0
\$\begingroup\$

I have a problem with a power amplifier from this application note.

enter image description here

I am wondering about the elements shown in the circuit in the "Application Circuit" section.

Can these elements be placed differently? What is their use? For example, why are TL1, TL2 and TL3 in this location? Can these transmission lines be located differently?

I didn't find an explanation of the use of these elements in the application note, and I'm not entirely sure why they are there and what they are for, and if all of them are needed.

\$\endgroup\$
3
\$\begingroup\$

Applications circuits are designed so that they will definitely work, and so that the manufacturer and the customer have a common reference circuit if any questions arise.

TL1 - TL3 act like inductors. TL1, TL2 & C1 form a low-pass network, presumably to hold down on harmonics. I'm not sure what the function of TL3 is -- possibly impedance matching, or maybe just to get the power to the edge of the board. C2 is a DC blocking capacitor.

These are probably not all needed if you're using this amplifier in some circuit of your own design that's not meant for the canonical 50 ohms in and AC-coupled 50 ohms out. But whether that's true for your circuit depends on what you're trying to build, and what tradeoffs make sense for your circuit.

\$\endgroup\$
5
  • \$\begingroup\$ Thank you so much for your reply, great comment! I plan to use this amplifier as a power amplifier for a WiFi network. I was thinking about input and output matching and adding a bandpass filter for the 2.4 GHz range. \$\endgroup\$
    – MagicMan
    Dec 28 '21 at 21:46
  • \$\begingroup\$ Then I have one more question - I understand that the input impedance matching must be at the input of the circuit (as the name would suggest), but in that case, when I add the bandpass filter to the circuit, should it be after the TL3 and C3 lines? And should the output impedance matching then be after that filter or before it? \$\endgroup\$
    – MagicMan
    Dec 28 '21 at 21:47
  • 1
    \$\begingroup\$ That's an excellent question. But stackexchange is a Q&A site, not a discussion forum. The policy here is to have a clearly stated question with clear answers. Please ask that as a separate question -- and hopefully you'll get some answers. \$\endgroup\$
    – TimWescott
    Dec 29 '21 at 0:46
  • \$\begingroup\$ okay, thank you so much \$\endgroup\$
    – MagicMan
    Dec 29 '21 at 0:57
  • \$\begingroup\$ Hello! I made another question for this exact problem: electronics.stackexchange.com/questions/604092/… :) \$\endgroup\$
    – MagicMan
    Jan 12 at 21:04
1
\$\begingroup\$

The C1 capacitor appears to function as termination to match the amplifier output, which is not exactly 50-ohms, to a 50-ohm transmission line.

As a result, its distance to the amplifier is somewhat important. On a Smith-chart view, the 0.036" transmission line leads to a rotation of the reflection coefficient as you traverse the line. With the "black box" being the terminated 50-ohm transmission line leading away from the amplifier, the impedance seen by the amplifier looks more like 8.91 - j14.9 ohms:

enter image description here

enter image description here

Those 0.036 inches made a detectable effect - if the capacitor were right on the pin, the amp would see an impedance of 9.46 - j19 ohms.

These diagrams come from this tool - note that I made sure to enter the correct frequency of ~2.4 GHz, and specified a permittivity of 3.48 because the datasheet specified Rogers 4350 substrate.

On the other hand, L1 and C2 are both very large, so they form a near-perfect bias tee and barely affect the impedance seen looking into the transmission line. Looking at the layout, the position of L1 might have been to get things to line up neatly in the layout; I cannot see a major effect of changing TL2 when playing with the Smith chart.

\$\endgroup\$
2
  • \$\begingroup\$ Thank you so much for your reply, great comment! I plan to use this amplifier as a power amplifier for a WiFi network. I was thinking about input and output matching and adding a bandpass filter for the 2.4 GHz range. \$\endgroup\$
    – MagicMan
    Dec 28 '21 at 21:47
  • \$\begingroup\$ @MagicMan Not a problem - I recommend using a Smith chart or similar tool to conduct a calculation with your band-pass filter, unless you happen to get one that's already matched to 50 ohms on both ends. \$\endgroup\$
    – nanofarad
    Dec 28 '21 at 21:48
1
\$\begingroup\$

Indeed, the cited document provides no explanation for the matching circuit components. But the other document, Designing w/ The HMC414MS8G PA Utilizing a Low Cost Laminated Printed Circuit Board, addresses the issue in full.

As you see from the title, the document discusses the catches of output matching when designing for low cost PCB. Specifically, a 50Ω microstrip line, when realized with 62 mil FR4's dielectric (\$ε_r=5.4\$), require 113-mil transmission line width, which can become an unwieldy requirement for some particular design. Reducing the line to a more manageable width of 35 mil increases the impedance to 82Ω, and a lowpass matching network TL1-C1 is added to transform the impedance (page 4/14 of "Designing w/ The HMC414MS8G PA ...").

The other problem with an FR4 PCB is parasitic inductance and resistance of vias, which also should be compensated by this added external circuitry.

For understanding of transmission line, capacitor and inductor component roles, this compensation circuit can be simulated with spice tline components. When transforming the component lengths into time delays, use FR4's dielectric value \$ε_r=5.4\$. An AC analysis plot shows understandable dependence on capacitor C1 value, whereas only linear-in-gain plot shows a weak dependence on L1 inductance: this 18 nH component functions mainly as a choke in the DC bias network. For this reason, you cannot get rid of this inductor, although it is not a significant part of the lowpass matching network (page 8/14 of "Designing w/ The HMC414MS8G PA ...").

All this said, you can see -- at least in simulation -- that this circuitry also improves the third order intercept parameter of the design, owing to its suppressed frequency response in the range of third harmonic (6.6GHz - 8.4GHz).

You can do without this added transmission-line-based compensation circuitry, if you use, for example, Rogers high frequency laminate RO4350 for your board, see page 3/14 of "Designing w/ The HMC414MS8G PA ...":

This laminate is specifically designed for RF/microwave circuits and displays stable properties over a broad range of environmental conditions. The board topology, shown in figure 1, is single layer with a ground plane height of 10 mils. The RF input and output circuitry is comprised of 50 Ω grounded coplanar transmission line with vias running along the transmission line to provide ground to the top metal.

\$\endgroup\$
1
  • \$\begingroup\$ Man, you're a lifesaver - I couldn't find this documentation, and you just posted it. Thank you also for your extensive comment, I am very grateful to you! \$\endgroup\$
    – MagicMan
    Dec 29 '21 at 12:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.