There is a large number of manufacturers currently selling large- diaphragm microphones from Chinese OEMS, and a large number of them are made at one Shanghai factory that produces a few variations on a standard design. A lot of other companies are importing the cases and bodies from the Shanghai factory and dropping in their own electronics. Still more companies are buying from other Chinese manufacturers that have bought tooling from the Shanghai factory or have copied their tooling.

As a result, there's a wide variety of very similar microphones on the US market, with wide variations in construction quality and sound quality depending on how they were made and for whom — and to some degree on the US company you buy from and how much you pay.

The same retrofit described in this article will work in all the mics that are designed around the Shanghai U87-like body and the Shanghai U87-like capsule. Will it turn a $99 mic into a Neumann U87? Of course not, but it will improve the sound considerably in many cases.

But not in all cases! And now we get to the disclaimers. Some mics built around this model, like the ones sold by Langevin and Groove Tubes, aren't going to benefit from this mod — their electronics are already "upgraded," and this mod will probably do much more harm than good. Furthermore, on some models you will lose some features. If your mic has the optional bass cut, either internal or external, you'll lose that. If you have the 10 dB pad, you'll lose that (and that's no loss, since with the new electronics you'll be able to get a much higher level into the mic before clipping it, even without the pad). If you have one of the dual-diaphragm models that give you an omni and a cardioid pattern, you will lose the pattern adjustment (and that is a substantial loss, I admit).

As you might expect, assembling a comprehensive list of mics that use these capsules wasn't easy, and I'm pretty sure I've missed a few. But the sidebar [in the printed magazine] will at least get you started on determining if that slick new mic you just bought at your local store might benefit from this mod.

If your model isn't listed, consult the photo — if your mic's innards look like this, the mod should work. And as always, remember that you're voiding your warranty, that the manufacturer won't service your mic any more, and so on.

The design

If you read my 'What's in a Mic' article from 9/01, you'll find this mic design very familar indeed. But unlike the Panasonic capsules, which are electrets that have built-in FET impedance converters, these capsules require an external polarization voltage to charge them up.

Click here for larger image

Schematic of the Circuit

Since they don't have an integral FET, we have to supply one. And because we are building the impedance converter, we have some circuitry on the board that has very, very high impedance and must be kept extremely clean to prevent any grease or residual flux from slightly conductive paths on the board and increasing the noise floor.

[snip] [Description of circuit removed per request of Scott Dorsey, please buy the back issue.]

R9 is a very, very high value resistor that allows only small amounts of that polarization voltage to reach the capsule. We need a very high value here to prevent as much signal as possible from being lost.

With that charge across the microphone capsule, vibrations will change the capacitance of the capsule and the amount of charge it stores will change, so the voltage measured at the input pin will change. Our whole goal here is to measure those changes.

The input amplifier stage is Q3, a FET with a very high impedance input. Now, the FET wants to see a negative voltage on the input gate in normal operation. Since we don't have a negative voltage supply, what we do is we make sure the gate is sitting at about ground level with RIO, we couple the input into the FET gate with C6, which blocks all DC, and then we lift the other pins of the FET up to a positive voltage so the gate is negative when it's idle.

We supply 12V DC from the 12V supply to the drain of the FET through RI, a 2.2K resistor, but then we also have another 2.2K resistor R2 to ground. Since there are a few volts of voltage drop across R2, it keeps the source of the FET a few volts above ground and so the gate is a few volts below it.

The whole thing about the FET is that as the voltage on the input changes, it allows more or less current to flow through it. So if there is a high level pulse sent in from the capsule, it will pull the source and drain voltages closer to one another. A positive pulse in the input causes it to conduct more, causing the voltage at the input of C3 to become higher and the voltage at the input of C2 to become lower. These two capacitors allow us to block those DC voltages, and if we looked at the other end of them, we'd see two AC signals that were exactly the opposite of one another, which is just what we need to drive the output.

But while I told you where the 35V supply for polarizing the capsule came from, I didn't tell you where the 12V supply came from. I'm not going to tell you about that yet, either.

[snip, again by request]

The input stage not only acts as an impedance converter to take the very high impedance input from the capsule, but it also acts as a splitter stage to turn that input into two opposing inputs. The output stage acts as an amplifier, giving both voltage gain and current gain to drive the output, and it uses the impressed voltage on the output as its own power supply, while also generating the voltage for the input.

Everything is very tightly interconnected, and each stage is doing several things at the same time. That's why I claim this circuit is just so incredibly ingenious. I want to meet the man who invented it.

Building it

Before you start, check the erratta sheet.

First, stuff the board. This is not all that easy to do since the board is rather tightly packed, but the layout diagram is fairly self-explanatory and should show you how to do it. Make sure you put the transistors, the zener diode, and the tantalum capacitors in the correct way. The stripe on the zener diode goes away from ground. [Note: the components go on the bottom of the board, away from the writing!]

[Follow this link to photos before and after]

Second, deflux the board. This is critical! Even a tiny amount of residual flux left on the input section of the board will increase the noise floor. Spray the board thoroughly with Chemtronics Flux-Off, then turn it around and do it again. Yes you can use 98% isopropanol and a Q-tip and a lot of elbow grease and probably get the same result, but Flux-Off is easier and more certain. If you do it by hand, make especially certain that there is no residue around the input section of the board where the FET and one G-Ohm resistors are.

Third, unscrew the base of the microphone, remove the four screws holding the PC boards in, remove the existing PC boards, and carefully desolder the leads to them. Unscrew the six Phillips screws holding the transformer can at the bottom of the chassis, remove the transformer, and put those screws back in.

Now you have an open chassis frame with a grey wire and a brown wire coming from the capsule, red, red/white, and black wires coming from the XLR connector in the base, and if your model has them, wires from the pad and external high pass switches (which you should tape out of the way or cut off, since these features won't be implemented). [Some mics have differing colors to the capsule. The wire going to the center point on the diaphragm goes to the cap. The wire going to the side of the capsule goes to ground. ]

Incidentally, you should be very careful removing and inserting these screws, because they are very soft and inexpensive metal and they can easily be stripped or twisted off. If they don't go in easily, loosen some of the other screws a bit and put a dab of lubricating oil on the ones that seem to be binding.

Screw the new board into place using two of the screws you removed when taking out the old boards. Connect the grey wire from the capsule to the ground pin on the board; this is the stator lead and is connected to the side of the capsule. Connect the brown wire from the capsule to the IN point on the board, this is connected to the center point on the capsule diaphragm.

The black wire goes to pin I on the XLR connector and should be hooked to the point on the board marked 1, which is a ground. The red wire and red/white wire should be hooked to the points marked 2 and 3 respectively.

Once again, now deflux it. Use the long nozzle on the Flux-Off spray to clean off the connections you made. Be verycareful not to get any of this stuff near the capsule.

Screw the base back together again and give it a listen! If you hear a lot more noise than you think you should hear, give it another hit with the flux remover and wait a few minutes for it to evaporate. Because the circuit is so high-impedance, it does not take much residual flux or atmospheric condensation to bring the noise floor way up! [If you hear only noise, your Q1 and Q2 transistors are not amplifying. You probably have them in backwards.]

If you can get a conformal coating, coat the board with it; if you can't, don't worry about it. The urethane conformal coats seem to be the most reliable for this sort of application, but one bottle will a number of boards. If you aren't doing a lot of boards, don't sweat it unless you're in a high humidity environment.

The overall sound is much better with the retrofit board, with a lot of the problems in the low end going away. (There is still a very large peak around 6K that varies in exact frequency from unit to unit. That's due to the capsule, and I considered trying to filter it out but the fact that it varies so much makes this difficult, and all in all it's part of what makes the sound of the mic anyway.) I think you'll find this mod an enormous improvement.

Parts list

Q1, Q2	2N5087 transistors	2N5087-ND
Q3	2SK170 FET	-- none --
Z1	12V zener diode	1N5242BDICT-ND
C1, C4	10 uF 16V tantalum cap	P2038-ND
C2, C3	1uF 100V film caps	EF1105-ND
C5	4.7 uF 50V tantalum cap	P2077-ND
C6	820 pF COG ceramic capacitor	P4860-ND
RI, R2	2.21 K resistors	2.21KXBK-ND
R3	332 ohm resistor	332XBK-ND
R4, R5, R7, R8	100K resistor	1000KXBK-NE
R6	6.8K resistor	6.81 KXBK-ND
R9,R10	1000M (1G) resistor	MOX200J-1000MEG-ND
Chemtronics flux remover pen		CW9100

Note: these part numbers were OCR scanned, and may contain typographical errors. Please see the original article for the real numbers.

The part numbers on the righthand side are from Digi-Key corporation, of 1-800-DIGI-KEY. Note that the resistor values used are slightly different than those listed on the schematic because they are standard 1% values. The parts will total less than $20.

You can try Cain Electronics of Hampton, VA, at 757/826-5535 for GC Print-Kote conformal coating and for Flux-Off flux remover (in a spray can rather than a pen). I found online sources for http://www.action-electronics.com/chemtron.htm

Scott is selling a kit that includes the Toshiba 25KI 70BL FET and the PC board for $20, postpaid. Send your check or money order to:

Kludge Audio
PO Box 1229
Williamsburg, VA 23187-1229

If you don't want to pay [Scott] for 'em, the layouts for the boards are printed here so you can make your own, but you are on your own for finding the FETs in small quantity. B&D Enterprises occasionally has them but does not have a consistent supply. You can also try NTS, their webpage is http://www.ntsproaudio.com/

Errata

There are some big problems with the parts numbers on the schematic in the Making Mikes Better article.

First of all, the parts list [in the magazine] IS correct and has the correct parts names and numbers.

The layout has a small problem; there are two C6es listed on it. The C6 on the top of the layout in the center between the FET and the input pin really is C6, the 820 pF cap. The C6 on the bottom left which is across the zener is really C4, the 10 uF tantalum cap.

The schematic has two caps marked C4, 10 uF tantalum. The one across the zener really is C4. The cap between the base of Q2 and the source of the FET is really C3, the 1 uF film cap. (Remember, C2 and C3 have to be the same type and value so the circuit can be symmetric).

But, the REAL problem, and I want to thank a reader for finding this one, is that the PC board layout is incorrect. On the schematic, the 330 ohm resistor R3 is tied to the lower side of R6 (the 6.8k resistor) which is held at 12V with the zener, but on the layout, it's tied to the top side of R6 which is pulled to 30V. This means there is 30V across C1 (which is bad) and too high a voltage across the FET (which is actually fine).

To fix this, remove the 330 ohm resistor completely, and take one of the spare 330 ohm resistors and use that as a replacement. Put one side of the resistor into the top hole of the original resistor, the one that goes to C1. Then take the other lead of the resistor and snake it across the board a bit, then tack-solder it to the bottom side of the 6.8K resistor, the side that connects to the zener.

If you use the published layouts to make your own board, or you got the original board from me, you will have to do this. This is only the case for the first 50 boards made, however. If your board says "REVISION II" in big letters in the lower lefthand corner, you have one of the newer boards which have had this change made on them and you can ignore this.

Go buy the reprint!