I tried to use the SA621 mixers in combination with the ADF4351 (35MHz-4.4GHz) and its clear the bandwith limitation of the SA612 will anyway block going much above 1.5GHz.
As you can see in my measurements the NanoVNA can be calibrated till 1.5GHz but noise becomes an increasing problem above 1.2Ghz and between .9GHz and 1.2GHz the board design and the unbalance in the bridge due to stray capacitance starts to have an impact.
But it is possible to build a VNA using exactly the same principles as the nanoVNA but using IAM81008 mixers and 2 ADF4351 generators. ADC and dsp where done using the line input of a PC and SW running on a PC.
Due to the single ended input of the IAM81008 mixers you have to use a very wideband balun but there are many designs available. There is even a readymade 3GHz bridge on ebay for about 10$, thats the bridge I used and works till around 2.8Ghz, above there is no directivity. S21 works nicely to 4.4GHz and I did not even test going to harmonics mode.....
So 2.4GHz is out of the question with the SA612 but I was able to tune a 2.6GHz cavity filter using the ADF4351's and IAM81008 setup. No PCB, just dead bug style build
More info can be found here:
The bridge does not have to be all 50ohm. Any setup will work but you have to design for port 1 to be 50 ohm otherwise the resistor measurements will be off. The NanoVNA bridge is very well designed. The only concern I have are the two 390/50 ohm resistors towards the SA612 input. The SA612 input is differential 3kohm so why not use that high resistance to not load the bridge? I did not test the low impedance driving of the SA612 so I can not comment.
Noise from the CPU is not important. I am using an arduino zero clone for setting up the ADF4351's and never experience any noise from the zero.
Attached a picture of all the HW for the GHz VNA (except 12 volt power supply). Its not nice but it works
