Look, I’m not saying datasheets are lies. But between the lab where Rohde & Schwarz writes its specs and the production floor where you connect an RF signal generator to a device, there’s a quiet gap. And it’s almost never about the generator itself.
I’m a quality/compliance manager at a mid-size defense supplier. I review roughly 200 unique test setups every year. In Q1 2024 alone, I rejected 12% of first-time measurement submissions because the numbers didn’t match the expected spec. The common denominator? Not the equipment—the connection.
Here’s the thing: most engineers assume a Rohde & Schwarz signal generator delivers whatever’s on the datasheet the moment they press “Output On.” And it does—inside its own calibrated world. But the moment that signal hits a cable, a connector, or a mismatched impedance boundary, the real-world performance starts to drift.
The Surface Problem: “My Generator Says -120 dBc/Hz, but I’m Measuring -115”
You’ve seen this. You’re testing a new RF front end. The phase noise spec on your SMB100B says -120 dBc/Hz at 10 kHz offset. Your spectrum analyzer shows -115. You re-run the test. Still -115. You swap the cable. -114. Now you’re blaming the generator.
I made this exact mistake in my second year. I’d set up a signal generator, attached a brand-new SMA cable, and ran a phase noise measurement. The result was 5 dB worse than the spec. I spent a morning re-checking the generator’s calibration. Turned out the generator was fine. The problem was a 3 cm connector adapter I’d thrown on without thinking.
The Deep Cause: Connectors Are the Weakest RF Link
Let me tell you what the datasheet doesn’t show. That beautiful Rohde & Schwarz signal generator—say the SMW200A for vector signal generation—is designed to a 50 Ω reference plane at its output port. The moment you attach an SMA connector with a dirty interface, or worse, a BNC-to-SMA adapter, the impedance changes slightly. At 6 GHz, a 0.1 dB insertion loss variation can shift your power flatness by 0.5 dB. At 20 GHz, a single bad connector can ruin phase noise measurements entirely.
Here’s a number that stuck with me: In a 2023 internal audit, we found that 60% of failed measurement submissions involved connector issues. Loose torque, worn mating cycles, or the wrong gender adapter. The generator was never the root cause.
What I mean is that the signal generator’s internal VCO and PLL are extremely stable. Rohde & Schwarz spends real engineering on that. But your measurement system is only as good as your worst physical connection. And most RF labs treat connectors as an afterthought.
The Price of Ignoring It
In mid-2023, we received a batch of 18 signal generators for a new production line. The spec called for absolute power accuracy of ±0.5 dB. During verification, one unit appeared to be off by 1.2 dB. The team flagged it as a return. That would have been a $22,000 redo—our customer needed four matched units, and the suspect generator would have broken the set.
Before sending it back, I asked our lead technician to re-torque the output connector and run the verification again. The measured power accuracy came back at 0.3 dB. The connector was hand-tightened, not torque-wrenched, in the first test. The issue cost us two days of delay and nearly triggered a return—but it highlighted a deeper: we had no connector inspection protocol.
The vendor (ours, not the generator maker) claimed it was “within industry practice” to hand-tighten. I rejected that logic. Now every contract for test equipment includes a torque verification step before any measurement is accepted.
What Actually Works: A Simple Verification Workflow
I’m not going to propose a multi-page quality manual. Here’s the short version of how we solved this:
- Torque every connector. SMA connectors should be tightened to 0.9 N·m (8 in-lb). We bought a $40 torque wrench and it eliminated 80% of our measurement variation.
- Check the reference plane. Measure the generator’s output with a power meter at the port first, not after 3 meters of cable. Then characterize the cable loss separately.
- Log the mating cycles. A typical SMA connector is rated for 500 cycles. We now tag connectors with a cycle counter. After 400, it gets replaced.
In Q4 2024, we ran a blind test: 10 engineers set up the same Rohde & Schwarz SGS100A to generate a 3 GHz carrier. The group that used torque wrenches achieved a power repeatability of ±0.1 dB. The group that didn’t? ±0.6 dB. The cost of the torque wrench: $40. The time saved in re-measurements: roughly 15 hours per month.
Personally, I’d rather work with a specialist vendor who knows their signal generator’s true spec—and helps me achieve it—than a generalist who just ships a box and hopes I own the right cables. Rohde & Schwarz publishes clear application notes on connector care. Use them.
A Final Thought on Signal Generator Selection
If you’re in the market for a Rohde & Schwarz signal generator, don’t just compare phase noise and bandwidth. Ask about output connector compatibility with your existing test station. The best generator in the world will disappoint you if you plug it into a worn adapter. Between a ruggedized Toughbook and a Dell in a harsh production environment, the choice matters—but the connector between your generator and the device matters more.
Prices as of January 2025: typical Rohde & Schwarz signal generators range from $8,000 (SGS100A) to $80,000 (SMW200A) depending on frequency and options. Verify current pricing at rohde-schwarz.com.
