Microphone build inside microphone

These mic use a nasty mic capsule that's (perhaps) a copy of the JLI-2555 - a 25mm (1") standard size for smaller LDCs, the other common one is 34mm. You could improve the performance of your own device simply by replacing that horrid capsule with the JLI-2555 and a decent JFET like the LSK170.

Bit of background to save you going all the way back through the thread and getting confused with some of my early work which was, frankly, poor. But I don't believe in covering my mistakes, I'd rather admit to them so others don't fall into the same trap. (I've added the meaty bits at the end, after TL;DR.)

There are two main formats in all condensor mics - electret (pre-polarized) and non-electret which require an external polarizing voltage around 50 - 80 volts, DC. All miniature condensor mics are electret, so they don't need a high voltage which makes the convenient and much easier to manufacture. They usually come with an internal JFET too.

The impedance (a complex form of resistance) of the mic is extremely high - in other words it cannot produce very much current at the output terminal - think pico-amperes. 

The reason without going into capacitor theory is because the capacitance of the capsule is tiny, around 50 pF so as the charge gets pushed and pulled (by the movement of the diaphragm in air) very little current flows.

You can get an idea from Ohm's law - assume 10V polarising voltage at 1pA works out at an impedance of 10 terra-ohms which sounds insane.

The solution is rather crafty. JFETs (all FETS) have a capacitive input - called the gate. Charge (measured as current) flowing in and out of the gate causes a current to flow in the channel from Drain to Source. JFETs work because they have a tiny diode formed which becomes a capacitor of a few pF (the actual value varies from device to device, even of the same part number. To explain why this happens means dipping into the quantum effects that cause all semiconductors to work and it's a bit of a mind bender so I'll skip over that.

OK, so to the resistors. Resistors provide current to the Drain-Source channel - which you can think of as being a small, variable resistor. Typically when the device is just sitting on your desk, the channel will have a resistance of around a few hundred ohms to perhaps a couple of K ohms.

After that things get a bit weird.

JFETs are all "depletion" mode devices which means that the effective resistance of the channel increases as the gate is brought more negative than the source end of the channel. Biasing a JFET is usually done with a small resistor from source to 0V (ground-reference). As current flows in the Source resistor, a voltage is "dropped" across it and that pushes the voltage at the source ABOVE ground, and that effectively means that even if the Gate is connected to 0V, it's still more negative than the Source.

Confusing isn't it - but you can show this in a simple simulation like Paul Falstad's excellent online tool.

I've drawn a simple "self-biased" JFET here https://tinyurl.com/28me67gu for you.

You can see the voltage at the Gate is 0V - but the voltage across the resistor is about 2.5V.

So the Gate (with respect to the Source) is now -2.5V.

The actual current you need is largely determined by the type of JFET but they're very low-power devices (unlike MOSFETs which can go to 100s of amps, particularly the closely related IGBJT). Most JFETs max out at a few 10s of milliamps and often will function at a lot less without noticeable effects. The idea operating current for any given JFET is around 75% of its operating max but in a mic, the "swing" is so small that the exact amount is not that critical unless you're looking for studio grade stuff and those are usually hand tuned to perfection on a device-by-device basis.

OK, so now we have that out of the way, why would you use two resistors rather than just one?

That's to get a pair of signals that are 180 degrees out of phase (opposites) which is what we need to drive a differential input stage. And we do that because differential input stages (sometimes called Instrumentation Amplifiers) reduce the amount of noise pickup from nearby electrical sources like mains wiring. I often chuckle when people complain about the radiation from their WiFI routers when every single mains wire in their home is throwing out bucket loads of the stuff. We're bathed in it 24/7.

When the pair of wires are simply hanging in space, a LOT of mains frequency (50 or 60Hz) gets impressed on there by magnetic induction - first demonstrated by Michael Faraday. But that's the sort of fundamentals you don't need to understand - and there are loads of resources online if you are masochistic enough to go look. 🙂

This unwanted signal appears on each wire equally - and is, therefore, called "common mode noise" or common mode interference. It's common mode because it's common - i.e. the same - on each conductor; and noise because it's something we certainly don't want.

The math is pretty simple we have two interference signals +Hum one one lead and +Hum  on the other(in phase) and two signals 180 degrees of out phase +Signal and -Signal. (We talk about phase in angles of arc because all non-DC signals are made up from sine waves which are closely related to circles, but for signals that are 180 degrees, we just flip the sign.)

Now, imagine adding these together. That gives us:

(Hum + Signal) + (Hum + -Signal) = 2 Hum

Our signal has gone and we've ended up with twice as much interference as we started with.

A differential stage calculates the difference between the two signals:

(Hum + Signal) - (Hum + -Signal) = 2 Signal

My algebra isn't the clearest but that's what's going on. The difference between the two causes the interference to zero out allowing the signal from our mic to appear. Remember that at this stage, it's still pretty weedy too - a few thousandths of a volt typically. 

Do you need to go to this trouble? Not always. A lot depends on the application and many low-end microphones such as tie-clip mics (Lavaliers) and smaller electrets are single ended. That means they only have a single "load" resistor on the Drain with the source grounded! Looking back on how JFETs work and this seems insane but it does work for very small changes in gate voltage caused by an electret capsule which can pull the gate very slightly negative during part of its swing.

Such designs keep the noise to a minimum by using a shield made of a lapped copper (cheap, nasty) or braid (expensive, far better) which forms a Faraday cage around the signal wire while also providing an electrical ground to complete the circuit.

And done right this can be very effective up to a few metres of cable, especially if it's a good quality with a braided screen.

In fact, some manufacturers produce "balanced" (another word for differential) mics that only have a single wire carrying the signal while the other one sits at ground potential. Using my wonky algebra, remembering that ground reference is zero:

(Hum + Signal) - (Hum + - 0) = 1 Signal

The preamp still cancels out the interference and we still get our signal returned from the mush, but you only get half the voltage. In many cases this is quite good enough though. Such systems are called "pseudo-balanced" or "pseudo-differential" which is something of a misnomer, because it's not really pseudo at all. The signals don't make the line balanced - it's the impedance of electronics driving it that does that. So long as the pre-amplifier "sees" identical impedances at each input the line is balanced.

I'm skipping over a lot of detail here because it does require some fundamental theory which few people need to know if they're not involved in the design of mics or preamps. It just helps to have a broad grasp of the idea for (and if) things go wrong.

Impedance matching matters in balanced systems because the voltages at the inputs must be as close to matched as possible. You can think of an impedance mismatch as multiplying the signals by differing amounts and because the preamp might be amplifying the difference signal by 100 or 1000x (or even more!) any error gets multiplied by the same amount. So the noise gets worse (it's common mode) and the signal gets lost because it isn't.

If you're still with me, well done, if not don't worry this is much easier to see demonstrated in a lab.

TL;DR

With a that short intro to the theory out of the way, let's look at the wire.

Motor wire is more properly called "ECW" or enamelled copper wire. This is a single strand of copper covered in an insulating lacquer (paint) so can be manufactured to fit in very tight spaces where thousands of turns are required. The alternative is TCW - tinned copper wire which is what it sounds like. We used that for bus-bars and short links for example, to save having to strip off plastic insulation. (You can get single or multi-strand copper wire with a soft plastic insulator from 7 strands up to dozens of strands. The common ones are 7 and 14 for light electronic work. Mains "flex" has a lot more as it's got to carry much more power. Beyond that we get into super-high voltage and current where the strands are spiral wound to get over some of the weirder things that electricity does.)

The thickness of the wire matters because the thinner it is, the more natural resistance it has. Thin copper strands are very flexible but terribly prone to breakage under even mild repetitive stress or strain. Thicker ones have lower resistance but are less flexible. For short runs, the fine stuff is perfectly adequate but the longer the run, the thicker it needs to be or the losses due to resistance start to have a detrimental effect on the input impedance. (Bet you wished you hadn't asked now.)

What matters more (if you're making up a pair like Matt did) that they are twisted together. This isn't for looks, it's to make sure that each lead picks up the same amount of interfering signal as they pass through the EM fields from the mains wiring. The screen (Matt, rather cleverly used solder wick) keeps most of it out but some will invariably get through.

The mic you have is a pure USB mic - the crystal is a huge giveaway even if it wasn't written on the PCB. I'm curious as to which chip that is though. I wonder if it's custom silicon or some low-cost USB endpoint. The ones I've worked with are much larger (up to 48 pins).

These mics do all the analogue work (including the ADC) in the case and pass the signal down the USB cable as a 90-ohm differential digital signal that's mostly immune to noise from mains, although it has to be protected from radio and other high-frequency sources.

There isn't a lot to do with these except replace the capsule for something with a decent pedigree - at the risk of sounding like a broken record, the JLI-2555 or one of the better clones, of which there are a few. But you can get the original part on Ali Express for about £10 plus duties. Changing the JFET is a fool's errand. Sure you CAN change the JFET for a better part but the vast majority of the signal quality (or lack thereof) comes from the capsule and nothing else. Any other lack of performance is going to be due to the ADC in the USB Codec - and not all of these devices are created equally.

As I found to my cost, a lot of them are ONLY available to huge corporations with people in the Maker Space stuck between the decent (but not stellar) CM108 or the PCM290x series from Texas. Very little else exists at a reasonable price and we're often left just making do with a proprietary device from UGreen, Soundblaster, etc.

I have a project (not the promised V2) on the back burner to fix this but my skills in C leave rather a lot to be desired, so V2 will still be limited to CD quality audio which is (quite frankly) far better than is needed unless you're in a studio. 

Gotta go, the Cat is getting hangry... 

@marcdraco 

Thank you for the in-depth information about how everything functions and works. I read it several times and searched some terms, but still can't fully comprehend some of the things or rather, can't fully imagine the picture it in my head. I'll do more thorough research and reading about all that.

I learned how to read schematics and thought that I can understand them, but I got a little lost in the one, which you linked. If I am not mistaken +10V from PCB is connected to drain of the JFET and JFET's source is connected with 2kOhm resistor to ground. The resistor filters +10V and turns it into +2.5V, which goes straight to JFET, which also provides +2.5V to capsule. According to datasheet of JLI-2555, its standard voltage is 1.5, so I guess 2.5V is good enough. The circle with V 0V is supposed to be the gate pin, which is connected to the center pin of the capsule. Since the 2kOhm resistor provides +2.5V to JFET, JFET provides -2.5V through the gate. If I mistook something here, please correct me on my mistakes.

After reading the provided information, from my understanding, the reason behind two resistors is to take away all the buzzing/hum and interference from both capsule and circuitry, and clear out signal, right? But copper wires and solder wick can do it too, so since I am going to copy Matt's solution, there is no need to use resistors at all. The thing about twisting wires between each other to balance the interference signal reminds me how the ethernet cable's wires are punched into a patch panel. In networking environment if the patch panel doesn't have keystones, the wires need to be twisted in order to ground them. After your explanation, I think that the twisting is done not only for ground, but also for balancing the interference signal of each other as well as from other cables' wires. Now that I think about it, are the wires on my microphone twisted between each other or not? On one hand, it seems like both capsule cables and headphone jack cables are not twisted. On the other hand, since they are covered with additional silicon layer, there is a chance that they might actually be twisted from the inside.

The chip next to the crystal is an 8 Pin ATMLH334 64DM B 3W0436E. The one, which is on the back is a 64 pin Trident UAC 3556B G7 A NNNN7311 255910.0002 234S Germany. The transistor is on the capsule is K596 B207.

From you explanation and my understanding, there is nothing to salvage from the board for the future build. Now, the thing I am curious about is whether it is worth to modify this microphone with JLI capsule (some people said , transistor (LSK170C TO-92 3L ROHS if it is good for the job) or modify it with APO Alice board, DC-DC Hex inverter (if it is even worth to go for full condenser capsule or if they both can even fit inside M-Audio’s case) and XLR 3 pin plug (probably will need to drill/sand the USB type B port to fit it properly). Matt’s pre-amp and ADC solution is great, but it will probably not fit the microphone’s body, which I wanted to use.

Personally, I use microphone mostly for vocal (voiceover, singing and etc.). I am quite satisfied with the sound, but on 80-85% levels of gain in microphone properties (OEM drivers are optimized for windows 7/8, so the gain malfunctions on Windows 10/11. M-Audio themselves advice to use Windows’ drivers instead) introduce microphone’s noise. The microphone itself is quite silent when you don’t increase the gain. From your own perspective, should I go for just capsule mod or the whole XLR conversion?

Posted by: @crave

↑

@marcdraco 

Thank you for the in-depth information about how everything functions and works. I read it several times and searched some terms, but still can't fully comprehend some of the things or rather, can't fully imagine the picture it in my head. I'll do more thorough research and reading about all that.

Yeah sorry about that, I've skipped some of the "theory" but it's almost a bottomless pit which eventually and invariably ends up at the quantum physics level - waaaay above my pay grade, although I understand the effects that are created.

Posted by: @crave

↑

I learned how to read schematics and thought that I can understand them, but I got a little lost in the one, which you linked. If I am not mistaken +10V from PCB is connected to drain of the JFET and JFET's source is connected with 2kOhm resistor to ground. The resistor filters +10V and turns it into +2.5V, which goes straight to JFET, which also provides +2.5V to capsule. According to datasheet of JLI-2555, its standard voltage is 1.5, so I guess 2.5V is good enough. The circle with V 0V is supposed to be the gate pin, which is connected to the center pin of the capsule. Since the 2kOhm resistor provides +2.5V to JFET, JFET provides -2.5V through the gate. If I mistook something here, please correct me on my mistakes.

That's a pretty good understanding, you've been confused by the poor quality datasheet (ones like that have caught me out more than once, I can tell ya!)

The JLI2555 is a special type of capacitor. It doesn't have standard voltage - it "simply" moves a tiny amount of charge (a moving charge is called current) when connected to a JFET. This is a difficult concept to explain without going into a lot of theory (some of which I've covered up this thread but it's pages back and I tend to keep the explanations shorter than they should be). 

Electrets are pre-charged capacitors that can't be discharged like a normal capacitor can, but they CAN move a tiny amount of charge and this is how they operate. A true condenser microphone (condenser is a very old term for a capacitor, they're the same thing) is a true capacitor - so it has to be charged with a polarising current in order to operate. The current comes from a biasing voltage via a high-value resistor, typically as high as 1 gigaohm.

Electrically, the gate of a JFET is also a tiny capacitor, even smaller than the tiny capacitor in the microphone! It's not a "true" capacitor as such though, it's created by a quantum effect that appears when a diode is biased (has a voltage applied) in reverse to its normal operation. This effect happens in all diodes but it's so small that we don't worry about it - particularly as diodes are normally wired the other way around as rectifiers.

This means that the impedance of the gate is staggeringly high - millions of ohms, because looking "into" the gate, the circuit sees a capacitor, albeit one that leaks a little bit. But that leakage is so minuscule that we rarely worry about it. MOSFETs are related to JFETs but their gates are "true" capacitors so the impedance at the gate of a MOSFET is even higher. While we could use a MOSFET in this application, they have other effects that make them a poor choice for audio.

So what happens is this - as sound hits the diaphragm in the electret, its capacitance changes very, very slightly; pushing or pulling a tiny amount of charge (a few million electrons) in and out of the gate.

As the charge on the gate changes, the effective resistance of the channel - that's the bar between the Drain and the Source - changes, so the current flowing through the resistors also changes. 

It's hard to visualise but resistors "convert" voltages into currents OR currents into voltages depending on how they are wired in the circuit.

Voltage is a theoretical parameter so it only really exists in circuits and measurements as an indicator of what's going on. It's a bit like the imaginary number in complex numbers, the root of -1. Even though root -1 is impossible, a LOT of mathematics simply won't work without it. The number "0" is an earlier example actually, for hundreds of years, early humans counted from 1. They had no concept of a 0 - or even a negative number for that matter.

One way to think of a JFET wired as you've seen is a string of three resistors making up two voltage dividers.

Most measurements are referenced to the ground, but that's not a requirement. We can measure a voltage "drop" (a poor word but it's what we got). A voltage drop is simply the voltage that "appears" across any two points in a circuit. This voltage is a function of the resistance (which can be "complex", so we call it an impedance) multiplied by the current flowing through that part of the circuit.

This is a tough one to wrap your head around, as we tend to think of voltage as being a physical property - it's often taught as such, but it's better to think of it as purely a convenience to make the maths work. I know a lot of more experienced engineers will scream at me for saying that heresy, but it's true. Voltage is more correctly called a "potential difference" - the key being "potential". EMF is another term which is even more confusing.

A voltage divider is simply two (or more) impedances/resistances in series between two points in a circuit. 

https://tinyurl.com/2ajnfudp

Shows a 5V source (a battery in this case) split into two parts with two identical resistors. If you hover over the resistors here you'll see the current in each is the same (2,5 mA) and indeed, the current through the battery is the same. This is the one of the simplest possible examples of Kirchoff's current law. Simplified, this says, "what goes in one end MUST come out the other end."

You can put any mess of resistors into that circuit but the total current leaving the positive terminal of the power source will always be the same amount of current returned because it's going around in a circle - from which we get the term "circuit".

A more complex example demonstrates this effect in more detail.

https://tinyurl.com/2yvcnr27

If you click on the centre resistance (a potentiometer) you'll see a slider on the right which will allow you to alter the resistance at that point and you can see how the voltages at the various points in the circuit change along with the current flowing around it.

In an electret mic circuit, the JFET (K596 is a popular part often used inside those electret capsules that are so cheap you might find one at the bottom of a packet of breakfast cereal) the JFET acts like a variable resistance.

Without looking at "how" it happens you can imagine the capsule/JFET is a sound controlled resistance, which changes the amount of current flowing in the circuit. The change in current at any time is directly correlated with the change in volume. 

The upper and lower resistors have the effect of converting the changing current into a changing voltage. Since the upper and lower resistors are the same value, they track the change at the same time.

Posted by: @crave

↑

After reading the provided information, from my understanding, the reason behind two resistors is to take away all the buzzing/hum and interference from both capsule and circuitry, and clear out signal, right? But copper wires and solder wick can do it too, so since I am going to copy Matt's solution, there is no need to use resistors at all. The thing about twisting wires between each other to balance the interference signal reminds me how the ethernet cable's wires are punched into a patch panel. In networking environment if the patch panel doesn't have keystones, the wires need to be twisted in order to ground them. After your explanation, I think that the twisting is done not only for ground, but also for balancing the interference signal of each other as well as from other cables' wires. Now that I think about it, are the wires on my microphone twisted between each other or not? On one hand, it seems like both capsule cables and headphone jack cables are not twisted. On the other hand, since they are covered with additional silicon layer, there is a chance that they might actually be twisted from the inside.

That's an excellent grasp, but not quite correct.

The two resistors produce two voltages in opposite phase. We don't NEED to do it this way, many circuits use a single resistor - usually from supply to Drain (and typically 2k2 by convention) but it can go from Source to ground with Drain wired direct to the supply (called Common Drain or Source Follower). It all depends on what we're trying to do.

A typical tie-clip mic will have a single resistor from supply (usually 1.5 to 3V) to the drain and "taps" the signal off via a capacitor. Dead simple, dead cheap and not very good quality.

This is a very low-level signal so we screen the signal with a mesh, usually in the cable itself. This is a form of Faraday cage that prevents interference (usually mains "hum") from getting mixed up with the signal.

A twisted pair carries two signals in opposing phase - as one rises, the other falls and vice versa. Ethernet is an excellent example of this, although USB uses it too. The voltage doesn't have to go negative, just that one signal is the opposite of the other.

The cables don't have to be twisted, but if they are, it just ensures that the each wire receives the same amount of interference which makes it easier to remove at the receiver. While Cat 5 and 6 is un-shielded, (UTP) you can get better quality stuff - STP, which is shielded for situations where there is a huge amount of interference and a twisted pair just won't cut it.

The best way to understand the "magic" of differential pairs and their ability to reject interference is to see them in action, but that requires, ideally, an oscilloscope a signal transmitter and receiver. In a typical demonstration it can be shown that the signal on each input is unrecognisable, but the output from the receiver is clean. I can talk theory about this until the cows come romping home, but there's nothing quite like seeing it for yourself. If I can find the time, I'll mock something up for you.

Posted by: @crave

↑

The chip next to the crystal is an 8 Pin ATMLH334 64DM B 3W0436E. The one, which is on the back is a 64 pin Trident UAC 3556B G7 A NNNN7311 255910.0002 234S Germany. The transistor is on the capsule is K596 B207.

That's kinds weird, it's a USB CODEC alright but the datasheet says it's a DAC (meant to convert USB digital audio into audio for speakers). There's probably a similar part number for the ADC, but hey-ho It matters not. 🙂

Posted by: @crave

↑

From you explanation and my understanding, there is nothing to salvage from the board for the future build. Now, the thing I am curious about is whether it is worth to modify this microphone with JLI capsule (some people said , transistor (LSK170C TO-92 3L ROHS if it is good for the job) or modify it with APO Alice board, DC-DC Hex inverter (if it is even worth to go for full condenser capsule or if they both can even fit inside M-Audio’s case) and XLR 3 pin plug (probably will need to drill/sand the USB type B port to fit it properly). Matt’s pre-amp and ADC solution is great, but it will probably not fit the microphone’s body, which I wanted to use.

If the board is still functional, it might be possible to use it as an ADC rather than buying one. The FET (likely the drain) will be connected to the audio input on the ADC but it's unclear what sort of voltage levels it's expecting. You'd need an oscilloscope to find this out.

Those capsules are horrible though. Not as horrible as they could be but the JLI2555 should be a direct replacement if it's a 25 mm saddle. The K596 is a common part in condensor mics and there's no reason it shouldn't work with the JLI capsule. The LSK170, while a better JFET, might not work optimally in a circuit optimised for the K596. Some users have reported success with a 2N3819 but if it's not busted, there's no point in replacing it.

Matt's original certainly won't go inside a mic body. I've done one that will but it's not the V2 which I'm just finishing off at the moment. (The load resistors for the JFET are on the board in Matt's version - it's quite clever and keeps the capsule section nice and small.)

Posted by: @crave

↑

Personally, I use microphone mostly for vocal (voiceover, singing and etc.). I am quite satisfied with the sound, but on 80-85% levels of gain in microphone properties (OEM drivers are optimized for windows 7/8, so the gain malfunctions on Windows 10/11. M-Audio themselves advice to use Windows’ drivers instead) introduce microphone’s noise. The microphone itself is quite silent when you don’t increase the gain. From your own perspective, should I go for just capsule mod or the whole XLR conversion?

If memory serves, the JLI capsule isn't as sensitive as those little cardiods but it has a much nicer sound so that might be all you need. Simply (cough) swap the saddle and capsule for a 2555 with a suitable saddle, wire it as the original and you're golden.

The question of what to do is the "ultimate question" (per Douglas Adam's and the famous "42!"). Deferring to Adams, the answer depends on what the question is. In this case, what exactly are you trying to do?

Matt made a unique, high-quality desktop mic with a Steampunk aesthetic that I personally love. So much so I spent all of my spare time and beer money over the last couple of years trying to better it - all without sending the BOM into the stratosphere. The final design (in small numbers) is more modular and can work out more expensive but it allows for far more customisation including DIY headsets and (later) Ambisonics.

DIY is wonderfully freeing - we're not constrained to whatever some faceless designer has decided is suitable for our application. All the way up to using a hex-inverter voltage booster to bias a traditional condenser, but that is more difficult and generally requires a couple of expensive parts, in particular a poly-type plastic capacitor to isolate the condenser input from the JFET gate. ANY leakage (and almost all capacitors leak a little bit) will cause the charge on the gate to rise over time, which will first appear as distortion and eventually will destroy the JFET.

Unpolarised LDCs (large diaphragm condensers) have a particular sound that won't necessarily match with what your expecting so you might end up with something that doesn't sound as impressive (or musically pleasant) as you are expecting. Rode Microphone's Doug Ford points out (as Matt did in his video) that designing microphones is as much an art as it is a science. And I know that some "experts" have grumbled that it's all science and Matt was wrong.

They're wrong and Matt is correct. (Sorry guys...)

The electronics is a science but human hearing is largely perceptual and impossible to quantify. A microphone with a perfectly flat response from 20Hz to 20KHz (which is as near impossible as it gets, but we can get close) will sound to the average listener as dull and boring. Measurement microphones are an example of this since they are specifically intended to *measure* sound accurately without any bias.

But humans are all different. Our pinnea (ears), the shape of our ear canal, age, and more all affect how we perceive sound. In the event, most vocal mics have a little peak rising from 3KHz to a max around 6KHz and then slowly falling off. (Ford describes it as "Wooo Hooo!)

This part of the design is the art.

Instrument microphones are different again and some classic microphones only produce useful output up to around 12KHz before falling off quite rapidly. A Swedish patent filed about 15 years ago also identifies a problem with circular diaphragms that means they have natural resonant peaks which make the sound less natural, and suggests (against conventional wisdom) that a better, more natural sound is achieved with a triangle. The inventor pointed out (quite rightly) that cymbals, bells, drums and so on are circular which makes them "louder" at their resonant frequency. A triangular shape produces more random reflections so they don't have a natural resonant peak.

It's been an amazing experience, I've learned far more about electronic theory than I ever did in years of college by going right down the rabbit hole and popping out (figuratively) on the other side of the planet. Digging into the (albeit simpler) quantum effects has been necessary to gain a deep understanding of what's going on. If you have a good understanding of calculus, all of this can be described and modelled mathematically, but if you're like me... it's mostly visual.

The problem with a purely mathematical approach (rather like simulation) is we're assuming perfect components. A 10K resistor is exactly 10,000 ohms; a 10 uF capacitor is exactly 10 micro Farads and so on.

When I lay out PCBs (now) I don't think of voltages and currents at first, I think of the electric and magnetic fields that appear around the copper tracks and how they are affected by the tracks and even the plastic that the board is made from.

Such hand-routing is a slow, arduous process (even with electronic CAD like KiCAD) and there's always a temptation to use an "auto-router" but auto-routers just lay out the board using the shortest possible connections between each components. This is often far from optimal and may even produce a board that simply doesn't work even though it's "wired" correctly.

TL;DR

I suspect the best, and lowest cost upgrade for your existing mic would be to swap out that nasty capsule for a JLI2555 on a saddle (a rubberised is better but I don't know if anyone produces one). A solid saddle means you'll need a support to prevent vibrations from getting into the body where they will pollute the audio. 

Beyond that, Matt's original or the up-coming V2 will almost definitely improve matters. The V2 development got held up after I got into a spat with CMedia Inc. over the CM6533... which was an expensive "mistake" because they didn't say that it needed firmware that is only available under licence.

Using an hex inverter charge pump with an unbiased capsule is more complicated than a simple swap (there are circuits online that describe this) but it's unlikely to make things better as the larger condensers tend to be "boomy" and lack high-end detail, particularly for musical use.