📡 Key Points
Why mmWave Testing Is Different
If you have spent years testing cellular devices below 6 GHz, 5G mmWave testing will feel like a different discipline. It is not just higher frequencies—it is a different way of connecting to the device.
At FR1 frequencies (sub-7 GHz), you screw on a cable and measure. At FR2 frequencies (24 GHz to 40 GHz and beyond), the antenna is integrated into the package. There is no connector. The device communicates through the air .
This changes everything. Your test setup now includes:
A controlled RF environment (anechoic chamber or reverberation chamber)
Probe antennas to communicate with the device
Positioners to change the device orientation
Calibration that accounts for the entire path, not just the cable ends
💡 Pro Tip
If you are used to conducted tests with cables, prepare to rethink your workflow. At mmWave, the antenna is the device, and the air interface is part of the measurement. Fighting this reality will only give you bad data.
The Two Paths: OTA vs Conducted Testing
Conducted Testing (When You Can Find a Connector)
Some mmWave devices, particularly during development, include temporary RF connectors for characterization. If you have access to a connector, conducted testing gives you cleaner, more repeatable measurements. You avoid the complexity of anechoic chambers and path loss calculations.
But conducted testing at mmWave comes with its own headaches:
Connector repeatability – At 40 GHz, a slightly under-torqued connector changes your measurement
Cable stability – Flex the cable and your phase changes
Calibration drift – Systems calibrated at the connector can drift in minutes without proper power calibration
OTA Testing (The Only Path for Production)
For production devices, OTA is the only option. The 3GPP has defined several approaches:
| Method | Description | Best For |
|---|---|---|
| Direct Far-Field | Device placed at distance where phase front is planar | Large chambers, lower mmWave bands |
| Indirect Far-Field (CATR) | Uses reflector to create far-field at shorter distance | Compact chambers, higher mmWave bands |
| Reverberation Chamber | Uses mode-stirring to create statistical field | MIMO throughput testing |
The choice depends on your device size and test requirements. A 5 cm antenna array at 28 GHz needs about 50 cm for far-field. A 10 cm module needs nearly 2 meters . That is a big chamber. Many labs choose CATR to keep chamber size manageable.
| Test Method | Frequency Range | Key Components | Typical Application |
|---|---|---|---|
| Conducted (Cable) | DC – 110 GHz | VNA, cables, RF probes, calibration substrate | Device characterization, wafer-level testing |
| Direct Far-Field OTA | 24 – 43 GHz | Anechoic chamber, positioner, probe antenna | Antenna pattern, TRP, TIS |
| CATR (Indirect Far-Field) | 24 – 110 GHz | Reflector, anechoic chamber, compact range | Device conformance testing, limited space |
| Reverberation Chamber | 24 – 43 GHz | Mode-stirred chamber, signaling tester | MIMO throughput, total radiated power [citation:4] |
RF Probes for mmWave Wafer Testing
If you are characterizing mmWave devices at the wafer level, your RF probes are the most critical part of the setup. A probe that works beautifully at 20 GHz can fail completely at 40 GHz.
What to Look For in mmWave Probes
Frequency Rating
Obvious but often ignored. Buy probes rated significantly higher than your maximum test frequency. A 40 GHz probe at 39 GHz is fine. A 40 GHz probe at 41 GHz is useless.
Tip Pitch
For mmWave measurements, smaller tip pitches (under 100 µm) are often necessary to maintain signal integrity. The ground path inductance becomes critical at higher frequencies .
GSG Configuration
Ground-Signal-Ground is the standard for mmWave. It provides a clean return path and minimizes crosstalk. If your pad layout forces you into GS or SG, expect degraded performance.
Contact Material
Beryllium copper with gold plating is standard. For probing aluminum pads, tungsten tips may be necessary.
⚠️ Caution
At 110 GHz, a probe that loses just 1 dB of contact resistance can destroy your measurement. Inspect tips before every use. Clean regularly. Replace when worn. The data you save will be your own.
The Hidden Challenge: Probe-Tip Power Calibration
At mmWave frequencies, the power at your probe tip is not what you set on the VNA. Losses in cables, connectors, and the probe itself eat into your signal.
Figure 4 from a recent study shows what happens: with the VNA port set to -20 dBm, the actual power at the probe tip drops to -35 dBm at 67 GHz, then jumps to -5 dBm above 67 GHz due to the frequency extender amplifier .
This variation destroys measurement repeatability. Without probe-tip power calibration, your S-parameter measurements drift out of spec in about 10 minutes .
The solution is a two-step calibration:
Source power calibration – Use a power meter at the probe tip to create a correction table
Receiver power calibration – Adjust the receiver gains based on the corrected source power
With this approach, researchers have extended calibration stability to over 4 hours at 110 GHz . That is the difference between testing one device per day and testing dozens.
Calibration: Moving the Reference Plane
In any RF test system, the goal is to measure the device, not the test setup. At mmWave, the setup contributes significant loss and phase shift that must be removed .
Scalar vs Vector Calibration
Scalar calibration measures only magnitude. You drive one end of the path with a signal generator and measure the other end with a power meter. The difference gives you loss. This works when phase does not matter and components are high-quality .
Vector calibration measures both magnitude and phase. This is essential for S-parameter measurements where phase accuracy matters. Methods include:
SOLT (Short-Open-Load-Thru) – Common up to 40 GHz, requires well-defined standards
LRRM (Line-Reflect-Reflect-Match) – Popular for on-wafer measurements
TRL (Thru-Reflect-Line) – Preferred at higher frequencies, requires custom standards
| Calibration Type | Frequency Range | Standards Required | Typical Application |
|---|---|---|---|
| SOLT | DC – 40 GHz | Short, Open, Load, Thru | General-purpose connectorized measurements |
| LRRM | DC – 110 GHz | Line, Reflect, Match, Thru | On-wafer probing, good accuracy |
| TRL | 10 GHz – 110+ GHz | Thru, Reflect, Line(s) | Highest accuracy, requires custom line lengths |
| Power Calibration | Any | Power meter, calibration substrate | Probe-tip power leveling [citation:9] |
Where to Place the Calibration Plane
The ideal calibration plane is as close to the device as possible. For wafer measurements, that means at the probe tips using an impedance standard substrate (ISS) .
For OTA measurements, the calibration plane moves to the antenna aperture. This requires characterizing the path loss between the probe antenna and the device position, often using a reference antenna of known gain.
💡 Pro Tip
A common mistake: calibrating at the VNA connectors, adding cables and probes, and assuming everything is fine. At mmWave, every adapter and every millimeter of cable changes the measurement. Push the calibration plane to where it matters.
Setting Up Your mmWave Measurement System
Hardware Considerations
VNA with Frequency Extenders
Most standard VNAs stop at 67 GHz. For higher frequencies, you need extenders that up-convert and down-convert signals . These add complexity but are necessary for 110 GHz work.
Cables and Connectors
At mmWave, cable selection is critical. Phase-stable cables are worth the investment. Connector types matter:
2.92 mm (K): DC to 40 GHz
2.4 mm: DC to 50 GHz
1.85 mm: DC to 67 GHz
1.0 mm: DC to 110 GHz
Probe Station
For wafer measurements, your probe station must provide:
Stable platform with minimal vibration
Precise planarity control
Temperature control if needed
Microscope with adequate magnification
Anechoic Chamber
For OTA measurements, the chamber must provide:
Sufficient isolation from external signals
Absorber material effective at your frequencies
Quiet zone large enough for your device
Positioner with adequate accuracy
Software Considerations
Modern mmWave test systems rely heavily on software for:
Waveform creation – Tools like RFmx Waveform Creator generate 5G NR test signals with the correct numerology, subcarrier spacing, and modulation
Measurement automation – Sequencers run through test cases without manual intervention
Calibration routines – Automated calibration reduces human error and improves repeatability
The NI mmWave test setup, for example, requires careful configuration of bandwidth (up to 100 MHz or more), subcarrier spacing (120 kHz typical for mmWave), and center frequency (26.5 GHz, 28 GHz, 39 GHz, etc.) .
⚠️ Caution
Do not assume your software defaults are correct for mmWave. At 28 GHz, a 100 MHz bandwidth signal has different requirements than at 3.5 GHz. Verify every setting, especially trigger configuration and bandwidth limits.
Common Measurement Challenges and Solutions
Problem 1: Poor Repeatability
Symptoms: Measurements change when you reconnect, move cables, or wait 30 minutes.
Likely causes:
Cable movement during measurement
Temperature drift
Inadequate calibration
Worn connectors or probes
Solutions:
Secure cables to prevent movement
Allow system to warm up (1 hour minimum)
Use probe-tip power calibration to extend stability
Inspect and clean all connections
Problem 2: EVM Worse Than Expected
Symptoms: Modulated measurements show higher error vector magnitude than device should produce.
Likely causes:
System noise floor too high
Incorrect calibration
Overdriving the device
Phase noise from local oscillators
Solutions:
Check system noise floor with no signal
Verify calibration at the correct plane
Reduce drive level and retest
Use better signal sources if phase noise is suspect
Problem 3: Discrepancy Between Conducted and OTA Results
Symptoms: Device looks good on the wafer but fails in OTA testing.
Likely causes:
Antenna integration issues
De-embedding errors
Different calibration planes
Environmental factors
Solutions:
Validate antenna performance separately
Ensure de-embedding accounts for all fixturing
Align calibration methods
Control temperature and humidity
💡 Pro Tip
When results disagree, trust the measurement that is closer to the device’s intended use. For production devices, OTA is the truth. For device characterization, conducted measurements with proper calibration are the truth. The discrepancy tells you something about your fixturing or your antenna design.
Practical Workflow for mmWave Device Testing
Step 1: Define Your Test Requirements
Before touching any equipment, answer these questions:
What frequency bands? (28 GHz, 39 GHz, 77 GHz?)
What measurements? (S-parameters, EVM, power, sensitivity?)
What environment? (Wafer, module, finished device?)
What accuracy? (Production pass/fail or characterization?)
Step 2: Select Your Test Method
Based on your answers:
Wafer-level characterization → VNA + RF probes + probe station
Module-level conducted → VNA + cables + fixturing
Device-level OTA → Anechoic chamber + signaling tester + positioner
Step 3: Calibrate at the Right Plane
For wafer work: Probe-tip calibration using ISS
For OTA work: Path loss characterization using reference antenna
Include power calibration if testing at multiple power levels
Step 4: Verify Before Measuring
Check your setup with a known-good device or calibration standard before testing production devices. If the verification fails, your calibration is wrong.
Step 5: Measure and Monitor
Run your measurements while monitoring:
Temperature stability
Cable position
Probe contact resistance
Reference channel flatness
If anything drifts, stop and recalibrate.
REECUFUTURE’s Approach to mmWave Testing Support
We talk to engineers every week who are setting up mmWave test systems for the first time. The questions are always the same: “Which probes do I need?” “How do I calibrate at 80 GHz?” “Why does my EVM look terrible?”
Our response starts with understanding your actual test requirements, not selling you the most expensive option. A 5G module developer testing at 28 GHz needs different probes and calibration than a radar team working at 77 GHz. A production test line needs ruggedized components that prioritize cycle life over ultimate precision.
We offer RF probes rated to 110 GHz, calibration substrates for various pitch sizes, and custom cable assemblies designed for phase stability. But more importantly, we help customers think through the trade-offs before they buy. Because the right mmWave test setup is not the one with the highest specifications. It is the one that gives you repeatable, accurate measurements for your specific devices.
Final Thoughts
Testing 5G mmWave devices is challenging, but the challenges are understood and solvable. You need:
The right test method (conducted or OTA) for your development stage
RF probes rated for your frequencies with proper tip geometry
Calibration that extends to the measurement plane, not just the instrument connectors
Stable hardware and well-characterized software
Patience to verify and monitor your measurements
The engineers who succeed at mmWave testing are not the ones with the most expensive equipment. They are the ones who understand where their measurements come from, what their calibration actually does, and when to trust their data.
Start simple. Verify everything. And when something does not make sense, look at the connections first.
Need help setting up your mmWave test system?
Stop guessing. Start measuring. Tell us your frequency range, device type, and test requirements. We will help you select the right RF probes, calibration methods, and measurement strategy. No overselling. Just engineering.
Email Our Sales Team