You've caught a technical error in my previous draft. I am glad you asked for the specific component, because trying to trace it out proves the mistake: **there is no isolated DC return line for the local oscillator.**
I incorrectly conflated the oscillator's bias with the AVC bus used on the RF and Mixer stages. The Local Oscillator (V4, 6SA7) does not receive AVC voltage; it relies entirely on its own grid-leak bias.
In the SX-28, the oscillator grid (Pin 5 of V4) develops a negative DC voltage across its grid-leak resistor (**R17**, a 50k or 100k resistor depending on your exact schematic variant). The bottom of R17 connects directly to the chassis ground. Because one side is hot with RF and the other side is a hard ground, there is no isolated DC node available to measure.
Since we can't measure from an isolated point, we have to create our own isolation to prevent the DMM leads from detuning the circuit on Band 6.
Here is the corrected HTML for that specific section. You can replace the previous "Band 6 Probe Trap" block with this one:
The Band 6 Probe Trap
Do not attempt to use the oscilloscope probe directly on the oscillator grid for Band 6 (21 to 43 MHz).
At these extreme frequencies, the 10 to 15 pF of parasitic capacitance introduced by your 10X scope probe acts like an AC short circuit (roughly 250 ohms of reactance at 42 MHz). Touching the probe to the grid will instantly drag the Q of the tank circuit down and kill the oscillation entirely, or cause it to start and stop erratically.
To adjust C7 (Band 6 Feedback): Because the Local Oscillator (V4) relies entirely on its own grid-leak bias (R17, which goes directly to ground), there is no internal isolated DC return line to measure from. You must create your own isolation.
Take a standard 1 Megohm to 2.2 Megohm axial resistor and clip it directly to the tip of your Digital Multimeter (DMM) positive probe. Keep the resistor leads as short as practically possible.
Touch the free end of this isolation resistor to Pin 5 of V4 (6SA7), and
connect the DMM ground to the chassis. The physical resistor introduces less
than 1 pF of parallel capacitance, blocking your meter's heavy capacitive load from detuning the RF tank while allowing the DC grid-leak bias to pass through. Sweep the main dial from the low end to the high end, and adjust C7 so this negative DC voltage remains as flat as possible.
This ensures the web page gives accurate theory and a practical bench solution without sending a restorer on a wild goose chase for a non-existent test point.
This recoagnized the error but the solution is impractical.
Page: /Alignment/Osc/Gemini_pages/Gemini_1403.shtml
Last modified: Monday, 27 Jul 2026