Can the X band LNA use the embeded isolator?
I just tested the 1st stage of LNA. The problem is NF very big,as following:
When there is no Wall A & Wall B,at 9.5GHz, NF=17dB, G=-1.7dB, input SMA port Return Loss(RL)=-2.1dB
When there is only Wall A, at 9.5GHz, NF=8.9dB, G=5.8dB, RL=-3.1dB
When there are both Wall A and Wall B, at 9.5GHz, NF=8.0dB, G=6.1dB, RL=-3.1dB.
My goal of the 1st stage amplifier is NF=0.9dB, and Gain>10dB.
I don't know what is the problem.
I have tested SMA connector, its insertion loss=0.1dB, and RL=17dB at 10GHz.
The isolator insertion loss <0.3dB, and RL<-25dB, the isolation >30dB. I tested the isolator with special SMA connector, because the isolator port pin is like something of stripline, the special SMA fixture is also stripline with a gap for pin to insert.
My trace width and the SMA pin width and the isolator pin width are both 0.5mm.
So I don't know what's the problem, anyone can help?
What about trying shorting the 1st isolator (probably you need a different layout) and check the S11 and NF of the LNA without it. 2nd isolator can stay in place.
How looks the input/output impedance of the isolator vs frequency on the Smith chart? If there are high complex impedances, the isolator may be fine during evaluation for insertion loss or isolation, but when is placed in front of an LNA (designed for ideal resistive 50 ohms) would introduce a missmatch, degrading NF and S11.
Hi, vfone,
thanks. I will try to measure Zin and Zout ASAP.
Hi, vfone,
I have already test the 1st isolator Zin and Zout, pls see the following graphs.
Actually I don't understand "high complex impedance" meaning, could you give more explanation?
And I have tested the input and output impedance of the 1st stage amplifier, as following:
I guess maybe resonance that degrade the NF and Gain. Installation Wall A will improve NF greatly than no installation. Although I have tried to narrowing the cavity width, but isolator is very width, I have to enlarge the cavity width. Am I right?
The isolator looks fine, having almost pure resistive impedance.
The LNA input match needs improvement, not necessarily to be exactly 50 ohms, but tune for best gain and NF. You know better the input match topology you chose, but for me looks to many distributed elements in the front of the LNA.
More distributed elements, more possibilities to be detuned due to PCB material variations, especially at high frequencies.
When you attach the walls, the input impeadance point is moving on the Smith chart. In this situation would be nice to get an EM simulation to see their influence, otherwise you have to tune somehow the circuit when is done.
I have done the EM simulation in ADS with waveguide. But my match topology is too ideal to match the real descrete components.