Evaluation of Homemade
Hardline Coax
1/26/07
I have wondered
how well home-made “hardline” coax could perform as an interconnect between
PCB’s in a project enclosure. It seemed
to me that for reasonably short lengths, you could make a coaxial interconnect
with just a small copper tube, through which you insert ordinary hookup
wire. At the ends, you solder the tubing
and the wire to a PCB. Or to connect to the outside world you solder it to the
backside of a bulkhead connector.
Having built the VectorAnalyzer60,
I decided to use it to test the performance of such hardline coax made of #22
Teflon coated stranded hookup wire inserted through 3/32” copper tubing (hobby
shop stock). The fit is reasonably snug,
but the slipperiness of Teflon makes it easy to assemble.
I used 12” tubing,
which is longer than I would normally use for an interconnect but should
magnify any deficiencies. I started out
with the tubing straight, and then wound it into a spiral around a 1 ¼” wooden
dowel. I don’t anticipate actually using
a spiral for an interconnect, but used the spiral to test the effects of
bending.
Photo 1—Homemade hardline
coax. Terminated
at small PCB on end. Attached to reflection port of
VectorAnalyzer80.
The wire inside
the tubing prevented the usual kinking from bending, and made a smooth
spiral. However, the winding undoubtedly
causes some distortion to the shape of the tubing.
I attached a 49.9
ohm 0.1% resistor to the far end of the coax, between the center conductor and
tubing, and attached the near end to the VectorAnalyzer60 for a reflection
test. The results are shown in Figure 1.
Figure 1—Straight and Spiral Hardline
Coax. The magnitude of the
return loss changed slightly
when the coax was wound into a broad spiral.
The magnitude of the
return loss of both shapes is respectable to
40 MHz. Phase above 40 MHz goes wacko on the spiral, probably
because of irregularities
caused by bending.
Obviously, the
coax did not have 50-ohm impedance. If
it did, the return loss magnitude should be higher than 40 db. Nevertheless, it has respectable return loss
up to 40 MHz when used with a 50-ohm termination, even in this 12” length. For shorter lengths, higher frequencies could
be tolerated.
So what is its
actual characteristic impedance? I used
Ansoft Designer (free edition) to simulate terminated coax, and adjusted the
coax impedance until I got a magnitude curve similar to Figure 1. The result was 40.5 ohms. So I terminated the coax in 40.7 ohms (the
best I could do) and got Figure 2.
Figure 2—Homemade coax, now
spiral, with 40.7 ohms termination.
The flatness of the
magnitude indicates that 40.7 ohms is close to the
characteristic impedance
of the coax, at least to 40 MHz.
If the coax had a
characteristic impedance of 40.7 ohms, the coax and the termination together
would appear to the VectorAnalyzer60 as a 40.7 ohm resistor, and would thus
have a flat return loss of 19.8 db.
Figure 2 indicates that the coax does indeed have a characteristic
impedance near 40.7 ohms up to 40 MHz.
Beyond that, things deteriorate.
It is likely that the bending of the coax causes irregularities in the
shape of the tubing, and areas where the wire insulation is compressed. That may cause the characteristic impedance
to vary with frequency, and perhaps even to be not perfectly resistive.
So we have
approximately 40 ohm coax hardline. But is
it extremely lossy? Figure 3 shows the
reflection coefficient of the 12” hardline coax terminated in a short. Ideally, the return loss should be 0 db. Any higher return loss represents signal loss
in the round trip through the coax.
Figure 3—Return Loss of
shorted coax. The phase starts near 180
degrees and declines, as
it should. The magnitude is at worst 0.2
db,
representing 0.2 db loss
in the round trip from input to termination
and back to input. That would be only 0.1 db per foot for a
one-way trip.
Conclusion
The homemade hardline
coax as described makes a fine interconnect in a 50-ohm system at up to 40 MHz
in lengths up to 12”, even when significantly bent. For shorter lengths, the allowable frequency would
increase proportionately. Where the
input/output terminations can be made 40 ohms, the performance should increase
substantially.
Part II
As a follow-up to
the above experiments, I made another coax out of 12” of 1/16” brass tube,
through which was inserted #30 silver-plated wire-wrap wire (probably Kynar
insulation).
Figure 4—Return Loss of brass
coax terminated in 49.9 ohms.
The return loss is much
better, and based on simulation the characteristic
impedance of the coax was
about 47 ohms. Bending the coax had some
effect, but this time
there was no dramatic phase change at
higher frequencies.
Figure 4 shows
that this coax has much better return loss, and has a characteristic impedance
near 47 ohms. While bending the coax
into a spiral had some effect there is no big change in phase. Most likely, the stronger, springier brass
was less susceptible than copper to flattening during the bending process. Note that the phase changes smoothly,
notwithstanding the apparent jump at the high end. Phase plots frequently give a false
impression of dramatic shifts when the phase crosses the +/-180 degree
boundary. If the final point were
graphed as -180.5 instead of +179.5, the jump would disappear.
One might expect
that this smaller, brass coax would have significant loss. The return loss of the shorted coax is shown
in Figure 5.
Figure 5—Return loss of
shorted brass coax.
The loss is much greater
than with the larger copper coax.
Indeed, Figure 5
shows that the loss is significantly increased, having a round-trip loss of 0.8
db at 80 MHz, which is about a 10% power loss.
Thus, while this smaller coax has a characteristic impedance quite close
to 50 ohms, its usability is probably limited to shorter interconnects and
lower frequencies.