开云体育

I'm trying to measure the (frequency dependent) characteristic impedance of a 3cm stripline using a nanoVNA-H
Attached the measurement using nanoVNA-App.

Using RFsim99 I calculated that a 3cm stripline with impedance of 65ohm and a phase velocity of 1.25e+8m/s gave the same S11

How can I calculate from the data shown in nanoVNA-App to the characteristic impedance of 65ohm? Would that be of 1GHz (50 + 85)/2 (the center of the impedance circle?) is 67ohm?

So using some points (one below, one at and one above the frequency of interest?) on the impedance circle, should it be possible to calculate the stripline impedance for any frequency?
--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

Erik,

try the 1/8 lambda method?

Arie Kleingeld

Op 27-10-2020 om 16:24 schreef Erik Kaashoek:

On Tue, 27 Oct 2020 at 15:24, Erik Kaashoek <erik@...> wrote:

I'm trying to measure the (frequency dependent) characteristic impedance
of a 3cm stripline using a nanoVNA-H

Is it really stripline, and not microstrip? See difference at



Dave

I know, but I like to have a method that works for "any" impedance/length and "any" frequency combination
This because the impedance is frequency dependent and I can not modify the length or target impedance
And I want to be able to "probe" any stripline on the PCB by isolating it, connect it to the VNA at one side and put a 50ohm resistor at the other side
Maybe too ambitious?

--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

On Tue, Oct 27, 2020 at 08:58 AM, Dr. David Kirkby, Kirkby Microwave Ltd wrote:

Is it really stripline, and not microstrip?

Good point. It will be a coplanar waveguide with ground or microstrip lines with signal side ground plane (see: )

--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

On Tue, Oct 27, 2020 at 04:59 PM, Erik Kaashoek wrote:

I know, but I like to have a method that works for "any" impedance/length and
"any" frequency combination

Look at how a transmission line transforms a load impedance:


The NanoVNA measurement gives you Zin = Rs + j*Xs, the impedance seen at the input of the transmission line. You know that the line is terminated with ZL = 50 ohms (at least that's what I think to see from your data). You have to solve this equation for beta and for Z0, the impedance of the line. You might wonder how you can solve one equation for two unknowns. But note that (neglecting losses) Z0 and beta are real, while your measurement gives you a complex Zin. I.e., in reality you have to equations, one for Rs and one for Zs. Also note that the equation becomes simpler if you can terminate your line with an approximate open (ZL = ∞) or short (ZL = 0).

Regards
Christian

PS: The equation might not be solvable analytically; among other things because tan() is a periodic function. When I did similar calculations, I already had an estimate for the parameters (impedance, phase constant) and I searched for the solution numerically. In my case, that turned out to be fairly stable.

Regards
Christian

Eric,
Take a look at message 17063. It discusses how to set up a NanoVNA to do TDR in a manner that allows direct reading of the impedance along a connected line. Using the reflection times, you can also get the velocity.
Last time I looked, NanoVNA-Saver did not do impedance readout in TDR correctly, but the firmware does.
--John Gord

HI Erik,

The characteristic Impedance of a transmission line is not frequency dependent.A coax is 50 Ohm at 1MHz or at 1000MHz.A TDR will measure Zo (Impedance) as well as transitions in the time domain which can be translated to distance.A VNA will measure Zo vs frequency, but it can measure like a TDR at CW.I don't know if the software in NanoVNA supports TDR.When you measure a length of coax it is doing exactly that.Look into that function.It should display the Zo for a reasonable lenght of transmission line.3 cm is very short, I don't know if it can measure that.
Best regards,
JoseN0GU

On Tue, Oct 27, 2020 at 01:19 PM, Jose Mihotek wrote:

HI Erik,

The characteristic Impedance of a transmission line is not frequency
dependent.A coax is 50 Ohm at 1MHz or at 1000MHz.

The characteristic impedance of a transmission line does vary with frequency. As shown in the formula below the characteristic impedance is based on several cable parameters that can change with frequency. In particular R and L will change with frequency due to the skin effect. At low frequencies the change in R can be quite significant. At frequencies beyond 100 MHz. there is little change.

Attached is a plot from the program TLDetails showing calculated characteristic impedance of RG8X with frequency. another example for microstrips is discussed here:

>>

Roger

On Tue, Oct 27, 2020 at 08:24 AM, Erik Kaashoek wrote:

I'm trying to measure the (frequency dependent) characteristic impedance of a
3cm stripline using a nanoVNA-H

Eric,

The following article may be of interest to you. It measures at multiple of 1/8 wavelength and I know you want to look at frequencies other than this but it might give you several data points to compare with other methods. What is also important is how they physically attach the SMA connectors and terminated the microstrip.



Roger

On 10/27/20 1:09 PM, Jose Mihotek via groups.io wrote:

HI Erik,
The characteristic Impedance of a transmission line is not frequency dependent.A coax is 50 Ohm at 1MHz or at 1000MHz.A TDR will measure Zo (Impedance) as well as transitions in the time domain which can be translated to distance.A VNA will measure Zo vs frequency, but it can measure like a TDR at CW.I don't know if the software in NanoVNA supports TDR.When you measure a length of coax it is doing exactly that.Look into that function.It should display the Zo for a reasonable lenght of transmission line.3 cm is very short, I don't know if it can measure that.

Well... A transmission line with frequency independent construction is not frequency dependent. But there are plenty of examples of transmission lines with non-constant impedance, even at a single frequency For example, there are tapered lines that have different impedances at each end. And actual coax is also frequency dependent.

Z0 = sqrt ((R+j*omega*L)/(G + j*omega*C))


So, for things like a air dielectric coaxial line at low frequency with constant sized outer and inner conductors, L per unit length and C per unit length are fixed by the geometry .

But real life coax has non-zero R and G, and those tend to vary with frequency, the former mostly from skin effect, the latter from dielectric losses.

Real life low loss coax has very low R and G, so you can probably ignore them for most practical purposes.

However, get up to microwave frequencies and this starts to make a difference, particularly for small diameters. And, of course, there's funky delay line coax, with a helical center conductor that's ferrite loaded and has a Z of around 1000 ohms.

Best regards,
JoseN0GU
On Tuesday, October 27, 2020, 03:50:11 PM EDT, John Gord via groups.io <johngord@...> wrote:
Eric,
Take a look at message 17063.? It discusses how to set up a NanoVNA to do TDR in a manner that allows direct reading of the impedance along a connected line.? Using the reflection times, you can also get the velocity.
Last time I looked, NanoVNA-Saver did not do impedance readout in TDR correctly, but the firmware does.
--John Gord
On Tue, Oct 27, 2020 at 08:24 AM, Erik Kaashoek wrote:

I'm trying to measure the (frequency dependent) characteristic impedance of a
3cm stripline using a nanoVNA-H
Attached the measurement using nanoVNA-App.

Using RFsim99 I calculated that a 3cm stripline with impedance of 65ohm and a
phase velocity of 1.25e+8m/s gave the same S11

How can I calculate from the data shown in nanoVNA-App to the characteristic
impedance of 65ohm? Would that be of 1GHz? (50 + 85)/2 (the center of the
impedance circle?) is 67ohm?

So using some points (one below, one at and one above the frequency of
interest?) on the impedance circle,? should it be possible to calculate the
stripline impedance for any frequency?
--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

And here is the 0-2GHz scan of a 3cm coplanar stripline on FR4 that seems to confirm the impedance is changing with frequency
The marker is at 1GHz



--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

Erik - you were right on the money in your original post. Look at your S11 plot. It draws a semi-circle from 50 ohms, around some center. That center is your transmission line's characteristic impedance. I'd have guessed around 65 ohms. Check out 1/4 wavelength microstrip transformers on the web to see this effect.

The frequency at which your S11 plot hits the real axis again at around 80 ohms, is your 1/8 wavelength. If you had a good measurement higher in frequency it would continue to rotate around the chart and pass through 50 ohms again at 1/4 wavelength. It would keep going around that circle over and over every 1/4 wavelength, forever.

As to what the Vp is, you can take the physical length of the line and back calculate the relative permitivity of the dielectric and/or the phase velocity from that 1/8 wavelength frequency. Strictly speaking that's not a perfect fit but it's good enough for what most of us will ever do.

On the variation vs frequency, I don't think you'll be able to measure that. I don't know FR4 material but I make t-lines all day long and they look so close to constant impedance from 50MHz to 50GHz that I have to have a careful setup to measure the dispersion. There's some loss due to Dk vs frequency that will push your impedance up, and maybe you'll see that in the few GHz range, but it looks like you're hitting the noise floor of the equipment enough at over 1GHz that the Zo variation will be lost in the measurement accuracy.

Roger's link to Bogatin's work is a good one. I forget that at very low frequency, lower than I've ever looked at a t-line, skin effect really kicks impedance up a lot.

On Wed, Oct 28, 2020 at 08:44 AM, RayC wrote:

Erik - you were right on the money in your original post. Look at your S11
plot. It draws a semi-circle from 50 ohms, around some center. That center is
your transmission line's characteristic impedance. I'd have guessed around 65
ohms. Check out 1/4 wavelength microstrip transformers on the web to see this
effect.

ah so, it's circulating around the lines impedance then :)

Like this then ? ..

65 = sqrt(50 * 84.6)

On Tue, Oct 27, 2020 at 03:24 PM, Erik Kaashoek wrote:

I'm trying to measure the (frequency dependent) characteristic impedance of a
3cm stripline using a nanoVNA-H
Attached the measurement using nanoVNA-App.

Can you post the S1P file from that measurement for me to test with ?

On Wed, Oct 28, 2020 at 02:45 AM, OneOfEleven wrote:

Can you post the S1P file from that measurement for me to test with ?

Here it is. Tell me if you need more

For the calculation maybe take 3 points some/many MHz apart and calculate the center of the circle that goes through these three points. This enables plotting the change of impedance over frequency (if any)

--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

On Wed, Oct 28, 2020 at 10:20 AM, Erik Kaashoek wrote:

On Wed, Oct 28, 2020 at 02:45 AM, OneOfEleven wrote:

Can you post the S1P file from that measurement for me to test with ?

Here it is. Tell me if you need more

For the calculation maybe take 3 points some/many MHz apart and calculate the
center of the circle that goes through these three points. This enables
plotting the change of impedance over frequency (if any)

Yes I was wondering if I could just maybe show the mean point on the smith chart as the impedance, but that would be accurate if you have a complete single circle, otherwise the mean will be offset.

Anyway, I'll have a think and a play with your data to see what I can do.

It maybe that I just take the nearest to 50R point and the farthest from 50R point and show the average of the two as the line impedance.

And here is 20cm not so good coax
This should show a nice increase of impedance with frequency


--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

Forgot the S11 plot of the bad coax


--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK