On Mon, 22 Jun 2009 09:51:15 +0100, Jim Lesurf
wrote:

In article , Brian Gaff
wrote:
Well, of course I cannot 'see' the graphs, but they do seem to not be
using real world tests, I mean not terminating things an testing
things in isolation..

That is one of the key points which the pdf does not deal with.

More generally, it says nothing about the termination and coupling at
either end of the lengths being compared. Let alone what these might be in
normal use situations.

WRT dogs in the night I can point out two things which strike me about the
graphs of poage 7 of the pdf.

A) That all the mains cables seem to show a common fall in level with
frequency at a rate of around 3dB per 100Mhz.

B) That all the mains cables show variations with frequency that indicate
the presence in the system of a pair of mismatch connectioned spaced 1 or 2
metres apart. (Hard to be precise about the distance as we have no clue as
to the propagation velocities.)

(A) looks like a common mode problem with the measurement system as it
seems doubtful that this variety of cables all show such a similar fall
with frequency.

(B) seems to indicate the the only noticable difference is that the
'PowerKord' cables have a worse match to the source and load than do the
ordinary cables.

Alas, nothing in the pdf tells us if that has any relevance in real use. No
mention is made of what the authors regard as the 'typical' mains socket
source impedance at RF, nor that of a 'typical' PSU. So for all we know, in
real use, the normal cables might reject more RF than the 'PowerKord'
examples if they happened to be a poorer match. The results depend on the
source and load used, and whose values are not specified or justified for
the context.

Hopefully, further details will allow us to assess the measured results.


Slainte,

Jim

This is all true, but of course all filters (of the non-absorptive
type) work by selective, controlled mismatch. But when that filter is
just a piece of cable, we have a situation where the attenuation is
not only unpredictable, but could quite easily result in an increase
in level when the impedance of the cable is somewhere intermediate
between the source and load impedances. In other words, all you can
say about cables used in this way is that the levels of RF will be
different at the two ends.

The overall slope of the cables (3dB per 100MHz) is about what I would
expect for a cable not designed for the transmission of RF. The
insulation will be pretty lossy, and the unshielded design will allow
a certain amount of radiation,

As to common/differential mode - who knows? Duncan doesn't describe
the experimental setup or the measurement protocol.

d