VNA setup
Set your frequency sweep and estimate the complete measurement time.
Set start, stop, and step for each segment. Uncheck a segment to leave it out.
Build round-number segments across decades, with an optional linear tail. Generate replaces the current segments.
Use measured timing factors and overheads when available. Blank overheads add zero; a blank IF factor means 1.
Point and segment overhead repeat each acquisition pass. Extra measurement time is added once per complete measurement. Avoid counting the same delay twice.
| # | Start | Stop | Step | Points | Last point | Δf/f start → stop | Avg points / decade | Points / decade start → stop |
|---|
Check for gaps, overlaps, duplicate points, and changes in spacing. Relative spacing suits decade tables; absolute step size suits linear tables.
| At | Transition | Step ratio | Δf/f-at-start ratio | Check |
|---|
Uses the IF bandwidth and averaging above. Enter a specified receiver floor to estimate the margin and trace noise.
Enter step or points; the other updates. The stop is included when the step divides the span: −20 dBm to −4 dBm in 0.1 dB steps is 161 points. Otherwise the sweep ends at the last point below the stop.
Order-of-magnitude only
| Item | Thumb | Why it matters |
|---|---|---|
| Acquisition time per pass | ≈ k N / IFBW | A complete corrected measurement may need multiple passes; 1001 points at 1 kHz is about 1 s per pass for k = 1 |
| IF bandwidth | ÷10 lowers the floor 10 dB and costs 10× the time | The trade you make on every single sweep |
| Averaging | N sweeps lower the floor 10 log₁₀ N | 16 averages buy 12 dB at 16× the time, the same trade as IF bandwidth |
| Trace noise | ≈ 6.1 dB × 10−SNR/20 rms | 60 dB above the floor is 0.006 dB rms; 20 dB above is 0.6 dB and the trace looks furry |
| More points | change resolution, not the noise floor | Adding points will not clean up a noisy trace |
| Narrow features | at least 5 points across the 3 dB bandwidth | Fewer and you will under-report the depth of a notch |
| Long cables or fixtures | Δf < 1/(2τ) | Otherwise the phase wraps between points and unwrapping is guesswork |
| Time-domain gating | range 1/Δf · resolution ≈ 1/span | 10 MHz steps see 100 ns round trip; 10 GHz of span resolves 100 ps; a segmented table cannot be transformed |
| Segmented sweeps | spend points across the DUT's range | A decade table spans 10 kHz to 10 GHz in 136 points |
N = ⌊(stop − start)/step⌋ + 1 for a linear segment: round down the number of intervals so no point passes the stop. Relative spacing Δf/f sets how well narrow features are resolved; points per decade is 1/log₁₀(1 + Δf/f). A log sweep with the same span needs ⌈ln(flast/ffirst)/ln(1 + r)⌉ + 1 points for relative spacing r. Acquisition time per pass is approximated by k N/IFBW. A complete measurement includes all acquisition passes and entered overhead; sweep averaging repeats that complete measurement. These are planning estimates, not guaranteed minimum times.
Segment count, total points, and per-segment IF bandwidth or power are instrument settings; check the analyzer's limits. This estimate assumes one IF bandwidth for all points. Automatic IF reduction, filter shape, point averaging, and other active channels are not modeled automatically.
Acquisition sequences and timing coefficients must be checked against the documentation for the instrument in use. Public examples are synthetic and do not identify an installed system.
Calculations stay in your browser. Shared links and exports include entered values; use nonsensitive examples.