Feed-through capacitors in RF

Hello, in my bridge construction http://qrp.gr/bridge/index.htm the two 1nF caps should be feed through types but I have used standard SMA connectors for taking out the DC and I have placed 1nF smd caps very close to the connectors.
Would that be ok or do these really need to be feed through?

PS. I have noticed that some small portion of AC is comming out of the DC outputs, could this be because I have not used feedthrough caps, or because of the leakage of the diodes?

Actually using an SMA connector for DC supply gives the advantage placing more decoupling capacitors in parallel (with various values to increase decoupling frequency range).
Usual values for HF, VHF, UHF range are: 100pF, 1nF, 10nF, 100nF.

Actually a good RF filter with a feedthru must be formed by a LC or RC combination. Capacitors alone are not real filters for RF signals. Therefore if you can, add inductors or small resistors in the DC plus RF line to reduce RF or AC. Commercial feedthrus include ferrite beads that form the inductors, and they specify RF loss over frequency.
A SMA connector might be "too open" for microwave signals even if you add multiple capacitors around.



If you operate diodes by feeding a RF signal to them, then you can expect harmonics to be generated as P/N junctions open and close. Then such harmonics can pass a SMA connector even with SMD capacitors blocking the opening.

No, no harmonics are generated. It is a tiny portion of the fundamental AC that pass though, and it's DC component is the wanted measurement result.
I do not think a single feedthrough, combined with a single ferrite bead, can block all frequencies this broadband bridge covers (dc to >1ghz).