Measuring GHz amplifiers with Nano-VNA

[szoftveres]

The problem is familiar: one wants to characterize a GHz amplifier, but -since the Nano-VNA is putting out a 0-300MHz square wave- the low-frequency high-amplitude square wave signal drives the amplifier into non-linear mode and thereby ruining the measurement.

The idea is to attach a simple PI- high-pass filter to port 0 of the Nano-VNA and include it in the calibration. The VNA will now be able to supply high frequencies, without the large amplitude square wave signal. The measurements will only be useful above the corner frequency of the filter.

I've made some small universal microstrip boards and built the filter on one of them. The two single-turn inductors are made from the legs of a through-hole resistor; the capacitor value is 10pf. The cutoff frequency is around 440MHz.

[EggertEnjoyer123]

It might be worth it to get a more expensive but better version of the NanoVNA

Consider taking a look at the LiteVNA.

[joeqsmith]

It's not only at higher frequencies that the squarewave can cause problems.  It can also wreck havoc at lower frequencies like I demonstrated in the attached video where I look at a 1MHz amplifier.   

Then there is the unleveled drive, lack of programmable attenuators.......  Still for less than $200, you can get the original NanoVNA along with the LiteVNA which would cover a lot of experiments.   As you suggest, you just need to be aware of their shortcomings when setting up a measurement.       

[szoftveres]

Are you suggesting that the LiteVNA is actually directly achieving e.g. 1.2GHz with its on-board generator, vs. using the 3rd or 5th harmonic of a lower frequency square-wave signal?

[joeqsmith]

Are you suggesting that the LiteVNA is actually directly achieving e.g. 1.2GHz with its on-board generator, vs. using the 3rd or 5th harmonic of a lower frequency square-wave signal?

yes, read the following:

https://groups.io/g/nanovna-users/message/31888

[szoftveres]

Noted, thanks for the info about the LiteVNA.

Slight improvement on the Nano-VNA filter is the capacitance reduction, from 10pf to 6.8pf.

The reason is that the sqrt(L/C) impedance with 6.8pf is closer to 50ohms than with the 10pf capacitor (~35ohms). Consequently the corner frequency went up to ~ 530MHz.
The result is more "rounded" corner (i.e. more damping due to better matching) and less loss at higher frequencies (the worst is -3.4dB at 1GHz).

Attenuation at 300MHz is close to -20dB.