Pi Matching circut for tube PA - what values?

I am working on my 80m and 40m CW transmitter and ran into a problem:

I need to match the high impedance of PL504 to 50ohm coax, a Pi match seems to be the best, but I don't know how to calculate the inductor - the calculators online give capacitor values of several nF, too big for tuning caps and sometimes they just say NaN.

Operation class is C and I'd like to get 20W of power at least.

• What is the best value for plate RF choke?
• Can it be made with a SMPS power core rated for 200kHz?
• Are there any general errors?

• What frequencies are we talking about? – Marcus Müller May 26 '19 at 12:06
• @MarcusMüller Ah yes, forgot about it, 3.5MHz and 7Mhz. – thegamebusterpl May 26 '19 at 12:10
• Hm, in that case, a core rated for 200 kHz will probably not work very well, will it. – Marcus Müller May 26 '19 at 12:13
• @MarcusMüller Even as a RFC? – thegamebusterpl May 26 '19 at 12:15
• oh wait, sorry, I mixed that up and thought it was the inductor in your $\Pi$ matcher – Marcus Müller May 26 '19 at 12:15

Inspection of figures 17.17 through 17.19 in the ARRL Handbook suggest that the components in your Pi network might need to be adjusted:

Jim Tonne, W4ENE, wrote a utility called PIEL to calculate output tank values. A Pi-L network - a Pi network with an L-network extension - is frequently used to match a wider range of impedances than a Pi network alone can handle.

To use the program, you must know the load represented by the plate of the tube. PIEL includes a "Plate load help" button to perform this calculation using your intended voltage swing, RMS output power and class of operation.

Assuming you want the plate to swing over half the 100V supply to produce 20W in class C, PIEL calculates the plate impedance to be 62.5$$\Omega$$. Entering this value in PIEL's main screen with an intermediate impedance of 50$$\Omega$$ to drop the L-section yields this Pi network:

Increasing the plate voltage swing and reducing the network Q reduces the capacitor values. You can experiment with these values until you find a network you can build with available components. For example, increasing the swing to 90V and reducing the Q to 3 produces this network:

But, you need to keep in mind that the network may no longer provide the harmonic rejection required to comply with FCC rules.

• Which one --if any-- has the correct pi-network values for the OP's circuit? – Mike Waters May 27 '19 at 0:40
• @MikeWaters It depends on the OP's goals: e.g., willingness to change design, need to use components on hand, harmonic attenuation provided by added networks (antenna matching unit), etc. This is a significant engineering question with a lot of "knobs" to turn. – Brian K1LI May 27 '19 at 15:30
• @BrianK1LI I want to use components at hand, I have 4 section 330pF, 360pF, 2x 15pF tuning caps, I also wind my own coils. I could use a diffrent design if it means I can use these tjning caps. The tuning section will be the part of a QRP 20W TX, shown in the OP. – thegamebusterpl May 31 '19 at 6:15
• @BrianK1LI Also - can I increase the inductance (to let's say, 8uH) and use slightly smaller capacitors? BTW. My call sign is SO9LEK – thegamebusterpl Jun 1 '19 at 14:24
• @thegamebusterpl I just added your call to your profile. You can edit that if you like. :-) – Mike Waters Jun 1 '19 at 23:51

DISCLAIMER: I offer these results only as a possible solution. Completing this design may depend on many other factors, thermal and electrical, steady state and transient, component- and layout-dependent parasitics. This is the subject of, at least, sizeable chapters if not entire text books, which I will not reproduce here. Your mileage may vary.

I entered the following values into the PiEl program:

- Plate voltage swing: 190 volts
- RF power: 12 watts, RMS
- Class C (k=2)

• Plate load result: 1.504k ohms

Main page

- Frequency: 3.5MHz