Noise Figure Application

The Noise Figure Application makes fast, easy, and accurate noise figure measurements. This function is available with Opt 028, S9x029A/B, or S9x027B.

The information presented in this topic pertains to Noise Figure measurements on BOTH Amplifiers and Converters unless stated otherwise.

• Noise Figure Hardware and Software Options Explained (028, S9x029A/B, S93027B,  H29 - obsolete)

• Features, Requirements, and Limitations

• Noise Concepts

• How the Noise Figure Application Works

• Scalar Noise Figure Measurements

• PNA-X Option H29 - Block Diagram

• PNA-X Option 029 - Block Diagram

• Noise Figure Option S9x027B

• The Noise Tuner Switch

• Noise Parameters that are Offered

• Using Noise Figure App

• • Connect Tuner and Noise Source

  • Create a Noise Figure Measurement

  • Make Noise Figure Settings

  • Perform Calibration (separate topic)

  • Save Noise Data

• Noise Model and the Noise Correlation Matrix

• Noise Figure Measurement Tips

• Using Noise Figure Traces in Equation Editor

• Noise Figure Automatic Switch Control

See Also

Noise Figure Calibration

Noise Figure on Converters (NFX)

Programming commands

PNA-X Noise Figure Options (VNA Configuration Guide)

High-Accuracy Noise Figure Measurements Using the PNA-X

Noise Figure and TRL Cal

See other VNA Applications

Noise Figure Hardware and Software Options Explained

See Also: PNA-X Noise Figure Options (VNA Configuration Guide) - Internet connection required

• 029 - (PNA-X ONLY) Includes low-noise receivers and noise tuner bypass switch to enable noise figure measurements to 50 GHz. Also includes Opt 028 capability.

• E29 - (PNA-X ONLY) Includes low-noise receivers and noise tuner bypass switch to enable noise figure measurements to 67 GHz. Only valid on N5247B. Also includes Opt 028 capability.

• 028 - Uses VNA receivers to measure noise figure.  A noise source is NOT used during calibration. Any two ports can be used. Use with DUTs that have sufficiently high gain and noise figure. Additional filtering may be necessary. Learn more.

• S93027B - (PNA-X ONLY) Allows PNA-X models with Opt 029 to use mechanical noise tuners. Learn more.

• S9x029A/B - Noise figure measurement application.

• H29 (obsolete) - Noise Figure on N5244A (43.5 GHz PNA-X model) and N5245A (50 GHz PNA-X model). Includes both:

• • Opt 029 hardware from the N5242A (noise measurements to 26.5 GHz)

  • Opt 028 (noise measurements using standard receiver to 50 GHz). See H29 specifications.

Note: Option H29 does not include an internal noise tuner.

50 GHz and 67 GHz Noise Figure Receivers

Beginning in October 2012, Option 029 (low-noise receivers) is available in the N5244B (43 GHz), N5245B (50 GHz), and N5247B  (67 GHz) models.

Beginning in Fall 2023, Option E29 (low-noise receivers to 67GHz) is available in the N5247B models.

• The low-noise receiver and the noise receiver path switch is added between the port 2 CPLR THRU jumper connector and the port 2 bias tee. See Option 029 block diagram below. When it is configured with a multiport test set (i.e. U3024AH10), the low-noise receiver cannot be used with test ports on the test set with standard multiport test set jumper connections. See modified test set connections to use low-noise receiver with multiport test set.

• These models (with Option 029) include a built-in (internal) Noise Tuner which can be selected at the beginning of the Noise Figure Calibration. The Noise Tuner switch is managed differently than in 26.5 GHz models. Learn more.

• When used with the N5247A, Noise Figure measurements between 50 GHz and 67 GHz are possible using the standard receivers (Opt 028) and a 67 GHz Power Sensor or with low-noise receivers with the E29 option. See Opt 028 Measurement tips.

• See limitations of the 50 GHz Noise Source.

• See Using Opt. 029 (low-noise receiver) with External Test Set

Noise Figure Application Features

• Cold noise method includes correction for imperfect system source match for highly accurate noise figure measurements.

• With Opt 029/S93029A/B you can measure devices with noise figure values ranging from about 0 to 50 dB and devices with GAIN ranging from about -40 to +60 dB. Learn more.

• With Opt 028/S93029A/B you can also measure noise figure using standard VNA receivers. Learn more.

• Measure noise figure of frequency translating devices. Learn more.

• During calibration, ENR values are interpolated for frequencies between the supplied data points.

Noise Figure Application Requirements

• Noise Tuner - Required for vector noise figure measurements. Not required for scalar noise figure.

• • ONLY the N4690 Series ECal modules are supported. The N4691D m-f is recommended.

  • Opt 029 provides an additional cable and adapter to connect the ECal module to the front-panel connectors. Learn more.

  • A built-in Noise Tuner is provided with the 50 GHz noise receiver models. Learn more.

• Power Meter - Required when calibrating NFX (Noise Figure on Converters).

• Recommended: An accurate thermometer. Learn more.

• MM Head: The N5290A or N5291A broadband millimeter heads may be used to measure noise figure. This feature must be used in conjunction with a broadband amplifier such as the U722xx series amplifiers, and the use of the macro feature (SENS:NOIS:SWE:MACR). Careful physical setup including filters may be necessary. The feature may be accessed by SCPI only (SENSe<ch>:NOISe:RECeiver <char>).

Limitations with the Noise Figure Application

The following features are NOT supported in a noise figure channel:

• FCA or Frequency Offset

• Analog sweep. All frequency sweeps are STEPPED.

• Independent IFBW, Power Levels, or Sweep Time in a segment table is NOT supported.

• External Test Set Control (Opt S93551A/B)

• Receiver calibration.

• Enhanced Response Cal

• ECal User Characterization.

• Some Fixturing Features

• Auto Port Extensions

• Auto Formatted Citifile data.

• External DC Devices

• Pulsed noise figure measurements are supported with the following limitations:

• • Minimum 300 microsecond pulse width using 24 MHz noise bandwidth

  • Narrower noise bandwidths cause larger minimum pulse widths

  • A drop-out may occur at start of sweep and at 6 GHz. This is corrected by a 1 ms pulse width at 24 MHz Noise BW.

• Option 205 or Option 425

Noise Concepts

The following conceptual information is a short summary taken from the Keysight Noise Figure App Note 57-1.

All electronic circuits have some degree of random noise. The most common form is thermal noise, which increases as the temperature of the circuit increases.

The signal-to-noise (S/N) ratio of components in a communications system is a very important parameter. To improve the S/N ratio, it is usually easier and more cost-effective to reduce noise than to increase signal power.  In order to reduce noise, an accurate method to measure noise is required.

Noise Figure

Noise Figure is the degradation in the signal-to-noise ratio as a signal passes through a device. For example, in the following images:

For consistency, noise measurements are calculated as if using a 1 Hz bandwidth, although measurements are almost always made at higher bandwidths.

The following formula shows the lowest possible noise power in dBm at 290° K (room temperature). The only way to measure noise lower than this is to make the measurement at a lower temperature.

• P = 10LOG(4.0 x 10 -21 watts/.001 watt)

• P = -174 dBm / Hz

How the Noise Figure Application Works

The goal of the noise figure application is to accurately measure the noise that is generated by the DUT.  This may be done using special low-noise receivers or using the standard VNA receivers depending on whether the VNA has Options 029 or 02. Learn more.  

The standard receivers are always calibrated using a power meter and a measurement of the receivers effective noise bandwidth.  The low-noise receivers can be calibrated using either a characterized noise source or using the same process as a standard VNA source. Learn more about the noise calibration process.

Some noise measurement error is caused by a poor source match presented to the DUT input. Therefore, during every measurement, the noise figure application uses an ECal module to present at least four different impedances at the input of the DUT.  This "Noise Tuner" is connected to the VNA port 1 or port 2 front-panel loops that are in the VNA internal source path (see block diagram below). From the measurements at various impedance states, the VNA calculates the noise out of the DUT as though the VNA were exactly 50 ohms. No assumptions are made regarding the input impedance of the DUT.

Here is how a vector noise figure measurement is made using Option 029. The sweep numbers are annotated on the VNA display as they occur.

1. With the noise tuner in the THRU state, S-parameter measurements are made to accurately characterize the gain of the DUT.  This requires sweeps in both forward and reverse directions. (sweep #1 and #2).

2. The noise measurements are performed next.  VNA source power is turned OFF and the noise tuner is switched to the first impedance state.

3. At each frequency, the noise receiver samples a large number of readings in order to attain one valid measurement. If Noise Averaging is selected, the specified number of measurements are made and averaged together to obtain one noise measurement. This continues for all frequencies (sweep #3).

4. The next noise tuner impedance state is switched IN and the noise measurements in step 3 are repeated. This occurs until measurements are made at all impedance states. At least four impedance states must be used. (sweeps #4, #5, #6+)

5. Calibration error terms are applied and calculations made to simulate the measurement with a perfect 50 ohm input impedance. The sweep result is plotted on the VNA display.

6. The VNA begins sweeping again with step 1.

Scalar Noise Figure Measurements

As described above, the noise tuner is switched to at least four different impedance states before a sweep is plotted. These sweeps are NOT made in a scalar noise figure measurement, resulting in much faster measurements. Of course, a scalar noise figure measurement is NOT as accurate as a vector noise figure measurement because scalar noise figure measurements assumes that all impedances are 50 ohms. Measurement accuracy can be improved by adding an attenuator as close to the DUT input as possible. This improves the effective system source match. The effect of the attenuator loss is removed during the calibration process.

With scalar noise figure, it is not necessary to connect the noise tuner. If a noise tuner remains connected, it is switched to the THRU state for scalar noise figure measurements. This results is a small amount of loss which slightly degrades measurement accuracy. To increase measurement accuracy, manually switch the noise tuner switch to the INTERNAL position. Learn how.

Select Scalar Noise at the first page of a Noise Figure calibration.

PNA-X Option H29 - Block Diagram with Noise Figure Components

PNA-X Option 029 - Block Diagram with Noise Figure Components

Noise Figure Option S9x027B

Option S9x027B allows a VNA with a low-noise receiver to use specialized mechanical tuners on the input port. These tuners are designed to have a large number of impedance states that are broadly distributed on a Smith chart. There are two situations in which a mechanical tuner is advantageous:

1. When measuring noise figure at low frequencies. While Ecals are good general-purpose tuners, they tend to have a suboptimal spread of impedance states at low frequencies.

2. When measuring NF of a device that is poorly matched. Ecal modules are well suited to measure the noise figure of devices with a match near 50 ohms, but do not perform as well with devices that have match far from 50 ohms.

Mechanical tuners also improve the quality of the Noise Parameter measurements NFmin, GammaOpt, and Rho.

The following tuners are supported by Option S9x027B. All are manufactured by Maury Microwave or Focus Microwaves. Use of a Focus Microwaves tuner is a Licensed Feature. Learn more about Licensed Features.

Maury Microwave Tuners

Focus Microwaves Tuners

To use a tuner supported by Option S9x027B, follow this procedure:

1. If using a Maury tuner:

   1. Download driver files for the tuner at the Web address shown below:
      https://www.maurymw.com/Support/downloads.php

   2. Run the driver installation program.

   3. Power up the tuner and connect it to the VNA via a USB cable.

2. If using a Focus Microwaves tuner:

   1. Download driver files for the tuner at the Web address shown below:
      https://focus2501.sharefile.com/share/view/s653a869bcc8f44c8abe1052ebea6a718

   2. Run the driver installation program.

   3. Power up the tuner and connect it to the VNA via a LAN to USB cable (USB-side connected to the VNA). Please see Focus Tuner documentation for set up and support. Once the tuner is defined in the FDCS Load Pull Explorer software, the VNA FW will automatically detect it.

3. The tuner should now be available for use. To verify this, start the VNA application and create a Noise Figure or Noise Figure Converters channel. Start the Calibration Wizard, and expand the selections for the Noise Tuner combo box. The attached tuner should appear in the displayed list. In the example below, a Maury Microwave MT982BL01 tuner was used.
   
   

4. Once the tuner has been recognized, it should be inserted into the signal path between port 1 and the DUT input. To increase measurement accuracy, it is best to have the tuner as close to the port 2 calibration plane as possible. This ensures that the spread of tuner impedances is as large as possible.

Calibrating with a Mechanical Tuner

No special steps are required to use a mechanical tuner for noise figure calibration. After verifying that the tuner is recognized by the PNA, simply perform a noise figure calibration in the usual way. The PNA firmware will measure all impedance states of the tuner during calibration. When the calibration is complete, the number of states used in vector NF correction can be set in the Noise Figure Setup dialog. In the example below, 15 states have been chosen, but up to 21 states are available.

Caveats when using a Mechanical Tuner

Mechanical tuners offer distinct advantages, but there are caveats to keep in mind:

• Mechanical tuners are slow. They take more time to move from one impedance state to another than Ecal modules do. This increases the amount of time required for corrected NF measurements.

• Because of the increased measurement time, interference can be more of an issue. Any external signals that can interfere with a noise figure measurement, such as cell phone traffic, have more opportunity to intrude. This is important to keep in mind if you are making lengthy measurements in an unshielded environment.

• The number of sweeps taken during a vector corrected measurement is usually equal to the number of impedance states, plus two (for the forward and reverse S-parameter measurements). If a mechanical tuner is used, the number of sweeps may be about twice as large. This is because (depending upon frequency range) the tuner may need to take two sweeps per impedance state, one for low band, the other for high band.

• If a Noise sweep is aborted for any reason, such as a change in stimulus conditions, the dialog below may appear. This is because the PNA is waiting for the tuner to complete an operation. The dialog should disappear after the pending operation is complete.
  
  

The Noise Tuner Switch while making S-parameter measurements

Because of the built-in Noise Tuner in the Option 029 50 GHz noise figure models, the Noise Tuner switch is managed differently than the 26.5 GHz noise figure models.

26.5 GHz Models

The default setting for the port 1 noise tuner switch is "External" as shown in the above diagram. This setting provides incident power through the front panel loops and the Noise Tuner when connected. When connected, the Noise Tuner may NOT be in the THRU state, which is necessary for accurate S-parameter measurements.

The switch is changed in any of the following ways:

• Set the switch to INTERNAL for the S-parameter channel using the path configuration dialog.

• Set the switch to INTERNAL for the S-parameter channel using the following commands:

• • SCPI - SENS:PATH:CONF:ELEM:STAT "Port1NoiseTuner", "Internal"

  • COM - PathConfiguration.Element("Port1NoiseTuner").Value = "Internal"

• Set the switch default to INTERNAL using a preference setting.

• Set the Noise Tuner (ECal module) to the THRU state using SCPI: CONT:ECAL:MOD:PATH:STATE.

Important Note: On the 26.5 GHz Opt 029 models, once you set this switch to "Internal", you must set it back to "External" to make noise figure measurements. The switch is NOT automatically set to "Internal" during a noise figure measurement.

50 GHz Models

The switch for the built-in tuner (Opt. 029) is set to "Internal" (Tuner) ONLY when making vector noise figure measurements. Otherwise, it is set to "Bypass" (the tuner). Therefore, you should NOT need to make switch settings. However, the switch can be changed in any of the following ways:

• Using the path configuration dialog.

• Using the following commands:

• • SCPI - SENS:PATH:CONF:ELEM:STAT "Port1NoiseTuner", "Bypass" (or "Internal")

  • COM - PathConfiguration.Element("Port1NoiseTuner").Value = "Bypass" (or "Internal")

• Manage the built-in Noise Tuner impedance states using the SCPI and COM commands and Element / Value settings.

Using the Noise Figure Application

Use the following general procedure to make noise figure measurements:

1. Connect Tuner and Noise Source.

2. Create a Noise Figure Measurement.

3. Make Noise Figure Settings.

4. For Opt 029 and H29, copy your Noise Source ENR file to the VNA "C:\Program Files(x86)\Keysight\Network Analyzer\Noise folder"

5. Perform Calibration  

6. Connect the DUT. Learn more about DUT input and output ports.

7. Measure Noise Figure.

8. Optional  Click File, then Save to save noise figure data. Learn more.

Connect Noise Tuner and Noise Source

• Connect the noise source to the 28V connector on the VNA rear panel. NOT required for Opt 028. The Noise Source is turned ON and OFF automatically as needed during a calibration. Connect the noise source to Port 2 reference plane when prompted during calibration.

•  Click Edit Commands to input commands for setting the noise source on into “Before sweep” and  “After sweep”. Other settings are ignored for noise source control.
   

• Connect the noise tuner (ECal module).  See Noise Tuner requirements.  

• 1. On the VNA front panel, remove the Port 1 jumper cable SOURCE OUT / CPLR THRU. Opt 028 allows noise figure measurements using any two VNA ports.

  2. Connect the noise tuner to the front-panel jumpers for the source (DUT input) port.

See the VNA Configuration Guide for recommended ECal modules, cables, and adapters.

Create a Noise Figure Measurement

1. On the VNA front panel, press Meas > S-Param > Meas Class....

2. Select Noise Figure Cold Source, then either:

   • OK delete the existing measurement, or

   • New Channel to create the measurement in a new channel.

3. A noise figure measurement is displayed. The following shows how to select or change displayed parameters.

Noise Parameters

Several noise parameters, as well as standard parameters, can be measured in the same Noise channel.

Noise Measurements that are offered

The following three categories of noise measurements can be made with the VNA:

1. Noise Figure is the amount of noise that the DUT is adding in a 50 ohm test setup. This is explained in detail in Noise Concepts.

2. Noise Power Parameters show the amount of noise coming out of the DUT in a 50 ohm test setup. With gain measurements of the DUT, these noise power parameters are used to calculate noise figure.

3. Noise Parameters are models of the noise that is generated in a DUT, similar to how S-parameters model how RF flows through a DUT.

• Noise Figure (NF) - Explained in Noise concepts.

• Excess Noise Ratio - Select when measuring the noise source. Compare with the ENR table to validate accuracy of the system. ENR is calculated as:

ENR (in dB) = 10 log10((Thot - Tcold) / T0), where T0 = 290K.

Learn more about the ENR table and Noise Source. Learn more about Noise Source ENR measurements.

• T-Effective - The effective temperature, in Kelvin, of the measured noise level. For example:

290°° K =  -174 dBm/Hz.

Noise Power Parameters

The Noise Power parameters below are offered in the following two formats:

• Available Noise Power  The calculated power that is based on an ideal impedance match at the output of the DUT. These parameters have always been offered in the VNA noise figure App.

• Incident Noise Power - An 'I' is appended to the end of the Available Noise Power parameter. The calculated power into a perfect 50 ohm noise receiver, regardless of the output impedance of the DUT.  

• SYSNPD / SYSNPDI - System Noise Power Density:  Total noise power available at the ADC, including the noise contributed by both the DUT and the internal noise receiver.  This is generally expressed as an absolute power measurement in dBm, but can also be expressed in Watts or Kelvin.   

• SYSRNP / SYSRNPI - System Relative Noise Power:  The noise temperature of the combined DUT and receiver relative to 290 Kelvin.  This is generally reported as a ratio in dB.  Therefore a perfectly quiet device would render a trace at 0 dB.

dB = 10 log10(T/290)

• DUTNPD / DUTNPDI - DUT Noise Power Density:  When correction is ON, this trace exhibits the available noise power, best described as the maximum power available from the DUT where the impedance of the noise port is equal to the output match of the DUT. To be more precise, this occurs when the noise port match is equal to the conjugate of the output match of the DUT.  The noise power contributed by the receiver is removed.

When correction is OFF, the trace exhibits what is more accurately described as delivered power.  Delivered power is the power actually seen by the ADC.  Any mismatch between the receiver and the DUT is ignored.  The noise power contributed by the receiver is removed.

This measurement is generally expressed in dBm, normalized to a 1 Hz bandwidth. For convenience, marker and trace readout shows dBm.

You could display the power in a different bandwidth using Equation Editor.

• DUTRNP / DUTRNPI - DUT Relative Noise Power:  This measurement is rendered as a ratio of the DUT temperature to 290 Kelvin.  It is generally expressed in dB.  The same comments apply with respect to available versus delivered power as described above for DUTNPD.

dB = 10 log10 ((System Temperature - Receiver Temperature) / 290)

Noise Model, Noise Parameters, and the Noise Correlation Matrix

Noise Parameters are models of the noise that is generated in a DUT, similar to how S-parameters model how RF flows through a DUT.  

Noise Model

The noise wave model of any linear 2-port network may be represented by the following image:

This shows a noiseless 2-port network with noise waves (an1 and bn1) added to the input terminals. The a1 a2 and b1 b2  are standard S-parameter waves.

The noise correlation matrix relates to the noise waves as follows:

Where:

• are time-averaged noise power in 1 Hz bandwidth.

•  are time-averaged cross correlation terms, correlation of an1 to bn1.

• Overbars represent time-averaging

• Star superscripts represent complex conjugation

Noise Parameters

• GammaOpt  (Optimum Complex Reflection Coefficient) - The optimal impedance for the noise figure measurement. Select the data format to display GammaOpt in terms of Log Mag, Lin Mag, Phase, Unwrapped Phase, Real, Imaginary, Polar, or on a Smith Chart.

• NFmin - The minimum noise figure that occurs at GammaOpt. NFmin is a scalar quantity that can be displayed as Log Mag, Lin Mag, or Real.

• Rn  (Noise Resistance) - Specifies the rate of change of the level of noise when varying the source impedance. Rn is a scalar quantity in units of ohms that should be displayed in Lin Mag or Real format.

Note: Rn, as a measurement parameter, is not normalized. When the value of Rn is written to an S2P file, it is normalized (Rn / Zo). When an S2P file is recalled, the normalized Rn is multiplied by the system impedance to obtain Rn.

• NCorr_11, NCorr_21, NCorr_12, NCorr_22 - The NCorr_11 and NCorr_22 terms are effective noise temperature, normalized to 290 K. Both terms are time-averaged, noise-wave powers referred to the input of the DUT, where NCorr_11 is the forward wave (noise going through the device towards the output), and NCorr_22 is the reverse noise wave (noise coming out of port 1 of the DUT, going back towards the source).

• • To convert to available noise power, multiply the terms by 290*k*B where:

  • • k = Boltzmanns constant

    • B = system bandwidth

Standard Parameters that are offered (Amplifiers-only)

• S-parameters: S11, S21, S22, S12

• Unratioed parameters using the following notation: (Receiver, source port). These parameters REPLACE the active measurement.  To do this (from front-panel ONLY), press Meas, then [More], then [Receivers].

• • (R1,1), (R2,2), (A,1), (A,2), (B,1), (B,2)

Save Noise Data

To save noise data, click File, then Save Data As   Then select from the following Save As Types:

• (*.prn), (*.cti), (*.csv), (*.mdf) - Noise Figure data can be saved ONLY with these choices. PRN saves only the active trace. CITI formatted, CSV Formatted, and MDF can save all displayed traces. Learn more about these formats.

• (*.s2p) - Saves S-parameter data only after performing a Noise calibration. This data is saved regardless of which noise measurement is active or displayed. Learn more about *.s2p data.

• Trace and Noise parameter (*.s2p) - Saves S-parameter data, then the Noise Parameters. This data is saved regardless of which noise measurement is active or displayed.  When the vector calibration is not enabled or if the noise parameters are not realizable, then the noise parameters have no calculated value. In this instance, the following values are displayed instead:

• • GammaOpt = 0

  • NFmin = 50 ohm noise figure

  • Rn= Z0 / 4 * (F - 1). This equation is how Rn is currently calculated for ill-conditioned data. F is the noise factor where F is related to the noise correlation value ct11 and the normalized noise temperature Tn by F = 1 + ct11 = 1 + Tn so that Rn = (Z0 / 4) x ct11

• NoiseCorr (*.nco) - Saves Noise Correlation data regardless of which noise measurement is active or displayed.  The *.nco file is a noise correlation matrix expressed in T-parameter form (Ct11, Ct21, Ct12, Ct22). These parameters are exactly the same as the Noise parameters NCorr_11, NCorr_21, NCorr_12, NCorr_22 that can be displayed as traces.

• When the vector calibration is not enabled, this data is set to -200 dBm.

Noise Figure Measurement Tips

Note: In this topic, the term Jitter is used to describe the trace-to-trace fluctuations in a measurement. In other topics, this is called 'trace noise'.

Option 029

See Opt 028 (NF with Standard Receiver)

• Measures devices with noise figure values ranging from about 0 to 50 dB and devices with GAIN ranging from about -40 to +60 dB.

• Highest noise figure accuracy is attained when the sum of device noise figure + GAIN is between 0 dB to +70 dB.

• For highest noise figure accuracy and stability, there should be the least amount of electrical loss possible between the DUT output and VNA Port 2.

Power level at the DUT Output - Opt 029

S-parameters are used to measure the gain of the DUT before each series of noise measurements. Jitter in the S-parameter measurements corresponds directly to jitter in the noise measurements.

For best noise figure accuracy, the power level out of the DUT should be between 15 dB and 20 dB below the compression point of the DUT during the S-parameter portion of the noise figure measurement.

To reduce jitter, the power level at the B receiver (port 2) should be above approximately -20 dBm. Much below this level, S-parameter measurements have more jitter. Power must be below +10 dBm as the B receiver starts to compress at this point, although there is no warning or annotation that shows this condition is occurring in S-parameter measurements.

The best way to monitor power at the B receiver is to display a B,1 measurement. With your DUT in place and powered ON, change the input power to the device and note the power at the B receiver.

• For low-gain DUTs, use 5 dB of source attenuation to improve the uncorrected match of port 1.  

• For high-gain DUTs, source and receiver attenuation may be required. Use the lowest possible attenuation values.

Attaining the optimum power level during calibration can also be challenging since calibration is performed without the DUT in place. Because of this, it is often necessary to set source power higher during the calibration than during the measurement. This will cause the 'CD' annotation on the status bar. Measurement results are accurate as long as the step attenuators and other configuration switches are in the same position and all receivers remain in their linear range (below +10 dBm).

It is best to find the optimum power and attenuation settings for both the calibration and subsequent noise measurements before performing a calibration.

IF Bandwidth

Jitter is further reduced by narrowing the IF bandwidth. If the calibration needs to be performed at a low source power, or with receiver attenuation due to high DUT gain, the IF bandwidth should be reduced during the calibration to reduce jitter. The IF bandwidth can then be increased to improve measurement speed.  The CD annotation can be ignored when changing IFBW after calibration.

Noise Settings

See Noise Figure dialog box help for a complete description of these important settings.

Temperature

Noise Figure measurements are extremely sensitive to temperature. As such, there are two settings that require an accurate temperature measurement: At the DUT input, and at the Noise Source connector.

Interference

When measuring the noise figure of an unshielded device, like an amplifier on a printed-circuit board, it is very common to pick up interference from external signals such as cellular phones, wireless LAN, or mobile radios. This interference shows up as non-repeatable spikes in the measurement, as shown below.

Usually, the interference adversely affects the noise figure measurement only at the frequency where it occurs. However, if the interference is large enough and present all of the time, it can cause the noise receivers to compress, which results in inaccurate measurements at many frequencies. In this case, the noise figure measurements should be done in a shielded environment like a screen room.

Option 028

Noise Figure of VNA receiver - Option 028 gives you the flexibility to measure noise figure using a standard VNA receiver. For best measurement accuracy, the DUT excess noise power, which is gain plus noise figure minus cable loss in dB (G + NF - Loss), should meet or exceed the noise figure of the receiver. This is generally not a problem with very high-gain devices such as converters with approximately 60 dB of gain.

If your DUT is NOT a very high-gain device, you can re-configure the VNA front panel loops to increase receiver sensitivity.

Re-configuring the front panel loops - This configuration reverses the main arm and coupled arm of the test-port coupler (see following images). This increases the signal to the receiver port by about 15 dB, while lowering the available port power by the same amount. This is a good tradeoff for noise figure measurements.

The following table shows the excess noise that is required at various frequencies. These values assume the front-panel loops have been re-configured as shown above:

For devices that do NOT meet this requirement, a low-noise amplifier (LNA) must be added to the receiver loop (see following image). This boosts the noise power at the receiver by the gain of the LNA. The disadvantage is the possibility of measurement drift and receiver compression. Any change in the gain of the LNA will have an impact on measurements that use the receiver with the LNA, so frequent calibration may be required. Care should also be taken when setting the channel power (used during the S-parameter portion of the calibration, and the gain portion of the DUT measurement) to ensure that the added gain of the LNA does not cause receiver damage or compression. A filter is also required on the output of the LNA. Learn more.

(Option 028)

Option  029  includes noise receivers with filtering to keep mixing-product noise out of the low-noise receivers

Using Noise Figure Traces in Equation Editor

In a Noise Power trace, the underlying unit is noise temperature.  

10*log10( temperature * 1000mw/w * 1.38e-23)

(1.38e-23 is Boltzmanns constant)

Any time you use Equation Editor on a Noise Power trace, the LogMag formatting will apply the above equation.  Therefore, first select REAL format and then generate the equation.

The following screen is an example showing three traces: DUTNPD (DUT Noise Power Density), NF (Noise Figure), and S11 with the equation set to "***=10". Note that formatting for noise figure measurements is different than noise power measurements or temperature measurements. For regular receiver power measurements, the conversion from linear magnitude to log magnitude uses 20*log10 (linear magnitude).