Instead of a large resistor to set the gate voltage, it uses the leakage current of the gate to channel diode and a second on-chip diode of opposite polarity. These diodes aren’t linear in their voltage / current ratio, with current increasing rapidly as the conduction threshold is approached at 600mV or so. That’s the main problem with using K596es with larger capsules, which can put out that much signal on loud sounds.

The “resistor” in the drawing is the combined leakage of the diodes and epoxy package.

Distortion rises rapidly with input voltage, reaching 1% at 30mV or so. Considering that a large diaphragm mic can be called upon to produce 100mV or more, the circuit has to minimize gate to source voltage. A source follower or charge amp configuration will do that. The source of a Schoeps phase splitter also closely follows the gate voltage.

So let’s modify some mics and see how the various circuit topologies compare. First off, what are typical voltages around the circuit? That will tell us what currents are flowing as well. In the stock mic, the JFET source is grounded, and gate rests at 0V as well. With this FET, source and gate MUST be kept at the same voltage.

FETs without built-in diodes can be biased to set a desired drain current. For the 596, we must live with what the factory gives us, at Idss. Ordinarily, we would like a higher current for lower noise, but as it turns out, the very low gate capacitance of the small chip partially makes up the difference by lowering the load on the capsule.

Can we get under 1% THD at loud levels out of a K596 hooked up as a follower? Can the rest of a Chinese mic circuit get loud? Are we polishing a turd? Measure and see.

Clipping begins on negative peaks at about 7.6V peak to peak. That’s plenty, though a few mics do better.

Second harmonic at 1V RMS (2.8V P/P, Purple) input to the gate is about -46dB or 0.5% and 3rd harmonic is about -60dB or 0.1%.  The mic has 6dB gain due to the balanced output.

At 100mV input, (blue) 2nd is -60dB below signal and 3rd is -90dB below. At 10mV (yellow) distortion is almost lost in the noise floor, and at 1mV (green) it is lost in the noise.

Charge amp 2nd harmonic

Schoeps 596 phase inverter 2nd

596 source follower 2nd

3rd harmonics

Charge amp

Schoeps

Follower

The Schoeps circuit has flat response rolling off 1dB at 20Hz using OEM 1uF coupling caps to PNP output transistors. The charge amp and follower circuits retain the PNP phase splitter and sharp 20Hz low cut filter of the stock mic.

Over all, using the K596 as a follower as seen here is the cleanest mod for typical inexpensive mics. For some harmonic color, a transformer-less charge amp (TL/C) can add spice and is a very easy mod. The K596 can drive the PNP output followers of a straight Schoeps circuit, but isn’t capable of the low distortion of some larger, higher current FETs.

There are FETs currently in production for mics, but they are in tiny surface mount packages. Most are like the 2SK596, low current, low capacitance types for internal mounting in small electret capsules. Hobbyists will either need to invest in equipment for surface mount components, or search for stocks of old parts. Obsolescence hasn’t stopped tube enthusiasts. In fact new tubes are being produced in boutique factories. Linear Systems is doing the same for JFETs, so there is hope, but expect to pay boutique prices instead of pennies for such parts.