Read on Twitter
After finding out just how bad most of my cheap SMA cables were, it's time to make some good ones...
First up I use a cheap coax stripping tool. Clamp it on, give it a few rotations, and it should give a nice clean cut with no nicks in the centre conductor. It takes a little adjustment of the tool to get it cutting just right.
Next the centre conductor is trimmed to 2.5mm, then de-burred with a file.
I like to do a dry fit before soldering anything and make sure the pin position is correct.
With that done, I clean any debris off the end with IPA and an acid brush. Then apply some flux & solder on the pin. Any excess solder can be shaved off with a sharp scalpel.
Quick PSA on an important part of the SMA spec — the pin shoulder & dielectric must not project past the reference plane. If they do, they'll damage any connector they're mated to (this is what the gauge above checks for).
With the connectors I'm using, the pin depth is set by seating the cable against the slug of dielectric already in the connector. One problem is that the dielectric can shift as the connector is heated.
So I made this gadget. This threads into the connector and seats against the reference plane, preventing the dielectric from smooshing outwards. It also sinks some heat so the rest of the connector doesn't overheat while soldering the shield.
Using my little jig/heatsink, I can now solder on the shield making sure to keep pressure on the cable so it seats correctly.
Final gauging and the dielectric & pin position are in spec 
and performance looks good too! I can only test to 3GHz, but S11 is < -30dB, S21 > ~-0.3dB, phase stability is reasonable (for a general-use cable)
For more info, this presentation from Berkeley EECS142 is great: http://rfic.eecs.berkeley.edu/142/labs/resources/Making_SMA_Conformal_Test_Cables.pptx - that's where I learned a lot of this from
