Homebrew MOSFET Linear with Harmonic and Parasitic Output
You have built a push-pull MOSFET linear amplifier using two MRF150 devices operated from a 50 V, 20 A switching supply. The target is 600 W PEP on 80 m. With 5 W of drive into a 50 ohm dummy load, the output is 550 W PEP. A spectrum analyzer shows the third harmonic at -27 dBc and a broad hump around 48 MHz that appears only when drive exceeds about 3 W. The output low-pass filter is a three-pole Chebyshev with a 4 MHz cutoff. The bias is set to 0.7 V per device, with 250 mA total quiescent current. The amplifier is stable into the dummy load but becomes unstable into a mismatched antenna. You have a 100 MHz oscilloscope, a 50 ohm dummy load, and a spectrum analyzer to 1 GHz.
1. Diagnose the harmonic and the 48 MHz hump. Which is a filtering problem and which is likely an instability?
Read the worked answer
The third harmonic at -27 dBc is far too high for a compliant transmitter and is a filtering problem. A three-pole Chebyshev with a 4 MHz cutoff is too close to the 80 m band edge and does not provide enough attenuation at 10.5 to 12 MHz. The 48 MHz hump cannot be a harmonic of an 80 m signal; it is far above the operating frequency and appears only above a drive threshold, which points to a parasitic oscillation or inadequate neutralization in the push-pull stage. The fact that it is stable into a dummy load but unstable into a mismatch also supports an RF-instability diagnosis rather than simple harmonic generation. The switching supply could contribute broadband noise, but a narrow hump at 48 MHz that depends on drive level is more characteristic of a parasitic mode.
2. What tests would you run to separate power-supply noise from RF oscillation?
Read the worked answer
First, terminate the amplifier input in 50 ohms and run it with no drive at full bias; look at the output on the spectrum analyzer. If the 48 MHz hump is present with no drive, suspect bias decoupling, grounding, or the switching supply. Next, replace the switching supply with a clean linear supply or a battery and repeat the test. Then apply drive slowly and note the exact threshold at which the hump appears; a sharp threshold suggests oscillation, while a gradual rise suggests supply hash or poor filtering. Use a scope probe with a short ground lead to check the gate and drain waveforms for RF bursts at 48 MHz, and add a 6 dB pad or reduce drive to see whether the oscillation disappears. Also check the supply output with a scope and a spectrum analyzer to see whether 48 MHz is present on the DC rail.
3. What design changes would you make to meet spurious-emission requirements and stabilize the amplifier?
Read the worked answer
Replace the three-pole 4 MHz filter with a sharper low-pass filter, such as a five- or seven-pole design with a cutoff around 4.5 MHz, and verify the harmonic attenuation on a dummy load at full power. For stability, add ferrite beads or small series resistors at the gates and drains, improve decoupling with low-inductance capacitors close to the devices, shorten ground returns, and shield the input from the output. Check neutralization and layout symmetry in the push-pull stage; MRF150s have enough gain at HF to oscillate if the layout is poor. Reduce the stage gain with a resistive pad or feedback if necessary. Test into a mismatched load only after the amplifier is stable into 50 ohms, and verify the 48 MHz hump is gone at all drive levels before connecting the antenna.