Single-sideband (SSB) transmission doubles spectral efficiency versus AM by suppressing the carrier and one sideband. Three methods generate SSB: the filter method (needs a razor-sharp crystal filter), the phasing method (needs a wideband 90° audio phase shifter — hard to build accurately), and the Weaver method (Donald Weaver, 1956), which sidesteps both problems using two mixing stages and lowpass filters.

How it works: The audio input (say 300–3000 Hz) first mixes with a subcarrier placed in the middle of the audio band — typically 1650 Hz — using two mixers driven by sine and cosine of that subcarrier. This produces I/Q baseband signals centered at DC. Each channel passes through an identical lowpass filter with cutoff at half the audio bandwidth (~1350 Hz). The filtered I and Q signals then mix with sine/cosine of the RF carrier, and the outputs are summed (or subtracted) to select USB or LSB.

Why it's clever: The lowpass filter cutoff sits at a benign frequency — no ultra-steep skirt required. And unlike the phasing method, there's no need for a broadband audio Hilbert transformer; the audio 90° shift is only needed at the single subcarrier frequency, trivially generated by an RC network or divide-by-4 flip-flops from a 6.6 kHz clock. Sideband suppression depends on amplitude and phase matching between the two channels, just like phasing SSB, but the matching burden shifts to two lowpass filters, which are far easier to match than wideband allpass networks.

Real-world example: Modern software-defined radios (SDRs) like the FlexRadio 6000 series or any GNU Radio SSB transmitter implement Weaver-style modulation digitally. The DSP performs I/Q multiplication and FIR lowpass filtering with matched coefficients — achieving 60+ dB sideband suppression that would require expensive crystal filters in analog. The historic Collins KWM-380 used analog Weaver modulation in the 1970s.

Rule of thumb: Sideband suppression in dB ≈ 20·log₁₀(2/√(Δa² + Δφ²)), where Δa is fractional amplitude mismatch and Δφ is phase mismatch in radians. For 40 dB suppression, you need amplitude matching within 1% and phase matching within 0.6°. For 60 dB, tighten both by 10× — which is why digital implementations dominate.

Gotcha: Any DC offset in the I/Q baseband signals leaks through as an unsuppressed carrier at the subcarrier frequency offset from the RF carrier — appearing as a whistle in the middle of your passband. AC-coupling the audio before the first mixer prevents this.