Adapting an HF attenuator to higher power

I want to make a 25W HF amplifier that can handle an input up to 8W. I found a schematic by ON6MU which can handle up to 2.5W input. There is an attenuator at the input that I copy here. R3-4 is 1W, R2 and R5 are both 0.5W.

What would be needed to change this circuit to allow for an up to 8W input? Would it be enough to take an R3-4 that can handle 5W and R2 and R5 that can handle 2W, or would I need to change values as well?

How would I actually calculate this?

The attenuator should work for 17m - 80m bands.

simulate this circuit – Schematic created using CircuitLab

The full schematic, for reference purposes:

• Why not make your 25 watt amplifier have more than 5 dB gain then you can input something smaller than 8 watts? I mean we're only talking 17 MHz as the maximum frequency and there's no need for 8 watts at the input surely? Please justify otherwise. – Andy aka May 9 '17 at 10:12
• @Andyaka I'd like to connect it to my QRP sets which output 5W and sometimes a bit more. I can reduce their output power by reducing their supply voltage, but this way it's easier to switch between with/without PA. – Keelan May 9 '17 at 10:15
• Looks to me like R2 would have to dissipate the most energy. – JRE May 9 '17 at 10:16
• I feel like a total noob here, but: at 17 MHz, why not simply go for a resistive voltage divider? You want to waste power, so the impedance mismatch at the PA can actually be your friend? – Marcus Müller May 9 '17 at 10:19
• @MarcusMüller this is over my head, but on the webpage I linked there's some explanation about that: "I made provisions to include an RF attenuator consisting out a Pi network of R2, R3/R4, R5 which gives a Forward Attenuation of 3.63 dB and a Input Return Loss of 23.23dB. There are numerous of reasons why I implemented it in this design. It improves overall linearity, achieves some "protection" and enhances stability of the drive input (being a transmitter, transceiver) and Q2 gate." – Keelan May 9 '17 at 10:21

You'll need 5 dB overall gain drop, so your options here seem to be the following ones:

1. SIMPLEST (and straightforward) OPTION): use an external 50 $\Omega$ 5 dB attenuator capable of handling 8W input power. Design and build it from scratch using one of the many available online calculators.

2. SIMPLE (but not straighforward) OPTION: Increase the input attenuation by 5 dB. This means changing the values of $R_2$, $R_3$, $R_4$ and $R_5$, and also means changing the power rating of the resistors because they will dissipate more power. The input attenuator also provides impedance matching, so you should recalculate these values taking into account $R_8$ and the input impedance of the MOSFET gate (according to the datasheet: 180 pF @ 1 MHz).

3. COMPLEX OPTION: Reduce the gain of the transistor by 5 dB (this is is also a more power-efficient option). You achieve this by changing the P1 potentiometer setting so that the MOSFET gate voltage is lower and the bias current is reduced. You can adjust the gain to a lower setting during the tuning process. In this case you don't have to change the values of $R_2$, $R_3$, $R_4$ and $R_5$, but you must change their power rating along with the power rating of $R_8$ and the current rating of $L_1$. You must also check that the gate-source voltage doesn't exceed the +/20V voltage rating of MOSFET. At 8W input, this a very real possibility and could mean this option is not feasible.

4. EVEN MORE COMPLEX (but possibly optimal) OPTION: A combination of option 2 and 3. Split the 5 dB gain drop between the input attenuator and the transistor gain itself.

For reference purposes: IRF510 datasheet.

• I think I will go with the first option (external attenuator). I calculated the values using this calculator (180 Ohm for shunt resistor; 30 Ohm for series resistor). This option is the simplest, but it also makes it the easiest to enable/disable this external attenuator, in case I later want to drive the PA with less input power (to waste less power). Thanks a lot for your excellent overview. – Keelan May 10 '17 at 7:53