Is my amplifier oscillates?
However, when I measured it with an Agilent PNA-X VNA, I found it looks weird...
The measured touchstone(s2p) file is linked below... It can be seen that both S11 and S22 is larger than 1 at some frequencies.
I think the VNA is well calibrated, and the output power of both ports are -30dBm which is the lowest level of the VNA.
Is my LNA oscillates?
I've measured it using a spectrum analyzer connected on the output port of the LNA, while a matched load is attach on the input port. And the result is the LNA doesn't oscillate.
From the s2p file that you posted the S11 and S22 looks pretty well in a range of -25dB to -30dB, and never goes positive for entire frequency range (1GHz to 4GHz).
S21 is not very good, having a maximum of only +3.6dB at 1.94 GHz.
Sorry, wrong s2p file. It was my fault.
The file is renewed. The previous one is measured with two 20dB ATT cascaded, so S11 and S22 are fine.
No, S11 and S22 are not fine because they are greater than 1 in magnitude.You LNA can oscillate but it's not guaranteed.You should check with SA upto Fmax. frequency of the transistor.
I believe your LNA is oscillating because S11 and S22 are positive in dB in some regions which are closer.
I've already test it by a SA. During the test, the input port was connected with a 50Ohm matched load while the output port was connected with the SA. I didn't find any oscillation signal only some interference signal which is -50dBm at the output port. And those signals were turned out to be GSM and WIFI signals.
My amplifier must be conditional stable, because when I unplugged the matched load at the input port, the amplifier oscillated with a 8dBm output at somewhere 1.3GHz
I don't think my amplifier oscillates when connected to a VNA. because if oscillated, the output signal must be 8dBm, not -50dBm. But, why are the S11 and S22 larger than unity?
1.3G may be GPS L band signal.
You're right, you have a conditionally stable/unstable LNA. To have a greater confidence in stability when testing on the SA, don't just connect a 50Ohm termination to the input but also short and open. Additionally I test LNA stability on the SA also with a resonant filter (highly reactive load) at the input. If LNA doesn't show any strong signals (like yours +8dBm) at the output on the SA on all four input loads, then you have (almost) stable LNA ;)
From the s2p file that you posted initially (I am attaching now again) looks that S11 and S22 are very good (< -25dB, and never go positive) , which is much better than the 20dB attenuators that you placed at the input and output.
So just providing a good 50 ohms termination at input and output of the LNA seems that your LNA gets stable.
That means, there is possible that the calibration of your VNA setup might be wrong, or bad coax cables, or something like this.
Looks pretty wonky. How about you share some photos of your design?
If you used braided coax in setup, replace it with semi-rigid coax with SMA, redo it.
YOu must have a lot of stray resonances in your setup and need shielding, ground plane, filtering ferrite beads, high SRF caps etc etc.
Really? That would be a good news to me....
Here are the s2p files of my attenuators. They all have 20dB attenuation.ATT_at_Output_Port.s2p.zipATT_at_Input_Port.s2p.zip
I found it difficult to calibrate with attenuators plugged using SOLT method, the ripple of measured S11 was even larger than 10dB.
If I directly measured my LNA without any attenuator, the S parameters would be the one in the first post. The S21 is even larger than 40dB. If it were true, the output power of the LNA would be -30dBm + 40dB = 10dBm, even larger than oscillation power.
I think -30dBm minimum output power is still too large. If only the minimum output power is lesser than -30dBm.
Stability test is not done by terminating 50 Ohm input and/or output.Instead, a technically correct stability is examined with Load Pull and Source Pull measurements. It means that, any combination of source and load impedances together shouldn't make this amplifier unstable.I know it's hard to realize and expensive technique but it's also technically correct.I can find a combination of impedance so that you can use your amplifier as a oscillator.Its' possible and in order to prevent this phenomena the stability tests should be done correctly.