K7ITM wrote:
I've as often seen the shields between attenuator sections, instead of
in the middle of a section. I can assure you, there is plenty of
coupling across the switch itself where you can't place a shield, and
the coupling is both inductive and capacitive. Maybe use two separate
SPDT switches with a shield between them and mechanically ganged?

The attenuator that G4PMK published in 'The VHF/UHF DX Book' was based
on earlier ARRL publications. It used common low-cost slide switches,
and 1% 0.25W metal film resistors mounted directly on the switch tags
with the shortest possible lead lengths. There were screens between the
switches.

The following is a selection of results (which I didn't have time to
type out last night). The 'Theory' column is the attenuation expected
from using available resistor values, and the '1MHz' column shows the
actual DC/LF result. Then read along each line to see how the
attenuation changes with increasing frequency.

Best viewed with a fixed pitch font...

Nominal Theory 1MHz 30MHz 145MHz 432MHz
dB
1 0.98 0.98 0.97 1.00 1.27
3 3.02 2.88 2.81 2.92 3.45
10 10.08 10.06 10.10 10.09 10.56
20 19.9 20.0 20.0 19.9 19.7

That's really pretty good; in fact some of the apparent error may be due
to the network analyser that was used for the measurements.

It seems that the main effect is a small *rise* in attenuation with
frequency, probably caused by the increasing inductive reactance of the
resistors. There doesn't seem to be any problem with internal coupling
for these particular switches.

Your mileage will definitely vary. This is partly because of the switch
construction, though garden-variety slide switches are all pretty much
the same. Another source of variation is that metal-film resistors of
the same value coming from different manufacturers may have
significantly different values of series inductance. However, very small
1% metal-film resistors are the best wire-ended components you can get
for this application. (My article in RadCom a few months ago has more
information. Don't even think of using carbon composition!)


It's instructive to take apart a good microwave relay to see how they
manage high isolation and constant impedance, but check the price of
such relays before you destroy one to just have a look at it. (e.g.,
Digikey 255-1579) With such relays, SMT resistors and careful board
layout, you can make a very decent step attenuator up through UHF at
least.

You certainly can. Part of the solution seems to be to use a physically
separate SPDT switch at each end of the attenuator, rather than DPDT
type where the switches may be too close together. Some commercial step
attenuators use individual switch contacts operated by insulating
push-rods, and a long rotating cam-shaft selects the combinations
required.

Yet another option is to use complete individual attenuators, shielded
from one another inside a rotating 'turret' block.



--
73 from Ian G/GM3SEK 'In Practice' columnist for RadCom (RSGB)
http://www.ifwtech.co.uk/g3sek