I downloaded the S2P parameter for HMC566LP4E and it seems that the amplifier is not very stable at high frequency.
In fact, K<1 when frequency is higher than 38.5 GHz.
I tried to test a circuit with HMC566LP4E followed by a QAT-3+ (3dB attenuator) and HMC1065LP4E.
I get a self-oscillation at 38.5 GHz as well, with the same amplitude and frequency even if I disable the HMC1065LP4E.
This self-oscillation also doesn't change if I install a bandpass filter before HMC566LP4E.
Do you have any workaround for this?