I think the +/- 18V was to be able to achieve 48Vdc (18Vac * 2 * sqrt(2) ~= 51Vdc after rectification).
Three volts is right on the edge of not having enough difference for a voltage regulator – assuming you use a packaged one, and it may have problems with normal mains variations – it works between breakfast and lunch, but not between dinner and bedtime. Maybe a passive regulator, a resistor feeding a shunt zener, heavily filtered. It doesn’t need to be much and it doesn’t have to be perfectly stable. It’s re-regulated inside the microphone. It can’t be super critical because it has to survive any length of microphone cable.
Phantom is an interesting design problem. You will need carefully matched resistors to inject the voltage and that may be expensive. You also can’t turn the voltage on and off without special consideration. That collapsing tension will go through the electronics like a nuclear blast. There is a protection circuit I need to find…
My supplies are both pre-packaged 15 volt units. Bypassed with mylar film .1ufd in and out, plus the large aluminum filter cap at the input. I put the diode front to back to absolutely avoid any reverse voltage. One of the two regulators, I think the negative one, demands the mylar caps for stability, so I put them on both. I did a sound check on the voltages and the ripple is in the -65 or better range.
The transformers are both Jensen. The input is a JT-115K-E60 with double shielding.
If you can’t be compulsive, life just isn’t worth living. I assume your chip matches the output impedance of the microphone with no transformer. That’s the holy grail of microphone amplifiers. So no input coil for you.
What are you going to follow the chip with? Generally, you can only get about 60dB of gain and the sound channel needs at least another 10 or 15 to be comfortable. Also, you may not want to run the chip at maximum gain all the time.
I have two portable mixers that only have 60dB gain, total, and they all run out. The knob hits maximum and I would kill to get a little more.
This is a write-up on phantom power regulation. They bring up a good point. The phantom power supply can’t be connected to anything else…