As per my knowledge of VNAs this should be a very easy REFLECTION measurement.
According to the results I get, ...it is not.
I know: the setup itself is questionable due to the differences between Golia (the R3361B) and David (the miniVNA). But this is what I have in hand and it is an experiment done for the purpose of learning something, not to make actual measurements.
Equipment:
ADVANTEST R3361B Spectrum Analyser;
vanilla miniVNA (no Pro, no lite, no tiny, ...nothing);
VNA-J software versione 3.4.8;
BNC adaptors from China (!);
All VNA calibrations are made for an OPEN over a 1-180MHz range with 6k points.
To make a prove of work I first calibrated the miniVNA for REFLECTION with a M2M BNC adapter followed by a F2F adapter. The calibration plan is expected to be located at the free side of that last adapter (far left end in the picture).
 |
| Cal plane for 50ohms load |
Then a 50 Ohms load was connected….
 |
| miniVNA with a 50ohms load connected |
Scanning the range I’ve got exactly what I expected:
 |
| 50ohms VNA plot |
What is worth noting here is the constant behavior over the entire range (forgetting the spikes at the far ends of the range).
I then replaced the F2F BNC adapter with a BNC2F_M adapter and re-calibrated the VNA:
 |
| BNC replaced by F |
then I connected the whole thing to the SA:
 |
| miniVNA connected to SA |
and here are the plots i've got from the VNA with different ATTENUATOR values:
 |
VNA plot with SA input ATT = 0 dB |
 |
| VNA plot with SA input ATT = 10 dB |
 |
| VNA plot with SA input ATT = 20 dB |
The 0db and 10db look very ugly. How is that possible that the SA fronend is so bad??
I posted the question to the eevblog forum and....
Quote from an eevblog forum fellow (credits to EggertEnjoyer123):
The
problem then probably is that you have a mixer with RF, LO, and IF
inside the spectrum analyzer, and the miniVNA is looking at the RF
port. If the IF has a bandpass filter after it, then you will see a
bad match if the mixed frequency is outside the passband. The signal
that gets mixed down will get reflected back at the mixer from the IF
filter, and it will get converted back up to the same RF frequency.
So you end up getting power reflected back at the VNA, which shows up
as bad return loss. The jumps are because the LO and RF sweep at
different rates so it will appear to jump around.
Try setting your spectrum analyzer to 100 MHz or
something, zero span mode.
Try setting your spectrum analyzer to 100 MHz or something, zero span mode.
He has guessed correctly about the SA frontend topology:
 |
| R3361B frontend and IF block diagram |
So I’ve set the SA to 90MHz and 0.SPAN and the results are surprising:
 |
VNA plot with SA input ATT=0dB & 0.SPAN |
 |
VNA plot with SA input ATT=10dB & 0.SPAN |
 |
VNA plot with SA input ATT=20dB & 0.SPAN |
I posted this because it was an interesting experience, very useful to get how often you have to learn from good fellows.
Ciao.
Emanuele.
