开云体育

Sorry, a typo lapsus had entered, a "-" instead of a "+":

evaluating the false formula, Instead of below

"from this we get: RC = (50 + j100 – (100 – j100)) / (50 + j100 – (100 – j100)) = -50 / (-50 + j 200)."

please read:

" from this we get: RC = (50 + j100 – (100 – j100)) / (50 + j100 + (100 – j100)) = -50 / (-50 + j 200)."

The result at the right was not influenced by the typo lapsus, however.

Sorry. I hope there were no other ones. (Let me know, if so.)

Further, instead of the so far headline "Presentation on the NanoVNA for the raileigh Radio Society"
where the above was just one point of so many, I think, we should better call the headline now
"False Return Loss calculations".

73, Hans,
DJ7BA

-----Ursprüngliche Nachricht-----
Von: [email protected] <[email protected]> Im Auftrag von DJ7BA
Gesendet: Dienstag, 2. Juni 2020 16:14
An: [email protected]
Betreff: Re: [nanovna-users] Presentation on the NanoVNA for the Raleigh Amateur Radio Society

Hi Dana,

thanks for your clarification request.
Sorry that I thought it was obvious, I was a bit too short, I think.

The following should take care, hopefully:

On



You can find a circuit diagram, that (disregarding the + and – DC signs) we will use here.

Zs and ZL each are a series of a resistance and a reactance, in the resonance example, XS and XL are reactances of of equal magnitude but opposite sign – as typical in a resonant serial LC circuit.



Let’s understand ZS as impedance of the source, (in my earlier example it is Z2 = R2 + j X2) and ZL as impedance of the Load, (in my earlier example it is Z1 = R1 + j X1) .

As was written in the simple resonance example, let be:

Generator (or Thévenin equivalent) impedance: Z2 = R2 + jX2 = 50 +j100 Ohm,
Load impedance: Z1 = R1 + jX1 = 100 –j100 Ohm.

meaning:

R1 = 50 Ohm
X1 = +100 Ohm ( represents omega * L)
R2 = 100 Ohm
X2 = -100 Ohm ( represents 1 / (omega * C)

we can use the (ATIS published) left part “false assumption” formula, neglecting also the magnitude signs, that are falsely arbitrary, too, as the reflection factor in general has a complex value, of course – just look at any Smith diagram):

as found here:





from this we get: RC = (50 + j100 – (100 – j100)) / (50 + j100 – (100 – j100)) = -50 / (-50 + j 200).

Using EXCEL (because of it’s ability to easily do complex math), this yields:

RC = 0.3333 –j 1.3333

or, if we don’t ignore the magnitude, as ATIS uses it:

RC = | 0.3333 –j 1.3333 | = 1.374

already here we see, that RC (or gamma, as it is sometimes called, too) in a passive lumped element circuit is > 1 ! That is wrong already.

Nevertheless carrying on with it, we get:

SWR = 1 + |RC| / ( 1 - | RC |) = 1 + 1.374 / (1 – 1.374) = -6.342 which, of course, is negative, the fact that the otherwise excellent article of Trevor S. Bird had quoted, not noticing that it yields a negative result, which he wanted to prove is not correct – just as you had said, too, as in a passive lumped element circuit RL must be positive.

I don’t know if you can use complex EXCEL, but I hope so.
Feel free to use – besides the simple resonance example – also any other complex values by entering these in the yellow EXCEL fields.

Just in case not - I also add a picture of the EXCEL. In some other work sheet programs, you may be able to make use of that, too.



73, Hans
DJ7BA








-----Ursprüngliche Nachricht-----
Von: [email protected] <[email protected]> Im Auftrag von Dana Whitlow
Gesendet: Dienstag, 2. Juni 2020 01:49
An: [email protected]
Betreff: Re: [nanovna-users] Presentation on the NanoVNA for the Raleigh Amateur Radio Society



Hans, I don't understand your "obvious example" circuit.

Could you please describe it in detail, perhaps expressed as a spice netlist in the body of your reply, for simplicity?



Thanks,



Dana







--
Diese E-Mail wurde von Avast Antivirus-Software auf Viren geprüft.






--
Diese E-Mail wurde von Avast Antivirus-Software auf Viren geprüft.

On this return loss discussion... sorry can't help myself...

See the attached. Hope that clarifies... again.

Alan

Thanks, Alan - wonderful source ! I love it. You made my day!

Congrats for having shown this. I can only agree.

From it's second very basic formula, (when leaving alone the '(db)',)
it is possible to derive the correct |Gamma| formula, as is used in the first one.

Resulting is:




with L for Load and S for Source, the asterisk meaning conjugate complex.

Just one little remark - knowing we cannot easily change words everybody uses:

The "incident" and "reflected" power is found on a transmission line.

At a complex impedance termination of a complex impedance source,
there is nothing reflected, however.

The power difference, instead, is that between "available" (but not always fully used)
real power and the real power actually delivered to and dissipated in the load’s real part.


In which exact IEEE document did you find it?
When was that published? Where and how can I get a copy?


Why does the probably best Smith diagram program available today,
(that I do not want to name in this context,) when using a complex source,
ignore the IEEE quote and rather stick to that false ATIS like formula?

Why does ATIS not change the false one? I think, time has come to do so.

Thanks again Alan - for this, it's the best source I ever was shown so far !


73, Hans
DJ7BA






-----Ursprüngliche Nachricht-----
Von: [email protected] <[email protected]> Im Auftrag von alan victor
Gesendet: Dienstag, 2. Juni 2020 18:31
An: [email protected]
Betreff: Re: [nanovna-users] out of "Presentation on the NanoVNA for the raileigh Radio Society", Now: "False Return Loss Calculations"



On this return loss discussion... sorry can't help myself...



See the attached. Hope that clarifies... again.



Alan











--
Diese E-Mail wurde von Avast Antivirus-Software auf Viren geprüft.

Does all of this mean that the RL on the NanoVNA and VNA Saver are wrong and should be positive??Deon

Hi everyone,

Interesting discussion. A point of clarification though is that conventionally, the definition of Gamma (reflection coefficient) is:

Gamma = (Z - Zo) / (Z + Zo) where Z is the impedance in question and Zo is the reference impedance for the calculation

The formula in the preceding post appears to have a Zs* in the numerator which is not consistent with the standard definition of Gamma. It's probably from a different context. But, as far as NanoVNA goes, I would think that any reported results should follow conventional definitions of parameters for consistency with standard measurement practices. On that note, the following article is a derivation from basic principles where formula (3.12.10) is the standard definition:
(Ellingson)/03%3A_Transmission_Lines/3.12%3A_Voltage_Reflection_Coefficient

And, for what it's worth, I couldn't resist the temptation to throw in a few points of support for some of the other posts. While the discussion seemingly is centered around an example where Zo = 100 - j100 rather than the more conventional Zo = 50, that is inconsequential to my points:

1) Since Gamma is the ratio of the output (= reflected signal) to the input (= incident signal) it is a "gain" signal as are all S-parameters. It must be < 1 for any passive circuit since a passive circuit cannot add energy.

2) By the definition of S-parameters, it turns out that Gamma = S11 (or S22). So, |S11| must also be < 1 for a passive circuit.

3) Expressing Gamma in dB's is common practice and would actually result in a return gain which for a passive circuit would have to be a negative number. Most VNA's report S11 in dB which would be 20*log10(|S11|) and it would be negative for a passive circuit. While many call this return loss, that is loose nomenclature. Strictly speaking it should really be called return gain in this context.

4) For convenience one can use return LOSS which is the negative of value in dB's or the reciprocal in linear units such as Gamma. Thus, for a passive circuit return loss would have to be a positive value, in dB. And, of course, that means it would be - 20*log10(|S11|) or equivalently + 20*log10( 1 / |S11| )

Definitely easy to get tripped up in the signs. I have probably missed the point of this discussion but just wanted to throw in my two cents. Sorry if it's gone off topic. I didn't want to add to the confusion but instead support some of the later comments on the topic.

Darrell - AF5FX

Hi Darrel,

From the formal and scientific point of view, this all can be discussed.

Well, as a radioamateur I don't really care about Plus or Minus. The nanoVNA is the most used VNA amongst hams in PA-land.
I know what loss is,? I know what gain is, I know what reflectioncoefficient etc is,? and I know how the dB-scale works.
That's enough for me to express myself to others and do my teaching to hamradio students.

:-)
My two eurocents...

73,

Arie PA3A

No, even though that could be a practical consequence. But that's not what I had in mind,
as - I think - it is possible to reverse the RL sign of the NanoVNA per gusto as individually wanted.

There are enough people of "another school", that strongly insist in the opposite of what I showed.

Trevor S. Bird, former IEEE chief editor, said, roughly a third of all RF papers turned in to the IEEE for publication,
had wrong sign RL. This, even though he had not yet seen my point in his article, as I showed it to him later.

However, none of them - to my knowledge - ever could show that the phenomena on a transmission line
of a certain characteristic wave impedance and the (Thèvenin equivalent of a) generator side impedance
are the same or else anyway share the same identical Gamma formula. This is just a very widely spread, false
assumption. Actually, as long as better power transfer is what we want, we have to use two different "Gammas",
one for each physical phenomenon.

That's what I wanted to demonstrate. So far I found a number of highest grade RF competence university
professors, who support this and who now share my with view me and have given the derivation as a proof.

Nevertheless, there are still so many (forgive me, would you?) "stubborn" adherents of the false assumption "school".

In respect of of them it is good to be able to change the sign of RL.

One - as I can see it, regrettable - consequence is: A programmer of the best general RF application programs available today stubbornly holds the opinion, that even negative SWR is correct, as such adherents of that school think, in a passive lumped
element circuit |Gamma| can be > 1.0. So, instead of a correct SWR, what you may get there is a numerically wrong and even
negative sign SWR. Sri. Who can convince such adherents of that school? I tried hard, but I couldn't.

I hope this helped the ever growing NanoVNA users to not just believe all the RF papers or even programs in every aspect,
but to sometimes check by derivation themselves, as errors are human, and are widespread, too.

73, Hans
DJ7BA



-----Ursprüngliche Nachricht-----
Von: [email protected] <[email protected]> Im Auftrag von randmental
Gesendet: Dienstag, 2. Juni 2020 20:40
An: [email protected]
Betreff: Re: [nanovna-users] out of "Presentation on the NanoVNA for the raileigh Radio Society", Now: "False Return Loss Calculations"

Does all of this mean that the RL on the NanoVNA and VNA Saver are wrong and should be positive? Deon

Yes, the "conventional" (meaning: widely published and often used) formula is:

Gamma = (Z - Zo) / (Z + Zo) where Z is the impedance in question and Zo is the reference impedance for the calculation

But ignoring what Zo actually means and assuming "Both, characterstic wave impedance of a transmission line, and Thévenin impedance (= generator side impedance) are the same or interchangeable, is an error.

Correct, instead, is (a properly different Gamma naming convention is missing yet, so I call them here Gamma1 and gamma2):

Gamma1 = (Z - Zo) / (Z + Zo) with Z = termination impedance, Z0 = characteristic wave impedance of transmission line

Gamma2 = (Z - Zs*) / (Z + Zs) with Z = termination impedance, Zs = source (or generator) side impedance.

Now to the very good point: "I anyway use 50 + j0 Ohm for the generator side":

In that case Zs = Zs*, and thus the difference between Zs and Zs* becomes none. This may be a reason, that the actual error
lasted so long.

The statement is ok, if only you want to match your transmitter to the rig end of the feed line. In such applications it really
doesn't matter.

But though very often used, this is not generally always what we might want to know.


Two examples of a complex generator impedance:

1. Think of a receiving antenna, having anything, but not 50 + j0 Ohms. That is quite common for most antenna heights,
especially, if it is also used far away from resonance. Here we have a generally complex generator impedance.


2. Think of a not well matched transmitter antenna's feedpoint after a lossy feed line.

You may think: "I have a tuner down in the shack at my transceiver. That should take care after adjustment to SWR = 1:1."

And that is, what many think.

What you don't see then is the SWR at the feedpoint. Your rig side 50 Ohm SWR meter sais 1:1, but up at the antenna feedpoint
we have a totally different story - especially on the low bands because of often electrically short antennas. We there may
perhaps have something like 10 - j 1000 Ohms feedpoint impedance trying to match the Thévenin equivalent up there
resulting from the 50 Ohm transmitter, the tuner and the (often quite lossy) feed line. All these influences together may
produce a feedpoint SWR of some 1:10 resulting in total system losses of some 10 dB or more.

The reason to use the above Gamma2 is: It is no longer referenced to 50 + j 0 Ohms, but there at the feedpoint must be
referenced to whatever Thévenin impedance you have there after tuner effect and transmission line losses.

The theroretically (that is: for lossless systems ONLY) valid theorem, "If matched in one place, the system is matched in all places"
does no longer hold in such cases.

Need an illustration?
This one is from DL1JWD:

Take a shortened G5RV (or Zepp) type antenna on 3.5 MHz: 2 x 10m dipole length, 15m CQ553 ladder feed line, and a tuner
for SWR=1:1 at the transmitter.

In spite of the nice SWR = 1:1 down there, up there we have an SWR of some 1:10 (referenced to the Thévenin generator (or TX) side impedance up there, that is relevant for good real power transfer at the feedpoint).

Did you expect that? Perhaps not. It is just a little beyond the "plug and play", I admit, and certainly not needed for most local QSOs.

So, as we can see, certainly not always, but sometimes we may need and want to use Gamma2.

This is what I prefer: Forget unnecessary (due to some false assumption) details that may make it work well enough sometimes.
Instead, use a formula that yields correct results.

In earlier times - when complex calculations were difficult to do by hand - some approximations would do and were helpful.
Today we have easy computer possibilities for complex calculations. No need to neglect the Gamma1 and Gamma2 difference.


73, Hans
DJ7BA


-----Ursprüngliche Nachricht-----
Von: [email protected] <[email protected]> Im Auftrag von af5fx
Gesendet: Mittwoch, 3. Juni 2020 17:36
An: [email protected]
Betreff: Re: [nanovna-users] out of "Presentation on the NanoVNA for the raileigh Radio Society", Now: "False Return Loss Calculations"

Hi everyone,

Interesting discussion. A point of clarification though is that conventionally, the definition of Gamma (reflection coefficient) is:

Gamma = (Z - Zo) / (Z + Zo) where Z is the impedance in question and Zo is the reference impedance for the calculation

The formula in the preceding post appears to have a Zs* in the numerator which is not consistent with the standard definition of Gamma. It's probably from a different context. But, as far as NanoVNA goes, I would think that any reported results should follow conventional definitions of parameters for consistency with standard measurement practices. On that note, the following article is a derivation from basic principles where formula (3.12.10) is the standard definition:
(Ellingson)/03%3A_Transmission_Lines/3.12%3A_Voltage_Reflection_Coefficient

And, for what it's worth, I couldn't resist the temptation to throw in a few points of support for some of the other posts. While the discussion seemingly is centered around an example where Zo = 100 - j100 rather than the more conventional Zo = 50, that is inconsequential to my points:

1) Since Gamma is the ratio of the output (= reflected signal) to the input (= incident signal) it is a "gain" signal as are all S-parameters. It must be < 1 for any passive circuit since a passive circuit cannot add energy.

2) By the definition of S-parameters, it turns out that Gamma = S11 (or S22). So, |S11| must also be < 1 for a passive circuit.

3) Expressing Gamma in dB's is common practice and would actually result in a return gain which for a passive circuit would have to be a negative number. Most VNA's report S11 in dB which would be 20*log10(|S11|) and it would be negative for a passive circuit. While many call this return loss, that is loose nomenclature. Strictly speaking it should really be called return gain in this context.

4) For convenience one can use return LOSS which is the negative of value in dB's or the reciprocal in linear units such as Gamma. Thus, for a passive circuit return loss would have to be a positive value, in dB. And, of course, that means it would be - 20*log10(|S11|) or equivalently + 20*log10( 1 / |S11| )

Definitely easy to get tripped up in the signs. I have probably missed the point of this discussion but just wanted to throw in my two cents. Sorry if it's gone off topic. I didn't want to add to the confusion but instead support some of the later comments on the topic.

Darrell - AF5FX




--
Diese E-Mail wurde von Avast Antivirus-Software auf Viren geprüft.

Hello all,

What you do in your own shop (or ham shack) is of course your own business.

But if you teach, please, please, teach that a passive load's return loss is a positive
number of dB, as it is the only one of the two practices that is logically consistent
with the meaning of the word "loss". Of course you should probably bring this whole
controversy to the attention of your students and warn them to take a close look at
numbers presented in papers etc to be sure what the author really means.

Now on to another aspect of return loss. I tend to define the system impedance Zo
as the impedance of transmission lines commonly used in the environment, and I
tend to assume that the line impedance is a pure real number. This assumption (or
"approximation" if you prefer) is excellent in normal RF applications, but falls apart
rather badly at very low frequencies.

Now here's my quandary: Only a pure real load impedance can absorb all the power
incident on it. So if one is dealing with a system Zo that is complex, it seems to me
that there is no load impedance that results in zero reflected power back into the line.
Is this reasoning too simplistic? When I took the transmission lines course at UMich
in the late 1960's, this issue was not addressed and I was not yet sharp enough to
think to ask.

Dana

On your other aspect of return loss:

Hi Dana and all,

Yes, indeed at low frequencies lossy transmission lines can have a complex characteristic line impedance.
In that case even some 2.141 > Gamma1 > 1 is possible, as the Gamma1 formula, correct for lines, for lines will yield.

This was described already in the late sixties by Robert A. Chipman in
"Theroy and problems of Transmission Lines" as well as by others.
At a close look one will see, that anyway that is not the invention of a "perpetuum mobile".

For describing lumped element mismatch caused Gamma2, application of the Gamma1 formula, however, leads to false results.

On transmission lines (Title and contents of Chipman's book is limited to these) we have actual reflections and
thus the Gamma1 formula is correct.

In lumped element mismatch caused cases "reflections" do not occur and
accordingly the Gamma1 formula does'nt apply - though looking similar.

So:

Let's distinguish these two cases rather than use the ATIS (former ANSI)
type of false assumption, that these were all the same.

By distinguishing the two Gamma cases, we avoid just one (less known) reason of a number of different possible
pitfalls, all leading to false negative Return Loss on passive lumped element mismatch circuits.

73, Hans
DJ7BA


--Ursprüngliche Nachricht-----
Von: [email protected] <[email protected]> Im Auftrag von Dana Whitlow
Gesendet: Mittwoch, 3. Juni 2020 23:18
An: [email protected]
Betreff: Re: [nanovna-users] out of "Presentation on the NanoVNA for the raileigh Radio Society", Now: "False Return Loss Calculations"

Hello all,

What you do in your own shop (or ham shack) is of course your own business.

But if you teach, please, please, teach that a passive load's return loss is a positive number of dB, as it is the only one of the two practices that is logically consistent with the meaning of the word "loss". Of course you should probably bring this whole controversy to the attention of your students and warn them to take a close look at numbers presented in papers etc to be sure what the author really means.

Now on to another aspect of return loss. I tend to define the system impedance Zo as the impedance of transmission lines commonly used in the environment, and I
tend to assume that the line impedance is a pure real number. This assumption (or
"approximation" if you prefer) is excellent in normal RF applications, but falls apart rather badly at very low frequencies.

Now here's my quandary: Only a pure real load impedance can absorb all the power incident on it. So if one is dealing with a system Zo that is complex, it seems to me that there is no load impedance that results in zero reflected power back into the line.
Is this reasoning too simplistic? When I took the transmission lines course at UMich
in the late 1960's, this issue was not addressed and I was not yet sharp enough to think to ask.

Dana




--
Diese E-Mail wurde von Avast Antivirus-Software auf Viren geprüft.

Yep, this is exactly what I do in courses.
If you know what loss is, and know what is meant by return loss, reading a text with a plus or a minus is no real bother.

But if they have to write up something, they'd better use the right sign in the context. :-)

Arie PA3A

Op 3-6-2020 om 23:18 schreef Dana Whitlow: