement Syntonic oscillator Phase Noise. Replace the synthesizer with the Syntonic source (with attenuator) and check that the power (corrected for the coupler loss) is the same as was recorded in the caption of Table 5.2. The frequency should be within 1.0 GHz. Be sure that the VCO has stabilized. You should readjust the phase shifter to insure that the DC component of v d is zero. Then follow the same procedure as above to get Λ(f m ) and complete Table 5.3. f m K d (db) V d (dbv) synthesizer 100 Hz 200 Hz 500 Hz 700 Hz 1 khz 2 khz 7 khz 5 khz 10 khz 20 khz 50 khz 70 khz L(f m ) synthesizer V d Syntonic (dbv) L(f m ) Syntonic Table 5.3 Measurements for determining phase noise using a delay line discriminator. 4 Measurement using the Phase Noise Meter To confirm your measurements you ll be using the HP E4407B spectrum analyzer s option B that enables the E4407 to perform as phase noise meter. It will plot the phase noise spectrum from 100Hz to 1MHz, which you can save to the floppy disk and use as a ground reference. Measure the signal generator as well as the Syntonic oscillator performance. 5 Report In your final report compare the phase noise measurements made three different ways; with the spectrum analyzer, with the simplified delay line discriminator, and with the Lab Exercise PN: Phase Noise Measurement - 5 -
phase noise meter. Do this with plots of Λ(f m ) in db versus f m with f m on a log scale. Plot several sets of results on a single graph such that comparisons can be easily made. Also compare the phase noise of the synthesizer to the phase noise of the Syntonic oscillator. Compare your measurements to the manufacturers specification. Lab Exercise PN: Phase Noise Measurement - 6 -