preliminary for nanoVNA release 0.2.3 and VNAR4.4
October 17, 2019
preliminary for nanoVNA release 0.2.3 and VNAR4.4
October 17, 2019
Previous versions for Ten-Tec VNA
Table of Contents
This manual provides instructions for using TAPR Network Analyzer (VNA) software with nanoVNA. Due care is required in test setup, calibration, and operational methods to fully realize nanoVNA accuracy. This manual applies to nanoVNA firmware release 0.2.3.
If you are not already familiar withA VNA performs measurements on one-port or two-port networks.
A two-port network has input and output (plus ground(s)).
A one-port network has only an input (and ground).
Network input is connected to nanoVNA CH0 (TX) connector,
and network output is connected to nanoVNA CH1 (RX) connector.
This VNA is capable of four measurements:
S21 Magnitude - gain or loss from two-port network input to output.
S21 Phase - transfer phase from two-port network input to output.
S11 Magnitude - return from network input port referenced to the signal sent to that port.
S11 Phase - the phase angle of return from network input port referenced to the signal sent to that network.
Other measurements are derived from these four.
See nanoVNA hardware
The nanoVNA places about a -10 dBm level signal (~0.1 milliwatts) onto the CH0 connector during measurements.
Overclocked nanoVNA Si5351 is somewhat temperature sensitive. For best performance, allow nanoVNA to warm for 10 minutes before calibrating or measuring, but also avoid overheating. Key nanoVNA temperature dependence is stimulus performance approaching 300/900 MHz.
To begin, connect nanoVNA to host by USB cable, turn it on, then launch the TAPR program.
nanoVNA measures neither signal magnitudes nor phase delays absolutely. All measurements are calculated relative to other measurements. In most cases, this other measurement will be an instrument or fixture calibration. Measurement accuracies thus directly depend on reference calibrations.
Return measurements are extremely sensitive to fixture configurations. Great care must be used making S11 measurements to obtain reasonably accurate data. For example, a single 50-ohm connector adaptor less than 1 inch long introduces measurable phase delay in return signals. Properly set up, nanoVNA can resolve connector adaptor lengths. Practically, calibrations should include connector adaptors as used in actual measurements.
VNA S21 dynamic range is limited by the receiver noise figure (mostly cross-talk). S11 dynamic range is limited by bridge directivity, then stimulus harmonics power and mixer response above 300MHz.
VNA basic resolution is 0.1 dB for amplitude measurements. Phase measurement resolution is 1 degree.
The group delay measurement simply differentiates successive phase measurements. Consequently, groupd delay calculations will be wrong if phase changes more than 180 degrees between successive measurement points.
The smallest frequency measurement interval is 1 hertz. Thus the span width (STOP Frequency - START Frequency) must be at least as great as the 101 points in the frequency grid. For example, the START frequency should be at least 101 Hertz greater than the STOP frequency.
nanoVNA CH1 return loss is about 15 dB. If measurement results critically depend upon a more accurate termination, be sure to use a highly accurate attenuator of at least 10 dB between the device under test (DUT) and the CH1 connector. The VNA fixture calibration can do the math for you automatically (run a fixture calibration 'through' with the attenuator in place).
Frequency source accuracy is determined by an on-board 26 MHz crystal oscillator,
The default frequency range on instrument startup is 50 KHz to 900 MHz. The instrument is capable of operating higher, but accuracy is degraded above 300 MHz.
The phase detectors are accurate only down to about 20 dB above the noise floor. Thus, the phase component of an S21 measurement with more than about 60 dB attenuation is not valid. The phase component of S11 measurements is only valid to about 40 dB return loss.
Some high quality commercial attenuators are not exactly 50.0 ohms in impedance. A resistance error of 0.5 ohms in the "50 ohm terminated" calibration step is enough to degrade return loss dynamic range. Similarly, quality of SHORTs and OPENs used in fixture calibration can degrade calibrations. This is one reason nanoVNA uses SMA rather than BNC connectors for CH1 and CH0 ports -- these connectors cause less discontinuity than other types.
Proper calibration needs:
Details:
A high-quality SHORT normally consists of a screw-on connector shell that shorts the center pin to the outside shell with minimal excess inductance. These are manufactured commercially. Alternatively, an SMA receptacle with a small disk (with a hole drilled in the center), fit over the center pin, and soldered between the pin and connector flange on the rear side minimizes excess inductance and thus makes an acceptable SHORT. Either can be easily attached to an SMA cable using an SMA barrel connector.
A high-quality OPEN consists of a connector with minimized length of center conductor projecting beyond the shell insulator. An open SMA connector at the end of a cable provides an adequate OPEN circuit for calibration steps because SMA connector designs provides this feature. A special SMA open provides best termination, but an open connector is adequate for many purposes.
A 40 dB attenuator is needed when running the detector calibration routine (described in section 3); a high-quality, accurate attenuator (or a combination of 2 SMA attenuators) should be used for this purpose.
One-meter and three-meter cables are needed for detector calibration (although exact cable lengths are not critical).
This TAPR software is multi-threaded, meaning that other Windows programs can run while this software acquires data in the background. However both fixture and instrument calibration are single-threaded code. This means that these two calibrations routines will stall if another Windows program has the focus, and will continue when the TAPR application regains focus. The progressbar may take awhile to update after the TAPR program regains focus.
Install TAPR software GitHub. TAPR requires Microsoft Windows© dotNET 1.1 run-time, but other applications and Microsoft service packs may have it already installed it.
Double-click on the downloaded VNAR4.4.zip,
which should open to reveal VNAR4.4.exe.
If that fails, the zip file was probably corrupted during download.
Launch VNAR4.4.exe
If the host computer (typically a laptop) hibernates or suspends, exit TAPR, then restart.
While VNAR4.3 software may run on Windows operating systems as shown, nanoVNA support is expected for no older than Windows 7:
| Windows | Support | Notes |
|---|---|---|
| Windows 98 -- Gold | Maybe | Supports only USB 1.0, not 1.1 |
| Windows 98 SE (Second Edition) | Yes | Tested |
| Windows ME | Probably | Not tested |
| Windows NT | Probably | Not tested |
| Windows 2000 | Yes | Tested |
| Windows XP | Yes | Tested |
| Windows 7 | nanoVNA | Tested? |
| Windows 8.1 | nanoVNA | Tested |
| Windows 10 | nanoVNA | Tested |
The current software distribution consists of:
| Filename | Function |
|---|---|
| VNAR4.4.zip | Installer ZIP file |
VNAR4.4.zip includes only VNAR4.4.exe; file locations can change...
| File name | Where found | Function |
|---|---|---|
| VNAR4.4.exe | Inside VNAR4.4.zip | Host program |
| Help.chm | GitHub Should be in folder with VNAR4.4.exe | Compiled help file |
VNAR4 software expects a reasonably current Windows
.NET 1.1 framework run time package (version 1.0 is not sufficient).
A virtual COM port driver is required for nanoVNA; see Appendix 3 for details.
After verifying TAPR program launch, quit it.
Connect nanoVNA to your computer using a USB cable and power on.
This may provoke Windows "New Hardware Detected Wizard".r
The wizard finds nanoVNA device and attempts to associate a device driver.
When prompted, let the wizard search to find the driver
(since it may have been installed for STM DFU utility).
After Windows has installed a driver and with nanoVNA connected and powered on, relaunch VNAR4.
If launched before nanoVNA connection, VNAR4 may not detect it.
To uninstall this TAPR VNA application, simply delete VNAR4 .exe and .zip files. See Appendix 3 for details.
Test fixtures (test setups) consist of connectors, cables, adaptors
and other things that affect S21
and S11 measurements.
Test fixtures have different physical properties.
For example, each interconnecting test cable length
affects the phase of all measurements (due to propagation time
delays specific to each cable). Therefore, each test setup
requires unique calibration data to compensate for test setup
and errors internal to nanoVNA.
TAPR software supports "fixture calibrations",
and you make one unique for each measurement set up,
saved with descriptive file names,
to later identify each fixture exactly!
Before starting Fixture calibration, have available
'SHORT', 'OPEN', '50-ohm TERMINATION', and a connector bullet (or barrel).
'OPEN' is just the open end of an SMA cable with nothing connected.
Fixture calibration data sets are built from four raw data measurements - 'Open', 'Short', 'Terminated', and 'Through'.
The following procedure makes a typical calibration data set:
'Short' calibration with a zero-ohm load terminating the CH0 cable
'Open' calibration with unterminated CH0 cable
'Terminated' calibration with CH0 cable terminated by 50-ohm load
"Through" calibration is performed with
connector barrel or bullet (usually a double-female connector
adaptor) connecting CH0 and CH1 cables
The Fixture Calibration dialog box lists the 4 individual steps needed to perform calibration. A green check mark appears next to each completed step. Follow steps in order, connecting SHORT, (OPEN,) TERMINATION, and THROUGH as described for each step before clicking that step. After completing all steps, with all check marks present, click the 'SAVE' button.
Interconnecting cables and connectors are left in place because all contribute measurement error; calibration removes both any time delay and amplitude changes they cause.
The calibration routine makes 1024 measurements evenly spaced from 200 KHz to 120 MHz, spaced about 117 KHz apart. The test fixture itself (without the DUT) is assumed to not have any high-Q resonances, which would introduc large uncertainties into measurements (whether or not calibrated). The DUT may of course have high-Q resonances, but NOT fixturing or interconnecting cables. When applying calibration, software linearly interpolates between corrections for those two calibration frequencies nearest actual measurement frequencies, to compensate measurements.
"Open," "Short," or "50-ohm" inaccuracies confound measurement compensation. At 100 MHz, nanoVNA is fairly sensitive to load inaccuracies.
Normally, many different calibration data sets are eventually taken and stored, one for each test setup. These can have any valid windows filename, and descriptive filename are helpful, but the extension is always .cal. The "Apply Fixture Calibration" checkbox is grayed-out until a valid Fixture Calibration Data set is loaded.
It's possible to partially update the Fixture calibration data. For example, changing the cable from a DUT to nanoVNA CH1 connector may not appreciably affect S11 measurement, but generally changes S21 measurement. In this case, load a previously saved calibration dataset and run just the one measurement ('Through') , then save the dataset with new or existing name as appropriate. The newly saved data set inherits S11 raw and derived errors from the loaded dataset, but uses the new 'through' measurements. In fact, this can be done for any or all four measurements.
The following three diagrams show fixture setup during fixture calibration process and during actual measurements using it. The following dialog box shows four steps needed for fixture calibration. Diagrams in this section show connections required to perform individual calibration steps.
The cables, connectors, and adaptors connecting both CH0 and CH1 connectors to the DUT are part of the test fixture. These same cables connectors and adaptors must be left in place during Fixture calibration steps, while the DUT itself is removed and replaced with various loads and connector barrels during Fixture Calibration steps.
Only the cables, connectors, and adaptors connecting CH0 to the calibration loads are part of the text fixture during the first three fixture calibration steps. The fourth step also uses the cables, connectors, and adaptors in the CH1 connection as well in order to complete the fixture calibration. Typically all cables, connectors, and adaptors are left in place, and only the DUT is removed during Fixture Calibration steps.
Once all four steps are completed, click the "Save Cal Results..." button to choose a file name and location. Use descriptive filenames, as many different fixture calibration files are likely.
Filenames can be long and may contain embedded spaces. The software automatically appends .cal to names. Filenames are not limited to 8.3 format, which is insufficiently descriptive.
Cables, connectors, and adaptors connecting both CH0 and CH1 connectors to the DUT are part of the test fixture. Additionally, the bullet connector is part of the calibration test fixture even though not present in actual measurements. This represents a source of uncompensated error. Thus that connector bullet must be short, low loss and impedance-matched to cables in order to minimize errors.
The following menu items available in the TAPR program are explained in the following sections.
The VNA is a reflection-transmission test set. It measures the forward direction half of an S-parameter data set (S11 and S21). Physically reverse DUTs in the test setup to measure their reverse direction parameters (S12 and S22).
To measure and export a complete 4 parameter set:
Forward and reverse storage arrays are both locked once stored, preventing accidental over-write. The menu items appear with a checkmark when locked. The only way to unlock them is to Export the file. Both forward and reverse arrays must be filled before the Export function can work.
Use the Trace menu to select how measurements are displayed.
S21 displays only on Rectangular display.
S11 displays on both Rectangular and Polar displays.
Traces supported by Rectangular display are:
| S11 Magnitude | Device Under Test (DUT) input return loss magnitude, in dB |
| S11 Phase | DUT input return loss phase angle, in degrees, from +180 to -180 |
| S21 Magnitude | DUT forward transfer gain (or loss) magnitude, in dB |
| S21 Phase | DUT forward transfer phase, in degrees |
| S21 Group Delay | DUT forward transfer gain (or loss) derived group delay |
| S11 Magnitude as SWR | |
| S11 as R Ohms | S11 real value converted to Ohms |
| S11 as jX Ohms | S11 quadrature value converted to Ohms |
| Raw Calibration Data - S21thru | Raw data taken from the through VNA connection CH0 to CH1 (through cables actually connecting to the DUT) |
S11 (as a polar complex number) can be displayed in Polar format,
along with raw calibration data (from a calibration file that is loaded)
as well as error parameters derived from calibration data.
Error compensation and raw data are only for informational purposes
and normally not displayed.
Fixture calibration data that can be formatted as Polar traces are here
The TDR mode displays the real part of the Inverse Fast Fourier Transform of a reflection measurement. Because the TDR mode requires a specific custom frequency grid, it ignores the Start and Stop settings. When switching to the TDR mode, a new Frequency Grid is created. Thus, a new sweep is required. Click the sweep button. A TDR interpretation of existing S11 sweep data is not meaningful and should be ignored, instead re-acquire the reflection measurement after you have selected the TDR mode. The 'Apply Calibration' checkbox sets the reference plane for the TDR analysis. When the 'Apply Calibration' box is checked, distances specified on the TDR display are with respect to the reference plane (the end of the cable calibrated with the Short, Open, and Termination loads).
The display Start time, display Stop time, and estimated velocity factor of a cable under test are specified in the TDR Setup dialog, reached from the Vertical Display menu. Additionally, distances from the Reference Plane to the Marker point can be displayed in METERS or FEET.
TDR mode resolution is limited by a VNA's maximum stimulus frequency. It is possible to infer between TDR points by visually interpolating. For example, if two consecutive TDR points have the same large time reflection, the actual time position will lie in between those two consecutive times.
The reflection and forward measurements are made with respect to the
Device Under Test (DUT).
The reference plane location is established
when the fixture calibration is run and loaded.
Use Vertical Scale to independently select vertical scale units per division for magnitude (in dB / division), maximum SWR display (minimum is always 1.0), group delay (seconds per division), and aperture size of the group delay computation.
Magnitude Display resolutions selections are:
10 dB / div 5 dB / div 2 dB / div 1 dB / div 0.5 dB / div
SWR Scale selections are:
SWR 1.0 to 11.0
SWR 1.0 to 6.0
SWR 1.0 to 3.0
SWR 1.0 to 2.0
Impedance Scale selections are:
1k Ohms / div
100 Ohms / div
10 Ohms / div
Right Scale Display selections are:
Phase / SWR / Delay / Ohms
[rectangular] Right Scale Display values adjust to selected vertical scale factors.
Polar Zoom Factor applies a zoom factor to polar displays.
The default value is 1.0, representing normal Smith Chart display,
with a reflection coefficient of 1.0 lying on the chart periphery.
Increasing zoom keeps the 50 + j0 point the centered
and displays impedances near 50 + j0 with more resolution.
A zoom factor of 2.0 places reflection coefficient values of 0.5 on the chart periphery.
There are four display zoom settings:
Zoom = 1.0 / 1.5 / 2.0 / 2.5
Group Delay Time Scale selections are:
1 millisecond / div
100 microsec / div
10 microsec / div
1 microsec / div
100 nanosec / div
10 nanosec / div
1 nanosec / div
100 picosec / div
Group Delay Aperture
See Appendix 4
for details on how group delay is derived
and how data can be
smoothed (or distorted) by aperture size changes.
Note that small frequency spans significantly magnify group delay measurement noise, due to small difference frequencies between adjacent samples. Larger aperture windows and grid point counts reduce measurement noise for narrow sweeps. However, large apertures may smooth data beyond the range of interest, so carefully select an appropriate aperture size.
Use TDR Setup for
The Marker menu can set up to five marker frequencies and
control display of associated parameters.
A marker value of 0 disables it.
Mouse left and right buttons perform several different actions.
Clicking left mouse button in Rectangular Display drops the first unused marker at the pointer frequency, which marker can be dragged left and right until the left button is released. Screen parameters update in real-time as markers are dragged. Existing markers can be grabbed and dragged, but it takes very precise aim and is pretty difficult to do in practice. It's usually easier to launch the Marker menu, disable that particular marker (by entering a value of zero) and then dropping and dragging a new marker. Alternatively, markers can be set coarsely by mouse, then adjusted precisely by menu.
When all five are in use, left-clicks do not create new markers but can drag exiting ones.Clicking Right mouse button pops up screen coordinates and electrical parameters for the pointer location (unrelated to traces). This works in both rectangular and polar modes. That popup remains until the pointer moves some pixels, then disappears.
Rectangular display frequencies are readily apparent, but not so much for Polar display. However, with marker parameters enabled, marker frequency displays in Polar display mode and updates in real time while being dragged. Markers move opposite to mouse movements when S11 trace is in the lower (capacitive) Polar display half.
Vertical (up/down) mouse movements do not change marker frequency.
Marker parameters are color-coded to match corresponding trace colors.
Parametric values
S11,
S21, and
SWR
are displayed only if corresponding trace displays are enabled.
Use FreqGrid to select Frequency Grid point count, the number of measurements made from Start to Stop Frequency. Allowed values are:
| 101 Points | Coarsest frequency display, but fastest measurement data acquisition |
| 201 Points | |
| 401 Points | |
| 1020 Points | Finest frequency display, but slowest measurement data acquisition |
The instrument defaults to 201 points at startup.
More points increase display resolution,
fewer points reduce measurement time period.
Current FreqGrid setting also affects S-parameter text file exportation,
which have FreqGrid individual frequency records.
Calibration runs are fixed in size at 1024 frequency points.
Checking 'Apply Calibration'
interpolates
grid frequency points to nearest calibration frequency points,
with nearest calibration point results applied to measurements.
With calibration points about 117 KHz apart, nearest is closer than 59 KHz.
Unless fixtures have resonances (which is very bad anyway),
interpolation errors for any frequency grid size will be
negligible except for very long cables.
Use Storage menu to store the current measurement set into
temporary memory. This temporary memory will retain that measurement
set until the program is terminated, or a new measurement set is
stored.
Setting a new stored measurement over-writes existing measurements in storage.
Storage menu has three options:
A common use of the storage function is to make a measurement, verify
it's useful, then 'store' the data.
Next, some change to the device under test is made and a new measurement is made.
Use Display Storage to compare on-screen the two measurements.
Markers are only attached to the active display set. MARKERS ARE
NOT ATTACHED TO THE STORED DATA. This helps to distinguish which data
on screen are active and which are storage.
To use markers on stored data,
retrieve that data to the active set using the Recall menu item.
Use Integration for averaging together multiple nanoVNA sweeps. 1x mean no integration. 2x averages the previous and current sweep. 4x, 8x, and 16x use exponential integration to average results. For example, at 16x, the display weights 15/16 of previous with 1/16 of new value. The integrator is RESET for selection changes and returns raw values from the next sweep. Subsequent sweeps will integrate. Changing the Frequency Grid size also resets the integrator.
Resetting the integrator prevents large offsets from taking long times to clear integrated values and avoids waits to see initial results from changes.
The help menu launches the HTML Help viewer with the TAPR help menu.
The About command opens an 'about TAPR' menu. This menu displays the current build number of the software
It also contains a button to launch your Internet Browser with the
address of the TAPR software update website. This will display a
page showing the current TAPR software available. You can
compare your build number with the latest build available number shown
on the web page. Be sure to REFRESH your web browser display.
If you wish, you may download the update from here (right-click, and Save as ...)
Exit the TAPR application before installing updates.
TAPR window controls change start & stop frequencies, reference & transmit signal levels, and sweep functions.
Those controls are not shown on printed outputs.
Frequency controls set sweep START and STOP frequencies. Arrows increment or decrement frequencies. Controls can be set to specific values by double-clicking, which opens numeric entry dialog boxes for directly typing in desired frequencies.
START and STOP controls display underscore characters beneath a digit. This digit will be incremented or decremented by up or down arrows. Left and right arrows change selected digit position. Controls will increment neither above 120,000,000 Hz, nor below 200,000 Hz.
A STOP frequency must be greater than START frequency by the frequency grid size. For exampe, if grid size is 400 points, then STOP must be at least 400 Hertz greater than START.
The apply calibration check box is grayed-out unless a fixture calibration data set has been loaded (see calibration menu). Checking the Apply Calibration box will compensate all readings by the corrections contained in the loaded fixture calibration set. Un-checking the box disables the reading corrections.
Measurements without any fixture calibration are not accurate, since there are uncompensated errors in cable length, fixture attenuation, etc. Further, some VNA instrument errors such as PC board trace lengths and amplitude variations are not removed if the measurement is uncalibrated. For best accuracy, measurements should have fixture calibration applied.
Reference Level adjusts how magnitude data are displayed on screen. The control value specifies the magnitude level at the top line of the display screen. When set to 0 dB, the display top line is 0 dB. Changing Reference Level optimizes trace locations.
For example, if a DUT attenuates approximately 22 dB, that will not display at a scale of 1 dB / division, because off the screen bottom. Reference Level set to -20 dB displays a range from -20 db to -30 db at 1 db/division, making the trace visible.
Reference Level set to positive values is useful when measuring DUTs (such as amplifiers) with output
magnitudes near or above 0 dB.
Note that VNA dynamic range extends only a few dB above 0 dB.
External attenuators should be used with DUTs having any amount of gain.
On-screen amplitude values displayed left-side automatically adjust when reference levels change. Small positive reference levels help make visible markers and marker numbers when measuring low-loss DUTs.
Transmit Level should normally be left at 0 dBm. Lower values usually degrade dynamic range and accuracy of both S11 and S21 measurements. However, reduced drive level to DUTs with gain may be useful. The control has about 50 dB of adjustment range, but nanoVNA dynamic range consumes 40 dB of that range. Thus, S21 magnitude measurements effectively support less than 10 dB of level change. Attenuating CH0 stimulation to DUTs using attenuators is better.
Magnitude values displayed automatically track changes in Transmit Level. By the way, one valid use for the Transmit Level control is to adjust output level when using the VNA as a signal generator and measured values are of no concern. In this case, the control can change transmit output signal level over a 10 dB range.
There are three buttons on the screen that control how the TAPR acquires data. These three buttons are:
The Sweep Speed button determines the sweep dwell time for each measured sample, and if nanoVNA data are checked for measurement glitches. If set to the Fast mode, the VNA sweeps as fast a possible, and does not check the reading data for measurement glitches. As the button is successively clicked, the sweep speed will be set to the following values:
The button recycles back to Fast after the 10 ms step. When the button displays any time value (but not Fast), more extensive comparison of the measured data is used to try and reject obviously defective individual data points (not always successfully). In the Fast mode these checks are bypassed, sometimes resulting in momentary glitches in the measured data. However, the Fast mode sweeps much than any of the timed modes.
The single sweep mode triggers nanoVNA one time, and holds the measured data after that single sweep.
The Free-run sweep button causes the VNA to trigger continuously (recurrently), so that the sweep updates as rapidly as possible. Pressing the Free-run sweep button a second time stops the recurrent sweeping. The frequency grid button is grayed-out when the VNA is sweeping, since the number of points cannot be changed during a sweep.
One very useful setup is to use a small frequency grid, for example 100 or 200 points, fast sweep mode, and recurrent sweep. This speeds display updates and may be particularly useful when tuning a DUT such as a filter or antenna in real-time.
The following general steps are used to measure a device under test,
Make sure that detector calibration has been run
(you do not get a warning message on program startup).
Connect cables and adapters from the device under test (DUT) to nanoVNA.
Either run a Fixture calibration on these cables and adaptors, or load a previously saved fixture calibration for these exact same cables and adaptors.
Remove any shorts, or through connector barrels, and replace with the DUT connected from CH1 to CH0.
Run a sweep of the DUT.
Enable the Fixture Calibration checkbox on screen. This checkbox can be selected or deselected at any time (before or after the sweep).
Let's look at an example DUT, a low pass filter. A single sweep of data captures all the data points shown in the following three screen shots. The different screen shots are just different views of the single data set.
This screen shows GREEN - the S21 transfer magnitude (forward gain/loss). The screen in set for 10 dB / division. RED is the S11 return loss of the filter, again at 10 dB / division. The filter has about 60 dB loss at 50 MHz, degrading a bit at 100 MHz. The return loss of the filter is about 10-15 dB in the filter passband (200 KHZ to 40 MHZ) and is 0 dB in the stopband (above 40 MHz).
This type of filter is called a "reflective filter" since it generates a stopband by reflecting the input signal back.
On the next screen, the S11 (RED) trace has been turned off and only the S21 (GREEN) trace is displayed. Five markers have been set, and the display of marker parameters has been enabled. Here we can see that the insertion loss of the filter is quite good, only about 0.1 dB in the passband, and that the -3 dB frequency is about 39.85 MHz.
The last figure shows input return S11 on a polar chart. Four markers (which must be set in the rectangular display mode) are selected, and marker parameter display is again enabled. The markers show the input magnitude and phase as well as the equivalent input impedance of the filter at each marker frequency. The filter spirals outwards from the origin, but stays between the SWR=1.5 and SWR=2.0 concentric bull's eyes until it reaches the stopband frequency, when it rapidly spirals out of the center toward the periphery of the chart. The light gray concentric SWR circles on the polar chart are:
Due to small ripple errors in the directional coupler, the stopband return loss is calculated as negative a few tenths of a dB (which is wrong). This is shown at marker 4. The polar chart is very non-linear and changes in return loss of tenth's of a dB cause the distance from the center of the chart to change a lot when the return loss is near zero dB, but these changes are minimal when the return loss is more than a few dB.
TAPR software currently compensates for frequency dependent errors.
Starting with version 1.4 the software models the coupler V/I tracking
error and coupler directivity.
It compensates readings by subtracting
the coupler directivity and applying a sinusoidal magnitude correction
vs. detected phase angle to S11 readings.
By the way, these plots are of a high-quality Bencher YA-1 low-pass-filter designed for HF amateur radio.
High quality variable attenuators may have several tenths of a dB loss at their 0 dB setting. This must be taken into account when characterizing filter insertion loss.
Attenuator resistance may not be exactly 50.0 ohms. This will degrade 50-ohm calibration.
Attenuators have finite length. Thus an attenuator inserted (or removed) after fixture calibration impacts phase angles of the setup from the Fixture correction because the distance to the 'SHORT' and 'OPEN' used for fixture calibration has changed.
VNA CH1 connector input impedance is only approximately 50 ohms. When measuring returns of networks with minimal attenuation, phase and amplitude components may include errors due to this CH1 connector deviation from an ideal 50 ohm load. To improve the measurement either:
Terminate the unit under test with a 50 ohm termination prior to S11 measurements, or
Insert 10 dB attenuation between DUT output and VNA CH1 connector. This will assure that the DUT sees more accurate 50 ohm termination impedance.
Some common measurements will result in the displays shown in this section. These can be used to verify that you have correctly calibrated and setup the VNA and the test cables. These results are typical, but you may realize slight differences depending on the cables, instrument measurement errors, etc.
Also, there may be minor variations between consecutive instrument sweeps due to measurement variation or other artifacts. These sweeps were performed in the 'Slow' mode.
3 meter cable terminated in 'Short' without fixture calibration applied.
3 meter cable terminated in 'Short' with fixture calibration applied.
3 meter cable terminated in ''Open' without fixture calibration applied.
3 meter cable terminated in ''Open' with fixture calibration applied.
1 meter cable terminated in 50-ohm load without fixture calibration applied.
Polar scale is zoomed 2.5:1 to show more details.
1 meter cable terminated in 50-ohm load with fixture calibration applied.
Polar scale is zoomed 2.5:1 to show more details.
Rectangular display of S11 for 1 meter cable terminated by 50-ohm load without fixture calibration applied.
Rectangular display of S11 for 1 meter cable terminated by 50-ohm load with fixture calibration applied.
This overview briefly covers reflection and transmission measurements using S-parameters and plotting them on Rectangular and Polar Displays.
Linear two port devices are characterized by ratios of signals returned for signals injected at their ports. The S-parameter model provides all linear characteristics for two port devices at discrete spot frequencies, specifically input return, forward gain, reverse gain, and output return as four complex numbers for each spot frequency. To characterize devices, S-parameters are usually measured for ranges of spot frequencies.
In the diagram below, a1 is the voltage input to port one, and a2 is the voltage input to port two; b1 is the voltage out of port one, while b2 is the voltage out of port two. S-parameters are defined as voltage ratios. S11 (input return) is the ratio b1/a1. S21 (transfer gain) is the ratio of b2/a1. The voltage exiting port one of the block is the sum of the input return loss times the input voltage plus the reverse gain times the voltage injected into port two of the block. S-parameters are defined in the following two equations:
VNAs are essentially transmission-reflection test sets. They apply stimulus a1 to port one of devices while measuring b1 and b2, then calculate S11 (because a2 is zero).
S11 and S21 are generally both complex numbers - in other words, devices reflect energy back to the input for any spot frequency with reflection magnitude and phase both generally non-zero. Similarly, two-port devices typically change both amplitude and phase of b2 from a1. A VNA measures and displays input reflection loss and forward transmission gain as complex numbers.
Polar chart help visualize input impedance real and reactive components. Since reflection ratios S11 are always <= 1.0 for passive networks, those values can be displayed in reactive vs real format on a polar charts. On the other hand, two port network forward gains typically include both loss and gain values, exceeding unity. That confounds polar charts, where polar chart circumference typically represents unity magnitude. Note that polar charts have no concept of frequency; S11 samples for any spot frequency may display anywhere on polar charts. VNAR4 software connects sample dots in lines; subsequently extracting frequency information along those lines wants markers.
S11 polar values are directly plotted, with S11 vectors originating at chart center, vector lengths being return loss magnitudes, and vector directions being S11 phase angles. The chart left side corresponds to +/-180 degrees and lower impedances, while the right side corresponds to 0 degrees and higher impedances.
Rectangular display works for both S11 and S21 values. Return losses are simply displayed vs. frequency, but sorting complex impedance gets harder. Two-port networks' transfer complex impedances are less interesting than their gains and phase delays vs. frequency, Rectangular display directly plots those S21 parameters.
Network transfer gain/loss ratio is labelled in dB on rectangular charts' right side. Return loss or transfer gain phase is plotted in degrees vs. Hertz. Rectangular chart top and bottom are +180 and -180 degrees, respectively, 'wrapping around' from bottom to top because +/-180 degrees are identical. Rectangular chart top and bottom both represent 180 degrees.
Phase values varying slightly above and below 180.0 degrees provoke confusing rectangular displays, since what might nominally be nearly straight traces can instead varying wildly up and down, alternately near top and bottom lines.
Network time delays are interesting, e.g. for TDR. Network group delay is the negative of rate of change of phase vs frequency and helps determine network electrical length for cables or stubs.
Group delay and spatial length measurement determine velocity factor. Cable phase velocity is elapsed time for light traveling a cable length divided by cable's electrical time delay. Light travels 11.8 inches per nanosecond (nearly 1 nanosecond per foot). 10 feet of free space delays light 120/11.8 ~ 10.17 nanoseconds. The velocity factor for 10 feet of cable with 15 nanoseconds electrical delay is:
10.17 nsec (free space delay) / 15 nsec (measured delay) = 0.68 velocity factor.
Network SWR at any specific frequency describes of how nearly that network input matches a 50 ohm non-reactive load when that network's output is terminated in 50 ohms. An SWR of 1.0 means that the network has exactly 50-ohm input impedance. SWR is a real number directly related to input return magnitude, ignoring input reflection phase. S11 is a complex number representing both magnitude and phase angle of input return.
An SWR of 2.0 means that an input impedance is either 25 ohms resistive, 100 ohms resistive, or some complex value with either of those magnitudes containing non-zero reactance. SWR cannot distinguish among those networks cases for non-unity values, while S11 can.
Each constant SWR value describes any impedance lying on a circle centered in polar charts. The polar chart center is 50+j0 ohms, with SWR of 1.0 (its circle is a point). All other SWR values are pointless. Higher SWR values correspond to circles of increasing radius centered on the chart, with the circle at polar chart periphery representing infinite SWR.
This table relates SWR to input return loss magnitude:
| SWR | Return Loss (dB) |
|---|---|
| 1.00 | infinite |
| 1.03 | 36.60 |
| 1.05 | 32.25 |
| 1.10 | 26.45 |
| 1.15 | 23.12 |
| 1.20 | 20.83 |
| 1.30 | 17.70 |
| 1.40 | 15.56 |
| 1.50 | 13.98 |
| 1.60 | 12.74 |
| 1.70 | 11.73 |
| 1.80 | 10.88 |
| 1.90 | 10.16 |
| 2.00 | 9.54 |
| 2.50 | 7.36 |
| 3.0 | 6.02 |
| 4.0 | 4.44 |
| 5.0 | 3.52 |
| 10.0 | 1.74 |
| 20.0 | 0.87 |
| infinite | 0.00 |
Detector calibration applies a range of signals from approximately 0 dBm down to -90 dBm to each magnitude detector (transmission and reflection) repeatedly over the range 200 KHz to 120 MHz. Calibration software models each detector response function (voltage vs. level) at each frequency. The model consists of linear and noise floor portions. Software models the linear portion by generating a linear least-squares regression estimator for amplitude measurements and a correlation coefficient. It then generates an empirical exponential noise floor estimator that fits a smooth curve for a complete linear + noise floor response.
The detector calibration routine similarly characterizes both phase detectors. Phase calibration is a bit more complex than amplitude calibration. It uses a fixed length of cable and varies frequency to develop linear frequency-dependent phase excitation to the detectors. It then finds phase detector voltage levels corresponding to - 180, 0, and +180 degrees (the positive peak, midpoint, and negative peak voltages respectively). Phase detectors have periodic error components, and the detector calibration routine uses a linear delay model to build a table to correct those periodic errors.
All those data sets are then saved in detector.ica in the TAPR program startup directory.
Fixture calibration measures received signals through a small bullet or barrel connector. It measures received magnitude and phase at 1024 points from 200 KHz to 120 MHZ and stores those data in a table in the Fixture calibration file (which is named during the Fixture calibration save operation). When the 'Apply Calibration' box is enabled, the calibration compensation routine divides actual S21 measurements by the nearest 'Through' calibration constant (which is a complex number) at each measurement frequency.
The calibration compensation for S11 derives three parameters of a virtual S- parameter error matrix computationally inserted between the VNA CH0 connector and the DUT. With three measurements and three unknowns, values of virtual S-parameter error matrix can be resolved. These three parameters are:
Et - the tracking error. This is (S21error * S12error) product of the virtual error matrix.
Ed - the directivity error. This is S11error of the virtual error matrix.
Es - source impedance mismatch. This is the S22error of the virtual error matrix.
The virtual error matrix is removed from S11 measurements (when 'Apply Calibration' is active) to compensate for these errors in reflection measurements.
Fixture calibration data sets include the following.
Values can be viewed (by the curious) from the Trace menu.
Polar display is only for reflection measurements (S11)
and related raw and derived calibration constants.
Rectangular display can show both S11 and
S21 measurements and related calibration constants.
| Error Compensation | Complex plot |
|---|---|
| Et - Tracking | amplitude and phase for cable from nanoVNA CH0 to the DUT |
| Es - Source Mismatch | error from imperfect source impedance termination |
| Ed - Directivity | due to finite bridge directivity |
| Raw Calibration Data | Data source during calibration (or loaded from a calibration file) |
|---|---|
| S11short | shorted CH0 during calibration |
| S11open | unterminated CH0 |
| S11term | CH0 terminated by "known" 50-ohm load |
This TAPR VNA application should work with the same driver as other nanoVNA software.
When you first apply power to nanoVNA and attach it to a Windows PC
via its USB cable, its USB device ID may not be found.
The New Hardware Detected wizard will ask if a disk or
file is available for the new device.
A recommended driver is:
STMicroelectronics Virtual COM Port Driver
The wizard creates a registry key associating nanoVNA's USB vendor
and product ID to a device driver.
When nanoVNA is subsequently plugged into that USB host port,
it avoids going through the selection process again.
nanoVNA uses the following USB IDs: Vendor_ID: 0483 Product_ID: 5740
The Wizard makes a registry entry in order to associate the device with the driver needed to communicate with it (usbser.sys). This registry can be removed (but need not be removed) if TAPR software is uninstalled. Registry keys are normally left in place, but if for some reason you want, it can be deleted using Regedit.
Different versions of windows have different Registry directory structures, and the keys are placed in different locations.
| Version | Registry Key Location |
|---|---|
| Win98, ME | \HKEY_LOCAL_MACHINE\Enum\USB |
| Win NT, 2000, XP | \HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Enum\USB |
Group Delay is an equivalent time delay through a device or network.
Mathematically, it is defined at the negative of the rate of change
(the slope) of phase φ vs. frequency ω:
GroupDelay = - dφ / dω
In a VNA, a measurement sweep consists of a number of individual discrete data points taken at single spot frequencies. Group delay is calculated by looking at two adjacent frequency samples - current samplen and next samplen+1, being taken at frequencies fn and fn+1. The difference in phase is calculated by subtracting the phase reading taken at frequency fn from the phase reading taken at frequency fn+1. The delay is then calculated by converting degrees into radians and the sign is changed. Unfortunately, this finite difference process can result in large amounts of noise in readings, due to small errors in measured phase, or from actual noise in measurements.
One way to reduce displayed noise is to average group delay readings over more samples. This delay reading is less noisy, but it may miss some important rapid changes in phase response. TAPR allows selecting the sample averaging count. The count of samples averaged is known as measurement aperture size.
An aperture size of 1 disables averaging, and measurements are taken between adjacent frequency samples. The aperture frequency window is thus fn+1 - fn.
An aperture size of 64 means that delays are calculated by subtracting the phase measured at frequency sample fn+32 from the phase measured at frequency sample fn-32. Thus the reading spans 64 frequency differences (65 samples). One drawback to this averaging technique is that the phase could change through several 360 degree rotations between fn-32 and fn+32. This would result in a group delay calculation being far less than the real group delay since the VNA cannot resolve phase changes exceeding 360 degrees. Thus aperture should be used with appropriate caution.
TAPR software's aperture window concept has difficulty for samples near limit (start and stop) frequencies. Lacking samples below a first measurement at start frequency fn, TAPR lacks phase value samples for frequencies < fn. For aperture windows of 64, software shortens effective aperture windows by using fn to fn+32 (rather than fn-32 to fn+32). Effective apertures are thus 32 (not 64) and skewed. The same shortening and skew happens at stop frequencies (no samples for > fn).
Software similarly shortens aperture windows for samples nearer than half aperture windows to limit frequencies, by a lesser amount for each sample farther from limits. Once the 32nd frequency sample distant from a limit frequency is reached, all samples necessary for 64-point apertures are present, and aperture windows are no longer shortened.
For aperture sizes of 4 and 16, worst case shortening is to effective aperture sizes of 2 and 8 samples, respectively, near limit frequencies.