Continue to Site

Is there a correlation between how bad the S22 is and how hard the antenna is to design?

A

Abdullah Sabry

Newbie level 4
Joined
Mar 12, 2025
Messages
7
Helped
0
Reputation
0
Reaction score
1
Trophy points
3
Activity points
47
I'm designing the frontend of a TX chain for a graduation project, currently designing a PA. I'm working in a narrow-band application (27GHz-31GHz) and I wanted to understand something.

Would having a poor S22 of the PA have any impact on the overall system?

From what I understand, the maximum power will be directly delivered to the antenna from a flipped chip configuration with minimal package losses. Say the PA output impedance is 100 ohms, and the required load impedance is 20 ohms so this would have an S22 of around -3.5 dB. Would the antenna gain/radiation efficiency be affected in this case?
 
Sort by date Sort by votes
If the PA output impedance is 100 ohms and the load impedance is 20 ohms, you may need a matching network to do the matching between these two impedances.
Then, if the antenna has 20 ohms impedance, you will get the maximum power transfer between the PA and the antenna.
 
Upvote 0 Downvote
vfone said:
If the PA output impedance is 100 ohms and the load impedance is 20 ohms, you may need a matching network to do the matching between these two impedances.
Then, if the antenna has 20 ohms impedance, you will get the maximum power transfer between the PA and the antenna.
Click to expand...
If I understand correctly but some PAs require a different impedance than the conjugate match for maximum power transfer, so I actually want to see 20 ohms even though the output impedance is 100 ohms. This causes mismatches at the antenna, and I'm mainly asking if this has a negative impact on the antenna performance.
 
Upvote 0 Downvote
Abdullah Sabry said:
I actually want to see 20 ohms even though the output impedance is 100 ohms. This causes mismatches at the antenna, and I'm mainly asking if this has a negative impact on the antenna performance.
Click to expand...
Without discussing how realistic the numbers are, bad S22 causes reflected waves to run back to the antenna. You may want to use a circulator or isolator to absorb reflected waves.
 
Upvote 0 Downvote
Abdullah Sabry said:
If I understand correctly but some PAs require a different impedance than the conjugate match for maximum power transfer
Click to expand...
Need to be careful with terminology here.

Conjugate matching guarantees maximum small signal power gain.

But power amplifiers are typically not intended to operate with conjugate matching (especially on the output). Typically a PA has an optimal load impedance Zopt at which it operates "optimally". What determines this "optimal" termination is up to you, the designer. But Zopt does not have to be the conjugate out Zout (the PA's output impedance).
--- Updated ---

FvM said:
Without discussing how realistic the numbers are, bad S22 causes reflected waves to run back to the antenna. You may want to use a circulator or isolator to absorb reflected waves.
Click to expand...
Can this actually affect antenna efficiency though? I don't think so, unless the path between the PA and the antenna radiates and acts as part of the antenna itself. I'm not an antenna expert though.
 
Upvote 1 Downvote
FvM said:
Without discussing how realistic the numbers are, bad S22 causes reflected waves to run back to the antenna. You may want to use a circulator or isolator to absorb reflected waves.
Click to expand...
From what I understand currently bad VSWR would mean lower efficiency because the TL to the antenna would have higher insertion loss, however it shouldn't affect the antenna's operation but I was just making sure this argument is correct.


mtwieg said:
Need to be careful with terminology here.

Conjugate matching guarantees maximum small signal power gain.

But power amplifiers are typically not intended to operate with conjugate matching (especially on the output). Typically a PA has an optimal load impedance Zopt at which it operates "optimally". What determines this "optimal" termination is up to you, the designer. But Zopt does not have to be the conjugate out Zout (the PA's output impedance).
--- Updated ---


Can this actually affect antenna efficiency though? I don't think so, unless the path between the PA and the antenna radiates and acts as part of the antenna itself. I'm not an antenna expert though.
Click to expand...
You're right, I think ideally an S22 of around -6 to -8 would be decent to get good loadpull matching with low VSWR causing no increased degradation of the packaging losses.

If the mismatch effect is taken into account when designing the PA then it won't be taken into account in the antenna's radiation efficiency from what I believe.

Thank you everyone for your help.
 
Last edited:
Upvote 0 Downvote
The intrinsic efficiency does not change, meaning the antenna's internal losses do not change. The beam aperture and gain should not change.

However the effective efficiency of the system is reduced by the forward insertion loss due to the PA's low s22 and higher Tx source impedance if unmatched.
 
Upvote 0 Downvote
D.A.(Tony)Stewart said:
The intrinsic efficiency does not change, meaning the antenna's internal losses do not change. The beam aperture and gain should not change.

However the effective efficiency of the system is reduced by the forward insertion loss due to the PA's low s22 and higher Tx source impedance if unmatched.
Click to expand...
That was just what I was looking for. Thank you for your clarification!
 
Upvote 0 Downvote
Abdullah Sabry said:
From what I understand currently bad VSWR would mean lower efficiency because the TL to the antenna would have higher insertion loss, however it shouldn't affect the antenna's operation but I was just making sure this argument is correct.
Click to expand...
This is only true if the antenna is not matched to the TL line (leading to large reflections within the TL).

D.A.(Tony)Stewart said:
The intrinsic efficiency does not change, meaning the antenna's internal losses do not change. The beam aperture and gain should not change.

However the effective efficiency of the system is reduced by the forward insertion loss due to the PA's low s22 and higher Tx source impedance if unmatched.
Click to expand...
No idea what you mean by "intrinsic" vs "effective" efficiency, AFAIK those aren't established terms for power amplifiers.

Let's make up some example numbers:

Scenario #1:
PA output impedance Rout = 50ohm
Z0 of transmission line = 50ohm
Transmission line is a quarter wavelength long
S21 of transmission line = 0.9 (-0.46dB)
Antenna impedance Ra = 50ohm
Let's assume the PA has a Popt of 1W into Ropt = 50ohm, and its input is being driven appropriately for those conditions.
In this case, there's no reflections anywhere.
The power delivered by the PA is 1W, and the power absorbed by the antenna is 0.9W.
Also, this result is completely independent of the PA's output impedance Rout. If Rout were changed to 10ohms or 1000ohms then the antenna power would not change (so long as the other conditions listed above are still true).

Scenario #2:
PA output impedance Rout = 50ohm
Z0 of transmission line = 50ohm
Transmission line is a quarter wavelength long
S21 of transmission line = 0.9 (-0.46dB)
Antenna impedance Ra = 25ohm
Let's assume the PA has a Popt of 1W into Ropt = 93ohm, and its input is being driven appropriately for those conditions.
The impedance seen by the PA is 93ohm (not 100ohms, due to the TL loss), which is equal to Ropt. So the PA still delivers 1W.
However, due to mismatch at the other end of the line (gamma=0.333), the effective insertion loss of the TL is now 0.9*(1-0.333^2)/(1-0.9^2*0.333^2) = 0.879 (-0.56dB).
So the total power to the antenna is 0.879W. Again, this has nothing to do with the PA's output impedance. If Rout were changed to 10ohms or 1000ohms then the antenna power would not change (so long as the other conditions listed above are still true).
 
Last edited:
Upvote 1 Downvote
I was referring to antenna losses (intrinsic loss , which is internal, antenna effective loss/gain depends on mismatch and aperture ratio for matched vs unmatched)

I just defined that as part of system loss but you can separate it in your loss budget.

D.A.(Tony)Stewart said:
The intrinsic efficiency does not change, meaning the antenna's internal losses do not change. The beam aperture and gain should not change.

However the effective efficiency of the system is reduced by the forward insertion loss due to the PA's low s22 and higher Tx source impedance if unmatched.
Click to expand...
No idea what you mean by "intrinsic" vs "effective" efficiency, AFAIK those aren't established terms for power amplifiers.
 
Upvote 0 Downvote
mtwieg said:
This is only true if the antenna is not matched to the TL line (leading to large reflections within the TL).


No idea what you mean by "intrinsic" vs "effective" efficiency, AFAIK those aren't established terms for power amplifiers.

Let's make up some example numbers:

Scenario #1:
PA output impedance Rout = 50ohm
Z0 of transmission line = 50ohm
Transmission line is a quarter wavelength long
S21 of transmission line = 0.9 (-0.46dB)
Antenna impedance Ra = 50ohm
Let's assume the PA has a Popt of 1W into Ropt = 50ohm, and its input is being driven appropriately for those conditions.
In this case, there's no reflections anywhere.
The power delivered by the PA is 1W, and the power absorbed by the antenna is 0.9W.
Also, this result is completely independent of the PA's output impedance Rout. If Rout were changed to 10ohms or 1000ohms then the antenna power would not change (so long as the other conditions listed above are still true).

Scenario #2:
PA output impedance Rout = 50ohm
Z0 of transmission line = 50ohm
Transmission line is a quarter wavelength long
S21 of transmission line = 0.9 (-0.46dB)
Antenna impedance Ra = 25ohm
Let's assume the PA has a Popt of 1W into Ropt = 93ohm, and its input is being driven appropriately for those conditions.
The impedance seen by the PA is 93ohm (not 100ohms, due to the TL loss), which is equal to Ropt. So the PA still delivers 1W.
However, due to mismatch at the other end of the line (gamma=0.333), the effective insertion loss of the TL is now 0.9*(1-0.333^2)/(1-0.9^2*0.333^2) = 0.879 (-0.56dB).
So the total power to the antenna is 0.879W. Again, this has nothing to do with the PA's output impedance. If Rout were changed to 10ohms or 1000ohms then the antenna power would not change (so long as the other conditions listed above are still true).
Click to expand...
So as long as the antenna is well matched to the TL connected to it then the PAs reflection shouldn't affect the antenna's performance. That sounds good, thank you for your insight!
 
Upvote 0 Downvote
Abdullah Sabry said:
So as long as the antenna is well matched to the TL connected to it then the PAs reflection shouldn't affect the antenna's performance.
Click to expand...
No, "The PA's reflection" is irrelevant no matter what.

The most important thing is that the PA be loaded with its optimal load impedance Zopt. The network connecting the PA to the antenna must transform the antenna impedance Za into Zopt.

The only reason the amplifier output impedance Zout might matter is if the antenna is a source of power when connected to the PA. For example, if the antenna is coupled to other antennas in a transmit array (I'm assuming this isn't the case for you).
--- Updated ---

D.A.(Tony)Stewart said:
I was referring to antenna losses (intrinsic loss , which is internal, antenna effective loss/gain depends on mismatch and aperture ratio for matched vs unmatched)

I just defined that as part of system loss but you can separate it in your loss budget.
Click to expand...
Could you give an example of how this depends on amplifier S22, either at a system level or component level?
 
Last edited:
Upvote 0 Downvote
mtwieg said:
(I'm assuming this isn't the case for you).
Click to expand...
This project is a beamformer IC and it is connected to an antenna array, it's a full duplex system with separate TX/RX chips and antennas. Would this change how important PA matching is?
 
Last edited:
Upvote 0 Downvote

S-Parameter Design

Learn practical S-parameter design techniques for RF and microwave applications to assist continued development of new high frequency circuit designs.
www.keysight.com

There are many different matrices including W-parameters. Let me try to summarize the above.

1742146215310.png



The total at any point , Vt, It

1742146535803.png



The traveling waves

1742146716795.png



The formation of scattering waves relative to impedance , Zo

1742146967058.png


The relationship to scattering parameters (S-parms)

1742147168205.png


The S-parm definitions and assumptions.
1742147312140.png


Using admittance
1742148231531.png



Using normalized shunt admittance.

1742147970957.png



Converting to T parameters which are independent of the other side of a 2 port.

1742148439893.png




The use of conjugate impedance matching with flow diagrams

1742148763907.png



The definition of delivered power

Pdel = |a|² –|b|²


The reflection also can be inverted if the magnitude of load Z is less than Zo or non-inverted if greater than Zo.
--- Updated ---

1742149725017.png


Ideally all interfaces should be conjugally matched.

If for some reason an interface path cannot be transformed efficiently which end do you match for Zo of the path to maximize output power?

1. The source?
2. the one which is higher? (source or load) for no reflection ...
3. The RMS median of both ?

RMS of both ends. Zo = sqrt (Zsource ²+ Zload ²)

e.g. sqrt( 100²+20²) = 102

e.g. without reactance factors or using s-parms for simplicity, Zs= 100 to Z.ant=20 ohm suggestions a path impedance of 102 ohms which seems to suggest and 2).

The simple answer is if we assume the PA input has zero reflections then s22 does not matter.
1742150498102.png
 
Last edited:
Upvote 0 Downvote
Abdullah Sabry said:
This project is a beamformer IC and it is connected to an antenna array, it's a full duplex system with separate TX/RX chips and antennas. Would this change how important PA matching is?
Click to expand...
Possibly. If there's significant coupling between antennas in an array, then that means the amount of RF power incident on a PA's output isn't just the reflection from its own output power, but also the power transmitted from other channels. If the coupling is strong enough, then this can affect the PAs' efficiency, linearity, or reliability.

In some applications, isolators are fitted to each PA to prevent the PA's from "seeing" this coupling (though the coupling does still affect the antennas).
 
Upvote 1 Downvote
mtwieg said:
In some applications, isolators are fitted to each PA to prevent the PA's from "seeing" this coupling (though the coupling does still affect the antennas).
Click to expand...
That was helpful. Thank you for your time!
--- Updated ---

D.A.(Tony)Stewart said:
The simple answer is if we assume the PA input has zero reflections then s22 does not matter.
Click to expand...
That makes sense that at least the input interface is matched. Thank you for your time!
 
Upvote 0 Downvote


Write your reply...

LaTeX Commands Quick-Menu:

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top