AF6AY wrote:
Hello Ian,
I don't get a chance to read Radio Communications often and didn't
see your article. As a professional in the design end, I'll offer
a few comments:
Hello Len
I beg your pardon, that was a different article by someone else. The one
that I wrote, on the same subject, was in Radcom for July 2005.
Even so, both articles were saying the same thing - and you have written
almost exactly the same again below! Nothing uncanny about that, of
course: we are all looking at the same resistors, and noticing the same
things.
There is still (needless) confusion in amateurism as to metal
film resistors' "inductance" in comparison to wire-wound
resistors which DO have considerable self-inductance.
It seems to be part of amateur radio's eternal quest to reduce the whole
of RF engineering to a series of simple one-liners, like "carbon
composition = non-inductive = good" and "spiral = inductive = bad".
The vital part that gets squeezed out is always "by how much"?
While there IS some self-inductance in metal film resistors
(due to laser-trimming and patterns of film on the usually
ceramic substrate), it is difficult as @#$%!!! to measure and
easier (but still grudge work) to model as a conductive strip
spiral-wound on the same physical dimensions.
I was able to make some measurements of (R+jX) using the N2PK VNA, with
a test jig made from PC board and copper foil, and some care in the
choice of open, short and 50R calibration standards. Sweeping from 60kHz
to 60MHz gave almost constant R, and a "good straight line" of X against
frequency. This implied that a simple series L-R model would be valid,
and that self-capacitance effects were not significant over the measured
frequency range.
For nearly all
amateur applications up to and including 6m, that won't be
noticeable. With some caveats, of course.
Exactly so.
Self-inductance of metal-film resistors will vary depending on
the manufacturer and their methods.
Yes, very much so.
All of these tubular metal film resistors are based on a cylindrical rod
of ceramic, with metal end caps connecting to the wire leads. A
continuous layer of resistive film is deposited on the surface of the
ceramic to form a continuous cylinder. A range of resistance values can
then be achieved by cutting away some of the metal film - normally they
remove a narrow spiral of material, leaving behind a broad spiral ribbon
of film.
The resistance value will depend on the width and the total number of
turns in this ribbon. The self-inductance will depend mostly on the
number of turns (along with the other body dimensions, of course). When
the number of turns gets much above about 10, the ribbon becomes quite
narrow, which makes it difficult to keep control over the resistance
tolerance. At that point, the manufacturer will switch to a base
material of higher resistivity, so the next-higher value of resistor
will drop back to having the minimum number of spiral turns; and so the
cycle repeats.
This has two important results:
1. Some values of resistor will have significantly more or less
inductance than others. All the way up the resistance range, from
typically1 ohm to 10Mohm, there will be a series of break-points where
the inductance flips between roughly the maximum and minimum possible
values.
2. These break-points will vary from one manufacturer to another - for a
given standard value of resistance, the inductance could be wildly
different (ask Elecraft about that one!)
For 3W resistors using 3-4 spiral turns, I measured about 4nH. This
correlated fairly well with the value calculated using the usual
inductance formula.
Even in the worst case of about 10 spiral turns, the inductance would
only be about 150nH, which is low enough for most RF applications up to
30MHz. For example, in a 50 ohm resistor, 150nH of series inductance
would increase the VSWR to about 1.75 at 30MHz.
But that is very much a worst case. If VSWR matters at all, you would
use a larger number of higher-value resistors connected in parallel, as
Len describes below. This would divide the effect of the inductance by
the number of resistors used.
So will construction which
adds varying self-capacitance from the end-caps (metal) holding
the wire leads. Self-capacitance is easier to measure on a Q-
Meter but is seldom over a half a pFd. That results in an
equivalent of a resistor in series with self-inductance, the
whole in parallel with self-capacitance. The effect on a
circuit depends on WHERE it is placed in the circuit.
And the frequency, of course.
Agreed with everything else too...
I've
found that carbon-composition resistors - in general - have a
slightly higher self-capacitance...but that depends on who
made them and what internal structures were involved (has to
be broken and observed if no X-Ray machine is handy).
As a dummy load consisting of many smaller resistors in series-
parallel, one can estimate the total capacitance and inductance
based on individual resistor models arranged in whatever
combination is planned. Offhand, I'd say that rarely does
that affect the dummy load's VSWR beyond 1.3 at 6m. In
arranging a series-parallel combination, there will probably
be more effect from whatever conductors' shape are in doing
the interconnects...less so if on a PCB, probably more if by
wires. A good rule-of-thumb is simply "make all connections
as short as possible, consistent with allowing air flow to
dissipate heat."
The only place to get paranoid about effects of self-inductance
and self-capacitance is in metrology. Metrology NEEDS to have
a minimum of each and to have accurate resistance values at
the rated frequencies. Everyday dummy loads for amateur radio
are far from lab-quality metrological stuff and don't need to
be in that precision range.
73, Len AF6AY
--
73 from Ian GM3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek