UD2 driver self-oscillation frequency
For a modified UD2.x with R7 removed and 50 kΩ added. Leave the target frequency empty to go forwards from a capacitor value, or fill it to go backwards.
τ = C · (R26 + R5‖R6) = C · 790 Ω half period = τ · ln(A/Δ) ≈ 4.57·τ f [kHz] ≈ 139 000 / (C + 10 pF) 10 pF: the board's stray, calibrated on Mathieu's 300 pF measuring 450 kHz
Accurate to ±10 to 15 %. The 4.57 is ln(A/Δ) and depends on an offset Δ ≈ 21 mV that was estimated, not measured. At Δ = 32 mV the constant would be 152 000 rather than 139 000.
Checked against hardware: Mathieu’s 300 pF measured 450 kHz and the formula with 10 pF of stray gives 448. Rafft’s board does not check it. His 200 to 330 pF gave 446 to 470 kHz, where this formula says 662 to 409, because his oscillator is a different circuit (a MAX913 comparator and other resistors). An earlier version of this page wrongly quoted his figures as a confirmation.
Use it for the first fit, then trim on the scope. The capacitor must be C0G/NP0. X5R and X7R drift.