Thanks Ed for the information. I need time to think about it.

At first sight your data falls in the same ball park as that generated
by program METALFLM below 500 MHz. Ball-park accuracy is all that
anybody can expect under the circumstances.

It is a fact that program METALFLM is the first of many programs I
have ever produced without the support of personal measurement
experience. I have very little practical experience of frequency
response of resistors above above 30 MHz and not very much below that.

But I DO have confidence in my assessment of L and C values as
calculated from physical resistor dimensions. Also I have confidence
in my ability to model equivalent circuits of distributed L, C and R
components.

In the end, all that's needed. are the limits of reflection
coefficients versus frequency for ordinary wire-end resistors mounted
on circuit boards.

And of course, rubbish in = rubbish out.

Cecil, is there an IEEE definition of "Ball Park Accuracy" ? I'm on
South African red. Gone off Californian white.
----
Reg, G4FGQ

By the way, the program has brought to light the fact that
board-mounted resistors in the range 100 to 400 ohms have a slightly
better frequency response than those around 50 ohms. Therefore,
wideband 50-ohm dummy loads are best made from a nunber of higher
value resistors connected in parallel. Avoid series connections. Which
is what most people do anyway.
----
RJE