On 4/9/2012 8:01 AM, Jim Lux wrote:
> One could spend quite a while nailing the uncertainty in the measurement
> and accounting for all the confounding factors..
> 1) If there is some mismatch in the system. with 1:1.5 on both ends,
> the power uncertainty is about 0.35dB. With 1:1.2, more like 0.07dB.
> 2) The impedance of coax isn't controlled all that tightly (is it 50
> ohms or 52 ohms nominal? What is it really?)
> 3) What about harmonic content? When you're getting to gnat's eyelash
> precision, a -20dB harmonic (which would be pretty bad) is a 1% error in
> 4) there's more...
Agreed on all counts.
Another of those complicating factors is the fact that the Zo of a
transmission line is NOT a pure resistance, nor is it constant with
frequency. Zo starts out highly reactive at audio frequencies,
gradually transitions to a resistive plus reactance with an absolute
value larger than nominal Zo, finally converging to it's nominal value
above the HF spectrum.
Driving and terminating a test sample with 50 ohms causes enough of a
mismatch at 1 or 2 MHz that there will be "ripple" in loss measurements
carefully taken over a frequency range. Been there, done that, have the
Tee-shirt. To get anything APPROACHING realistic loss data in the 1-2
MHz range you've got to be measuring at least 1,000 ft of coax, and even
then, I don't consider my measurements, carefully done at multiple
frequencies with an HP generator and HP spectrum analyzer as voltmeter,
any better than about +/- 20%. My best estimate of the loss of
Commscope 3227, #10 solid copper center, foil plus braid essentially
equivalent to LMR400, is about 0.22 dB/100 ft at 1.8 MHz. That's the
only coax I've had 1,000 ft of to measure.
73, Jim K9YC
TowerTalk mailing list