R
Member
A fresh VNA calibration feels definitive, but it is only a snapshot of the setup at that moment. It assumes cables, connectors and environment stay exactly as they were during calibration. In practice they do not. Cable flex changes phase and loss, particularly at higher frequencies. Connector wear, different mating torque or debris shifts repeatability, and temperature drift can move traces slowly over a session.
My routine is simple. Right after calibration, measure a known check standard in the same configuration as the device measurement, and log S11 and S21 magnitude and phase across the band, with cable positions and the adapter stack. That is the baseline. Re-measure it before critical measurements and after any cable move, reconnection or adapter change, without moving cables between the check and the device measurement.
Re-calibrate when the change is systematic, not a single noisy point. Unexpected ripple, a sustained shift in S11 or S21 beyond usual repeatability, or drift growing with frequency or time signals a real problem. A lone spike that vanishes on a repeat sweep is usually noise or a loose connection: reseat and re-check. If the shift persists, re-calibrate.
Two traps are common. First, verifying with the same kit used for calibration proves little, because errors common to both steps can cancel and hide a flawed calibration. Use an independent check standard. Second, skipping verification after swapping adapters is risky, because a new adapter changes the reference plane and can introduce mismatch that looks like device behaviour.
What check standard or routine do you trust on your bench?
My routine is simple. Right after calibration, measure a known check standard in the same configuration as the device measurement, and log S11 and S21 magnitude and phase across the band, with cable positions and the adapter stack. That is the baseline. Re-measure it before critical measurements and after any cable move, reconnection or adapter change, without moving cables between the check and the device measurement.
Re-calibrate when the change is systematic, not a single noisy point. Unexpected ripple, a sustained shift in S11 or S21 beyond usual repeatability, or drift growing with frequency or time signals a real problem. A lone spike that vanishes on a repeat sweep is usually noise or a loose connection: reseat and re-check. If the shift persists, re-calibrate.
Two traps are common. First, verifying with the same kit used for calibration proves little, because errors common to both steps can cancel and hide a flawed calibration. Use an independent check standard. Second, skipping verification after swapping adapters is risky, because a new adapter changes the reference plane and can introduce mismatch that looks like device behaviour.
What check standard or routine do you trust on your bench?
