On Mar 12, 1:37*pm, K7ITM wrote:
On Mar 12, 9:24*am, "JC" wrote:

In a lossy coax the lost energy is, I suppose, heating up the dielectric.
To try *to visualize that I stripped off 30 cm of dielectric from an old
RG58 cable and put it in a 900 W 2450 MHz standard microwave oven together
with a 100cc cup of water as dummy load.
2 minutes after switching on the water was boiling but the polyethylene was
only slightly *warmer due to the proximity to the boiling water., Can I
conclude that RG58 dielectric has no loss at 2350 MHz ?
Certainly not ( it is well known that all the PE food containers used in
such ovens are not heated ), but what is wrong in this test ? how does it
differ from the dielectric heated in an actual operating lossy cable ?
JC

Others have set you straight about most of the loss being due to
heating the conductors (I^2*R loss) rather than dielectric loss. *Look
in the thread "Two coax as substitute for open line" thread for my
posting on 25 February; it contains a formula for line loss that lets
you see how the two loss mechanisms stack up as a function of
impedance, frequency, conductor size and dielectric loss tangent.

An interesting point to note: *If you buy line of a certain impedance
and diameter, you'll note that if the line uses solid polyethylene
dielectric its loss is higher than line of otherwise the same
construction using foam polyethylene dielectric. *The reason for that
is NOT that the foam dielectric is less lossy, but rather that the
lower effective relative dielectric constant of the foam requires a
larger diameter center conductor to get the same impedance, and the
larger center conductor has lower loss.

If you assume copper conductors and dielectric with a dissipation
factor of 0.0002 (which should be close to what either polyethylene or
PTFE of high quality is, up to a few GHz), you'll find that RG-213
size coax with a 0.285" outer conductor ID and solid 0.081" inner
conductor (appropriate for solid polyethylene 50 ohm line) yields the
following _approximate_ losses, in dB/100ft:

* * * * * Total * * Copper * * Dielectric
1MHz * * *0.138 * * 0.137 * * *0.001
10MHz * * 0.437 * * 0.433 * * *0.004
100MHz * *1.383 * * 1.370 * * *0.013
200MHz * *1.957 * * 1.938 * * *0.018
500MHz * *3.094 * * 3.064 * * *0.030
1GHz * * *4.376 * * 4.334 * * *0.042
2GHz * * *6.188 * * 6.129 * * *0.059
5GHz * * *9.784 * * 9.690 * * *0.094

You can see that even at 5GHz, the dielectric loss in this particular
line is quite small compared with the copper loss. *It would be
appropriate to use a bit higher dielectric dissipation factor in the
GHz region, but even if it's ten times as large as what I used here,
the dielectric loss is less than 10% of the total, at 5GHz. *The
calculation I used here is idealized, but the non-idealities tend to
be unrelated to dielectric loss: *things like conductors that aren't
smooth copper (braid; stranded center conductor) and small variations
in impedance along the line that cause additional apparent and real
losses. *It does depend on the dielectric not becoming "contaminated,"
but modern cable construction seems to do a good job minimizing that,
if you use the cable in reasonable environments.

Cheers,
Tom

Oh, crap. Let's try that again. I looked at the table above and it
did NOT look right. Wondered why the ratio of copper to dielectric
loss didn't get worse with increasing frequency. Made a mistake in
the spreadsheet that calculated it. Should have spotted it before I
posted it. This is probably better:

Total Copper Dielectric
1MHz 0.138 0.137 0.001
10MHz 0.442 0.433 0.008
100MHz 1.454 1.370 0.084
200MHz 2.105 1.938 0.167
500MHz 3.482 3.064 0.418
1GHz 5.169 4.334 0.836
2GHz 7.800 6.129 1.671
5GHz 13.869 9.690 4.179

So the contribution of dielectric loss by the time you get to 5GHz is
significant, but not dominant if the dielectric is high quality and
uncontaminated.