开云体育

Found several articles and several answers/opinions on this topic :)

If I want to do 2 port measurement in a setup where the input and output of the DUT do NOT have common ground (let's say an RF transformer, or a simple current balun), how will internal common ground on nanoVNA ports affect measurements.

Let's be even more specific - want to do some simple testing of an HF balun. S21 for differential mode and for the common mode suppression. Just for the sake of the argument, balun will be wound with a short (comparing to lambda) piece of coaxial cable, either wound in air or on ferrite toroid.

In the CMR mode I'l short both sides of the balun, connect to "hot" ends of both nanoVNA ports, and "rely" on the grounds being connected internally - is that correct setup?

In the differential mode how will that internal common ground affect measurements as it will effectively short the current that is supposed to go through the shield of the coaxial cable making the balun?

So, in one case (CMRR measurement) that internal common ground helps. but in the other (differential) it does not.

All would be easier if ports on nanoVNA are not bonded internally and I have an option to connect them or not :)

I measure through loss
And cm surpression
In both cases a common ground does not hurt so i connect both grounds together
And i measure swr
with the balun loaded with its nominal impedance.. 50 ohms.. Or 200 (for a 1 to 4).. Or for whatever it is made
Those three tests give you a good idea how well your balun or choke works
How much loss it has
How good it suppresses cm current
And in what frequency range you can use it... Whatelse do you need to know?? Maybe how much power it can handle.. Grin
Greetz sigi dg9bfc

Am 07.07.2022 22:25 schrieb "Miro, N9LR via groups.io" <m_kisacanin@...>:

[Siegfried said: In both cases a common ground does not hurt so i connect both grounds together]

That's what I was hoping for :)

That common/internal ground (bonding) can't affect when measuring against 50ohm load (1 port measurement), but I was concerned how will that affect S21 (through loss).

Any "science" beyond empirical finding that internal bonding won't hurt? A gut filling tells me that there much be more to it when I have two ground paths in parallel - one internal and one through DUT.

When it comes to "how much power" - that would be great for another release - nanoVNA with legal power output (1k5W) and temperature sensor built in :) + other port that can handle these levels too! And battery to support all that!

As you said, short the input terminals and the output terminals and measure S21 between the terminals with them floating. With the phase and magnitude of S21 the common mode Z can be calculated. There are files in this discussion group that do the calculation. Save the s2p data and insert it into the spreadsheets.
For insertion loss measure S11 (return loss format) looking into the DUT with the output terminals of the DUT shorted. Half the return loss is the loss thru the DUT.

I do this a little differently, for simplicity, as discussed in some of the
message threads (search for CMC, and you will find lots of methods/info
from discussions several months ago).
For insertion loss, with a coaxial coax choke like what you described, what
I care about is the loss through the center conductor when the shield is
working as a return path - so I just hook it directly to the coax port 0
and 1 connectors of the nano, and display the S21 loss curve in dB. A good
choke will have just a small insertion loss, typically less than .1 dB at
HF frequencies.
For common-mode rejection, what I care about is how much the shield
attenuates the signal - so I connect the shield only to the center
conductors of the port 0 and 1, and do another S21 loss curve. Then I see
20-40 dB of attenuation in a curve across 1-30MHz for my HF chokes. If I
want to see the impedance, I can also change the display to show R+jX and
get an estimate of the impedance, hoping for k's of ohms resistive and a
small reactive part. The measurement and values are not perfect, but is
easy to do. You can see a similar loss curve if you just hook the shield
across port 0, and display an S11 loss curve - but displaying the impedance
isn't correct in that setup, since it is the impedance of the reflection,
not of the choke.

Some really good suggestions, but let me highlight "my problem" here :)

I'll only focus on measuring S21 in differential mode using 2 port setup!

On the port 0 I connect shield and center conductor (as with any coaxial transmission line), on the other port I do the same. Simple connection, nothing fancy. Shield to shield, hot to hot :)

Now I have internally bonded ground between shields at Port 0 and Port 1, and i ALSO have shield of the coaxial cable connecting grounds of those two ports!!! THAT DOES NOT SEEM RIGHT!

Chances are that DUT (the current balun in this case) will lessen the impact of the internally bonded grounds by acting as the current balun, but what if CMRR is small - parallel ground paths will add some measuring error!

I still need to take a look at the S11 concept suggested by WB2UAQ with shorted output and 1 port measurement of S11, but that still does not give me S21 in differential mode!

So, how will internal ground bonding affect 2 port setup for measuring S21 in differential mode?

Miro, the nanovna is designed to measure this way - think of the
calibration you do for a thru S21 measurement; it calibrates the error due
to the cable and ground paths.
For measuring the choke, there is no need to worry about this issue. Other
issues are present which have a much larger effect on the
resolution/accuracy of an S21 measurement - it is quite accurate for small
values of Z, but gets less and less accurate as Z exceeds several k-ohms.
But it is entirely sufficient to get a reasonable measure of the
effectiveness of the choke, especially for the values of insertion loss and
common-mode S21 attenuation.

On Thu, Jul 7, 2022 at 7:23 PM Miro, N9LR via groups.io <m_kisacanin=
[email protected]> wrote:

Some really good suggestions, but let me highlight "my problem" here :)

I'll only focus on measuring S21 in differential mode using 2 port setup!

On the port 0 I connect shield and center conductor (as with any coaxial
transmission line), on the other port I do the same. Simple connection,
nothing fancy. Shield to shield, hot to hot :)

Now I have internally bonded ground between shields at Port 0 and Port 1,
and i ALSO have shield of the coaxial cable connecting grounds of those two
ports!!! THAT DOES NOT SEEM RIGHT!

Chances are that DUT (the current balun in this case) will lessen the
impact of the internally bonded grounds by acting as the current balun, but
what if CMRR is small - parallel ground paths will add some measuring error!

I still need to take a look at the S11 concept suggested by WB2UAQ with
shorted output and 1 port measurement of S11, but that still does not give
me S21 in differential mode!

So, how will internal ground bonding affect 2 port setup for
measuring S21 in differential mode?

Hello Mito,

In the differential mode how will that internal common ground affect

measurements as it will effectively short the current that is supposed to
go through the shield of the coaxial cable making the balun?

Now I have internally bonded ground between shields at Port 0 and Port 1,

and i ALSO have shield of the coaxial cable connecting grounds of those two
ports!!! THAT DOES NOT SEEM RIGHT!

I see sometimes a transformer when measuring the differential mode:
; DM Measurement
b).
I though don't think this solves your question... So you have a good
question.

All the best,

Victor


Op do 7 jul. 2022 om 22:26 schreef Miro, N9LR via groups.io <m_kisacanin=
[email protected]>:

Found several articles and several answers/opinions on this topic :)

If I want to do 2 port measurement in a setup where the input and output
of the DUT do NOT have common ground (let's say an RF transformer, or a
simple current balun), how will internal common ground on nanoVNA ports
affect measurements.

Let's be even more specific - want to do some simple testing of an HF
balun. S21 for differential mode and for the common mode suppression. Just
for the sake of the argument, balun will be wound with a short (comparing
to lambda) piece of coaxial cable, either wound in air or on ferrite toroid.

In the CMR mode I'l short both sides of the balun, connect to "hot" ends
of both nanoVNA ports, and "rely" on the grounds being connected internally
- is that correct setup?

In the differential mode how will that internal common ground affect
measurements as it will effectively short the current that is supposed to
go through the shield of the coaxial cable making the balun?

So, in one case (CMRR measurement) that internal common ground helps. but
in the other (differential) it does not.

All would be easier if ports on nanoVNA are not bonded internally and I
have an option to connect them or not :)

You can measure the insertion loss of a common mode choke made from coaxial cable that is air wound or wound on a ferrite core by connecting the shield end of the choke between the center conductors (hot leads) of the two connectors on the nanoVNA after doing a standard calibration of nanoVNA using the standard method.
The common grounds between the connectors will not affect the insertion loss between the center conductors of the two connectors.
The center conductor of the common mode choke does not need to be connected to anything. You are just measuring the insertion loss of the shield connections to the common mode coaxial choke.

I say that, to follow what you are discussing, which seems interesting to me, there is a cruel lack of diagrams.
F1AMM (Fran?ois)

On Fri, Jul 8, 2022 at 08:47 AM, F1AMM wrote:

I say that, to follow what you are discussing, which seems interesting to me, there is a cruel lack of diagrams.

You are right :) Here is a quick diagram (see attached) that shows:
* 2 port setup
* internally bonded grounds for two nanoVNA ports
* two "available" paths the return current can take (through DUT and through internally bonded ground)

Standard and required practice with anything RF: BOTH ENDS of the shield
of a coaxial cable must be appropriately connected at their respective
ends / ports.

Here is my setup for measuring CMRR. Notice the "ground" is corried
through the test fixture. Also see the attachment in case this site does
not support images in the text.

[image: image.png]

Dave - W?LEV



On Fri, Jul 8, 2022 at 2:23 AM Miro, N9LR via groups.io <m_kisacanin=
[email protected]> wrote:

Some really good suggestions, but let me highlight "my problem" here :)

I'll only focus on measuring S21 in differential mode using 2 port setup!

On the port 0 I connect shield and center conductor (as with any coaxial
transmission line), on the other port I do the same. Simple connection,
nothing fancy. Shield to shield, hot to hot :)

Now I have internally bonded ground between shields at Port 0 and Port 1,
and i ALSO have shield of the coaxial cable connecting grounds of those two
ports!!! THAT DOES NOT SEEM RIGHT!

Chances are that DUT (the current balun in this case) will lessen the
impact of the internally bonded grounds by acting as the current balun, but
what if CMRR is small - parallel ground paths will add some measuring error!

I still need to take a look at the S11 concept suggested by WB2UAQ with
shorted output and 1 port measurement of S11, but that still does not give
me S21 in differential mode!

So, how will internal ground bonding affect 2 port setup for
measuring S21 in differential mode?

--
*Dave - W?LEV*
*Just Let Darwin Work*


--
Dave - W?LEV

"On the port 0 I connect shield and center conductor (as with any coaxial transmission line), on the other port I do the same. Simple connection, nothing fancy. Shield to shield, hot to hot :)"

"Now I have internally bonded ground between shields at Port 0 and Port 1, and i ALSO have shield of the coaxial cable connecting grounds of those two ports!!! THAT DOES NOT SEEM RIGHT!"

Hi Miro,

With a properly connected and calibrated S21 setup, the internal ground connection between ports in the Nano is not a problem. There is almost no RF current along that path, and calibration will account for the very very small amount that may exist. I know this seems counter-intuitive.

Imagine a large double-sided PCB. The bottom of the board is a solid copper plane. The top has a single copper trace that starts at one edge and takes a big U-shape path across the board and returns to a nearby edge. Inject RF at the starting edge between the top trace and bottom plane (call it Port 1) and terminate the end edge with a resistor between the trace and bottom plane (call this Port 2).

For DC or low frequency AC, the return current path in this circuit is on the bottom copper plane directly immediately between the two ports and diffused widely over the rest of the bottom surface. As the signal frequency is increased, the electrical and magnetic fields begin to concentrate almost entirely in the small space between the top trace and the bottom copper plane. At RF frequencies, the return current path on the bottom plane is directly below the trace and NOT between the ports even if they are close.

It's the same in the Nano. With typical coax connections between the ports and through the device under test, there is almost no RF current in the ground connection between the Nano ports. Almost all of the RF return current flows thought the port connector, coax shield, and DUT.

RF return current always flows via the path of least reactance. Here's more info:


Dave NU8A

Hello all of you,

Op vr 8 jul. 2022 om 19:58 schreef DP <dpoinsett@...>:

For DC or low frequency AC, the return current path in this circuit is on
the bottom copper plane directly immediately between the two ports and
diffused widely over the rest of the bottom surface. As the signal
frequency is increased, the electrical and magnetic fields begin to
concentrate almost entirely in the small space between the top trace and
the bottom copper plane. At RF frequencies, the return current path on the
bottom plane is directly below the trace and NOT between the ports even if
they are close.

Perhaps this is related: When transporting non-DC currents in a coax, both
the shield and the core need to be transporting the TE mode, and thus the
coax has to be the return path and not the connection between the port-1
and port-2 of the NanoVNA.

RF return current always flows via the path of least reactance. Here's
more info:

Nice article!

Thanks,

Victor

The diagram from Miro F9LR is very generalized.

It is not specific for the case where the DUT is a common mode current choke (current balun).

It might be helpful to review what a common mode choke is to establish how to correctly connect it to a nanoVNA.

A common mode choke is a transformer having two seperate, isolated windings, each having a start point and an endpoint. It is intended for use in a balanced transmission line.

The nanoVNA channels are 'unbalanced' and share a common (cold) connection.

The common mode choke has 4 wires or terminals, these must be connected correctly to configure it as the DUT. The wiring arrangement is different for measuring thru performance of the common mode signal and the differential mode signal.

The 4 wires, I will call winding 1 start, winding 1 end, winding 2 start and winding 2 end.

Now the common signal means that the same signal must be applied to both windings in the same orientation or polarity. Therefore winding 1 start must be connected to winding 2 start and winding 1 end connected to winding 2 end.

Respectively, the joined wires or terminals are connected to nanoVNA channel 0 and channel 1, to measure the common mode performance.

To measure the differential mode performance, the windings must be wired in series and provide an anti-phase arrangement - that is winding 1 end is joined to winding 2 end and nano VNA connections are channel 0 & channel 1 to winding 1 start and winding 2 start.

I hope that helps

Ed G8FAX

Hello Ed,

To measure the differential mode performance, the windings must be wired
in series and provide an anti-phase arrangement - that is winding 1 end is
joined to winding 2 end and nano VNA connections are channel 0 & channel 1
to winding 1 start and winding 2 start.

I have not seen this configuration before (I use the configuration
mentioned here: ).
Any link to an article about this way of measuring DM?

All the best,

Victor

Respectfully, I think measuring a CM choke is being overly complicated in this discussion. The common mode Z is measured just as if the CM choke was an inductor or capacitor. Either measure it with an S11 measurement (shield to shield leaving the center conductor alone or short the input terminals and output terminals together) across Port 0 or do an S21 Thru method getting the insertion loss and the phase angle (input shield to port 0 center conductor and output shield to port 1 center conductor leaving the center conductor of the coax alone. If bifilar-wound, short the input terminals together at port 0 and the output terminals together at port 1 OR determine which conductor is common to in and out and use that conductor only).
If it is necessary to use port 0 and 1 to measure insertion loss just do an S21 with the CM choke connected as if the core wasn't there as the core is immaterial for a CM choke's loss (not so for a bifilar-wound CM choke because Zo is not well controlled so there will be loss due to mismatch).
73

Tnx Dave!

Nice and short write-up of what's happening.

73

Arie PA3A

Op 8-7-2022 om 19:57 schreef DP:

Actually, that's a common, but incorrect statement. Current - RF or otherwise - always flows along ALL available paths, split according to the inverse of the resistance.

To make it simple, let's say there are two paths. One has 10 ohms resistance and one has 5 ohms resistance. If you apply 10V to the circuit, 1amp will flow though the first path and 2 amps though the second path.

The same goes for milliohm paths like those found in grounds.

Jerry, AI0K

Jerry, Precisely correct, Interestingly, when I first heard that: "always takes the path of least resistance"?statement it was within the context of the propensity of a lightning strike. The tallest, or best grounded?item within range (path of least resistance) typically having a greater chance of being struck during a?lightning storm. However, with regard to normal electrical flow, and function you are 100% correct,?and make a very important point. Current?always flows along ALL available paths, split according?to the?inverse of?the resistance.