Bias and coupling

The capsule's supply

The capsule's output is joined to two resistors: 10 kΩ up to VCC and 51 kΩ down to GND. On their own they would make a divider and hold the node at 51/61 of VCC, about 4.2 V from 5V or 2.8 V from 3V3.

The capsule is not on its own. Its built-in transistor draws a current from that node, and the node sits lower by that current times the two resistors in parallel, about 8.4 kΩ. When sound moves the membrane the current rises and falls a little, and the node's voltage rises and falls with it. That small movement is the signal.

Slide the current and watch the node. The maker's figure for this capsule's current could not be found, so the slider covers 0 to 500 µA, the range a comparable capsule of the same size is rated for. At 200 µA the node sits near 2.5 V from 5V, and near 1.1 V from 3V3.

That is the one cost of 3V3 on this block. The capsule's own transistor needs some voltage across it to work, and a capsule that draws a few hundred microamps is left with little of it. This is worked out from the resistor values, not measured, but if a block seems deaf on 3V3 and lively on 5V, this is the likely reason.

The capacitor in the middle

The transistor's base sits at about 0.65 V, the voltage any small silicon transistor's base settles at when it conducts. The capsule's node sits somewhere else entirely. Join them with a wire and one would drag the other.

So they are joined by C6, 100 nF. A capacitor lets a change through and blocks a steady voltage: it charges once to the difference between its two ends and then holds it. After that, when the capsule's node moves up by a millivolt, the base moves up by nearly a millivolt too, but the two resting levels stay independent. That is called coupling.

It costs something at low frequencies. The slower a change, the more time the capacitor has to charge and cancel it, so rumble and slow drift are turned down. Three filters works out where.