As you will most likely know, a condenser microphone capsule can only generate an AC voltage if it is biased with a polarization voltage. This polarization voltage can be permanent, as is the case with pre-polarized (electret) capsules. Or it can be applied externally to the capsule, sometimes referred to as a “true condenser” microphone. An external polarization voltage can be derived directly from the Phantom power supply voltage on the XLR signal pins, or can be supplied by an an internal or external voltage source. In this article I will limit myself to the internal voltage source, as we usually encounter it in condenser microphones that operate on P48 Phantom power. And more specifically, I will focus on the implementation with a CMOS oscillator, because in my opinion that is the most reliable implementation. A commonly used generator is based on the Hartley oscillator, but I have had several microphones where, for no apparent reason, this oscillator did not work. Since I consider reliability to be one of the most important features of a well-designed microphone circuit, I focus primarily on a type of oscillator whose operation is guaranteed by its design: the Schmitt Trigger Relaxation Oscillator. It is also the most simple execution of a (digital) oscillator, so why choose something else?
A popular circuit charge pump circuit used in many DIY microphone designs is the one described by Rory Holmes. However, I found the circuits described to be needlessly complex if the only requirement is that the circuit should be able to supply a high voltage. The idea of Rory’s circuit is that it should be able to provide some current to a circuit that must be powered by the charge pump. Rory’s CD40106-based circuit also limits the number of multiplier stages to the number of gates available in the CD40106. In most cases, it will be sufficient, but if you want a higher voltage than the circuit can deliver, you’ll have to add another CD40106. If you would just use a Capacitor-Diode ladder network as described in Rory’s introduction and tie that to a CMOS oscillator, you can infinitely add multiplier stages to obtain higher voltages, without having to add more gates.
Another commonly known circuit can be found in Rode’s NT1-A’s circuit. It also uses a CD40106, but in a different configuration than in Rory’s circuit. You can infinitely add multiplier stages to this design, if needed. But I just fail to understand why gates have been put in parallel: it will only draw additional power. It is not required to make this circuit work properly, as some people claim (sorry, I cannot find those posts on the web anymore).
The charge pump circuit described here uses two Toshiba TC4S584F Schmitt-trigger gates instead of the commonly used CD40106. As depicted here, it quadruples the supply voltage, less the diode drops. At 18V, you’ll get a ~69.5V polarization voltage from the output. Power is derived from the internal power supply of the microphone. An example of this circuit is depicted below, taken from the KM84++ design.
The output of the charge pump is rectified and filtered by D5B and C6 and additionally filtered by R8 and C10 before it goes to the capsule via the x-GOhm capsule bias resistor (not shown). With the trimmer potmeter in the supply line of the circuit, you can adjust the output voltage and, hence, the sensitivity of different microphones, thus allowing you to make matched pairs.
The table below lists the output voltage of the quadrupler circuit shown above as a function of the supply voltage. It also lists the current drawn from the power supply. The supply current increases as the square of the supply voltage, so if there’s not enough supply current available, it can be beneficial to lower the supply voltage to the charge pump circuit and add some more stages to obtain the desired output voltage. Output voltage rises linearly with the supply voltage Vdd.
An important point to watch out for in any oscillator circuit is that the high-frequency currents and EM fields do not end up in your audio path. They can otherwise cause direct or indirect intermodulation products. So standard EMC guidelines apply here! Briefly summarized, these are
- Keep the circuit compact: keep current loop areas as small as possible. This in practice means that decoupling capacitors should be placed as close as possible to ICs and the return current paths should be adjacent to the signal lines. This is best achieved through a ground layer.
- High-frequency ground currents of the oscillator should not share a common ground path with the audio circuit. Attach the oscillator ground plane to the audio ground at one point.
- Keep the oscillator circuit as far away as possible from the audio circuits and traces, though that may prove difficult in compact SDC designs.
What I mean by “compact” is best illustrated by a universal generator board that I designed for use in LDCs and can be placed as a daughterboard on the main PCBA. It measures only 10 x 20 mm and can generate almost 80V from the 18V zener power supply that is also integrated into the PCBA. I will make this PCBA available through PCBWay. You can check availability and find the link here.
