Silicon, germanium and LED clipping diodes

What you're hearing

Each diode has its own knee. A diode passes almost nothing until the voltage across it nears its knee, then conducts hard. One way to compare knees is the voltage at which each passes 1 mA. In the engine's models, that's 0.58 V for the 1N4148, 0.29 V for the 1N34A and 1.56 V for the red LED. Through the 2.2 kΩ resistor, the diodes hold every peak a little above those voltages, so each one clips at a different height.

Silicon: a wide bandgap and a small leak. Silicon's bandgap is 1.14 eV, the energy an electron needs to break free and carry current. Few electrons manage it at room temperature, so the diode's saturation current, Is, is tiny: 2.52 nA in this model. It takes about 0.58 V to push 1 mA through. Each tenfold rise in current then needs about 0.10 V more (n·Vt × ln 10, with n = 1.752).

Germanium: a narrow bandgap, so an earlier knee. Germanium's bandgap is 0.67 eV, about half of silicon's. Far more electrons are free, so its Is is 200 nA, about 80 times silicon's, and 1 mA flows at only 0.29 V. This model also has 7 Ω of series resistance, so as the current grows, its tops keep leaning upwards instead of going flat.

LED: a wide bandgap that makes light. A red LED's material has a bandgap high enough that each electron crossing it gives off a photon of red light. Red light runs from 610 nm to 760 nm, about 2.0 eV down to 1.6 eV per photon. A red LED's forward voltage, 1.63 V to 2.03 V, matches that. In this model, 1 mA flows at 1.56 V. Its n is high (3.73), so each tenfold rise in current needs 0.22 V more. That's a wide knee in volts, but narrow next to a 1.56 V clipping level.

Compared at the same level. Swapping diodes changes the volume a lot: germanium is 5.4 dB quieter than silicon, and the LED 8.3 dB louder. So we set the output to match the peaks, as the Circuits 1 lesson does, and compare the shape alone.

What the table says you hear. Germanium starts bending at about a quarter of silicon's input level. It also needs 10.3 dB more drive to go from a light squash to a heavy one, against 7.2 dB for silicon, so its clipping comes in more gradually: softer. The LED stays clean until 1.44 V in and then goes from light to heavy in 6.5 dB, the quickest of the three: harder, and with more headroom before it starts. At the same 6 V in, germanium is driven furthest past its knee and the LED least. So at one drive setting, germanium sounds the most distorted and the LED the cleanest.

Is germanium warmer? Partly. Its knee is earlier and more gradual, as the table shows. Its tops also keep rising as you push: from 6 V to 40 V in, its peaks rise 3.6 dB, against 1.3 dB for silicon and 1.1 dB for the LED. Even harmonics, often called warm, come from the two halves of the wave clipping differently. Matched diodes don't do that, whatever they're made of. Real parts are never perfectly matched, so a real pedal has a little of it. In SynthMaker every 1N34A is identical, so you build the mismatch on purpose: Circuits 1's germanium bass uses one diode one way and two the other. Much of the rest of the argument comes down to level. A straight swap from silicon to germanium drops the peaks by 5.4 dB, which is why fair comparisons match the level first. The diodes here also have no temperature, ageing or part-to-part spread, so this page shows the knee itself, not how real parts vary.

Try this

• Turn input from 0.5 V to 40 V and watch the Scope. With silicon, the peaks only rise from 0.43 V to 0.72 V. Past about 1 V in, the wave is squashed by half or more.
• Hold input at 1 V and listen to the tops. The peaks are 0.52 V, already rounded but not yet flat: the knee itself. Then go up to 6 V to hear the tops flatten.
• Switch between the three buttons above the circuit. Each sets its own output level, so the peaks match: at the same input, germanium is fuzzier and the LED cleaner and more open.
• Open it in SynthMaker to change the parts. Type 1N34A or LED in place of 1N4148 on D1 and D2, and set output to 0.35 V or 1.7 V to match the level.

The parts list

circuit Diodes  setting input: 6 V, from 0.5 V to 40 V  setting output: 0.65 V, from 0.1 V to 2 V  R1 in out 2.2k  C1 out gnd 10n  D1 out gnd 1N4148  D2 gnd out 1N4148end

How we checked it

Checked against ngspice, a standard circuit simulator: this circuit with each pair of diodes, driven by a 1 kHz sine at 4 V and 6 V, is within 0.2% of ngspice across the audible band (up to 20 kHz) at the 2× rate it runs at.

Build it step by step

• The lesson: A distortion pedal from six parts (opens in a new tab). Ned swaps the diodes for germanium, then red LEDs, matching the level each time, and ends with a lopsided germanium bass.
• The video: This distortion pedal is six lines of text: a real circuit, simulated. It loads only when tapped.

This distortion pedal is six lines of text: a real circuit, simulated

Next and previous

• Start of the chapter: How a diode clipper works

All circuits

Further reading

• The diode model (Shockley's equation with Is, n and a series resistance, as SPICE uses it): ngspice documentation; background in Wikipedia: Shockley diode equation.
• Bandgaps of silicon and germanium: Wikipedia: Band gap (its table of materials).
• LED colours, materials and forward voltages: Wikipedia: Light-emitting diode (its table of colours).
• D. T. Yeh, J. S. Abel and J. O. Smith, "Simulation of the diode limiter in guitar distortion circuits by numerical solution of ordinary differential equations", DAFx 2007. PDF