That circuit will need some tweaking to get right. The 30uH (which seems high to me, but maybe needed for VHF) look like high impedance to the PIN diodes, so the layout must be done correctly to ensure that the inductor doesn't load down the signal path.

You would be much better getting a single solid state switch and then putting additional SPDT on each signal path to increase the isolation.

The solid state switch is basically the same as this discrete solution but uses FETs instead of PINs. There will be series and shunt FET on each input. The selected path will turn the series FET on while the other paths will be shorted by the shunt FET.

Also, have you checked out Mini-Circuits to find ICs for switching? They are kinda expensive, but they are readily available.

For instance, this part has typically 65dB isolation at 1GHz. You would need at least two of them for a SP4T and then maybe a different SPDT to combine them. $15/piece though.

Mmm.. those look rather excellent, apart from being the wrong impedance (50 rather than 75 ohms). If I could get them without the cost doubling due to shipping, I would.

Got an easy way to cure the impedance thing?

MythTV svn trunk is still a train wreck (too many new features all at once, and a likely multi-year release cycle).

But.. I backported vdpau to 0.21-fixes (the stable stream), and so did several other people! The best backport (which I'm now using) is here.

Seems to work without much fuss on the fanless 8300GS card here, though the GPU on it is too slow to do the best "Advanced" modes at 1080i. But no problem with the "Temporal" de-interlacing modes and it still looks better than the best software method.

When playing back 1080i content, the playback chews up about 10% of a 2.6GHz Core2 CPU. It used to demand 100% with the best software method.

The guy that did this backport even has packages for Mythbuntu, so the average dude(-ess) shouldn't have to fuss about with messy source code and patches.

Cheers!

For a switch, the impedance really doesn't matter. The impedance really only matters when one port is de-selected, and the 75 ohm cable is terminated with 50 ohms. This is still 1.5 VSWR/ 14dB return loss. The extra loss in the series path should be pretty small, probably an extra 0.3dB due to mismatch.