Neewer BM800 Upgrade - DIY Perks Style.

Hey! new to the forum. i have a couple of questions, if you dont mind,

I want to build a mic using this diaphragm (RK-87): https://www.aliexpress.com/item/1005007682928721.html

please correct me if this is a bad base, but it seems good, im sort of building something similar, gutting a cheap mic.

but from what i understand i need phantom power, but id preferably want to have usb, and works (or aux and usb power),

but in a post you mentioned something about the voltage from that one being 15v and thats too low for such a "bigger" diaphragm? could i know what would be the best approach?

would the diaphragm i chose work with the usb and with the op amp and all that stuff? or better stick with the JLI2555? and if so i dont understand which is the latest version of the schematic, if you could please show me,

to go with the usb all in one without external stuff thingy, thats.. if itll work, or aux and usb power only, as long as i dont need to plug in some external stuff

Hope that made enough sense, have a nice evening

Good morning, the monster has finally given in an fed the cat which means I'm out of bed.

All perfect sense, yes. Getting 48V (which is typical for an unbiased capsule) from 5V is possible but it's very difficult. The only examples I've seen to date seem to use custom digital chips which allow for huge numbers of MOSFETs and small capacitors in a small space.This sounds a bit strange on the surface but the reason is that very efficient voltage multipliers - and consider they have to provide current AND voltage (vs. a typical UHT generators trade current for voltage). 2-5mA might not seem a lot but a multiplication factor of around x10 at 100% efficiency - which is impossible - would require 2-500mA. In reality efficiency is considerably less than this even on a dedicated chip. Doing this on a PCB means the efficiency drops dramatically because of the losses involved in the interconnects.

This is why we've seen such an explosion in electret (pre-biased or permanently biased) technology since the first practical devices appeared in the early 1960s. Modern electrets give exceptionally good performance when compared to unbiased condensor capsules and if you want a standard "sized"  LDC at 34mm JLC offer the JLI-3413AU ( https://www.jlielectronics.com/content/JLI-3413AU01.pdf) should give excellent performance with the right JFET.

A lot depends on what you're trying to achieve, but based on the performance of the 25mm JLI used in Matt's original, I would expect it to be very good indeed and quite possibly better than an Chinese RK47 clone.

If IS possible to run a JLI2555/JLI3413 from a purely 5V supply using nothing more than a very simple bias circuit but the linearity of such adapters is generally poor. The latest and best performing version of the V2 circuits (and the designs that evolved from that) are at my github: https://github.com/marcdraco as Sarah, Michelle2 and Varee. There you'll find all the circuit diagrams and PCB layouts. A stripped down version of Michelle 2 which removes the "fancy" isolated side for professional use, will be up soon. I haven't tested this one in production but it's simply a cut-down version on a smaller board.

Varee can run from 12-15V or 48V (it's the same design but the 15V version doesn't have a regulator to limit the voltage to the FET). This use a complex self-settling bias circuit to keep the JFET operating more linearly than a simple common source or common drain configuration. Note that the 1G resistor from gate to ground is required.

The voltage on an unbiased condensor does a couple of things. It causes a field to form between the two plates just as in a normal capacitor - which pulls the two plates together (only a little but it's there). This affects the sensitivity because the amount of charge differential between the plates determines how much charge moves as the plates move. It also affects the frequency response by causing the diaphragm to tighten up like a drum. At lower voltages this means the diaphragm will be "loose" and boomy, too high and it gets tighter causing peaking in higher frequencies. The actual pretension is determined by the designers.

If you want the *simple* way of doing things (common source with a 2k2 load resistor) you can do that with the Sarah PCB - but note that one is designed be a "blank" project supporting Matt's original FET only or common source/common drain. I'm not proving detailed instruction at until Matt's done with the Varee/Michelle pairing but it's reasonably easy to do if you follow the circuits provided by JLI and others.

@marcdraco 

thanks for the quick reply! from what i read, i think ill go with the original JLI2555 used in the video. and thanks for all the info,

i think i know what i should do now, but ill just confirm if i did understand everything.

for the JLI2555 i need a preamp, so Sarah 2.0 needs to be added to the back of the diaphram, and then from that to Dorit or Michelle 2.2 if i want improved sound isolation

(From silkscreen mark "1" on Sarah2.0 to ve+ on dorit and same with pin 2 and ve-, gnd to gnd). (will go with dorit).

(I also have a question on what "Expansion" is for?)

Populate R3 & R5 with 680R Ohm.

if i cant get my hands on a THAT1512, i can use SSM2017, SSM2019, and INA217 as a drop in replacement.

Am i missing anything? sorry for all the questions and what not, i dont want to mess this up.

And ofc at the end add some premade usb sound card to connect pic port to it and yeh-

Sorry I've been on phone most of the day so I hope this gets to you in time. 😉

All electrets need some sort of pre-amp. All the JFET heads do (from the simple mod that Matt used all the way through to Varee (which you'll see is relatively complex for such a small board) only produce a high-current, low-voltage (low impedance) output that rejects interference by swamping it.

The "balanced" lines go a step further to reject line noise but ignore that for now.

Dorit and Michelle 2 are just souped-up versions of Matt's original pre-amp that replace the huge input caps with "capacitor multipliers" and add a lot of extra RF filtering and some extra protection to (a) prevent the THAT from blowing up and (b) prevent the THAT from blowing up something else further downstream (like your ADC!). This isn't a design flaw - because the circuit works as it's supposed to - provided the screens are in place. The issue comes when you're trying stuff out and then... BOOM. Dorit and Michelle have protection to stop that.

Michelle 2 is probably too much for most users, hence why I did Dorit with the THT resistors. You'd use the 2k2 resistors (marked on the PCB) which provide + and - voltage to power a FET the way that Matt's original did it OR two 680R resistors which provide a P12 compatible phantom power to power the Varee capsule adaptor.

This is quite an expensive option and it's a case of diminishing returns. While a Varee and Michelle 2 set up for fully isolated power has the best noise rejection, MOST users will be fine with just the simpler (Sarah) head wired for a single FET using 2 x 2K2 resistors. These circuits are deceptively simple (to me anyway) but I know what it's like trying to read one for the first time and I have documented them (deliberately) as yet so Matt has time to do a video.

Sarah can even be used in Common Source mode, allowing you to hook a JLI2555 or JLI3413 (or similar 25 or 34mm LDCs) electrets to a standard USB microphone adaptor or the mic jack on a typical PC. Many laptops don't have this facility but most desktops should. You'll find the "standard" wiring all over the web. There are three wires: ground (screen); signal and bias (which is actually a supply voltage about 1.5V). A capacitor - which you can fit to Sarah at home if you have the soldering skills or have assembled at (for instance) JLC.

I've got some boards here to make up at some point because I suspect the load resistor is at the PC end it's been that long since I fiddled with one, I've actually forgotten! I know - you can't get the staff these days. I'm just overloaded with stuff at my end. :/

The expansion port it just that. It's a way of piggy-backing extra functions on the board BEFORE the signal is passed to the Line input for recording by the ADC, It was originally meant for a (still only partly tested) board called Mary which is a Baxendall volume control that allows you to use a LINEAR pot for gain (or a set of fixed resistors) vs. a LOG track. If that's not clear, hit me up and I'll explain. It's on a 2.54" pitch so you can make your own expansions (filters etc.) on a small piece of Vero/Stripboard. As designed you have to cut a small link to use it - which allows the signal to travel in and out. Limited power is available at the port - described (in very little detail); you've got a raw 5V USB which can be noisy, plus the (approx. 13V) smoothed lines from the NMA0515 (or directly compatible). The NMA0515 is only 1W - but you can get 2W and 3W from other manufacturers. My experience has been that the less powerful NMAs are better made however.

While I haven't TRIED the SSM2017/19 or other audio chips, any AUDIO GRADE, 8-pin instrumentation amplifier will drop into that socket but beware that they likely have a different gain calculation - necessitating a change in the gain resistor selection if you go that route. My personal preference is to set the gain to around 40dB and leave it there, but there's little to complain with Matt's approach of using a selection of resistors on a rotary switch.

The inputs on Dorit/Michelle also carry the Phantom power. So you send DC "up" the lines to the capsule adaptor or just the JFET and you pick of the (albeit) weedy signal a of few mV of AC by passing it through the DC blocking capacitors. It's essential to either use bipolar devices here OR make sure you get them the right way around if you don't. C5 is the one you're looking for. If you're using a simple JFET carrier PCB and Matt's remote FET, (using the 2K2 resistors - to "CPWr" - Cold) the NEGATIVE end faces out to the microphone. As shown in the circuit diagram, C5 is arranged for Varee (or a P12 microphone which carries positive supply on BOTH leads).

So you fit R5 OR R64, not both! Being through hole, they are easy relatively easy to remove later on if you want to go full commando and use a Varee or P12 mic. The smaller resistor is there because these designs need a little more current than a single FET to drive all the various bits and bobs - 3-5 mA or thereabouts.

I should point out that while Dorit was developed from Michelle (on the same platform) and shares the same power supply, etc. it's still considered experimental. Michelle can works almost identically but will cost a little more, primarily for the shipping due to being a physically larger (and thicker) board - Dorit is 2-layer, Michelle is 4-layer for slightly better RFI rejection.

There are a couple of errors showing on the DFI check but you can verify yourself that they're not going to affect anything.

There's a KiCAD extension in the store that will generate the Gerbers, drill files, BOM and POS files for you. Saves a huge amount of time but there's always a chance I goofed on the odd part so please do double-check!

Thanks so much for all the detailed explanation (maybe a bit too much to take in at once tho)

well ofc ill go with the most simple and easy option

so i can use just sarah, with a usb mic interface?

is the image above what you meant?

if so, its bacically this in a simplified form:

(J1 is the actual condencer)

xlr to a usb interface or streight to the computer, and that 1.5v from the aux will power it--- is that right?

Posted by: @marcdraco

↑

MOST users will be fine with just the simpler (Sarah) head wired for a single FET using 2 x 2K2 resistors

you mentioned here something about 2k2 resistors, can you expand if im wrong with the schematic i replicated?

im mainly asking because condensers require active power but isnt that 1.5v too low? or as you mentioned earlier if its too low the sound out will sound "boomy"

or just Dorit straight to the capsule, and add the usb mic interface, did i understand correctly?

condenser membrane >> Dorit >> usb mic interface

edit:

so after looking into it, it actually might be good,

https://www.aliexpress.com/item/32736596135.html

thease sound cards have that bias or smth on the mic like you mentioned except its 5v! not 1.5 ish

plus one of thease >> https://www.mouser.com/datasheet/2/308/2SK596S-D-516274.pdf

Sorry, I do tend to write rather long explanations so other people can glean their own answers. Comes from doing it professionally for years

That's very close but not quite. 1.5V is enough to get a "microphone" JFET into bias because they have a very low Vgs Off (Pinch off). The LSK170 is good for around 30 volts and the 2SK208 goes all the way to 50 (Vds). More voltage = more current = more sensitivity.

But that won't work as you have it here. The D/S taps are meant primarily for Matt's original or Dorit in "Perks" mode - that is with the 2K2 resistors slinging the channel (D -> S) between two supply voltages.

To work with Dorit (or Matt's board) can leave the capacitors off and use 0R resistors (links) for R5 and R5 and don't place any of the other parts. In this way the board is a fancy carrier to hold the JFET stable and provide more mechanical strength. In addition you get the bonus of the guard trace. This is one way of wiring Sarah and then Dorit/V1 will send and receive signals directly from the Drain and Source pins.

If you want to cheap out (and let's be honest here, who doesn't!) and just make something that works with reasonable performance, you're going to do something like this:

In effect, Sarah does the work here of mounting the 2K2 resistor and a DC blocking capacitor directly. The suggested value of 100uF is clearly guesswork (but we've all done that in a pinch too).

The 2.2V (which is likely to be 5V but anything from about 1.5 will work) comes from what is often incorrectly called the "bias" voltage and is supplied on  one of the pins, the middle connection of a 3.5 TRS (tip-ring-sleeve). This is more accurately called Vdd because it's the supply voltage.

The circuit warrants a little bit of description though because it's not obvious.

And here's a link to the actual simulation which runs live showing current and voltage.

https://tinyurl.com/28b5uyj2

This is actually what's going on inside those cheap "electret" capsules you get 20 for £1 on eBay. The JFET and the bias resistor (which is actually created by leakage inside the capsule!) are all internal and the source terminal is connected to the case. Which leaves us just two connections. (Some, famously the Panasonic WM51a) had a link from the source to the case which many people used to do things like the "Linkwitz mod" but that's not necessary here.

What happens in essence is that the microphone causes charge to flow in the 1G resistor - this charge is permanent but by squashing the capacitor slightly, which is what the air pressure does) the value of the capacitor changes and charge flows back and forth through the resistance.

This causes a voltage to appear at across the resistor which is where the "field effect" takes over. JFETs don't draw any current from the circuit (unlike conventional junction transistors) so the microphone isn't "loaded".

This seems like magic but you'll just have to trust me. I've put a much more detailed explanation later.

Another way to talk about this is to say that the capsule has very high impedance - which means that even if it generates 0.001 volts, the amount of current available is so small (in the order of picoamps) and ohm's law gives us an impedance of:

R = 0.001V / 0.000000001A

= 1G

This is a very rough approximation to give you an idea of the problem. The amplifying stage (the JFET) can't afford to "steal" current from that because there simply isn't enough to go around. JFETs employ the idea of "fields" (hence "field effect") by using the presence of an field that forms in the capacitance of the gate-channel junction. The actual impedance of a JFET is in the order of 10s of gigaohms. (For MOSFETs it's even higher in the terraohms but they don't lend themselves to audio very well.)

Mind blown yet? 😉

Fields are the most fundamental building blocks of all electronics and they're rarely covered until you're so far into the weeds that you feel you'll never see daylight again. But, and this is also going to bake your noodle, as the amazing teacher Rick Hartley points out, "the energy is in the fields" and worse, if you've only done this at high school or college level, "the fields are in the insulation".

BOOM. Right? Definitely mind blowing.

We're taught that electricity flows in conductors but it's more accurate to say that electric energy is guided by conductors - and it's energy fields that move the electrons about the place.

OK so back to the problem at hand.

Throwing away all that weird theoretical stuff we can say this:

When a voltage "tickles" (creates a field) at the gate, current flows in the channel from the Drain to the Source. Since we're converting a voltage to a current this is called a "transconductance effect" - I mention this because you'll see that mentioned when you read about FETs. (The gate controls the conductance in the channel.)

The more intense the field, the more current flows in the channel. If that's a varying voltage (as it is with a microphone capsule) then the current in the channel wiggles in unison. There are other factors that throw all of this into disarray but that's beyond the weeds, and right into the river and off your brain floats, waving a white flag. It's this weirdness that Varee's design attempts to solve but for general purpose use, it's not really necessary.

TL;DR

In very simple terms the JFET is a voltage controlled resistor. The combination of the load resistor (conventionally 2K2) and the varying resistor causes a varying voltage to appear at the Drain terminal (with respect to the ground or Source).

This is our audio signal and all that's left is to put a capacitor in series to block the DC that appears at the drain. If we don't have that, the DC that the signal is riding on - you can see that in the Falstad simulation above - will upset the biasing on the next stage. 

The datasheet for the 2SK596S shows a similar circuit with a 1K load resistor and a 4.5V supply which is where On Semi feels their particular FET operates most efficiently. Distortion and noise in audio grade JFETs is fairly low anyway.

The AliExpress link didn't work but I'll assume it's one of those tiny USB adapters with a separate microphone and a line input socket.

Boom and Bias

The confusion here comes from the fact we're using electrets. These devices have a permanent charge that won't leak away. Manufacturers expect a lifespan of over a century. Early ones used carnauba wax ( https://en.wikipedia.org/wiki/Carnauba_wax) which can be charged by exposing it to a strong electromagnetic field as it transitions from liquid to solid. The "tension" on the diaphragm is set by the factory but that's determined by the material properties alone.

Non-electret condensor mics, the original form, require a bias voltage to keep them charged in operation. These are the ones that get boomy or tinny if the biasing voltage is outside of the manufacturer's specification. These are things like the RK47 and apart from the trouble of having supply a 48V (ish) bias voltage, they are exceptional fragile and require very careful handling. There are a few specialists who repair these things - and some people just do this

But note how much care this chap takes to make sure he doesn't damage that diaphragm. I couldn't find a detailed repair video (they do exist) showing someone remove strip the whole thing apart and even replace the diaphragm. That's definitely something for special tooling and a wealth of experience.

This really isn't the sort of thing most hobbyists shouldn't attempt simply because of the ease of damaging an expensive capsule and in truth, there's not whole hill of beans separating a good quality electret LDC from a more conventional LDC.

A Bit More Theory

Fields are everywhere - most of us have played with magnets at some time and we've seen how like poles repel and unlike poles attract.

But there are electrical fields too - (the EM fields around a wire run at right angles - electric in one plane and magnetic in the other).

When we "charge" a capacitor what we're really doing is stripping electrons from one plate and putting them on the other. We do this by applying a voltage - and you can see this for yourself by getting a large electrolytic, >100uF and connecting the terminals to a battery. (Strictly you should put a small resistor in the circuit but it won't harm anything for this demonstration. A very large capacitor will spark though as they appear like a short circuit at first.

And there's the first bit of weirdness.

We have to bits of metal - that are "joined" by a layer of insulation... where the **** is that short circuit coming from - there's no actual circuit is there?

To understand what's going on, you need to understand that when we hook up a capacitor to a DC voltage the voltage source pulls electrons the same way as if you'd connected the circuit up as a short.

For a brief moment, electrons are dragged from the positive plate and moved to the negative side. The larger those plates are, the more free electrons are available - this is part of what determines the size of the capacitor.

Over time (as the capacitor charges) the amount number of free electrons starts to drop off so the amount of current flowing also falls off - this is the capacitor's charging cycle and you'll see from the graphs online that it's a curve.

But here's the clever bit (so I'll take this a bit slower)

Those two plates are really, really close. In the order of micrometers apart  - separated by a thin, insulating layer.

When we remove the voltage source, the electrons stay where they are.

Why?

Like the opposite poles on a magnet, the two sides of the capacitor have opposite electrical charge and they are literally stuck by the electrical force (at a quantum level). We don't see this effect visually because it's so ridiculously small, but you can see the same thing happen if you rub a balloon on your hair (or someone who has hair).

In this case the electrons moved from one surface (the plastic in the balloon, IIRC) which causes the surface of the balloon to be slightly positively charged and the negatively charged hair tends to pull toward it until the charge bleeds away after a few seconds.

This is what's going on in a conventional condensor mic. The bias voltage pulls electrons from one of the plates causing the slightly flexible diaphragm to get pulled toward the fixed plate. We're charging the capacitor (which is, incidentally, tiny typically a 30-50 nF for an LDC). Like the baloon (or a real-life capacitor) the charge bleeds away due to leakage so the "bias" voltage has to remain all the time.

Sorry for another lengthy one, but I hope more people can learn this fundamental stuff because as you progress in this hobby (if you want to) the important of fields will inform more of how you design and build stuff (especially PCBs) and get them right first time.

@marcdraco 

Don't be sorry!

Very mind blowing stuff indeed! surprised I'm still living after that explosion ;>

Thank you so much for all the information! i feel 67% smarter now!

I decided to buy the parts and i can play around and see if this will work

If i think the quality isn't the best or some buzz is coming out then ill try and make something more advanced.

Again, thank you for the info! Wish you a wonderful evening, or whenever you'll read this.

Thank you, that's very kind.

That circuit won't work with a JLI2555 or similar because they don't have a JFET fitted internally - it does work with those "lucky bag" capsules that measure about 10mm or less across though. 😉 The WM61a is a good example of a mic with an internal FET and it's good too. They're discontinued now though and getting increasingly expensive. There are others like the JLI-6022CP1033-G45 which is specified from 100Hz (quite sufficient for normal speech) all the way into the ultrasonics which might make for a bat detector, a project that I've been meaning to create for a while now.

Mind you there are plenty of bats in my particular belfry. 🙂

Another example would be the JLI-9750AL26-HD which is more tailored to audio and also has an internal JFET (this is sometimes called an "impedance converter" but it means the same thing. It's show as a little triangle which is symbol for any amplifying bock.

Sarah's job is to provide a mount for the required JFET. There's room for a 2SK208 in SMD or an LSK170a on there and that basically turns it into a more "traditional" capsule - the advantage being YOU get to pick your own JFET which is almost guaranteed to be better than the nasty thing that comes with those others.

One word of caution - LDC (Large Diaphragm Condenser) electret capsules like the JLI2555 and similar (Ali and eBay sell the 25mm electret, fet-less ones for a lot less but they're not as good). Ironically, they are more sensitive though which is ... odd. I can only assume they're poorer quality but you can't really tell without a full studio test setup. For the price (£8-£10) they're very cheap to play around with. In my experience the JLI, which exudes quality, has quite weak tabs so once you've soldered it to the adaptor it's a nightmare to get off again. For this reason all the gubbins that make it work are on the outside so you can get at them. This applies to both Sarah and Varee but there's nothing saying you can't wire this like Matt did.

The difference is you'd hook the gate terminal direct to the centre pin and the source to the shield.

And the gotcha is.... there's EMI no screen.

Which means that the JFET will collect EMF interference - audibly at 50/60Hz which is the mains frequency as there are fields spreading out from all of your house wiring and it will pick it up.

The reason we don't need to screen those "el-cheapo" lucky-bag mics (even the MEMS ones which are a whole different and quite fascinating departure again) is that the JFET is surrounded by the case and circuit board. I've experimented a little (nowhere near enough) by popping the FET on the inside of the PCB but it's fiddly and results are "m'eh". I expect the thing doesn't create enough of a Faraday shield because it needs "breathing" holes for the cardioid pattern.

One thing you might want to try is to take the supply voltage direct from the PC or USB adaptor. This saves faffing around with a separate power supply. Another interesting experiment you might want to try is to use a 4.5V cell pack from one of those "pound store" LED torches. The nasty ones with the multiple white LEDs solder in the top. I've made actual microphones (only "omnidirectional ones) in these case and the three AAA cells will provide perhaps 100s of hours of use before they expire. This is because the actual current drain is minimal.

Check back with me if you hit any snags. For experimental purposes though, you might want to grab a pack of cheapo ones - I just looked on Amazon - (search for electret microphone capsules there or on eBay). Sourcingmap - a supplier of generics - have a pack of eight, 10mm SDC (small diaphragm condensers) for under £7. The ones without pins cost even less but they are harder to work with.

@marcdraco

Quick response to not keep you waiting,

Yeah I got a fet with it too. And I'll do like Matt did, I thought I mentioned that, but guess not

So diaphragm > FET > the picture below > usb interface

OR

diaphragm > fet > aux

Ofc adding some protection from mains buzz

Anddd maybe filtering caps and stuff, will play around

Hope I won't get some devastating news thar there's some obvious thing wrongs, (I saw the reply and 1st thing I saw was "that circuit won't work"... traumatic- 1 minute after buying the parts 😢

Posted by: @marcdraco

↑

One thing you might want to try is to take the supply voltage direct from the PC or USB adaptor.

I don't quite understand what you mean here- what do I do with it? Add it to l and r channel of the mic (middle and tip of aux)?

(I've got a bunch of thoes tiny mics and even some weird ones that somehow work: Ive opened them up and it's thoes plastic speakers that... are used as a microphone- first time seeing such a thing-)

For all I know such a microphone is super common but before being interested in microphone I only knew about electric mics (thoes tiny ones)

Mic is very high passed tho- mostly top end.

As for the bat detector- I've made the opposite! A directional speaker! Using ultrasonic, it is quite the fun project if you ever get the time!

Andddd this isn't "keep it short" well, have a lovely evening!

There are some quite esoteric microphone technologies once you start digging. It's a huge area and we're barely scratching the surface night now. Early telephones (you hear the effect to this day in movies when people are making phone calls and the sound is all scratchy) used a variable resistor made from carbon granules. Now that was crude!

Some years ago I made (a poor) ribbon mic. Like my other hardware attempts, I fail at the physical stuff. The ribbon mic "capsule" is actually called a motor because it uses a piece of aluminum foil surrounded by a very strong magnetic field. As the foil moves, the magnetic field causes an incredibly small amount of current flow which (without some very tricky, hand picked transistor circuitry) requires a specialist transformer to interface to its amplifier.

Ribbons are still prized today and cost scary amounts of money. Most are incredibly fragile too although I did hear of a development using a new material (presumably a fancy polymer) that works as well but is very resilient. They have an interesting pattern to - with a two lobes, the area they are sensitive to sound - spreading out in front and behind but are almost totally "deaf" from the side. This made them popular in the early days of cinema with the early cameras that rattled like my bones falling out of bed on a morning.

I've put it to one side for now until I can find a reliable way of constructing one that we can all enjoy. The difficulty I found wasn't so much making the ribbon (I made some tools on a 3D printer for that), rather the adjusters so set the motor tension and the distance the two super-magnets sit from each side.

There are other technologies too - even for condensers, there's an alternative method of interfacing which uses a radio frequency oscillator and a bridge (patented by Peter Baxendall in the 1960s). I've got the parts for this but getting the physical size down is proving tricky. 🙂

I've even tinkered with a home-made condensers but, since electret material is hard to make, it has to be biased with quite a high voltage and it's nothing more than a curiosity. Certainly not a project for us here. Next up, I'm going to do a "Shotgun" hyper-directional mic assuming I can find the time.

Some of the early work on this project provided some interesting designs with MEMS mics with ASIC chips on board. At the time I felt they didn't perform as well because the "affordable" ones were only omnidirectional which limits their usefulness for speech. This is something else I can revisit at a later date.

Have your mind blown (again) at the huge selection of technologies - including some that even have the ADC on the mic capsule that are small enough to fit into... well, take a look!

https://ecmicrophones.com/solutions/

Speakers as Microphones

Mics and speakers share a lot in common, you can generally make one do the work of the other because they are the opposite device. All devices like this that change something in the physical world to electric signals or vice versa are "transducers" but not all transducers are microphones/speakers. And yes, there are condenser/capacitor speakers but they are super expensive and usually very large indeed.

Most speakers use a coil and fixed magnet, just like a dynamic mic does but dynamic microphones are even less sensitive than condensers so they tend to be limited to purely vocal use where a speaker is expected to stand next to or even sing into the microphone, perhaps on stage.

When the cat decides he's had his snooze, I'll check the exact wiring on one of those super-small ADCs for compact use like the UGreen USB sound card because these are ideal for a quick and easy solution that might not be up to the standard Matt's set (VERY high) but will still give excellent bang for your buck (under a tenner).

OK, rather than edit that post (and now the cat has nipped to watch the magpies eating his wasted meat 🤣 ) I've had a few minutes to look at "simple" USB audio adaptors.

To my great surprise the voltage is quite variable across four different devices from 2.5V to 5V DC. But what I did come across was a UGreen TRRS device with a 384ksps and (claimed) high fidelity output for under £15!

The upshot of this is (don't tell @diyperks cough, cough or my bank manager 😳 😱 ) is we can potentially made a really, really good microphone based on the Sarah adaptor for considerably less work and less money than the original.

While the new designs are still valid, this is a faster and cheaper way of converting one of those nasty BM800s into something highly performant. This also suggests that it's worth revisiting the Varee design to get that to operate at such a low voltage. I doubt (so I'm going to see) how low it can go in real terms but 5V should be enough. It will require some work at my end, but this is quite exciting.

@marcdraco wow your really making me want to make a load of diferent mics now!

fascinating stuff!

you mention

Posted by: @marcdraco

↑

UGreen TRRS device with a 384ksps and (claimed) high fidelity output for under £15!

is this the item? if not could i get a link 🥺 - ill make sure not to tell diyperks or your bank manager!

https://www.aliexpress.com/item/1005006608451329.html

i bought 2 adapters that seem to have decent reviews

and

i will mention if any of them are good (no mic noise) in case someone want a cheap one (one's £3 and other is £2) once they arrive

This little fella I think.

Now there is a little, tiny, gotcha here.

The 3.5mm socket is a TRRS (supposedly TRS compatible) adapter.

So let's see what the heck all that means. Here's a TRS (Tip, Ring, Sleeve) plug which is used for stereo headsets and microphones.

You can see there are three bands on this plug. From the left we count Tip, Ring, Sleeve - which is rather obvious when you see it. (*We used to call them "trash" plugs because while the 1.25 inch ones were robust and long lived, and are still used for professional audio, the 3.5mm and 2.5mm ones are a nightmare to work with.)

The sleeve of these plugs is (by convention and standard) always ground. The others have different functions depending on what the designer is attempting. The simplest form of jack plugs (which date all the way back to the earliest days of telephony) only has a the tip and the sleeve and has either mono sound or can be used with either a dynamic microphone or a self-powered condenser such as the type you're looking at for your project.

It gets interesting when we move to TRS - because a PC mic input puts a "bias" (actually the supply) voltage on the ring connection. Moving up to TRRS things get a little weird.

In this case the DC and signal are carried in a "phantom" power arrangement. This means the DC comes out of the sleeve and with the signal going back on the same line. This isn't difficult, it just means connecting the output capacitor to the power input but that also means the signal gets back onto the power rail which causes more issues.

As we're chatting here, I'm developing an intermediate version of Varee that will drive an LDC direct into one of these devices.

Sorry for the late reply, thanks for the heads up,

you mention:

Posted by: @marcdraco

↑

In this case the DC and signal are carried in a "phantom" power arrangement. This means the DC comes out of the sleeve and with the signal going back on the same line. This isn't difficult, it just means connecting the output capacitor to the power input but that also means the signal gets back onto the power rail which causes more issues.

im not sure this is a difficult issue to fix, and i wouldn't of known this could be an issue without you saying, since my motherboard on my computer can take stereo mic's meaning both the tip and sleeve had voltage, and when i plugged in a mic like this:

https://www.jlielectronics.com/microphone-capsules/jli-160a-t/

(with a fet)

and it was working just fine

anyway thanks, ended up canceling the other usb aux thingy ones from hope the one you mentioned is good enough, i mean.. it is more expensiveeeee- so its betterrrrrr- righttttt?

Well it looks better on paper - the specifications are better than the audiograbber we used for the original, 384Ksps which is a considerable improvement and based on my experience of the more recent ASICs, it looks like it could be a real performer and overall it's way cheaper. What could go wrong... EEEK.

Quick update (I'm not sure if the edit button will send and email) so I'll pop any thoughts in here.

The UGreen is SMALL. Very, very small. "You just won't believe how mindbogglingly small it is. I mean you might think that a the chip on an Audiograbber  is small, but that's just peanuts to the UGreen". (With apologies to Douglas Adams for butchering that quote."

But yeah, it's small. Essentially nothing more than a TRRS socket and a USB C connector joined by a few mms of wire (that you really don't want to be slicing and dicing).

I suspect we're going to need some way to extend the USB cable because that's what we want to come out of the mic - keeping all those electronics screened and conveniently connected.

As for the voltage measurement I promised - I got a "cheap" Amazon TRRS male-to-male extension, cut it in half and then the problems started. The plastic they used on this one is so rubberized and flexible, cutting this stuff needs a razor blade and that's mucho-ouchy if you slip.

The usually "safety first" me made a similar mistake with a boxcutter (Stanley knife) a few months back wiring some 15A cable. That cost me a knuckle when the knife slipped, it took weeks to heal up requiring some fairly unpleasant removal of dead tissue (debridement) and a lot of antiseptic. This is what I get for being in a hurry.

So it looks like we're going to have to solder a TRRS plug ourselves unless someone finds a pre-wired one that has wire that an end-user with basic skills can strip. I've done it - and now I've got to find the UGreen because - you guessed it - I've lost it again.

I can imagen that pain, awful. As for the trrs connector, if the wire inside is thst un-solderable string, you can split the connector. Like at the end the plastic bit in half and solder it yourself. And for extending the usb port, my plan was a female usb connector, and a female usbc connector (that will be the actual connector / interface for the mic if that makes sence

(Replacing original xlr with usb-c)

Solder that together and shove it all in

I've also made very awfully designed stand for the capsule that uses rubber o' rings to keep it suspended for hopefully any weird sounds when tapping the desk, even tho the bm800 has some sort of thingy for that. 3d printed