Connect with us

Query on phase shift in Network Analyzer Interconnects...

Discussion in 'Electronic Design' started by Paul Burridge, Feb 11, 2006.

Scroll to continue with content
  1. The RG-214 cable I ordered from Germany to make up some patch leads
    for my VNA has finally arrived. It was advertised as "high-end audio
    cable" which is what it's commonly sold as over there (those wacky
    Germans!) but it *is* the real deal low-loss 50 ohm UHF cable,
    nevertheless (just as well since it took 3 months to arrive).

    Now I have to make up these leads with N-type connectors to replace
    the stock leads that didn't come with the analyzer. These hook up the
    VNA to the Transmission/Reflection Test Set (which I do have). I see
    from HP's tech spec that the interconnects I need to copy comprise
    what was originally sold as the "11851A RF cable kit" and it consists
    of four cables. Now here's the tricky bit:

    "Kit includes three 24" 50 ohm cables, phase-matched to 4 degrees @
    1300Mhz and one 34" cable. The 34" cable is used to phase balance the
    transmission path of the Transmission/Reflection Test Set."

    So does this mean the cable lengths are critical and if so, how best
    to go about making them up?

    thanks,
    p.
     
  2. Rich Grise

    Rich Grise Guest

    I'd brush up on my N connector soldering, so I can get it in a consistent
    place relative to the end of the cable, and just measure four cables of 24"
    minus 2X whatever the length of the rest of the N connector is, and
    possibly put them in some kind of jig.

    It doesn't sound like it's the _length_ that's critical, but the _matching_
    of those lengths.

    About the 34" cable, I have no idea what the rest of that sentence means. ;-)

    Cheers!
    Rich
     
  3. K7ITM

    K7ITM Guest

    Yes, you probably do want the cables to be the same length
    electrically, especially if you don't have an automatic way to apply a
    calibration correction, just for convenience if nothing else. 4
    degrees at 1300MHz is about 1.5mm, assuming solid polyethylene
    dielectric. You may be able to achieve that sort of matching if you
    are really careful in your construction. What sort of connectors are
    you using? If they are they type where you solder or crimp the pin to
    the center conductor and put the line into the body of the connector
    until the dielectric hits the dielectric of the connector, you should
    be able to come close to that sort of match in physical length by very
    accurately matching the length of the dielectric, and cutting it very
    cleanly perpendicular to the axis of the coax. If you're using good
    coax, the variation in velocity factor from piece to piece should be
    small, and so as long as you don't try to make really long cables, you
    should do well.

    Note that you can calibrate out lead length differences. A difference
    leads presumably to a phase error that ramps linearly with frequency.
    Does your VNA have a way to remove/add a time or at least a constant
    phase to the readout? And you can perhaps arrange to measure the
    electrical length differences. If you can do that, then you can make
    up, say, four of the 24" (nom) cables, and pick two matching ones to go
    to the two receiver inputs. Note that the length from the source to
    the coupler (T/R test set) is NOT CRITICAL at all. So make some cables
    as close to the same as you can, and pick the best-matched two for the
    receiver cables. That should let you make good reflection
    measurements. Then you can either adjust the length of the longer
    cable, or calibrate out its length error, for transmission
    measurements, assuming you care about phase. Often you'd have to
    account for the length of connectors and other cables to your DUT
    anyway.

    I've made some measurements on some rather expensive test cables and
    found the electrical length to vary more than I'd have expected, though
    they weren't matched sets. I'd hate to have to select a matched set
    from such cables, though, as it would take quite a few cables to assure
    that I'd get two that matched close enough! For several hundred
    dollars per cable, I'd have expected they'd hold them closer to the
    nominal length.

    Cheers,
    Tom

    (Note that some extra care is taken in the design and construction of
    the coupler to get accurately matched paths to the two receiver
    outputs!)
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-