The standard PI and T attenuator circuits are very simple - just three resistors each.
There are a number of calculators available on the web to calculate the resistance
values for each attenuation step.
If you use a rotary switch with two wafers, you can connect the circuits between
the two decks. But that may not be optimum from a shielding perspective.
One question to ask first is how accurate you want to the attenuation to be: for
measuring relative signal levels (antenna patterns, etc.) you really want 1dB
resolution or better, meaning that even the highest attenuation sections have to
be within that accuracy. For this reason most commercial units switch a number
of 20dB (or smaller) in and out for higher values, rather than trying to make a
single pad with 50 dB of attenuation. But it is more difficult to do this with standard
rotary switches, which is why most boxes are built with DPDT toggle or slide switches,
one for each stage.
If the attenuator is just for knocking down the signal for transmitter hunting, then
you don't need that sort of accuracy. Depending on the resolution of your S-meter,
10dB steps are probably adequate, and I've found that a practical maximum is
around 60dB: beyond that the signal will get in through the case of your radio unless
you take special precautions to shield it. That's why the "Offset Attenuator" ( or
"Active Attenuator") has become so popular for transmitter hunting, as it avoids that
problem.) But you can probably develop a resistive attenuator using your switches
that takes some liberties with the input and output impedances to simplify the circuit.
The best way to shield a PI attenuator is to build it on a piece of PCB with a shield
across the center, with the series resistor passing through a snug hole in the shield
wall. Then put another shield case over the top of everything, including connecting
it to the center divider. For the T you run basically put the common junction in the
hole through the divider, but the shunt resistor has to be on one side or the other.
I've seen a number of attenuators that put each individual stage in a separate
shielded compartment, but where you really need the shielding is across the center
of each stage to keep the input signals from finding their way to the output through
One approach might be to disassemble your switch and insert a piece of PCB in place
of one of the wafers. Build the attenuator secions (preferably using Ts) passing through
holes in this shield, connected to the terminals on the wafers. You have to make sure
that the center shield is grounded, and that the coupling of signal from input to output
is minimal - you might get 40dB out of such a circuit, since it would be difficult to shield
the circuit any better than that.
There may be a better circuit, depending on the details of your switches, but shielding
is going to be a problem if you are trying to get too much attenuation in a single switch.