matching network

[hunas2127]

Are components like the RF choke, bypass capacitor, and gate bias resistor included in the matching network?

After obtaining the optimal load and source impedances (Zload_opt, Zin_opt) for a transistor through load-pull and source-pull simulations, how should these bias-related components be considered when matching to 50 ohms?

If they are to be included in the matching network, how should components like a bypass capacitor — which is connected in parallel — be treated on the Smith chart, since parallel elements can’t be directly plotted in the same way?

 

[vfone]

Those LC components are indirectly part of the matching network, and their value and their SRF could affect the impedance matching over frequency.

 

[FvM]

Example circuit isn't well suited to discuss the question. In a meaningful bias circuit, C1 and C2 have sufficient capacitance to consider the nodes as RF short, so they can be ignored in calculation. If the condition isn't met, the complete supply network has to be included in impedance analysis, effectively impossible.
R1 = 50 ohm doesn't look like a useful choice, either.

Designation "DC_Feed" suggests a choke with X >> Z0, so matching analysis can ignore it. A real DC choke has parasitic elements, e.g. parallel capacitance that may be considered in matching fine tuning but should not dominate matching network behaviour.

Alternatively drain bias can use a finite RF inductance that's dedicated part of the matching network. In this case, DC short at its cold end is often supported by a transmission line stub or multiple paralleled capacitors. The idea is that only the finite inductance appears in Smith chart.

Thus C1 and C2 shouldn't be encircled.

 

[vfone]

In my opinion the original question is very pertinent. In a real world, in a well-analyzed circuit, all the components (out of those that are part of the main matching network) can influence the impedance matching vs frequency. Since no frequency range is described, the question is general, and is nothing wrong with this..

 

[volker@muehlhaus]

R1 = 50 ohm doesn't look like a useful choice, either.

... especially because it will present a 50 Ohm load to RF input signals, because SRC1 has 0 Ohm source resistance wideband from DC to GHz. This is very different from real world behaviour of bias path.

 

[FvM]

the question is general, and is nothing wrong with this..

Agreed. OP is asking "do all circuit components affect matching" and the general answer is yes. Next question is, "if so, how do you calculate it"? General answer could be: by exact numerical simulation.
I believe, art of RF design is answering the question differently, you'll design the circuit so that only a few components significantly affect matching, e.g. not C1 and C2 in above example, preferably neither R1 or DC_Feed. Then matching can be determined with intuitive methods like Smith chart.

 

[dick_freebird]

For sure, the more components you allow to matter the worse the optimization effort (and probably outcome).

 

[hunas2127]

동의합니다. 질문자님은 "모든 회로 구성 요소가 매칭에 영향을 미치는가?"라고 질문하셨는데, 일반적인 답변은 "예"입니다. 다음 질문은 "그렇다면 어떻게 계산하시나요?"입니다. 일반적인 답변은 정확한 수치 시뮬레이션을 통해 계산하는 것입니다.
RF 설계 기술은 이 질문에 대한 답을 다르게 제시한다고 생각합니다. 몇 가지 부품만 매칭에 큰 영향을 미치도록 회로를 설계해야 합니다. 예를 들어, 위의 예에서 C1과 C2는 제외되고, R1과 DC_Feed도 가급적이면 제외됩니다. 그러면 스미스 차트와 같은 직관적인 방법을 통해 매칭을 판단할 수 있습니다.

Does this mean that after designing only R1 and the DC feed, the values for bypass capacitors like C1 and C2 should be determined later by observing the Smith chart?

--- Updated Apr 28, 2025 ---

... 특히 SRC1은 DC에서 GHz까지 광대역으로 0옴 소스 저항을 가지므로 RF 입력 신호 에 50옴 부하를 가하게 됩니다 . 이는 바이어스 경로의 실제 동작과는 매우 다릅니다.

I added the resistor on the gate bias side because I heard it helps prevent oscillation.
Through load-pull measurements, I confirmed that the input impedance is very small.
In this case, doesn't it mean that the resistor has little effect on the combined impedance?
For example, 50 + j0 in parallel with 2 + j1 results in 1.94 - j0.92.
On the drain side, a large current flows, so a transmission line is typically used.
However, on the gate side, the current is very small, so I judged that it would have little impact.
If this understanding is incorrect, I would appreciate it if you could explain it to me.