开云体育

"the logmag is extremely narrow and touchy to be resonant"

Yes, that one. And you should be seeing a pretty decent return loss value.

You will be measuring S11, a one port measurement. Set the display for log magnitude dB. So the sweep should show an excellent return loss at the desired frequency set for the PI MATCH, say 7.1 MHz for the 40 meter band. The value of the return loss might be 20 dB or better, it will be clearly a notched like response. See the picture of what I obtained on my amplifier. Your sweep should be similar. I think I was running the sweep over a span of 5 or so MHz at 7, or 5 to 9 MHz for the 40 meter band.

That’s a good way to put it Roland.

Thanks for the link.

--
VE6WGM

Did you calibrate on saver and not just on the NanoVNA?

On 14/10/21 10:24 pm, Jim Lux wrote:

But more curiosity than anything else.? The idea of multiple
simultaneous measurements (no relays) is intriguing.? One could build a
N port analyzer with N sources and N^2 receivers, and choose the
stimulus frequencies carefully so that no two are at the same frequency
at the same time.? If the receiver is a wideband SDR, then conceivably,
the N^2 receivers are implemented in only N pieces of hardware, and the
selectivity is in software. With 16 bit ADCs and some averaging, one
might be able to get 80-90 dB dynamic range, or better.

Simultaneous is probably overkill — particularly because non-linearities would break it — but yes, a single receiver at each port would make sense.

There's no shame in mixing, so long as the clocks are synchronous (easy to do in a single instrument).


- Roland 9V1RT

On 14/10/21 10:14 pm, Jim Lux wrote:

What I'd be more interested in, somewhere down the road, is figuring out
whether the NanoVNA (or a similar device) can be used to make
measurements of the S22 of a power amplifier, at higher power (for use
in phased arrays, where there is mutual coupling among the antennas).

Questions I've not even begun to think about... (although, yes, directional arrays are things that I expect to want to do at low UHF within the next 12 months).


- Roland 9V1RT

Hi Alan,

OK I have been using for traces. a smith chart, and a swr line, and a log mag,
Smith is well a smith,

the SWR seems broad, where the logmag is extremely narrow and touchy to be resonant with the SWR trace.

What trace should i use to take this measurement you are describing?

Joe

On 15/10/21 1:55 am, Gregg Messenger wrote:

It is easy to come to the conclusion that the S11 phase angle as seen on the nanoVNA (CH0 PHASE 90..) is the relationship in a device under test between the voltage and current in that device, and I once fell down that hole myself.

I'd suggest that although use of the NanoVNA should not be limited to those with engineering degrees, it would be helpful for most beginners to at least understand what the S11 and S21 parameters refer to. It seems that many new users have never even heard of scattering parameters, let alone understand what they are about. As usual, Wikipedia has a good high-level article <> on this.

Anyone using a VNA — Nano or otherwise — would benefit from having read and understood about the first third of that article and perhaps at least skimmed the rest. Trying to use a VNA without understanding this would be a little like trying to drive a car without understanding what speed was, let alone what a speedometer does. It's possible to proceed without this knowledge, sure, but doing so makes life needlessly difficult.


- Roland 9V1RT

I should add, this takes the pi net as a single tuned resonator, that the match at fo is good,say 20 dB return loss and for a lossless network,
S11^2+S21^2=1. See how to get Q from a single tuned resonator by measurement of its return loss.

Hi Joe,

The Q you will measure is the operating Q and that probably is the one you want. That is to say, we have the Q on the plate side of the tube and the Q on the antenna side... that is the 50 ohm side and where you would place the VNA. So, with the pi net properly terminated on the tube plate side, use the vna and measure the return loss from its maximum best case, say 20 db until it drops to within 3 dB of its worse case. That total BW should be the operating BW of the pi net. Its value should be 1/2(Q_tube side+Q_ant side). So for example if the tube side Q were 12 and the antenna side 2, operating Q is 7. Usually the antenna side Q is quite small and the Q operating is set by the tube side Q.

I have a NanoVNA-H4,

Using it, It it possible to be able to measure the "Q" of a Pi output network of a HF frequency Amplifier?

I can measure the match it is making for between the tubes and the output just fine.

But I'd like to verify the circuits "Q"

Joe WB9SBD

On Wed, Oct 13, 2021 at 07:46 AM, Jim Lux wrote:

Sort of like if you have a transfer switch (or swap cables) and a single
1 way VNA, you make a sweep, flip the switch, make the sweep the other way.

I'm working on some python code to do the sweeps alternately, but
haven't had time to finish it.

Jim,

You are doing some interesting experiments. Explained very well and helped me understand the issues. I look forward to seeing how you progress.

Thanks - Roger

Very good. Thanks.

It is easy to come to the conclusion that the S11 phase angle as seen on the nanoVNA (CH0 PHASE 90..) is the relationship in a device under test between the voltage and current in that device, and I once fell down that hole myself. Haha! After finding that the S11 phase readings on the nanoVNA didn't make sense when thought of as representing the voltage vs current relationship in the device under test I quickly knew that I must have missed something in my understanding of how the VNA works.

After doing some studying on the fundamentals of the VNA, I found that the phase angle that the nanoVNA measures and displays as “CH0 PHASE 90” is in fact the phase angle between the incident and reflected waves being measured by the VNA at it's port.

Are they related? I once had an engineer who frequents nanoVNA forums try to emphatically tell me that there was no relationship.. and argued profusely producing charts and graphs and explaining and explaining, to no avail, how he was so certain of this. I see he has now changed his tune after having given it some time and thought, but, YES, there is a relationship.

Are they the same thing? NO.

The VNA works by producing a stimulus wave at it’s port, then measures the amplitude and phase of the returning wave that bounces back off of the device under test. From this, the VNA is able to calculate all of it’s (S11) information.

Think of it as a device used to poke at an unknown black box.. you can tap and shake the box and listen to the sound it makes and can make educated guesses as to what’s inside.. the VNA is similar, but a bit more capable than simply tapping and listening and guessing.

The VNA looks at the amplitude and phase of the signal that comes back, in relation to the signal it has sent out. The S11 phase displayed on the nanoVNA is simply this relationship between those two things, however, this is not the same as the phase relationship between the voltage and current in the device under test.

If you wish to read a bit more on how the S11 phase (as displayed on the nanoVNA) and the current vs voltage phase in the device under test is mathematically related, a conversation took place here back in 2018 that I think explains it well.. (See reply #4)



It seems there are some who tend to want to confound and obfuscate the nanoVNA and it’s function in their ‘explanations’ inundating people with graphs and math to impress you with how smart they are.. with the end result being that the person asking the question winds up more confused than when they started.

Things can be explained simply and clearly if one chooses to do so . Often times, I find it's best to only use as much complexity as is necessary to sufficiently convey one's thoughts.

Regarding how the nanoVNA works.. I explain this very simply and quickly in my video, here at 1:15


--
VE6WGM

I'm still struggling with my faulty nanoVNA SAA-2N :-( In addition to the SAA-2N I have got a nanoVNA-F, a nanoVNA-H4 and a small "very tiny...". The other units are working ok, but I really want to get my SAA-2N back in business! Today I tested the S21 measurements with the SAA-2N. I measured a 18MHz bandpass filter, and the unit seems to work ok in this mode :-)
The next thing I wanted to check was the level of the "output signal" on Port 0. I also have a handy spectrum analyzer, the tinySA. I put the SAA-2N in "CW mode" at a frequency of 8MHz. Connected the tinySA and observed a lot of harmonics, but the "main peak" was at about -23dBm!!! I think this level is too low! I did the same measurement on the H4 - result -8dBm and the nanoVNA-F had +3dBm. The result on the "very tiny..." was -7dBm.
When putting this together, I my first thought was the Si5351, but when I have done the rest of the measurements, I think one of the RF-switches, MXD8641, must be broken. Can someone please tell me if my assumptions correct?

Will be grateful for any response!

Karl Jan - LA3FY

On 10/14/21 9:36 AM, alan victor wrote:

Hi Jim,

Do you think there is adequate dynamic range in this 2 vna system to measure a run of the mill device, like a 2N3904? At HF the s21 might be 20 dB while the s12 is -40 dB.

Regards, Alan

Sure, I think so.

The bare NanoVNA has about 80 dB dynamic range.? If you lose 10-15 dB due to dividers, etc. then you're at 60 dB.


It would be interesting to compare the "resistive divider" vs "accept the mismatch"

Looking at the flowgraph for a 50 ohm system for the use a T approach - the source comes out, and it sees 25 ohms (the other VNA's input and the DUT in parallel), so the voltage will be 25/75 as opposed to 50/50 (1/3 of source vs 1/2 of source) - that's 3.5 dB lower incident on the DUT (and potentially reflected back).

On the receiver port(s), the same thing applies (if the DUT is actually 50 ohms), so overall it's a 7dB reduction in power into the receiver.

If the DUT isn't 50 ohms, it could be better or worse in terms of SNR of the measurement.


if we look at the "6 dB resistive divider" then the SNR is 12dB down, but now the DUT sees closer to 50 ohms.? For what it's worth, this is an application where the isolation between the ports isn't a huge deal (it comes out in the calibration).? And for the "put 50 ohms in series" scheme you'd see the source voltage at the DUT being

( (50 || 100)/ (100 + (50 || 100) = 0.25? (e.g. 6dB)

and the voltage at the receiver being 1/2 * ((100 || 100) / (50 + (100 || 100)) = 1/2 * 50/100 = 1/4 - also 6dB

SO whether you use a 3 resistor divider or a 2 resistor scheme, you see 12dB hit in SNR.

Another interesting approach would be to rearrange some of the resistors in the NanoVNA to make it a 100 ohm device - the actual receiver input is high Z and the actual RF source is low Z.? The CH1 input is a resistive pad, it would be trivial to make it a 100 ohm input by changing the 3 resistors.? The 50 ohm bridge on the CH0 port? is also fairly easy to change.

Hi Jim,

Do you think there is adequate dynamic range in this 2 vna system to measure a run of the mill device, like a 2N3904? At HF the s21 might be 20 dB while the s12 is -40 dB.

Regards, Alan

On 10/14/21 6:03 AM, Roland Turner via groups.io wrote:

On 14/10/21 8:14 pm, Joe Smith via groups.io wrote:

Some background my be found at these two links:



/g/NanoVNAV2/topic/t_check_with_labview_nanovna/85036914?p=,,,20,0,0,0::recentpostdate/sticky,,,20,2,0,85036914,previd=9223372036854775807,nextid=1630672957048663752&previd=9223372036854775807&nextid=1630672957048663752

OK, so a cost-and-integration-effort vs. isolation tradeoff. Interesting. Thanks.

I very rarely want to be able to measure all four parameters, but did have an experience recently where the iterative process of tuning a duplexer meant many dozens of cable changes during the process and hands that were sore for days afterwards. A transfer switch — or two analysers — would have been less painful. Although in that particular case the isolation requirements were so high that a suitable transfer switch would indeed have been unworkably expensive and NanoVNAs would generally not have been sensitive enough.

Getting what one pays for...

Exactly... It's what things like 3 and 4 port analyzers were developed for - Simultaneous measurements, with high dynamic range, without needing to cable/uncable or do the math to convert multiple 2 port measurements to N port measurements, etc.

Tuning a ferrite circulator, especially with multiple ferrites, would be another "it would be nice to have a 3 port analyzer" application.


I need to build a little circuit which has a very different S11 and S22 to try it out - a single LC low pass would work nicely. Or even an asymmetric resistive divider (1 series R, 1 shunt R)


But more curiosity than anything else.? The idea of multiple simultaneous measurements (no relays) is intriguing.? One could build a N port analyzer with N sources and N^2 receivers, and choose the stimulus frequencies carefully so that no two are at the same frequency at the same time.? If the receiver is a wideband SDR, then conceivably, the N^2 receivers are implemented in only N pieces of hardware, and the selectivity is in software. With 16 bit ADCs and some averaging, one might be able to get 80-90 dB dynamic range, or better.

On 10/13/21 9:41 PM, Roland Turner via groups.io wrote:

Jim,

On 13/10/21 10:46 pm, Jim Lux wrote:

On 10/13/21 7:33 AM, btomek@... wrote:

How is the sweep synchronized in both of these devices? Unless it

doesn't have to be synchronized, but it seems to me that it does.

You have to turn off the sweep on one while making measurements on the
other. Otherwise there's interference, which shows up in the S21 trace.

Sort of like if you have a transfer switch (or swap cables) and a single
1 way VNA, you make a sweep, flip the switch, make the sweep the other way.

I'm working on some python code to do the sweeps alternately, but
haven't had time to finish it.

What benefit does this approach offer over using a DPDT/transfer switch? Is this just a cost concern?

Pretty much the cost - An extra NanoVNA is another $50, brand new - There are no $50 transfer switches, except used (maybe), and you need to make driver circuits, etc.

The cabling is going to be about the same either way (especially if you build your own resistive splitter) - the performance will be worse (less dynamic range with the pair of NanoVNAs, or source/load impedances that aren't 50 ohms, which might make your DUT unhappy).

I note for example that a transfer switch approach would solve concurrent access to multiple NanoVNAs, impedance matching, simultaneous sweeping, etc.

Sure, but either requires some coding to keep track of which measurement is being made when.

Starting and stopping the sweep is trivially easy driving it from a computer.


I was just intrigued because it was an unusual solution to a problem and decided to try it out.

The last time I needed simultaneous measurements in both directions was when I was adjusting a 3 section filter - S11 shows the resonances of the "input" side better, and S22 shows the resonances of the "output" side better. And as anyone who has adjusted a multi section filter has found, all of the adjustments interact, so it was easier seeing all 4 parameters together. My experience is almost certainly different from others, and I'm sure there are systematic ways of tuning these filters with a "one way" analyzer.


What I'd be more interested in, somewhere down the road, is figuring out whether the NanoVNA (or a similar device) can be used to make measurements of the S22 of a power amplifier, at higher power (for use in phased arrays, where there is mutual coupling among the antennas).

Have nanovna -H I think firmware 1070
Saver ver 3.9
When I measure swr on saver for a known antenna it never shows above 1.13 when I know it higher
on vna screen looks close to correct
what did I mess up and how to fix?
Thank you so much
Dave