High-Accuracy Noise Figure Measurements Using The PNA-X .
High-Accuracy Noise FigureMeasurements Using thePNA-X Series Network AnalyzerApplication Note
Table of ContentsOverview of Noise Figure . 3What is noise figure? . 3Importance of noise figure accuracy . 5Noise Figure Measurement Techniques . 6Y-factor method . 7Cold source method . 8Accuracy Limitations . 9Assumptions of the Y-factor method . 9Noise figure measurement uncertainty contributions . 10PNA-X’s Unique Approach . 19Option choices . 19Correcting for noise-parameter effects . 21Measurement comparisons of PNA-X and Y-factor method . 23Scalar noise calibration . 30Sweep considerations . 31Using the standard receivers for measuring noise figure . 32Noise-power parameters . 38Measuring frequency converters . 38Measuring differential devices. 40Measuring noise parameters . 42Calibration Overview . 43Vector noise calibration . 43Standard receiver noise calibration . 45Scalar noise calibration . 46Calibration for frequency converters. 46On-wafer calibration . 47Moving the noise calibration plane . 49Practical Measurement Considerations . 50Ambient temperature setting . 50Noise averaging. 50Optimizing S-parameter power level . 54Optimizing power sensor level during calibration . 55Compression and damage levels . 56Interference . 58Additional Resources . 59Application notes . 59Magazine articles. 59Papers . 59Web . 592
Overview of Noise FigureWhat is noise figure?Noise figure is a figure-of-merit that describes the amount of excess noisepresent in a system. Minimizing noise figure reduces system impairmentsthat result from noise. In our personal lives, noise degrades the image qualityof TV pictures, and adversely impacts the voice quality of cell phone calls. Inmilitary applications like radar, receiver noise limits the effective range of thesystem. With digital communications, noise increases the bit-error rate. Systemdesigners always try to optimize the overall signal-to-noise ratio (SNR) of thesystem. This can be done by increasing the signal, or by reducing noise. In atransmit/receive system like a radar system, one possibility is to increase theradar’s transmitted power by using bigger, more powerful amplifiers, and/or byusing larger antennas. Decreasing the path loss between the transmitter andreceiver also helps increase SNR, but path loss is often defined by the operatingenvironment and cannot be controlled by the system designer. SNR can also beincreased by decreasing receiver-contributed noise, which is usually determinedby the quality of the low-noise amplifier (LNA) at the front end of the receiver. Ingeneral, it is easier and less expensive to decrease receiver noise (and achievea better noise figure) than to increase transmitter power.The definition of noise figure is simple and intuitive. The noise factor (F) of anetwork is defined as the input SNR divided by the output SNR:F (Si/Ni)/(So/No), whereSi input signal powerSo output signal powerNi input noise powerNo output noise powerNoise figure (NF) is simply the noise factor expressed in decibels: NF 10*log (F)This definition is true for any electrical network, including those that shift thefrequency of the input signal to a different output frequency, such as an up ordown converter.To better understand the concept of noise figure, consider an amplifier wherethe output signal is equal to the input signal multiplied by the gain of theamplifier. If the amplifier is perfect, the output noise is also equal to the inputnoise multiplied by the amplifier’s gain, resulting in the same SNR at both theinput and output of the amplifier. For any real-world amplifier however, theoutput noise is larger than the input noise multiplied by the gain, so the SNR atthe output is smaller than that at the input, resulting in F being greater than one,or NF being greater than 0 dB.It is important to note that when measuring and comparing noise figures, thetest system is assumed to provide perfect 50-ohm terminations at the input andoutput of the device-under-test (DUT). In real-world scenarios however, this isnever the case. Later, we will discuss the accuracy implications if our test systemis not exactly 50 ohms, and we will show how calibration and measurementmethods can overcome the errors produced from an imperfect 50-ohm sourcematch.3
Another way to express the amount of noise added by an amplifier or systemis in terms of effective input temperature (Te). To understand this parameter,recall that the amount of noise available from a passive termination can beexpressed as kTB, where k is Boltzmann’s constant, T is the temperature of thetermination in Kelvin, and B is the system bandwidth. For a given bandwidth, theamount of noise is proportional to temperature. Therefore, the amount of noiseproduced by a device can be expressed as an equivalent noise temperature,normalized to a 1 Hz bandwidth. For example, the amount of electrical noisecoming out of a commercial noise source with a 15 dB excess-noise ratio (ENR)is equivalent to a termination at 8880 K. The noise factor of any real devicecan be expressed as an effective input noise temperature. While Te is not theactual physical temperature of the amplifier or converter, it is the equivalenttemperature (in degrees Kelvin) of an input termination connected to a perfect(noise-free) device that would produce the same amount of additional noise atthe output. Te is related to the noise factor as:Te 290*(F - 1)A plot of Te versus noise figure is shown in Figure 1. While the majority of LNAsare described using noise figure, Te is often used for LNAs that have noisefigures that are less than 1 dB. Te is also useful for mathematical calculationsinvolving noise powers.Figure 1. Effective noise temperature versus noise figure4
Importance of noise figure accuracyOne of the goals of this application note is to give the reader a betterunderstanding of accuracy issues related to noise figure measurements.Measurement accuracy is important in both R&D and manufacturing environments.In R&D, better noise figure accuracy means that there will be a better correlationbetween simulations and measurements, helping designers refine circuitmodels faster. But higher accuracy also means that a system designer canbetter optimize transmit/receive systems like those used in radar applications.When assigning performance values to all of the individual components of thesystem, the system designer must add a guard band based on measurementaccuracy, since a component designer will measure their device to verify itsperformance. For noise figure, improved measurement accuracy and smallerguard bands mean the LNA can have better specifications, which in turn meansthat lower-power transmit amplifiers can be used for the same overall systemSNR. This translates to smaller, lighter, and cheaper transmitters, all of which isvery important for airborne and spaceborne applications.In manufacturing, improved measurement accuracy also allows use of smallerguard bands, which provides better correlation among multiple test stations.This means fewer products must be reworked, resulting in higher yields andimproved throughput, and lower test costs. Smaller guard bands also allowbetter device specifications, yielding more competitive products that commandhigher prices or attain higher market share.5
Noise Figure Measurement TechniquesThere are two main techniques for making noise figure measurements.The most commonly used method is called the Y-factor or hot/cold-sourcetechnique, and it is used with Agilent’s noise figure analyzers, and spectrumanalyzer-based solutions.The Y-factor method uses a calibrated noise source consisting of a speciallydesigned noise diode that can be turned on or off, followed by an attenuatorto provide a good output match (Figure 2). When the diode is off (i.e., no biascurrent is present), the noise source presents a room-temperature terminationto the DUT. When the diode is reversed biased, it undergoes avalanche breakdown, which creates considerable electrical noise over and above that providedby a room-temperature termination. This amount of extra noise is characterizedas an “excess noise ratio” or ENR, and for a given noise source, ENR variesversus frequency. Typical noise sources have nominal ENR values that rangefrom 5 dB to 15 dB, depending on the value of the internal attenuator. Using thenoise source, two noise-power measurements are made at the output of theDUT, and the ratio of the two measurements, which is called the Y-factor, is usedto calculate noise figure. The Y-factor method also yields the scalar gain of the DUT.346C 10 MHz – 26.5 GHzNoise sourceExcess noise ratio (ENR) Thot - Tcold290 K 28VDiode off Þ TcoldDiode on Þ ThotFigure 2. Schematic of an excess noise sourceThe second approach for measuring noise figure is the cold source method,which is also sometimes called the direct-noise method. It relies on a single,cold (typically room temperature) termination at the input of the DUT, and anindependent measurement of the DUT’s gain. This method is often used withvector network analyzers (VNAs), because multiple measurements, such asS-parameters, compression, and noise figure, can be performed on an amplifieror converter with a single set of connections.6
Y-factor methodLet’s take a closer look at the Y-factor technique. Using the noise source, twonoise-power measurements are made. One measurement is made with thenoise source in its cold state (noise diode off) and the other measurement ismade with the noise source in its hot state (noise diode on). From these twomeasurements, and from knowing the ENR values of the noise source, twovariables can be calculated: the scalar gain and the noise figure of the amplifierunder test.When the DUT is measured, the noise contribution of the test instrument’snoise receiver is also measured. To remove the effects of this additional noise,a calibration step is done prior to the actual measurement, where the noisesource is connected to the test instrument to determine the noise figure ofthe internal noise receiver. A simple mathematical expression can be used toextract the noise figure of the DUT from the overall system noise measurement.This step is referred to as second-stage noise correction, as the DUT’s measured noise figure is corrected based on the gain and noise figure of a secondstage, which in this case is the test instrument’s noise receiver.If the output noise power of an amplifier is plotted versus input noise, it followsa straight line as shown in Figure 3, as long as the amplifier is linear. This is agood assumption for LNAs, since their purpose is to amplify low-level signalsthat are far from the amplifier’s compression region. If the input noise was zero,there would still be some noise coming out of the amplifier, due to noise generation processes within the amplifier’s active circuitry. This amplifier-generatednoise is what is characterized with a noise figure measurement. Graphically, itis easy to see why two measurements of noise power can be used to solve boththe amplifier’s gain (the slope of the line) and the noise figure (derived from theY-intercept point).Noise power outPout (hot)DUTD PoutD PinPout (cold)Pin (cold) amplifier gainNoise power inNoise added by amplifierFigure 3. Graphical representation of Y-factor method7Pin (hot)
Cold source methodLet’s now take a closer look at the cold-source technique. This method is verysimple in concept. A room temperature (the so-called “cold” termination) isplaced on the DUT’s input, and a single noise power measurement is made. Thenoise measured is due to the amplified input noise plus the noise contributedby the amplifier or converter. If the gain of the amplifier (or conversion gainof the converter) is accurately known, then the amplified input noise can besubtracted from the measurement, leaving only the noise contributed by theDUT. From this, noise figure can be calculated. In order to achieve accuratecold-source measurements, the gain of the DUT must be known with a highdegree of precision. A vector network analyzer can provide the necessary levelof precision using two-port vector-error correction and other advanced calibration methods. Therefore, the cold-source method lends itself well to a vectornetwork analyzer-based solution.Just as was done with the Y-factor method, a calibration step is required tocharacterize the noise figure and gain of the instrument’s noise receivers. Thiscan be accomplished using a noise source, as is done with the Y-factor method,or by using a power meter and a frequency sweep to determine the receiver’seffective noise bandwidth. Note that a noise source or power meter is ONLYused during calibration with the cold-source method, and not during themeasurement of the DUT.Noise power outFigure 4 shows a graphical representation of output versus input noise power.In this case, the slope of the line is known from an independent gain measurement of the DUT. To define the line, only one power measurement is needed toestablish the Y-intercept point, which allows derivation of the noise figure ofthe DUT.DUTslope known amplifier gainPout (cold)Pin (cold)Noise power inNoise added by amplifierFigure 4. Graphical representation of cold-source methodNote that when using a VNA, vector-error correction can be used whenmeasuring the gain of the DUT, which gives a more accurate value than thatobtained from the Y-factor method. Vector-error correction requires that allfour S-parameters of the DUT are measured, which requires two sweeps of theanalyzer (forward and reverse). The corrected values for S11 and S22 of the DUTare used to correct for other errors in the measurement, as discussed later.An adaptation of this approach has been developed for measuring frequencyconverters, where the input and output frequencies are different.8
Accuracy LimitationsAssumptions of the Y-factor methodThe accuracy of the Y-factor method relies on several assumptions about theamplifier and the test system. The validity of these assumptions varies, depending on the S-parameters and noise parameters of the test system and the DUT.The first assumption is that the noise source presents a good 50-ohm matchto the DUT. This is a reasonable (but not perfect) assumption when the noisesource is connected directly to the DUT, especially when using low-ENR noisesources which tend to have better match than high-ENR sources. However,this is usually not a good assumption when an electrical network is presentbetween the noise source and the DUT. The deviation from a perfect 50-ohmsource match can contribute a large amount of measurement error, as will beshown later.There are several examples where it is impractical or impossible to connect thenoise source directly to the input of the DUT. First of all, many devices used inaerospace/defense applications and commercial microwave communicationsare not connectorized. For example, many transmit/receive modules used inphased-array radar systems have microstrip input and output lines, requiringtest fixtures to interface to commercial coaxial-based test equipment. Anotherexample is that of microwave monolithic integrated circuits (MMICs), whichare often tested while still on the wafer on which they were fabricated, beforebeing sealed into hermetic packages. In this case, coaxial-to-coplanar testprobes must be used to connect test equipment to the DUT. In both of theseexamples, the noise source cannot be connected directly to the DUT’s input.Even when the devices being tested have coaxial connectors, many times theyare measured with automated test equipment (ATE), allowing the connectionof multiple test instruments for full characterization of the DUT. For example, anetwork analyzer might be used to measure S-parameters and gain compression, while a spectrum analyzer is used in conjunction with signal generatorsand a noise source to measure intermodulation distortion and noise figure. Inthis scenario, a switch matrix is used between the test equipment and the DUT.Again, when measuring noise figure, the noise source cannot be connecteddirectly to the input of the amplifier.In these cases, when the noise source cannot be connected directly to theDUT’s input, the addition of cables, adapters, switches, test fixtures, and/or probes adds loss, as well as extra reflections, which causes the effectivesource match of the test system to degrade. While the impact of loss can bemitigated by applying a scalar correction to the ENR values of the noise source,the effects of source-match degradation are not easily removed, causing acorresponding decrease in measurement accuracy.The second assumption is that the output match of the noise source does notchange between its hot and cold states. In reality, there will be some difference,since the impedance of the noise diode is different between its biased andunbiased conditions. This change is smaller for low-ENR noise sources, whichhave more attenuation between the diode and the output connector.9
The third assumption is that the noise figure of the test instrument’s noisereceiver is the same when the noise source is con
extract the noise figure of the DUT from the overall system noise measurement. This step is referred to as second-stage noise correction, as the DUT’s mea-sured noise figure is corrected based on the gain and noise figure of a second stage, which in this case is the test instrument’s noise receiver.
Noise Figure Overview of Noise Measurement Methods 4 White Paper Noise Measurements The noise contribution from circuit elements is usually defined in terms of noise figure, noise factor or noise temperature. These are terms that quantify the amount of noise that a circuit element adds to a signal.
INSTRUMENTATION FOR NOISE MEASUREMENTS 7.1 PURPOSES OF MEASUREMENTS There are many reasons to make noise measurements. Noise data contains amplitude, frequency, time or phase information, which allows us to: 1. Identify and locate dominant noise sources 2. Optimize selection of noise control devices, methods, materials
7 LNA Metrics: Noise Figure Noise factor is defined by the ratio of output SNR and input SNR. Noise figure is the dB form of noise factor. Noise figure shows the degradation of signal's SNR due to the circuits that the signal passes. Noise factor of cascaded system: LNA's noise factor directly appears in the total noise factor of the system.
Figure 1: Power spectral density of white noise overlaid by flicker noise. Figure 2: Flicker noise generated from white noise. 1.1 The nature of flicker noise Looking at processes generating flicker noise in the time domain instead of the frequency domain gives us much more insight into the nature of flicker noise.
the noise figure of the receiver. Noise figure has nothing to do with modulation or demodula-tion. It is independent of the modulation format and of the fidelity of modulators and demodulators. Noise figure is, therefore, a more general concept than noise-quieting used to indicate the sensitivity of FM receivers or BER used in digital .
A noise factor of 1, no degradation in signal to noise, produces a noise figure of 0 dB. HF receiver noise figures will range from about 10 to 20 dB. VHF and UHF receivers will often exhibit a lower noise figure, 8 to 15 dB, to take advantage of the lower atmospheric noise environment found there. The receiver noise figure is primarily a .
Where F rx is the receiver noise figure, F LNA is the LNA noise figure. F MIX is the mixer noise figure, GLNA is the LNA gain. Assuming that the LNA and mixer are the dominant contributors of noise, if the mixer noise figure is assumed to be 10dB, the maximum noise figure of the LNA is 4.5dB. Maximum swing at ADC output is 0.5V.
Academic Phrasebank. Setting objectives –student tip “It is important that your chosen research question is clearly focused and well written; answering it must be achievable in the time available to you!” Your research question and the objectives to achieve it. Your research title Your dissertation title is important as it tells your reader what your dissertation is about ! Titles often .
