I am very impressed by the NanoV2+4 performance and user interface.  However I want to point out the limitation in measuring a low noise amplifier.  I would like to measure teh S11 and S21 of an amplifiers with 40 dB gain and output  P1dB of +12 dBm; thus an input power of <-28dBm is required.  I see the option in Stimulus/CFG Sweep//ADF4350  to reduce the  power input to the DUT by 9dB and when I select this and add a 12 dB pad on Port 1 before calibration i measure a pulsed on power of -32 dBm at 1500 MHz. S11 measures correctly at around -20dB *until  I connect Port 2 to the amplifier output - it then increases by about 10dB!* This increase in S11  does not occur if I terminate the LNA output.  As I connect Port 2 to the amplifier output I see this large change in the indicated S11.  This is occurring when the Port 2 input is > +5dBm.  My conclusion is that there is a non-linear cross talk from Port 2 to Port 1  when the Port 2 is > +5dBm.  I suspect this occurs at the SW/RECV switch at the input of he AD342 mixer.  The switch isolation fails at >+5 dBm input from Port 2.

I suggest that Rev 3 of the Nanovna include larger attenuation steps in the source power - perhaps steps of 0, 10,  20,  and 30dB.  The high level cross talk problem I describe above should be corrected.  A specification of maximum power into a Port is needed both from a criteria of damage and of correct operation.

use an attenuator on port 2 ... lower input = lower crosstalk

with an amp with 40 db ... you have to have also around 40 db in line
(say 20 db on both ports)

dg9bfc sigi

Am 13.07.2021 um 02:19 schrieb Weinreb, Sander:

> I am very impressed by the NanoV2+4 performance and user interface. 
> However I want to point out the limitation in measuring a low noise
> amplifier.  I would like to measure teh S11 and S21 of an amplifiers
> with 40 dB gain and output  P1dB of +12 dBm; thus an input power of
> <-28dBm is required.  I see the option in Stimulus/CFG Sweep//ADF4350 
> to reduce the  power input to the DUT by 9dB and when I select this
> and add a 12 dB pad on Port 1 before calibration i measure a pulsed on
> power of -32 dBm at 1500 MHz. S11 measures correctly at around
> -20dB*until I connect Port 2 to the amplifier output - it then
> increases by about 10dB!*  This increase in S11  does not occur if I
> terminate the LNA output.  As I connect Port 2 to the amplifier output
> I see this large change in the indicated S11.  This is occurring when
> the Port 2 input is > +5dBm.  My conclusion is that there is a
> non-linear cross talk from Port 2 to Port 1  when the Port 2 is >
> +5dBm.  I suspect this occurs at the SW/RECV switch at the input of he
> AD342 mixer.  The switch isolation fails at >+5 dBm input from Port 2.
>
> I suggest that Rev 3 of the Nanovna include larger attenuation steps
> in the source power - perhaps steps of 0, 10,  20,  and 30dB.  The
> high level cross talk problem I describe above should be corrected.  A
> specification of maximum power into a Port is needed both from a
> criteria of damage and of correct operation.
>

I had a similar problem measuring a 50 dB LNA. I just put a 50 dB attenuator in front of LNA to assure a low input level at LNA and a level at VNA P2 similar to the level of the calibration. With this configuration you can measure LNA Gain but not LNA S11.

Putting an attenuator at P1 output (and calibrating P1 with the attenuator) may work to measure S11 but only with a low attenuator. In fact with the attenuator on port 1 you reduce the level of the S11 reflected power and then you increase the noise of S11 measurement. I made a test with MAX power at ADF4350 and no attenuator on P1 vs MIN power at ADF4350 and 10 dB attenuator on Port 1. I measured 3 dB, 6 dB and 10 dB attenuators put on P1 (or after the 10 dB attenuator in the second case) with the other end left open: these tests represent well a 6 dB, 12 dB and 20 dB return loss device. Attached are the pictures of the 6 tests (left column in MAX power, right in MIN power+attenuator). It is clear that the S11 measurements with the second configuration are more noisy but still usable. Note the 5 dB/Div and the 10xAVG (10xAVG used also during the two calibrations).

With this configuration, the output of NanoVNA (MIN power + 10 dB Attenuator) should be around -30 dBm (at frequencies > 140 MHz) and should be useful to test LNAs up to around 30/40 dB gain (depending on P1dB > 5/15 dBm, just to remain in LNA linear region).

Of course if you need to measure S21 contemporary to S11 you have to put another attenuator at output of LNA to obtain a Nano VNA input level of about -10 dBm, then assuring linear operation of P2. In this case I'd not calibrate S21 with this second attenuator to not reduce the dynamic of the VNA but I'd just sum the attenuation value to the measured S21.

Marco.