|
I thought it might be interesting to start a post dealing with how to accurately measure components like inductors, capacitors and resistors on a NanoVNA. As an RF engineer I have made mistakes over the years using trial and error methods. By reading technical books/articles and posts by from others skilled in this area I learned about some of the pitfalls one can easily make and how to avoid them. So I hope some of you will jump in and share your tips and knowledge.
It is easy to make a mistake or draw a false conclusion about a component's characteristics when using a VNA. This can be due to many factors including the following:
- limitations of the NanoVNA hardware and software
- limitations of the test jig
- quality of calibration load and calibration method
- excess lead length
- misconceptions about the component under test
- insufficient technical knowledge
Rather than discuss each of these individually it might be more interesting to do some tests on actual components and point out the pitfalls and things to watch out for when making these measurements.
Here is a test I did on a 47 pF SMD capacitor using a NanoVNA-H4 with the DisLord 1.0.45 firmware. The PC software used was the NanoVNA app by OneOfEleven. Both of these individuals have done an excellent job developing this software and thanks for sharing it with the user community.
The test fixture used was a female SMA connector with a small header row attached. Cal loads were made using male pins and a SMD 49.9 ohm cal load. The idea originally came from a post by Owen Duffy on his blog. A annotated photo of the setup is attached.
A sweep from .05 to 900 MHz. was done and the reactance plotted (graph attached). A graph converting the reactance to "apparent capacitance" is also shown. A pitfall made by those new to VNA's is to assume this is the actual capacitance vs. frequency of the part. This is not the case because a physical capacitor also has some inductance associated with it as shown in the simplified capacitor model. This results in positive inductive reactance adding to the negative capacitive reactance and a faster rise in total reactance with frequency than just the capacitor alone. In the graph below note the rapid reactance increase with frequency as we get close to the self-resonant frequency (SRF) of the capacitor.
Now one might think it is reasonable to measure at a very low frequency in order to get the actual capacitance but this is not possible because the capacitance reactance is so high at low frequencies that it cannot be actually measured by the VNA. By looking at the markers you can see that we can only start making accurate measurements around 2 MHz. for this part. Another instrument in my lab a DE-5000 can make these measurements at low frequencies (below 100 kHz.) and the part measures close to 47 pF in this range. The frequency at which accurate measurements can be made will be based on the capacitance of the device under test. Larger capacitors can be measured at a lower frequency and smaller ones like 10 pF at a higher frequency.
If we wish to determine the frequency of self resonance we can look for the frequency where the reactance is zero. This is when the capacitive reactance and inductive reactance are equal and opposite in sign and result in zero at the SRF. This is also the frequency where the S11 phase angle abruptly changes from - degrees to + degrees for this example. For those familiar with Smith charts this is a data point on the horizontal line of the Smith chart.
Roger
|
Hello Roger,
Thank you very much for the professional description.
I think alligator leads are not useful in VNA measurements above 5 MHz.
73, Rudi DL5FA
|
This is very good. Thank you for sharing!
--
VE6WGM
|
Roger, I think you basically said it all with this list of pitfalls: - limitations of the NanoVNA hardware and software
- limitations of the test jig
- quality of calibration load and calibration method
- excess lead length
- misconceptions about the component under test
- insufficient technical knowledge I can only add a little about my own experiences measuring components: When trying to measure through20cm long SMA cables and connectors, I often got strange results. It seems that the NanoVNA's measuring range gets restricted when it has to compensate for the effects of those cables. I get much more reliable and credible results when using the test jig I published here, several days ago, which has only a very short piece of coax cable and large contact surfaces. So I would suggest to make and use any sort of test jig that has as short a connection as possible to the NanoVNA, so there are less parasitics to calibrate out. About calibration: The best, of course, is having a perfect short and a perfect 50? load for calibration... But not having those, one can get closer to the truth about a component by adding the known (or estimated) imperfections of the test loads to the measured data. For example, for my basic simple test jig I use a shorting wire stuck to a clothespin, which is pretty good, having maybe 1 or 2 nH. But my 50? load is two leaded quarter-watt 100? resistors in parallel, and these obviously have some significant inductance. How much? Well... measuring any single resistor of the same size and type, with the same lead length, and of reasonable value, reports roughly 5nH of ESL. Now, it's logical that two resistors in parallel, reasonably separated, should have half as much inductance as a single one. So I conclude that a single quarter-watt resistor has around 10nH, and my 50? test load, having two of them in parallel, has 5nH. And that's why a single resistor measured in my jig reads 5nH. A poor calibration load, but entirely possible to compensate for! With this jig I have been playing around over the last days, measuring resistors, capacitors, inductors, transformers (main inductance and leakage inductance), and crystals. It works very well. I use the NanoVNA MOD v3 software, and find it most practical to use the series RLC display for most component testing. While named RLC, it actually gives RXZ curves, not RLC... But the L or C values are easily obtained by sliding the mouse over the graph. This allows easily seeing over what frequency range the capacitance or inductance stays pretty constant, and that's likely to be the range in which the nanoVNA is reading reasonably correct while at the same time the component value is still reasonably pure. With relatively high impedance parts I switch into the parallel RLC display, which then makes more sense because the parasitics in high impedance parts tend to be dominated by parallel capacitance. It's important to understand what one is measuring, of course. When I measure leaded parts, I include as much lead length in the measurement as I will need later when mounting the part on the board. Usually the shortest length possible. Since my test jig has wide connection areas, its own inductance is reasonably low, and the exact spot on the jig where each lead connects is rather uncritical. A while ago I was comparing the performance of quarter-watt carbon film and metal film resistors, at frequencies into several hundred MHz. It works nicely, and is very useful to dispell long-standing myths, to select the best parts for a project, and even to find out whether a given part is actually usable for the intended application. About alligators: They add quite a bit of inductance, which can be calibrated out to some extent, but limiting the NanoVNA's measuring range. What's worse is that the flying leads can end up in different positions between calibration and use. So indeed I would only use them in situations where 10 or 20nH uncertainty, and 1pF or so uncertainty, are acceptable. A fixed, solid test jig is always better. I have become a fan of clothespins. For measuring SMDs, an easy test jig could be made from a single clothespin, with two copper foil (or PCB pieces) glued to it, and a short piece of coax cable with an SMA plug soldered to them. The short circuit calibration would be done simply with nothing in the clothespin, just a direct contact. A small 50? SMD resistor would be used as 50? calibration load. And a small piece of plastic, such as teflon or polyethylene, the same thickness as the length of the SMD to be tested, as an open calibrator, or simply doing the open calibration while holding the clothespin slightly open. This jig should allow pretty accurate testing of SMDs. Just the tiny inductance of the 50? calibrator would need to be compensated for after the measurement. The obvious advantage of such a clothespin SMD testing jig, of course, is that the part can still be used after testing... I would prefer not using an SMD in a circuit after having soldered it to a test connector, and then desoldering it... And I hate wasting parts.
|
Manfred,
Thank you for your detailed and informative post. Lots of good information and tips on how to measure components. I have not seen the same problem with cables up to 30 cm but I will keep what you posted in mind. I agree with you about alligator clips being a poor test jig. It is so easy to construct a test jig that I never use them.
The NanoVNA MOD v3 software has some nice plotting features but I prefer the NanoVNA app by OneOfEleven that works with the 1.0.45 firmware by DiSlord. The firmware does 401 points on the -H4 instead of 101, faster data transmission and some nice new user features on the NanoVNA. The NanoVNA app is the best PC program I have used to date with nice scaling and plotting features, trace smoothing, calibration averaging and the ability to do firmware updates and much more.
That clothespin idea is really interesting for testing SMD parts. I am going to build a jig using this method. If you have one already built could you please post some pictures.?
You mentioned the serial and parallel impedance measurement of components. This is not familiar to some users and I will be doing a post later showing how this can be useful when measuring higher impedance parts.
Thanks for joining the discussion and sharing your experience.
Regards - Roger
|
The use of a small piece of Teflon or other plastic between the contacts of the clothespin for doing the OPEN calibration may be a bit problematic.
It will increase the capacitance between the contacts which will probably lead to a less accurate calibration . I would suggest finding a way to hold the contacts open without the inclusion of the plastic.
For doing cals up to about 1Ghz for a nanoVNA it probably isn’t that big of a deal, but just thought I’d mention it. (I’m used to doing vna cals up to 30 or 40 GHz where it is a big deal)
Ray WB6TPU
toggle quoted message
Show quoted text
On Feb 19, 2021, at 6:42 AM, Manfred Mornhinweg <manfred@...> wrote:
have become a fan of clothespins. For measuring SMDs, an easy test jig could be made from a single clothespin, with two copper foil (or PCB pieces) glued to it, and a short piece of coax cable with an SMA plug soldered to them. The short circuit calibration would be done simply with nothing in the clothespin, just a direct contact. A small 50? SMD resistor would be used as 50? calibration load. And a small piece of plastic, such as teflon or polyethylene, the same thickness as the length of the SMD to be tested, as an open calibrator, or simply doing the open calibration while holding the clothespin slightly open. This jig should allow pretty accurate testing of SMDs. Just the tiny inductance of the 50? calibrator would need to be compensated for after the measurement.
|
Hi Roger et al,
I am not about to admit to how many of those misconceptions I have fallen for .... publicly .... :)
I have been 'safe' in the analogue, low frequency, ideal component electronic world for awhile, but kept bumping into non-ideal component problems. I purchased my nanoVNA primarily as a challenge to that environment. It, and this Group, have been great teachers. Thanks.
Re: your Freq Phase S11 graph ... can you elaborate please? ....
* The Commentary refers to a misconception around S Parameters and Impedance phase ... is the distinction here Reflection co-efficient Gamma phase => Incident and Reflected voltages .. versus .. Impedance phase => Voltage and Current .... ???
* Can you describe the 'drivers' of the 3 key points to this graph ie 0 deg, -145deg, and +/-180deg?? 360deg rotation at self resonance??
I am unsure as to how many more 'nanoTraps' I have yet to climb out of .... :)
|
Forgot to my Signature
--
Cheers,
Ian
Melbourne, Australia
|
Where Roger and I have differed in the past is with regards to the concept of the S11 phase as measured by the nanoVNA.
While Roger has firmly stated in previous conversations that there is no relationship between the S11 phase and the phase change between voltage and current observed in reactive circuits, I still hold to my original thought that the voltage vs current phase of a reactive device is in fact related to S11 phase. It simply must be!
As the reactive component causes a change in phase of the voltage and current, so to does it cause a change in the S11 phase as measured by the nanoVNA.
Please see “Purely Reactive Load” in this reference:
(Ellingson)/03%3A_Transmission_Lines/3.12%3A_Voltage_Reflection_Coefficient
(Sorry, the link is not fully recognized, so one will need to cut and paste the link manually into a browser. I have included a screen shot with this post.)
The formula under the “Purely Reactive Load” heading for the voltage reflection coefficient clearly includes reactance. Where a reactive component is present, there will also be a change in phase between voltage and current at any given frequency.
Thus, I maintain that S11 phase is in fact related to voltage vs current phase.
Please feel free to prove me wrong.
--
VE6WGM
|
Gregg-
I concur with your understanding 100%.
Reflection phase with an open is zero degrees and is 180 degrees with a short. Reactive loads (inductive or capacitive) will have phase angles somewhere between 0 and 180 dependent on the reactance at a particular frequency which depends on reactance (Zl +jX or - jX).
See the last 2 sentences in the book section on Purely reactive loads that was cited.
Ray WB6TPU
Senior Signal Integrity Staff Engineer
(Retired)
toggle quoted message
Show quoted text
On Feb 19, 2021, at 7:52 PM, Gregg Messenger <techgreg@...> wrote:
?Where Roger and I have differed in the past is with regards to the concept of the S11 phase as measured by the nanoVNA.
While Roger has firmly stated in previous conversations that there is no relationship between the S11 phase and the phase change between voltage and current observed in reactive circuits, I still hold to my original thought that the voltage vs current phase of a reactive device is in fact related to S11 phase. It simply must be!
As the reactive component causes a change in phase of the voltage and current, so to does it cause a change in the S11 phase as measured by the nanoVNA.
Please see “Purely Reactive Load” in this reference:
(Ellingson)/03%3A_Transmission_Lines/3.12%3A_Voltage_Reflection_Coefficient
(Sorry, the link is not fully recognized, so one will need to cut and paste the link manually into a browser. I have included a screen shot with this post.)
The formula under the “Purely Reactive Load” heading for the voltage reflection coefficient clearly includes reactance. Where a reactive component is present, there will also be a change in phase between voltage and current at any given frequency.
Thus, I maintain that S11 phase is in fact related to voltage vs current phase.
Please feel free to prove me wrong.
--
VE6WGM
-
|
Maybe the misunderstanding is in the fact that the reflectioncoefficient has another phaseangle than the phaseangle between current and voltage on the load. These are two different things. The relation is in the already quoted formula.
73,
Arie PA3A
|
Roger, The NanoVNA MOD v3 software has some nice plotting features but I prefer the
NanoVNA app by OneOfEleven that works with the 1.0.45 firmware by DiSlord. I didn't know about that software. I just downloaded it, but it doesn't seem to work with standard firmware, and I'm hesitant about changing to a different firmware, given that the stock one works well. Also, I didn't find a way to change what the graph displays. Is it really limited to only displaying S11 and S21? Among all the options given by the NanoVNA, and by the PC software I'm using, I find RLC graphs most useful for component testing. The firmware does 401 points on the -H4 instead of 101, faster data
transmission and some nice new user features on the NanoVNA. The NanoVNA app
is the best PC program I have used to date with nice scaling and plotting
features, trace smoothing, calibration averaging and the ability to do
firmware updates and much more. I only have the NanoVNA-H, and I wouldn't like to put special firmware on it. So it seems that the software you use is not for me! I can't use NanoVNA Saver either, because it doesn't run on Windows XP, and that's what I use. Good that there is a choice of programs. I would love to get more than 101 points, and I would also love an actual RLC display, rather than the RXZ display given when selecting RLC. A true RLC display would be great to directly observe the variations of component values against frequency. Maybe some day I end up writing my own program for this purpose. Is there a clear description of the NanoVNA's communication protocol? That clothespin idea is really interesting for testing SMD parts. I am going
to build a jig using this method. If you have one already built could you
please post some pictures.? I haven't built it yet. It's just an idea. So far I have only built that very simple component test jig using a small piece of PCB, using clothespins for calibration and for holding component leads and SMDs against the board. It serves me well enough for now. Ray, The use of a small piece of Teflon or other plastic between the contacts of the clothespin for doing the OPEN calibration may be a bit problematic.
It will increase the capacitance between the contacts which will probably lead to a less accurate calibration . I would suggest finding a way to hold
the contacts open without the inclusion of the plastic. Well, there is no need to place the plastic "open" calibrator between the contacts! A clothespin jaws are much larger than an SMD. The jig could easily be made with small contact areas, something like 3mm diameter. The "open" calibrator can be placed in the clothespin several mm away from those contact areas. That should bring the "open" calibration error down to less than 0.1pF, which is probably good enough for most of us. The traces leading to the contact areas should of course run well away from each other, to reduce capacitance miscalibration if the clothespin flexes and distorts in use.
|
If you explore NanoVNA-App you will find lots of things. Right click on the graph window to find numerous options. Clicking on the text in the upper left corner of the graph window will allow you to select the graph type. The New Graph button at the far right of the top button row will do just that. Do upgrade your firmware. You will benefit from new features and bug fixes. I am using Dislord 1.0.45 on my H. There is no "standard" firmware, only latest and obsolete. :) 73 -Jim NU0C On Sat, 20 Feb 2021 06:33:06 -0800 "Manfred Mornhinweg" <manfred@...> wrote: Roger,
The NanoVNA MOD v3 software has some nice plotting features but I prefer the
NanoVNA app by OneOfEleven that works with the 1.0.45 firmware by DiSlord. I didn't know about that software. I just downloaded it, but it doesn't seem to work with standard firmware, and I'm hesitant about changing to a different firmware, given that the stock one works well.
Also, I didn't find a way to change what the graph displays. Is it really limited to only displaying S11 and S21? Among all the options given by the NanoVNA, and by the PC software I'm using, I find RLC graphs most useful for component testing.
The firmware does 401 points on the -H4 instead of 101, faster data
transmission and some nice new user features on the NanoVNA. The NanoVNA app
is the best PC program I have used to date with nice scaling and plotting
features, trace smoothing, calibration averaging and the ability to do
firmware updates and much more. I only have the NanoVNA-H, and I wouldn't like to put special firmware on it. So it seems that the software you use is not for me!
I can't use NanoVNA Saver either, because it doesn't run on Windows XP, and that's what I use.
Good that there is a choice of programs.
I would love to get more than 101 points, and I would also love an actual RLC display, rather than the RXZ display given when selecting RLC. A true RLC display would be great to directly observe the variations of component values against frequency.
Maybe some day I end up writing my own program for this purpose. Is there a clear description of the NanoVNA's communication protocol?
That clothespin idea is really interesting for testing SMD parts. I am going
to build a jig using this method. If you have one already built could you
please post some pictures.? I haven't built it yet. It's just an idea. So far I have only built that very simple component test jig using a small piece of PCB, using clothespins for calibration and for holding component leads and SMDs against the board. It serves me well enough for now.
Ray,
The use of a small piece of Teflon or other plastic between the contacts of the clothespin for doing the OPEN calibration may be a bit problematic.
It will increase the capacitance between the contacts which will probably lead to a less accurate calibration . I would suggest finding a way to hold
the contacts open without the inclusion of the plastic. Well, there is no need to place the plastic "open" calibrator between the contacts! A clothespin jaws are much larger than an SMD. The jig could easily be made with small contact areas, something like 3mm diameter. The "open" calibrator can be placed in the clothespin several mm away from those contact areas. That should bring the "open" calibration error down to less than 0.1pF, which is probably good enough for most of us.
The traces leading to the contact areas should of course run well away from each other, to reduce capacitance miscalibration if the clothespin flexes and distorts in use.
|
I don't need to "prove" you wrong. Without phase, reactance just becomes a
non-dissipative circuit element that doesn't do much outside of resistive
losses. One can't have a reactive component without phase.
Z = R ± jX
Without phase information, there is no ± jX term and no reactive component.
Dave - W?LEV
toggle quoted message
Show quoted text
On Sat, Feb 20, 2021 at 3:45 AM Gregg Messenger <techgreg@...> wrote: Where Roger and I have differed in the past is with regards to the concept
of the S11 phase as measured by the nanoVNA.
While Roger has firmly stated in previous conversations that there is no
relationship between the S11 phase and the phase change between voltage and
current observed in reactive circuits, I still hold to my original thought
that the voltage vs current phase of a reactive device is in fact related
to S11 phase. It simply must be!
As the reactive component causes a change in phase of the voltage and
current, so to does it cause a change in the S11 phase as measured by the
nanoVNA.
Please see “Purely Reactive Load” in this reference:
(Ellingson)/03%3A_Transmission_Lines/3.12%3A_Voltage_Reflection_Coefficient
The formula under the “Purely Reactive Load” heading for the voltage
reflection coefficient clearly includes reactance. Where a reactive
component is present, there will also be a change in phase between voltage
and current at any given frequency.
Thus, I maintain that S11 phase is in fact related to voltage vs current
phase.
Please feel free to prove me wrong.
--
VE6WGM
--
*Dave - W?LEV*
*Just Let Darwin Work*
|
Hi Manfred,
You note that you cannot use NanoVNA Saver under Windows XP.
I inherited a laptop running Windows XP from my daughter.
I just tried that if I copy the NanoVNA-App to the C: / root directory, it will log in without further ado.
I remembered that the creator of the program mentioned this, but I never tried it.
You may also want to update the firmware because the program and firmware were developed together.
So one cannot be used without the other.
--
*** If you are not part of the solution, then you are the problem. ( ) ***
|
On Sat, Feb 20, 2021 at 01:21 AM, Arie Kleingeld PA3A wrote:
Maybe the misunderstanding is in the fact that the reflectioncoefficient
has another phaseangle than the phaseangle between current and voltage
on the load. These are two different things. The relation is in the
already quoted formula.
Arie, Thank you for your comment. That is exactly the misunderstanding I was referring to in my previous post. I am also on Facebook and I have seen several posts there that stated that the "S11 phase angle" displayed on the NanoVNA and the "impedance phase angle" of the device under test were the same thing. Several YouTube videos have also made the same assertion which is wrong and confuses those new to this subject. That is why I stated the following in one of my drawings in the post above.... "This is the S11 phase plot which is the angle of the reflection coefficient gamma. A common misconception by those new to S parameters is to confuse this with the impedance phase of the device under test (DUT)" I do not see why some object to this statement which is all I said on this subject. I will write another post later today to clarify why this is true. Roger Roger
|
This is a friendly group and I do not wish to get into some kind of argument with others. If I have posted something in this group which is wrong or confusing please quote what I said and tell me why it is incorrect and I will do my best to answer your question or if proven wrong admit my error. But please quote what I said NOT what someone else thought I said or meant about the subject at hand.
In a previous post on a graph I wrote "This is the S11 phase plot which is the angle of the reflection coefficient gamma. A common misconception by those new to S parameters is to confuse this with the impedance phase of the device under test (DUT)". Here is why I believe that to be true.
If we have a complex impedance attached to our NanoVNA and wish to measure its characteristics we can do this with an S11 or S21 measurement. For the purposes of this discussion I will focus on the S11 measurement.
Let us say our device under test (DUT) consists of a 50 ohm resistor in series with a 1.6 uH inductor. We perform the measurement at 5 MHz. The inductive reactance is X = 2*pi*freq*L which is equal to 50 ohms. So the complex impedance R+jX is 50+j50. At 20 MHz. the same DUT will have a complex impedance of 50+j200.
It is well known that the "impedance phase angle" is equal to the arctangent of X/R and this will range between +90 degrees to -90 degrees depending on the values of X and R and whether X is inductive or capacitive. For the case of 50+j50 the phase angle is arctan 1 or 45 degrees and fro 50 + j200 is arctan 4 or 76 degrees. This is shown on the attached diagram. We could easily do a plot of impedance phase versus frequency for the DUT and get a graph.
Now when we connect this example DUT to the CH0 port of a NanoVNA and do a measurement over the range of 5 to 20 MHz. we can set the traces to display a number of parameter plots. One that is very useful is to plot the "reflection coefficient" Gamma (Γ) . Gamma Γ = (Z - Zo) / (Z + Zo) with Z being a complex number and Zo typically set at 50 or 75. Note that Gamma will also be a complex number in the form a+jb or as magnitude @ an angle. The Magnitude is = sqrt(a^2+b^2) and the angle is the arctan (b/a). The angle is often referred to as the "S11 phase angle" and that can be displayed on the NanoVNA or in PC programs that work with the PC. There are calculators that calculate the reflection coefficient in both forms for a given R and X and the phase angle . Here is one that is easy to use.
If R and X are plotted on a Smith chart the vector from the origin to that point will have a length equal to the magnitude of the refection coefficient and the angle will be the S11 phase angle. I have plotted the 50+j50 and 50+j200 on a Smith chart and attached them to this post.
Here is a comparison of the "impedance phase angle" and "reflection coefficient phase angle" for the two cases above. One can clearly see that they are different which is what I originally stated. Yes reactance does have an effect on both and this is not in dispute. But they are different measurements!
Impedance phase reflection coefficient phase
50+j50 45 63.4
50+j200 76 26.6
It would be nice if the "impedance phase angle" could be added to the graphing capability of the NanoVNA or one of the PC programs. Someone requested this on the NanoVNA saver GitHub page last month.
Roger
|
Roger, you are indeed correct. I also appreciated Arie's comment. Gregg is also correct for the reflected signal measurement - and in saying that this is related to the V/I/phase in the DUT; indeed since the VNA is providing the signal to the DUT, all V/I/phase in the DUT and those reflected will be related - but that relationship is complex. S11 does not measure the device itself, it measures the reflection that device causes in the feedline between the VNA and the device. (An S21 measurement will measure some characteristics of the device itself.) A very simple example of this is when the DUT is an antenna. The VNA S11 is a measurement of the reflected signal in the feedline, and ideally we design this to be a perfect match at 50 ohms purely resistive, which will have a reflected phase angle of 0. But the V/I/phase in the antenna itself varies greatly along the antenna, and the phase is necessarily non-zero except at the feedpoint. On Sat, Feb 20, 2021 at 11:14 AM Roger Need via groups.io <sailtamarack= [email protected]> wrote: On Sat, Feb 20, 2021 at 01:21 AM, Arie Kleingeld PA3A wrote:
Maybe the misunderstanding is in the fact that the reflectioncoefficient
has another phaseangle than the phaseangle between current and voltage
on the load. These are two different things. The relation is in the
already quoted formula.
Arie,
Thank you for your comment. That is exactly the misunderstanding I was
referring to in my previous post. I am also on Facebook and I have seen
several posts there that stated that the "S11 phase angle" displayed on the
NanoVNA and the "impedance phase angle" of the device under test were the
same thing. Several YouTube videos have also made the same assertion which
is wrong and confuses those new to this subject. That is why I stated the
following in one of my drawings in the post above....
"This is the S11 phase plot which is the angle of the reflection
coefficient gamma. A common misconception by those new to S parameters is
to confuse this with the impedance phase of the device under test (DUT)"
I do not see why some object to this statement which is all I said on this
subject. I will write another post later today to clarify why this is true.
Roger
Roger
|
Roger-
Your latest message very clearly described the differences between a reflection coefficient phase angle and the phase angle associated with a complex impedance.
Sounds like most of the recent responders are in violent agreement.
Regards,
-Ray
WB6TPU
toggle quoted message
Show quoted text
On Feb 20, 2021, at 12:12 PM, Roger Need via groups.io <sailtamarack@...> wrote:
?This is a friendly group and I do not wish to get into some kind of argument with others. If I have posted something in this group which is wrong or confusing please quote what I said and tell me why it is incorrect and I will do my best to answer your question or if proven wrong admit my error. But please quote what I said NOT what someone else thought I said or meant about the subject at hand.
In a previous post on a graph I wrote "This is the S11 phase plot which is the angle of the reflection coefficient gamma. A common misconception by those new to S parameters is to confuse this with the impedance phase of the device under test (DUT)". Here is why I believe that to be true.
If we have a complex impedance attached to our NanoVNA and wish to measure its characteristics we can do this with an S11 or S21 measurement. For the purposes of this discussion I will focus on the S11 measurement.
Let us say our device under test (DUT) consists of a 50 ohm resistor in series with a 1.6 uH inductor. We perform the measurement at 5 MHz. The inductive reactance is X = 2*pi*freq*L which is equal to 50 ohms. So the complex impedance R+jX is 50+j50. At 20 MHz. the same DUT will have a complex impedance of 50+j200.
It is well known that the "impedance phase angle" is equal to the arctangent of X/R and this will range between +90 degrees to -90 degrees depending on the values of X and R and whether X is inductive or capacitive. For the case of 50+j50 the phase angle is arctan 1 or 45 degrees and fro 50 + j200 is arctan 4 or 76 degrees. This is shown on the attached diagram. We could easily do a plot of impedance phase versus frequency for the DUT and get a graph.
Now when we connect this example DUT to the CH0 port of a NanoVNA and do a measurement over the range of 5 to 20 MHz. we can set the traces to display a number of parameter plots. One that is very useful is to plot the "reflection coefficient" Gamma (Γ) . Gamma Γ = (Z - Zo) / (Z + Zo) with Z being a complex number and Zo typically set at 50 or 75. Note that Gamma will also be a complex number in the form a+jb or as magnitude @ an angle. The Magnitude is = sqrt(a^2+b^2) and the angle is the arctan (b/a). The angle is often referred to as the "S11 phase angle" and that can be displayed on the NanoVNA or in PC programs that work with the PC. There are calculators that calculate the reflection coefficient in both forms for a given R and X and the phase angle . Here is one that is easy to use.
If R and X are plotted on a Smith chart the vector from the origin to that point will have a length equal to the magnitude of the refection coefficient and the angle will be the S11 phase angle. I have plotted the 50+j50 and 50+j200 on a Smith chart and attached them to this post.
Here is a comparison of the "impedance phase angle" and "reflection coefficient phase angle" for the two cases above. One can clearly see that they are different which is what I originally stated. Yes reactance does have an effect on both and this is not in dispute. But they are different measurements!
Impedance phase reflection coefficient phase
50+j50 45 63.4
50+j200 76 26.6
It would be nice if the "impedance phase angle" could be added to the graphing capability of the NanoVNA or one of the PC programs. Someone requested this on the NanoVNA saver GitHub page last month.
Roger
|
Hi Roger et al,
Ooooppppssss .... I think I blindly stepped into something .... or 'stepped onto' something ... :)
I guess it is going to take me awhile to understand that issue .... and the shape of that S11 Freq Phase graph.
Maybe some differences of opinions but great info for people like me. I have found that a lot of differences between experts usually comes down to definitions/terminology.
--
Cheers,
Ian
Melbourne, Australia
|
Roger Need
Gregg Messenger VE6WGM
Arie Kleingeld PA3A
Jim Shorney
Gyula Molnar
