Chris Jones wrote:
Paul Burridge wrote:

Hi all,

I need a couple of accessories to enable me to make some phase
measurements with my HP network analyzer. I'd thought I could pick
these up on ebay easily enough, but note the lack of availably with
surprise and dismay.
I need to therefore contrive two precision parts:

Firstly, 50 ohm load that's essentially non-reactive up to 1.3Ghz.
Power handling only need be a few tens of miliwatts. N-type
connection.

Secondly, a 50 ohm power splitter (one feed-in; three outputs) N-type
connections, again flat up to 1.3Ghz. No switching needed, thankfully.

If I can't source these parts elsewhere, how feasible is it to make
them up and can anyone point me to any designs on the web that might
fit the bill?

I'm aware that the introduction of any stray reactances into the
devices will render all subsequent measurements invalid so I need to
get these parts right. At least 1.3Ghz capability should be achievable
for a hobbyist with care.

Thanks,
P.

You can make a pretty good 50 Ohm termination with a PCB-mounting SMA
connector. Cut off the centre pin of the PCB-end of the connector leaving
0.5mm or less protruding (careful of your eyes, the pin can go shooting off
pretty fast, it's hard metal), and then solder two 0.1% 100 Ohm 0603
resistors between the centre pin and the outer (ground) part of the
connector. The resistors should be diametrically opposite.

I made one with 1% resistors and got the following:
s11 -30dB up to 6GHx and
s11 -47dB up to 500MHz

It helps to tweak how flat you lie the resistors on the teflon at the back
of the connector, but without a working VNA you just have to accept what
you get.

If you want a termination with a N connector, then you could use a really
good adapter with the SMA termination I mentioned above, or work out
something similar with a N connector however I have never tried that since
I mostly use SMA anyway.

I think a very accurate / flat power divider would be fairly hard to make
well unless you can get boards made with microwave substrates. If you can
work out your measurement setup such that the flatness etc of the divider
is not so important, then that would help.

Two SMD 100R resistors in parallel is good; six 300R resistors spread
symmetrically around the centre pin would be better (if you can fit
them in).

You can buy 50R terminating resistors on a circular alumina substrate
with a hole in the middle, where the conducting pat runs from the
(metalised) rim of the inner hole to the (metalised) circumference.
Finding them isn't easy, but they do offer the theoretical minimum of
inductance.

You should be able make a pretty accurate and flat power divider in a
metal box between two bulk-head-mounted coax sockets by stringing
together 0603 or 0805 0.1% SMD resistors with bits of 22swg wire in a
neat and symmetrical birds nest. A fiddly job, and you'd probably want
to use continuous solder seams to seal the box when you were finished
to avoid creating a slot antena or two.

The 0.1% SMD resistors are available off the shelf from Farnell - IIRR
you have buy them in multiples of five, but they aren't that expensive
and if want to built a low-inductance symmetrical structure you need to
put three or four resistors in parallel (six would be nicer, but you'd
need to be a Jim Williams level artist with the soldering iron to keep
it neat).

--
Bill Sloman, Nijmegen