Cascaded RF Amplifiers Stability in ADS

[melisandez]

I designed the setup I shared in the photo to create a 70 dB gain block. But it oscillates. I caged each amplifier with EMI shielding to prevent oscillation. I put attenuators between them to increase stability. But I couldn't make it stable.
By the way, in the measurements I made without the emi shield, I see in-band oscillation. When there is a shield, there is no in-band oscillation, but I see the gain at 42 dB levels. It is about 30 dB less than I expected. I think this also oscillates out-band, since there are filters in between, I cannot see the oscillating signal at the output.

Later, I ran stability analyses on the ADS for troubleshooting. I put additional isolators between the amplifiers, made the gnd very strong with vias, made a perfect 50 ohm match, removed the filter because there might be leakage from the SIW filter gnd and put a 50 ohm trace instead, but I can't fix it.

Has anyone encountered such a problem before?
Is it not appropriate to have such a gain structure on a single PCB?

 

[volker@muehlhaus]

What about coupling through the DC lines?

 

[melisandez]

What about coupling through the DC lines?

It may be possible as you say, but I had done a good filtering in the bias design. Also, when I use the measured s2p data taken from the evalboard of a single amplifier as de-embedded while simulating on ADS, I get the same result. However, the amplifier is stable when measured alone. Independent of the DC part.

The reverse isolation of an amplifier is 45 dB, and it starts to oscillate when the total gain exceeds approximately 45 dB. I started to think that it is due to a physical limit on the layout.