The one "large" open item on the Hybrid Wireless is the final design for the 12GE5 Final RF Amp stage. It is also the only module not already fully built and tested.
The urge to use copilot AI loomed large and so there I went. Perhaps (and most likely) the results would be different using ChatGPT, or Gemini or the ham's best buddy, Claude.
Whether AI or not, the best answer to a question always starts with the premise of the input. So, I asked about a design given the plate voltage (500VDC), the screen voltage (280VDC), the bias point (-30VDC), the operating frequency (7 MHz), the class of operation (AB) and that the output stage was a Pi Network.
From Copilot
Provide a design of a linear RF amplifier stage using a 12GE5 with a plate voltage of 500VDC, a screen voltage of 280VDC and a bias of -30VDC operating at 7 MHz using a Pi-Network output circuit. Also provide an expected output power assuming class AB operation.
Absolutely! Let’s walk through a practical design for a linear RF amplifier stage using the 12GE5 beam power tube, tailored for 7 MHz (40-meter band) operation in Class AB mode.
📐 Design Overview
✅ Tube Parameters
Tube: 12GE5 (beam power tetrode)
Plate Voltage (Vᵖ): 500 VDC
Screen Voltage (Vg2): 280 VDC
Grid Bias (Vg1): -30 VDC
Operating Frequency: 7 MHz
Mode: Class AB linear amplifier
⚙️ Circuit Design
1. Input Stage
Drive Requirement: ~10–15W RF input for full output
Input Matching: Use a tuned grid circuit with a low-Q parallel LC tank resonant at 7 MHz
Grid Leak Bias: Fixed bias via -30V supply, with 10kΩ grid resistor to ground
2. Plate Circuit
Load Impedance: For Class AB, target ~2.5–3.5 kΩ plate load
Tank Circuit: Series LC tuned to 7 MHz
DC Blocking: High-voltage capacitor (e.g., 1–2 kV rated, 1000 pF)
3. Pi-Network Output
Used to match the high plate impedance to a 50Ω load.
| Component | Typical Value (for 7 MHz) |
|---|
| C1 (tune cap) | 100–300 pF variable |
| L (coil) | ~2.5–3.5 µH (air-wound) |
| C2 (load cap) | 500–1000 pF variable |
The response did say I should expect 30 to 35 watts PEP output with an efficiency of around 60%. The data above is very consistent with what I thought the values of the components and the tank circuit should be and the expected output.
But then the problem hit the fan when I was asked would you like a schematic. I said yes, and it did generate a schematic which I won't show here as it was totally wrong.
What was generated was a schematic of a Triode, no plate blocking capacitor, a large value resistor in the Cathode circuit (the Cathode has a meter in the circuit which for AC is bypassed to ground) and some capacitors labeled with resistor values.
I inputted that the schematic was wrong and then got "Oh you are so right". I stopped there.
So, I saw the only value of the AI was as a check on what I already knew but of no use in providing me a build schematic. The AI between my ears is really good. Maybe Claude would have done better.
Your mileage may vary but anything from the AI world must be suspect as to correctness. Not unlike what we are told by certain people. You must ask am I really better off as you go through the grocery store checkout line.
Them that know can make things go.
73's
Pete N6QW
