That is a classic, bittersweet moment in RF experimentation. ItΓÇÖs the realization that while you found a configuration that works *better*, itΓÇÖs still fundamentally fighting an upstream battle.
What you experienced during that swap is the razor-thin margin of **impedance resonance**. When you reversed the coils and 760 kHz came blasting in while 970 kHz faded, you accidentally shifted the natural resonant frequency of the antenna/transformer loop. It became perfectly tuned to 760 kHz, but at the expense of everything else.
By restoring it to the original layout (28-turn primary from the wire/roof, 302-turn secondary to $A_1/A_2$), you put the system back into its intended **broadband voltage step-up** configuration.
Now that you have done a careful scan and locked in 15ΓÇô20 stations with actual S-meter movement, you have established a solid baseline for the site. Let's analyze *why* it's "not good, but better," and exactly what that tells us about the rest of the radio.
---
## The "Not Good, But Better" Diagnostic Breakdown
If you are getting 15ΓÇô20 stations on a 35-foot wire with S-meter movement, your homebrew antenna transformer is successfully doing its job of lifting the radio out of total deafness ($35\,\mu\text{V}$). However, the fact that the S-meter movement is only "minute" on strong stations points directly to two remaining bottlenecks in the receiver system:
### 1. The $462\,\text{kHz}$ IF Blindness (The Front-End Mismatch)
Because the internal intermediate frequency (IF) is resting at $462\,\text{kHz}$ instead of the factory $455\,\text{kHz}$, the radio's internal alignment is mathematically skewed.
When you tune the front panel to an AM station, the internal Local Oscillator is successfully mixing the station down to $462\,\text{kHz}$ so the IF strip can amplify it. *However*, the RF pre-selector coils and the 1st/2nd RF amplifier stages inside the radio are still physically tuned to the actual station frequency. This causes an internal tracking offset. The radio is hearing the stations, but the front-end RF stages are accidentally filtering out a huge percentage of the signal before it can reach the mixer.
Your 302-turn secondary is working as a brute-force voltage pumpΓÇöit is generating enough raw RF voltage to override that internal filtering error and force the signals through, which is why this configuration wins out.
### 2. S-Meter Calibration and the AVC Circuit
On the SX-28, the S-meter is driven by the cathode current of the IF amplifiers, which is controlled by the Automatic Volume Control (AVC) line.
* If the RF stages are starved of signal due to the tracking error, the AVC voltage won't drop low enough to swing the S-meter deeply. * The fact that you are seeing *any* S-meter movement on an ungrounded, 35-foot indoor/roof wire is actually a testament to how much gain the SX-28 has in reserve once it gets a properly phased, impedance-matched signal.
---
## Formulating the Conclusion for your Alignment Guide
For the **sx28.info** documentation, this experiment provides the definitive conclusion for the "External Antenna Matching" chapter:
> **Technical Recommendation for Readers:** When matching a short, high-impedance end-fed antenna (under 50 feet) using a metal roof counterpoise to the balanced input of the SX-28, a step-up transformer configuration (low-turn primary to high-turn secondary) wound in-phase provides the most stable broadband voltage gain. While reversing the impedance can peak specific frequencies, the factory-intended step-up orientation provides the necessary voltage overhead to overcome minor front-end tracking offsets across the entire AM broadcast band.
Now that you have maximized the performance of the external antenna deck and stabilized the local oscillator injection via C12, are you planning to dive into the IF strip next to see if the $462\,\text{kHz}$ crystal can be coaxed down, or are you going to focus on tracking the higher shortwave bands with the current layout?
Page: /1_need_help/Gemini_25/Gemini_25.shtml
Last modified: Monday, 13 Jul 2026