New thread - was 10GHz transverter ...

It occurred after the last post that using an ADF4351 to multiply up from the LB was fruitless as:

a) it would have to be specifically programmed to take a 108MHz ref input (and not even sure it can go that high, and 

b) We're not interested in the added PN of the ADF4351 anyway

c) Its PLL would do some filtering of the reference anyway.

So I looked out and old G8ACE designed 108MHz *24 multiplier module, reprogrammed my  dual output LB for 108MHz on both outputs (to be able to get some no-compromise divider values)  and plotted the resulting 2582MHz output

Here are the plots, with a rang eof resolution bandwidths   

Phase noise is sort of constantish out to a few 100 kHz either side and, oddly, slightly lower on the LF side - but I suppose that could be due to filtering in the multiplier - it is quite high-Q.  I did a quick tweak but only superficial, it was already on that frequency.

At 2.6GHz, 300kHz ish spacing, it's around -65dbc in 100kHz so -115dBc/Hz

At 10GHz, after a further X4 multiplication that would be -103dBc/Hz

To put it in perspective, in a 3kHz SSB bandwidth at that same 300kHz spacing, PN would be -68dBc.  near enough borne-out by the third plot, actually set for 3kHz RBW. Not marvelous, is it?

On SB3SCS I use a loop bandwidth of 50kHz or thereabouts on the PLL so the further out PN is cleaned up.

This is one instance where a PLL synth will offer better results than a direct multiplier

Look at the plots and judge for yourself.

Of course, this is using just one set of LB registers to get 108MHz and unfortunately, I didn't capture the values it chose, and have now redstored mine to the frequencies I want here.