AnsweredAssumed Answered

Practical cal without a real cal kit?

Question asked by alexw on Oct 5, 2009
Howdy, y'all,

    I'm building a physics apparatus, and I'm trying to roughly characterize my system with network analyzers up to a few GHz; it's been a learning experience. I've borrowed a 3577A/35677A (ancient, but DC to 200MHz) and an 8703A (principally optical, but using it here as an electrical VNA). However, no-one seems to have a real cal kit, of any type (it's an academic environment). I have a motley collection of SMA, BNC, and type-F adapters. I've obtained an SMA short and a few 50 ohm terminators of uncertain provenance, plus some new Mini-Circuits ANNE-50L+ (SMA term, spec'ed > 26 dB, typically > 30 dB return loss to 4GHz). I have a couple of cables that I'm told are "good", but nothing known to be cal-worthy. I have no true SMA "open". In general, my DUT will be several meters of untried coax away, for measuring S11 and S21.

    What's the best that I can do, or hope for?

    I've done full 1-port and 2-port calibrations with what I have, but being new to this, I'm not sure what are real data and what are measurement (or calibration) artifacts, though it's clear that cable movement stability is important. I've read through the whole forum; niceties like air lines or known-equal-adapters are intriguing but unavailable. As a crude test, doing a response cal on the 8703A with e.g. the unknown-quality short, then applying my SMA terminators, gives S11 of ~-40dB to 2 GHz and oscillations up to -35dB by 5GHz, which doesn't seem terrible -- but that's without long cables.

Are there steps I can take to improve my calibration, quantify my measurement uncertainty, and get the most out of the equipment I've got?  I've access to machining and fabricating, but our electrical shop has little in the way of GHz-class equipment or expertise.