A network analyzer measures how electrical signals behave as they pass through or reflect off a device, cable, antenna, or filter. The term covers two distinct tools: vector network analyzers (VNAs) used in RF and electronics work, and packet analyzers like Wireshark used in computer networking. This guide covers practical setup and usage for both, starting with the hardware VNA since it involves more hands-on technique.
What a Vector Network Analyzer Actually Measures
A VNA sends a known signal into a device under test and measures what comes back out. Internally, it contains four key blocks: a signal source (typically a synthesized oscillator), signal-separation hardware, a tuned receiver, and a processor with display. The signal-separation stage uses directional couplers or bridges to split the outgoing signal from the reflected signal, so the instrument can track both simultaneously. The tuned-receiver design gives VNAs dynamic range of 100 dB or better, meaning they can detect extremely weak signals alongside strong ones.
The core output of a VNA is a set of S-parameters, which describe how power moves between the ports of your device. For a two-port device:
- S11: How much power reflects back from Port 1. This is the reflection coefficient, also called return loss. A well-matched antenna or filter will have a very low S11 at its operating frequency.
- S21: How much power passes from Port 1 through to Port 2. This is insertion loss (or gain, if the device amplifies).
- S12: Power transferred in the reverse direction, from Port 2 to Port 1. Useful for checking isolation or reverse gain.
- S22: Reflected power at Port 2, the mirror of S11 for the other side of the device.
The naming convention is straightforward: S followed by two numbers, where the first is the receiving port and the second is the source port. S21 means “received at port 2, transmitted from port 1.”
Calibration Before Every Measurement
Calibration is the single most important step in VNA use, and skipping it is the fastest way to get misleading data. The instrument needs to mathematically remove the effects of your cables, adapters, and connectors so it only measures the device itself.
A standard calibration kit includes three physical standards: an open, a short, and a known load (usually 50 ohms). You connect each standard to the end of your test cables, and the VNA measures the known response of each one. From those three measurements, it calculates error-correction terms that apply to every subsequent sweep. For two-port measurements, you also perform a “through” calibration by connecting the two test cables directly to each other.
Calibrate at the exact point where your device will connect. If you calibrate at the cable ends but then add an adapter to reach your device, that adapter’s loss and phase shift will contaminate your data. Temperature changes can also drift your calibration, so recalibrate if the room temperature shifts noticeably or if you’ve been running for several hours.
Connecting Your Device Without Damaging It
RF connectors are precision components, and overtightening or cross-threading them can permanently damage both the connector and the VNA’s test port (which is expensive to repair). Always use a torque wrench set to the correct specification for your connector type:
- Type N (50/75 ohm): 12 lb-in (135 N-cm)
- 3.5 mm, 2.4 mm, 2.92 mm (K): 8 lb-in (90 N-cm)
- SMA (standard): 9 lb-in (100 N-cm)
- TNC: 5 lb-in (56 N-cm)
When mating connectors, align them carefully and turn only the outer coupling nut. Never rotate the cable or the device body, as this twists the center pin and wears the contact surfaces. Torque values can vary between manufacturers, so check the documentation for your specific calibration kit and cables.
Setting Up the Sweep
Three settings define your measurement: frequency range, number of points, and IF bandwidth.
Set your start and stop frequencies to cover the range you care about, with some margin on either side. If you’re testing a 2.4 GHz bandpass filter, you might sweep from 1 GHz to 4 GHz to see the full passband and the rejection on both sides. The number of sweep points determines frequency resolution. More points give finer detail but slow the sweep down.
IF bandwidth controls the tradeoff between speed and noise. A narrower IF bandwidth reduces noise on the trace, which matters when you’re measuring very low signal levels or need high accuracy. A wider IF bandwidth sweeps faster but introduces more noise. For most routine measurements, start with a moderate IF bandwidth (1 kHz to 10 kHz) and narrow it if your trace looks noisy, or widen it if you need faster updates while tuning something in real time.
Set the output power level appropriately for your device. Passive components like cables and filters can handle full power, but active devices like amplifiers may compress or saturate if you drive them too hard. Start with a low power level for amplifiers and increase gradually.
Reading a Smith Chart
Most VNAs can display S11 data on a Smith chart, which plots impedance in a way that reveals whether your device looks resistive, inductive, or capacitive at each frequency. The chart is circular, with pure resistance along a horizontal line through the center.
The center point represents a perfect 50-ohm match (assuming a 50-ohm system). Points to the left of center have lower impedance; points to the right have higher impedance. The upper half of the chart represents inductive behavior (the kind you’d see from a coil or a transmission line that’s too long), while the lower half represents capacitive behavior (from a gap, a stub that’s too short, or actual capacitance in the circuit). Points along the outer edge of the circle mean all power is being reflected, a total mismatch.
As the VNA sweeps across frequency, the S11 trace draws a path on the Smith chart. A well-matched antenna will loop through or near the center of the chart at its resonant frequency. If the trace sits in the upper half, adding a series capacitor might pull it toward the center. If it sits in the lower half, a series inductor could help. This visual feedback makes the Smith chart especially useful when you’re tuning matching networks by hand.
Using a Packet Analyzer (Wireshark)
If you searched “network analyzer” looking for a tool to inspect computer network traffic, you’re likely looking for Wireshark, the most widely used packet capture tool. It captures data flowing through a network interface and lets you inspect individual packets, filter by protocol, and trace communication problems.
To start a capture, select the network interface you want to monitor (Wi-Fi, Ethernet, or loopback) and click the capture button. Traffic fills the screen quickly on a busy network, so capture filters let you limit what gets recorded. Filters use a simple syntax that combines primitives with “and,” “or,” and “not”:
- tcp port 443 captures only HTTPS traffic
- host 10.0.0.5 captures all traffic to or from a specific IP
- tcp port 23 and not src host 10.0.0.5 captures Telnet traffic that didn’t originate from that IP
- dst net 192.168.1.0 mask 255.255.255.0 captures traffic headed to a specific subnet
You can also filter by direction using “src” or “dst” before “host,” “net,” or “port.” For Ethernet-level filtering, prepend “ether” to filter on MAC addresses instead of IP addresses.
Capture filters apply before packets are saved, reducing file size and helping you focus. Display filters, applied after capture, let you drill deeper into data you’ve already recorded. The syntax differs: display filters use a dot notation like “tcp.port == 443” rather than the libpcap syntax used for capture filters. Keeping these two filter systems straight avoids confusion when you’re troubleshooting.
Common Mistakes to Avoid
With a VNA, the most frequent error is forgetting to calibrate, or calibrating at the wrong reference plane. The second most common is using damaged or dirty connectors. Inspect connector faces with a magnifier before connecting, and clean them with isopropyl alcohol if you see debris. A single fiber or metal shaving on a connector face can shift your measurements significantly at microwave frequencies.
With Wireshark, the most common issue is capturing too much data without filtering, then struggling to find the packets that matter. Start with a focused capture filter rather than recording everything. If you’re diagnosing a specific connection problem, filter on the relevant IP address and port from the start. On switched networks, remember that your interface only sees traffic destined for your machine unless you configure port mirroring on the switch or use a network tap.