A 5-Step Signal Generator Verification Checklist I Use Before Critical RF Tests

Who This Checklist Is For

If you're an RF engineer setting up a crucial EMC pre-compliance scan, a production line final test, or a design verification for a new wireless module, you need your signal source to be reliable. I've been on the other side of that—reviewing test setups where a flawed source led to a month of rework.

This checklist is for the engineer about to push 'start' on a measurement that costs more than a few hours of lab time to redo. It's built around the realities I see in our quality audits at Rohde & Schwarz, specifically for our signal generator family (like the rohde schwarz signal generator SMB100A or the SMW200A).

Here are the 5 steps I run through—verbally, with a clipboard—before any high-stakes test.

Step 1: Confirm Calibration Status, Not Just Dates

I don't just look at the calibration sticker. I check the actual measurement drift reported in the last calibration certificate. A date is an administrative flag; the drift values tell me if the unit is stable.

Checkpoint: Is the reported frequency accuracy within 0.5 ppm? If it's at the edge of tolerance, I'd schedule a recalibration after this test. I've rejected a signal generator in a Q1 2024 audit where the certificate showed a drift of 0.9 ppm over 12 months. The vendor's cal lab said it was 'within spec' for their equipment, but for our 5G NR noise figure measurements, that drift introduced unacceptable uncertainty. We sent it back.

When I first started in quality, I assumed any 'in date' gear was fine. (I do not anymore.) That naive assumption cost us a $22,000 redo on a batch of pre-compliance measurements.

Step 2: Verify Output Power Stability Over 30 Minutes

Power instability is a silent killer. I run a simple 30-minute test: set the generator to a mid-level power (e.g., 0 dBm at 2 GHz), log the output power every minute with a calibrated power sensor (e.g., our R&S NRP-Z series), and look at the peak-to-peak deviation.

Checkpoint: Is the deviation less than ±0.1 dB? If you see more than ±0.15 dB, you've got a thermal settling issue. I wish I had tracked this more carefully on a prototype batch in 2023. The issue was a marginal Peltier cooler in the source module. Without this step, it looked like the DUT had intermittent gain problems. It didn't—the source was drifting.

Step 3: The Overlooked One — Phase Noise at the Modulation Bandwidth

Most checklists stop at power accuracy and harmonics. They ignore phase noise at the offset you actually modulate at. For a 64-QAM signal with a 20 MHz bandwidth, the phase noise at 100 kHz offset is critical. People check it at 1 MHz offset and call it good.

Set the signal generator to your intended carrier frequency. Use a spectrum analyzer with a low phase noise floor (like our R&S FSW) to measure the phase noise at the modulation bandwidth edge.

Checkpoint: For a typical test, you want < -130 dBc/Hz at 100 kHz offset at 2 GHz. If you only checked at 10 MHz offset, you missed the phase error contributions that degrade your EVM. I only believed this advice after ignoring it and seeing a 5% EVM bump on a test batch that we couldn't explain. The source was 'clean' by conventional metrics. It wasn't clean enough.

(This has become a talking point in my team's discussions about industry_evolution. What was best practice in 2020—checking phase noise at one offset—may not apply in 2025 when we're packing in higher-order modulations.)

Step 4: Check Harmonics and Spurious Under Load

Don't check harmonic performance into a perfect 50 Ohm load. Check it into the same cable and DUT impedance you'll use for the actual test. A mismatch can cause the source's output amplifier to generate harmonics that aren't present in the 50 Ohm test setup.

Checkpoint: Use a circulator or an attenuator close to the source if you suspect a high VSWR from the DUT. I've rejected roughly 12% of first deliveries in 2024 where the setup instructions didn't specify this. The technician dutifully set up the test, the 'source check' looked fine, but the setup failed verification due to a 2nd harmonic at -25 dBc generated by a 2:1 mismatch.

Step 5: Validate the CW vs. Modulated Level Accuracy

A signal generator can report accurate CW power but have internal attenuator tracking errors once you switch on complex modulation. I set the generator to a complex I/Q modulation (like LTE-Advanced), measure the RMS power with the power sensor, and compare it to the CW reading at the same average power.

Checkpoint: The difference should be < 0.05 dB. If it's more, the generator's crest factor compensation is off. I don't have hard data on industry-wide failure rates for this, but based on our 5 years of audits, my sense is that this affects about 5-8% of standard-performance generators. On a rohde schwarz signal generator, this error is typically negligible, but it's still worth a 2-minute check.

Common Pitfalls & Notes

• Temperature Drift: Don't run these checks cold. Let the generator warm up for 30 minutes. A cold source has an error budget you can't quantify.
• Cable Fatigue: A flexible test cable at 6 GHz can have 0.5 dB insertion loss variation just from cable movement (ugh). Use phase-stable cables or lock everything down.
• Reference Source: The internal reference oscillator drifts with age. For tests requiring < 0.1 ppm frequency accuracy, use an external 10 MHz reference.
• Pricing: For a verification setup like this (power sensor, low-phase noise analyzer, calibrated cables), budget roughly $15,000–$25,000 (based on major test equipment distributor quotes, February 2024; verify current pricing). It's not a line item you want to skip.