Antenna - S21 Analysis

Status
Not open for further replies.

Ubbe

Member
Joined
Sep 8, 2006
Messages
9,257
Location
Stockholm, Sweden
The BCD436 most likely use the same filters as in a BCD536 and here are some plots I did of the impedance variations at the antenna port:

/Ubbe
 

ScubaJungle

Active Member
Joined
Mar 11, 2020
Messages
618
Location
West Central Florida
No worries - that's kinda what I was kinda going to suggest - walk away and come back later after a good rest and/or sleep. For you this is just a hobby and shouldn't get overly frustrating. You aren't trying to prepare for an all day RF engineering interview at Qualcomm, Hughes, or TI tomorrow (been there and done that). Have fun and learn along the way and you'll be ahead of the game.

-Mike
Wow - that sounds like one hell of an interview lineup! I ordered a UHF to SMA connector so I can see how the LMR400 cable compares. Looking back on it - I dont see any obvious setup failures - calibration was seemingly done right (results look as they did for previous antenna sweeps), and I dont see any physical connection failures. I think it really may be cheap coax. It is thinner than a USB-C android power cord - I suppose they didnt care about the quality as it is only a few feet?


Awesome, @Ubbe, thanks.
 

prcguy

Member
Premium Subscriber
Joined
Jun 30, 2006
Messages
15,692
Location
So Cal - Richardson, TX - Tewksbury, MA
I actually had a career at Hughes but if your in the northern San Diego area might I suggest interviewing at ViaSat, excellent company to work for, kind of like Google but with satellites, RF and everything in between.

No worries - that's kinda what I was kinda going to suggest - walk away and come back later after a good rest and/or sleep. For you this is just a hobby and shouldn't get overly frustrating. You aren't trying to prepare for an all day RF engineering interview at Qualcomm, Hughes, or TI tomorrow (been there and done that). Have fun and learn along the way and you'll be ahead of the game.

-Mike
 

ScubaJungle

Active Member
Joined
Mar 11, 2020
Messages
618
Location
West Central Florida
So the only other coax in my apartment that I could connect both ends to is a piece of coax used for the TV, which was in the apartment already. I suppose it is RG6, or something similar(?)

I took a few sweeps, and this is what I got:
86665

Then, I swept the dipole coax, and this is what I got:

86666

Not sure if this helps at all, but here is the Smith Chart also:
86667


Theres so much conflicting info online about this its crazy. A lot of them are using the S11 measurement for coax loss? Could that be right, considering the S11 chart looks a lot more reasonable at -2dB< for the dipole coax
 

Mike_G_D

Member
Joined
Dec 19, 2002
Messages
1,242
Location
Vista, CA
I'm actually on the clock and working so can't elaborate much. But something is really off on those graphs if I'm reading them right.

For coax most are interested in loss (negative gain) as it's a passive element that is used in the whole antenna system - the link between the receiver and the antenna. Looking at the impedance is only really needed when troubleshooting or evaluating unknown cable but you have to have the other end terminated in the correct resistive system impedance.

Also - why S21 and not S12? Doesn't really matter much as they should be identical but just usual to do a S12 sourced from port one and read by port 2. Anyway, that's not the big deal. What is REALLY throwing me is seeing your loss figures get better as the frequency increases - that really should NOT happen unless the coax is not properly connected and the signal from the VNA is actually radiating out of the coax and getting into the second port and that radiation is getting better with higher frequencies - which it could.

-Mike
 

ScubaJungle

Active Member
Joined
Mar 11, 2020
Messages
618
Location
West Central Florida
I'm actually on the clock and working so can't elaborate much. But something is really off on those graphs if I'm reading them right.

For coax most are interested in loss (negative gain) as it's a passive element that is used in the whole antenna system - the link between the receiver and the antenna. Looking at the impedance is only really needed when troubleshooting or evaluating unknown cable but you have to have the other end terminated in the correct resistive system impedance.

Also - why S21 and not S12? Doesn't really matter much as they should be identical but just usual to do a S12 sourced from port one and read by port 2. Anyway, that's not the big deal. What is REALLY throwing me is seeing your loss figures get better as the frequency increases - that really should NOT happen unless the coax is not properly connected and the signal from the VNA is actually radiating out of the coax and getting into the second port and that radiation is getting better with higher frequencies - which it could.

-Mike
No problem, thanks for the response either way!
I meant to include in my last post that this software does not provide a graph/any options for S12 measurements.
From what I have seen on youtube, other NanoVNA graphs look similar with thru loss calculations, so what I am thinking is perhaps since this is cheap and doesn't have as many features as a proper VNA would, that it shows what should be S12 under the S11 measurements? Im not sure, kind of just guessing at this point, but since there is absolutely nothing about S12 in the software at all, and the YouTube videos use S11, Im thinking that may be the simplest explanation for the strange graphs - although I have no idea what is up with the S21 gain going positive as frequency increases.
Ill continue doing research and see if I can find something that explains this better. Since the other coax seems to show the same trend, I am thinking now that the coax may not be that bad, and it is just something weird about this device - as I doubt both coax cables are so equally horrendous. Of course - it's possible that it is my fault also, but I am following the online calibration directions to a T.
 

Mike_G_D

Member
Joined
Dec 19, 2002
Messages
1,242
Location
Vista, CA
Let's go through a few basics...

Your S11 and S22 graphs SHOULD look like nice fairly flat straight lines all through the frequency range; granted, you might see something a little off if the characteristic impedance of your coax is not 50 ohms (like using 75 ohm lines which the one that came with the TV rabbit ears dipole very well might be) but nothing like what your graphs are showing.

But the S21 (and S12) should be really close to the 0 dB loss line at the lowest frequency and get worse steadily all the way up to the highest frequency. I don't understand the "jaggies" that I am seeing - looks like some kind of test system glitch maybe something not setup right in the VNA.

Now for calibration:

Set you cable to test aside and start your calibration procedure:

1) Set up your start and end frequencies;
2) Right at the VNA test port 1 - Perform single port Open, then Short, then Load on Port 1;
3) Right at the VNA test port 2 - Perform single port Open, then Short, then Load on Port 2;
4) Connect the Calibrated Thru between Port 1 and Port 2 of the VNA and perform a through S12 and S21 calibration; Zero your graph scale at this point (thru then becomes the test reference as a "0 dB loss" reference);
5) Make sure that you store your calibration data if possible.

Now disconnect the thru and connect the cable to test in its place - one end to port 1 and the opposite end to port 2. If using adapters use the absolute minimum of adapters as possible as the system will include those as part of the cable to be tested (they will add a little loss to the readings but if their quality is good it shouldn't be significant).

Run your S12 and S21 measurements. Let's not worry about S11 and S22 yet for the coax.

-Mike
 

Mike_G_D

Member
Joined
Dec 19, 2002
Messages
1,242
Location
Vista, CA
I just read your last post as it came after I finished writing my last response.

Ok, that is weird that the VNA has no S12! But, that's basically ok as the S21 is just the mirror image of that so you'll just have to use the S21. S21 means sourced from 2 as seen by 1 which, for a cable, should be no different then the other way round.

You really don't need S11 or S22 for a cable of known characteristic impedance at least not for loss measurements.

So just use S21 in your case.

-Mike
 

Mike_G_D

Member
Joined
Dec 19, 2002
Messages
1,242
Location
Vista, CA
Couple more things -

Looking at your last results:

1) Though we don't really need the S11 and S22 measurements what you show for those makes no sense either; you SHOULD see a REALLY HIGH return loss as the system SHOULD be matched so that the reflected energy is very slight. The Smith Chart view looks really terrible! It looks like the other end of the coax is shorted at the opposite end, kinda. On a Smith Chart normalized to 50 ohms system impedance you want your results to be at the center. Perfect like a 50 ohm resistive load like your calibrated load standard should show just a point at the very center of the chart. Getting away from that perfect point you want to be as close to the center as possible (for a 50 ohm match). The horizontal line is the resistance line or "Real" part while the "curves" going above and below the center line show positive and negative reactance, respectively. Capacitive reactance goes below the center line (-j) and inductive reactance goes above the center line (+j). Those curves are really sections of complete circles. You can get the whole skinny on Smith Charts on-line, of course, too much to go into here.

2) I am now wondering if your VNA just doesn't have sufficient isolation between the two test ports for the level of test signal applied? On that VNA is there any way to reduce the level of the test signal applied at either or both ports? If possible, try turning down the applied test level at the source port (so if doing S21 you would reduce the level at port 2). Reason why is so as to try and minimize any radiated energy that might be getting back into the VNA's opposite port from the source port. So, in other words, as the loss of the coax goes up with frequency the radiated signal level on the outside of the coax is getting stronger and getting back into the receive port and screwing up the readings so as to make the thru loss look like it's getting better at the higher frequencies. With these new small form factor low-cost VNA's I can see that as a possibility - those ports are pretty close together on the device based on the pictures I have seen. If the shielding and isolation between the two ports isn't very good it could really mess up the results. I don't think that explains the really odd S11 and S22 measurements I'm seeing from your results though but could explain the odd S21 results, at least in part.

-Mike
 

Ubbe

Member
Joined
Sep 8, 2006
Messages
9,257
Location
Stockholm, Sweden
It's just a bug in the program. It shows S11 as S21 and the other way around and the colors are also wrong. Switch the black curve with the turqoise and all is well.

It says S11 Return loss when it's actually the S21 Gain diagram.
It says S21 Gain when it is the S11 Return loss.
The S11&S21 LogMag diagram should have the black and turqoise curves interchanged.
As every function are the wrong one, also the smith chart then use the wrong parameters when doing calculations for the plot.

It's probably software for another VNA that have been used. Check the VNA type and model number and try and find the correct software for it.

/Ubbe
 

Mike_G_D

Member
Joined
Dec 19, 2002
Messages
1,242
Location
Vista, CA
It's just a bug in the program. It shows S11 as S21 and the other way around and the colors are also wrong. Switch the black curve with the turqoise and all is well.

It says S11 Return loss when it's actually the S21 Gain diagram.
It says S21 Gain when it is the S11 Return loss.
The S11&S21 LogMag diagram should have the black and turqoise curves interchanged.
As every function are the wrong one, also the smith chart then use the wrong parameters when doing calculations for the plot.

It's probably software for another VNA that have been used. Check the VNA type and model number and try and find the correct software for it.

/Ubbe
I have to admit - I hadn't even considered this as a possible problem! But it makes sense. Thanks Ubbe!
 
Status
Not open for further replies.
Top