开云体育

Jim,

I am no RF expert but you really did a great job in your response. I seriously appreciated your thoughtful and informative response. I just joined this group and must admit I am very impressed with the shared and learned knowledge available. Having worked in electronics and design most of my life I learned many people guard knowledge like gold and share it even less. I just wanted to say thank you and thank those who freely and patiently share your experience and knowledge with others.

Darren

Hi All,

I have a 6 meter amp project I've been slowly working on. It's modifying
a Larcan lo-lo 1Kw amp used in a VHF channel 4 TV station to 6 meters.?
It's based on the work of K1WHS.

I got a low pass filter/dual directional detector from W6PQL and I have
the low pass filter finished.? I used my NanoVHA H4 to sweep the
filter.? Maybe I did it right.? Here's a screenshot from the sweep. Does
it look about right or in the ballpark?


73

Stan
KM4HQE

On 7/19/22 11:44 AM, F1AMM wrote:

It puts a CW signal out at a frequency, and uses a bridge to measure the impedance (or reflection
coefficient) at that frequency by comparing the amplitude and phase of the signal from the oscillator vs whats at the device.

This part of the text (once translated into French) is not clear (understandable). Can you be more descriptive please.

There's no pulses - each measurement is made with a steady state signal.

For S11 measurements, it's like feeding a zero ohm output impedance signal generator through a 50 ohm resistor to the input of the unit under test. That forms a voltage divider.

You measure the signal generator output and the voltage (including phase) at the junction.

If the UUT is 50 ohms, then the voltage at the junction is exactly 1/2 and the phase is the same.


If the voltage isn't 1/2 or the phase difference isn't zero, then the UUT isn't 50 ohms, and you can calculate what it actually is.

That's sort of over simplified - there isn't such a thing as a perfect generator or a perfect 50 ohm resistor, and there's physical sizes involved, etc.

So what you do is measure 3 known impedances: 0, 50, and infinite.
From that you can calculate correction factors to turn raw measurements into calibrated numbers.

Furthermore, because the VNA comes from a world of microwaves, using directional couplers, the whole system is designed and the calibration algorithms are developed in terms of S parameters, which work better for those kinds of things. S parameters are defined in terms of incident and reflected waves, but that's just an alternate representation of the current and voltage at the various points.

If you were doing power engineering, you'd be working in Watts and Vars.
If you're doing low frequency systems people often work in Admittance and Impedance (e.g. Smith charts).

The S parameter heritage means that most of the papers and analytical tools is in terms of incident and reflected (a and b) waves, reflection coefficients, transmission coefficients, etc. So the calibration process tends to focus on developing ways to take your measurements of the standards and apply that mathematically to measurements of an unknown and produce calibrated S parameters, which is the de facto standard.


For instance, you'll see T parameters or ABCD chain matrices as alternate forms. I'm sure someone, somewhere, has developed calibration equations for those representations.

And there are other ways to do the measurement - A so-called 6 port analyzer requires only precise power measurements, with no phase measurement. That's handy when you need measurements for which mixers or phase detectors and such aren't available. (mm wave and terahertz, for instance)

I use the .Sp1 files to recalculate everything, which makes it possible to simulate from the frequency response of a real object (an antenna for example); for example to test an adapter, on paper (actually in Excel)

Yes, and if you want to fool with Python, scikit-rf has a whole set of tools to work with all the different representations, plot them, convert back and forth, and even do a variety of VNA calibrations.

On Mon, Jul 18, 2022 at 02:06 AM, Diane BONKOUNGOU wrote:

Hello Roger,
I have two coaxial cables that came with the NanoVNA. I cut one of them to
make the solder. I use the cable I didn't cut to do the calibration and
after connecting the one I cut, I correct the delay. That's why the delay
is so low. When I do the calibration directly at the input of the NanoVNA
and connect my SMA cable, the delay correction is difficult for me, the
phase(pink curve) is triangular like in the attached picture. When I made
the calibration at the cable I didn't cut and I connect the cable I have
cut the phase is just shifted I don't have triangular thing on the curve.
Best regards.

Diane,

Unfortunately you are not using the edelay correctly. Once you have done your SOL directly at the NanoVNA connector you connect your test cable and do NOT attach the open end to anything. You then add edelay until the trace on the Smith chart rotates back towards the far right side on the horizontal axis of the Smith Chart ( the spot where you had a dot with no cable connected). Then you connect it to your device-under-test (DUT) and make your measurement. I suggest you first try attaching the end to some SMD resistors (50 ohm, 100 ohm and 25 ohm) to make sure you get are getting reasonable SWR, Return Loss and impedance numbers. The impedance numbers may not be what you expect because at the frequencies you are using even a slight extension to the DUT will result in a reflection coefficient phase shift.

BTW - I am not a fan of using edelay to move the reference plane to the end of the cable. It works OK at HF frequencies because the cable attenuation is very low and a few cm of cable is only a tiny fraction of a wavelength. But at the frequencies you are using results may not be great. I suggest you do your calibration at the end of your test cable using SMD parts and get a better reference plane. Little pin jacks work well for me (photo attached).

My last point is that the antenna you are using is an unbalanced antenna. One side is connected to the ground plane of the PCB board it will be used on. You have to do the tests just as it will be used in the final product. The PCB ground plane layout will have a big effect on the antenna characteristics because it forms part of the antenna. You also have to concerned that you don't affect the antenna when you attach your measurement cable. You need a way to prevent the outer surface of the test cable shield from becoming part of the antenna system. You will know this is happening if results change when you grab the cable with your hand.

Roger

On Mon, Jul 18, 2022 at 12:27 AM, Observer wrote:

Its a fact, that inductance value, vary according to frequency.

Yes that is true for real-world inductors that you test. Air coils have minimal variation but those with a core of powdered iron or ferrite material have more change with ferrite having the most. (see my posted graphs in another message.)

How does the nanovna measure ?

The NanoVNA has an internal bridge and the analog signals from the bridge are digitized by an A/D and then processed by the firmware. The first thing the firmware does is calculate what is known as the S11 parameter which is a complex (real and imaginary) number that is known as the "reflection coefficient". There are mathematical formulas that take the complex reflection coefficient and convert that to other useful parameters like VSWR, Return Loss and complex impedance (R+/-jX). Using the menu you can select which parameters you want the on-screen trace to display vs frequency.

The NanoVNA can display the reactance X vs frequency as a trace and can also output impedance (R+/- jX) as a marker on the Smith chart trace. The Smith chart trace chart is very useful and as you move the marker along with the rocker switch it can display various parameters for that frequency. You select which ones in the marker menu. It does have an option to automatically display inductance for +X measurements and capacitance for -X measurements. BUT users need to be aware that these calculations make a big assumption! The calculation from reactance to inductance/capacitance uses a simple formula that assumes the reactance measured is the result of only inductance OR capacitance and not a combination of the two. This means that for an inductor that you are testing the display inductance does not take into account any parasitic capacitance in the component.

If you use an external PC program like NanoVNA Saver or NanoVNA app you can plot inductance vs frequency for a component under test. But the calculation is still done using a basic conversion from reactance to inductance. A real world inductor is actually composed of resistance, inductance and capacitance. A simplified model of a typical inductor is attached. So the measurement made is not of the "actual inductance" L of the component but the "apparent inductance" that is calculated assuming all of the reactance X is due to inductance alone. Actual inductance and apparent inductance are very close for an inductor at low frequencies but as the frequency goes up the parasitic capacitance will have considerably more effect on the reactance X and the apparent inductance will increase rapidly until the self resonant frequency (SRF) of the component is reached. After this point the component will look like a capacitor.

Attached are some plots of a mult-ilayer inductor that has considerable self capacitance. The "apparent inductance" calculated by the program is not the actual L of the component. It is possible to calculate the actual L and inter-winding capacitance C using regression analysis or by attaching another small known capacitor across the inductor, calculating the new SRF and then doing some algebra on two equations with one unknown.

Does it use the saved selected frequency range inside the nano ?

It uses whatever ever stimulus frequency range you have selected.

Roger

It puts a CW signal out at a frequency, and uses a bridge to measure the impedance (or reflection
coefficient) at that frequency by comparing the amplitude and phase of the signal from the oscillator vs whats at the device.

This part of the text (once translated into French) is not clear (understandable). Can you be more descriptive please.

I use the .Sp1 files to recalculate everything, which makes it possible to simulate from the frequency response of a real object (an antenna for example); for example to test an adapter, on paper (actually in Excel)

--
F1AMM Fran?ois

-----Message d'origine-----

De la part de Jim Lux
mardi 19 juillet 2022 15:00

No Pulse, just analysis of peaks and nulls in the response.
i.e. 1/4 wave response and 1/2 wave response are opposite.

You know, I've had AN11276 for years but never looked at it because I don't design inlays. Thanks very much for bringing it into the conversation. Of course that's the right place to look for ideas.

If you are still curious, here is an old Type 2 NDEF spec, downloaded from NFC Forum legitimately in 2013 when the specs were still public.

Thanks for the help.

On 7/19/22 5:54 AM, DougVL wrote:

On Mon, Jul 18, 2022 at 03:27 AM, Observer wrote:

How does the nanovna measure ?

I believe it works like this:
It sends a pulse of RF at the Nano's 'start frequency' and analyzes the phase and amplitude of the return signal (like an echo?). Then it sends a pulse at the next frequency step, and so on for the number of steps programmed (usually 99, I think).
For each echo pulse, it calculates the L, C, etc. values from the phase difference between the transmitted pulse and the received pulse.
To see the data that actually results, look at the contents of an s1p or s2p file that the Nano can save (maybe only using the 'NanoVNA-Saver' program). (The -F nano can save s1p/s2p files to its own internal storage.)

Not actually pulses, it's a CW measurement. It puts a CW signal out at a frequency, and uses a bridge to measure the impedance (or reflection coefficient) at that frequency by comparing the amplitude and phase of the signal from the oscillator vs whats at the device. Or, for a Through measurement comparing the signal going in the input and the signal coming out of the output.

The clever part is that one can measure some calibration standards (short, 50 ohm, open), and then the VNA can relate the measurements it makes to those you'd get with an ideal 50 ohm system, which is what it displays.

It takes a few milliseconds to do this, so it just steps through all 101 (or 201 or 401) frequencies in a couple seconds, and there's your data.

On Mon, Jul 18, 2022 at 03:27 AM, Observer wrote:

How does the nanovna measure ?

I believe it works like this:
It sends a pulse of RF at the Nano's 'start frequency' and analyzes the phase and amplitude of the return signal (like an echo?). Then it sends a pulse at the next frequency step, and so on for the number of steps programmed (usually 99, I think).
For each echo pulse, it calculates the L, C, etc. values from the phase difference between the transmitted pulse and the received pulse.
To see the data that actually results, look at the contents of an s1p or s2p file that the Nano can save (maybe only using the 'NanoVNA-Saver' program). (The -F nano can save s1p/s2p files to its own internal storage.)
--
Doug, K8RFT

On 7/19/22 5:09 AM, Miro, N9LR via groups.io wrote:

On Mon, Jul 18, 2022 at 05:40 PM, Jim Lux wrote:

I use the T1-1T-65X - 6 pin DIP package 200 kHz to 80 MHz - it's 1:1 with a
center tap. If you need 2:1 the T2-1T-65X has a secondary 2x the primary, with
a center tap.

You have a typo in the component name - it's "X65", not "65X", but great find!

Oops, yes..

I used those because I had an application where I wanted to be able to change the turns ratio, so I have a 6 pin DIP socket, and I can plug in the T1, the T4 (2:1), or the T16 (4:1).

And it's easier to solder to DIP pins if you're dead bugging it, than to some tiny SMT pad.

As someone mentioned minicircuits does have a minimum order, so one might need to do some scrounging around for a source or alternate parts if you only want one or two.

On 7/18/22 11:15 PM, Larry Martin wrote:

And of course, the RF interface spec for the chip is something you have to buy from NFC Forum. (for the princely sum of $600)

Yeah, the NFC Forum restricted those specs a while ago, for the good of mankind I'm sure. The 2011 and 2013 versions are still out in the wild on Google. But NFC Forum is not the chip company. This public NXP datasheet for the slightly newer Ntag213 includes the radio interface:

Well the data sheet doesn't say much about the electrical properties of the interface other than to say it's 50pF and compliant with the $600 spec..

However, this ap note on antenna design..



In particular Figure 5.

The document goes on to describe how to measure these antennas, including coupling with a 1 turn loop, just like you're doing.

But they talk about it in the context of making impedance measurements, not looking at the return loss compared to 50 ohms.

It's just a transformer with high leakage inductance...

So it sounds like your answer to my question is that I should SPICE model the system comprising the NR-loop and the tag. After a reality check, I should be able to deduce SPICE properties like the tag's internal capacitance, etc, based on the model, assuming I can quantify the coupling. Am I hearing you right?

That's a way.. the app note give a bunch of ways to look at it. I'm not sure I'd use SPICE.
It kind of depends on what you're trying to do. If you're trying to characterize different models of tags, it might be more a matter of turning S11 measurements in to lumped circuit equivalents, and going from there.

On Mon, Jul 18, 2022 at 05:40 PM, Jim Lux wrote:

I use the T1-1T-65X - 6 pin DIP package 200 kHz to 80 MHz - it's 1:1 with a
center tap. If you need 2:1 the T2-1T-65X has a secondary 2x the primary, with
a center tap.

You have a typo in the component name - it's "X65", not "65X", but great find!

And of course, the RF interface spec for the chip is something you have to buy from NFC Forum. (for the princely sum of $600)

Yeah, the NFC Forum restricted those specs a while ago, for the good of mankind I'm sure. The 2011 and 2013 versions are still out in the wild on Google. But NFC Forum is not the chip company. This public NXP datasheet for the slightly newer Ntag213 includes the radio interface:

It's just a transformer with high leakage inductance...

So it sounds like your answer to my question is that I should SPICE model the system comprising the NR-loop and the tag. After a reality check, I should be able to deduce SPICE properties like the tag's internal capacitance, etc, based on the model, assuming I can quantify the coupling. Am I hearing you right?

Jim Lux wrote: “google for "woodward balun balance quality 1983" “

Thanks, I no longer have academic access, so will need to buy or get site of the paper some other way. However, there might be alternatives/better solutions to pursue.

I found “Analysis and Performance of Antenna Baluns” by Lotter Kock

Victor Reijs might be interested in this. This models a balun as a delta-connected network of impedances, provides a good explanation of how the common mode current occurs (another question I think Victor has) as well as giving several means of testing, including back-back and the “Woodward” method, tabulating their usefulness.


Jim Lux wrote: “ I use the T1-1T-65X - 6 pin DIP package 200 kHz to 80 MHz - it's 1:1 with a center tap.? If you need 2:1 the T2-1T-65X has a secondary 2x the primary, with a center tap.
The 622-kk81 is a non catalog part. I'd find something that's in stock. If you need 3 isolated windings, I'm sure they've got something.”
It appears ordering direct from Mini-Circuits requires a minimum order value of $50. With shipping costs and import duties, the cost of 622-kk81 would be high, so thanks Jim for alternative part you use.

Neither part is available from Digikey or Mouser, so a look though what they have in stock to see if there is yet another alternative might be more fruitful and cost effective.
Kind regards

Ed G8FAX

On 7/18/22 1:42 PM, Larry Martin wrote:

Jim:

Can you give a sample part #, and I'll go look at the data sheet.

The tag in the photo is an NXP NTAG203 chip in a Smartrac Bullseye inlay, about 10 years old. Smartrac never published the exact part number of Ntag203 in inlay X, and doesn't really exist anymore since merging with Avery. Google shows many old hits on both devices.

And of course, the RF interface spec for the chip is something you have to buy from NFC Forum. (for the princely sum of $600)

For a active reader passive tag, don't the tags just rectify the transmitted signal to produce DC to run the internal tag chip?

Sure, which to me argues against transmission line effects within the tag, i.e. behind the secondary winding, being significant to the VNA trace.
I will read up on non-ideal RF transformers and see if anything "sparks" an idea.

It's just a transformer with high leakage inductance. That is, you can probably model it as an ideal transformer with some turns ratio, in series with a bulk inductance. Model wise, you can put the leakage L on either side of the transformer.

Typically, you'd put some C in the circuit to counter the leakage inductance.


Consider the "transmit direction"

You've got some voltage source that wants to push power to the tag. So you've got a big inductance in series with the transformer, which powers the tag. So the voltage at the tag is Xload/(Xload + Xleakage) - it's a voltage divider. To get the voltage higher, you add some C, so it's Xload/(Xload + Xleakage - Xcapacitance)

It's similar, in principle, to power factor correction; if that helps.

Same thing in the "receive direction" - the tag is pushing a voltage through the transfomer to the reader. Again, the extra leakage L is in series - it's L, so it's not lossy, but it still is a voltage drop. So if you can cancel it with a capacitor, then all is good.

Naturally, just like with PFC - a particular value of capacitor only works at one frequency - but hey, these are single frequency devices, and the "compensation" doesn't have to be perfect.

On 7/18/22 12:23 PM, Victor Reijs wrote:

With ref 113 being “O. M. Woodward, “Balance Quality Measurements on
Baluns,” IEEE Transactions on Microwave Theoiy and Techniques, vol. 31, no.
10, pp. 821-824, Oct. 1983.” I have not located this, so no further comment
on the technique at present.

Indeed that is the reference I would like to read also. Does someone have a
copy?

google for "woodward balun balance quality 1983"

The paper works through the math of using a VNA to do the measurement as a 3 port device.

I did a little research in to commercial baluns, “mini-Circuits” produce a

very wide range of rf transformers (500+), there may be something suitable
there, I have yet to explore possibilities

Yes, and they publish N-port VNA measurements. 4 port for most of the baluns.

I think the 1:1:1 T-622-KK81 can do the function.

I use the T1-1T-65X - 6 pin DIP package 200 kHz to 80 MHz - it's 1:1 with a center tap. If you need 2:1 the T2-1T-65X has a secondary 2x the primary, with a center tap.


The 622-kk81 is a non catalog part. I'd find something that's in stock. If you need 3 isolated windings, I'm sure they've got something.

I agree with Jim Lux. The inductance of an "ideal" inductor does not vary with frequency but any physical ("real world") inductor will have its inductance vary with frequency. In the case of an air coil the change will be small due to the skin effect and current distribution resulting from it but can still be measured as Jim pointed out in his reference.

In the case of an inductor made with some type of core material the change in inductance with frequency is much more noticeable especially with high permeability ferrite cores. With powdered iron or brass it will be much less than it is with a ferrite material. Attached are plots of 5 turns on a powdered iron toroid and 5 turns on a ferrite mix 43 toroid.

Roger

The answer(s) to your question depends on what you want to know about it and what type of Tuner it is.

For example, I have a homemade manual Tuner that I built many years ago, and, having learned a lot more about impedance by now, wanted to know what the characteristics of the Tuner were as far as impedance and the effect that it has when in line and in bypass. I calibrated the vna to the end of my patch cable to a 50 ohm dummy load over 2-30 MHz, then connected the load to the output of the Tuner and the cable to the input of the Tuner. It was pretty bad.... both in line and bypass. I cleaned up the wiring and improved it, although it is still not ideal.

The other way I use the vna with the Tuner is to find and record the settings for a range of frequencies in each band. For that, I use both channels displaying swr, logmag, phase and the Smith Chart. By observing the changes in the Smith Chart, I can dial in the settings for the desired frequency and tweak the response around it.. I like to try to get the logmag down to -40 dB at that frequency if I can, but that is only at the exact frequency.

Jim:

Can you give a sample part #, and I'll go look at the data sheet.

The tag in the photo is an NXP NTAG203 chip in a Smartrac Bullseye inlay, about 10 years old. Smartrac never published the exact part number of Ntag203 in inlay X, and doesn't really exist anymore since merging with Avery. Google shows many old hits on both devices.

For a active reader passive tag, don't the tags just rectify the transmitted signal to produce DC to run the internal tag chip?

Sure, which to me argues against transmission line effects within the tag, i.e. behind the secondary winding, being significant to the VNA trace.

I will read up on non-ideal RF transformers and see if anything "sparks" an idea.
Larry