[Sticky] USB-C Microphone (official topic)

Cats will sit. It's a cat thing. 🙂

I'll sort that stuff later today, my "easy" job of restoring my backups to the new machine has hit a bit of a snag. Not a computer snag, the family needing "dad's bwanes".

My kids grew up with the Internet as part of life and yet they still come to me... I started before the ZX-80. I remember that being launched but I could never afford one at the time. Then Clive did it again a year later with the ZX-81! Piffle.

Ok hilarity aside. I have had a couple of JLI-2555 copies from Ali and even a real one (amazingly). What I really need to do is hook up a couple of alternatives so everyone can hear the difference. Since no one is going to be impressed by my Northern English patios, so I've asked a mate to ask his bandmate if he will afford us a couple of hours to record him on an acoustic guitar.

This *appears* to be a genuine unit - the main giveaway is that you can see the holes where the rear pressure wave enters the mic. The spacer is a light tan colour and the mic looks to be steel or something similar. I'm not 100% sure what it is. Ali Express price £4.34 - up to around £20 (weird that but hey-ho).

Now the copies aren't drilled (they have a plate on the back with drilled holes but it doesn't go into the body) - this particular one is listed at almost £14 and it's a fair bet that it's an omni. Those rear vents are essential. These little gold omnis are surprisingly sensitive though so they might find a place.

You can see the venting on the 34mm capsules - although it's worth mentioning that these come in electret and non-polarised versions. Electrets don't need a biasing voltage, but the (generally more expensive ones) do need a DC bias of 40-80 volts. They often look the same so make sure you're getting an electret. I'll do a P48 compatible generator but like anything, I'm slooooooow so rather than show everyone the progress, I figured it's better to just drop them with little fanfare and let Matt do his magic.

I do have an experimental DIY capsule. (I'll put a pic up when I have this laptop sorted) but that will need a biasing voltage. I have a 60 volt generator (from 5V) using a charge pump bucket brigade that works quite well but the capsule is kinda tricky to build and (frankly) I'm not in the same solar system, let alone the same league as Matt when it comes to physical assembly. I'll stick to CAD and electronics. 🙂

In theory it should perform quite well but (not unlike a ribbon mic) the materials are quite difficult to work with because they are thin and quite fragile. 

 hello everyone!, as you guys can see from the images, this is the alpha, the upgraded v1. Huge thanks to @marcdraco for the pcb file and for helping me. Below are some sound test!

I read the updates for the last few months. Will V2 finally see the light of day?
I've been waiting for this for so long, I already want to make a new version 😎 

Posted by: @polaryetti

↑

I read the updates for the last few months. Will V2 finally see the light of day?
I've been waiting for this for so long, I already want to make a new version 😎 

This is entirely my fault for the glacial development time but yeah, it's ready now and the quality of this thing is boggling - I had no idea the quality of modern parts until I got tired of fighting with the USB whine that plagued mine and started over.

I'm very grateful to @muzammil17 for being very patient with me while I tinkered with the "hand-solderable" (THT) version "Perks Alpha" that uses most of the original BOM with a few additions for protecting the downstream digitiser. Muzammil built his with very little guidance from me and his results stand on their own. There are a few minor issues with the silkscreen, but as you can hear the result above. Performance is at least on a par with the original but the PCB is just nicer to work with.

There are parts of the world where taxes, etc. make the full V2 impractical which is where the Alpha comes in since it's a simple, two layer board with that's entirely through-hole parts.

I dropped @diyperks the SMD kits last week so things are moving. The full SMD "Songbird Michelle" outperforms the original design in pretty much every way possible but is more modular too (it also comes in two versions but enough of that for now).

As this is going to be the basis for the new model some parts are still under embargo but if anyone wants to jump the queue such as it is, I can supply the project files as KiCAD 9 project or just the Gerbers, BOM and position files ready for JLCPCB.

Unlike the Alpha, Michelle contains the entire device and only requires an electret microphone and JFET to operate. Total cost (taxes and postage allowing) is similar to the earlier version, but due to the way JLC works, you have to make a minimum of two*.

Additionally, Songbird Michelle has a stereo I2S digital output for things like (but not limited to) headphone monitors, VU meters and spectrum analysers. I haven't developed any code for these as yet but there's nothing to stop anyone in the community from doing their own. This will be the true power of the new one. USB-C output is mono only however, this is a hard limitation of the USB chip. In theory, it should be possible to use I2S with something like a Pi Pico (RP2040) or one of several other "Arduino compatible" boards to pull stereo to USB - this is something I'm leaving to people better qualified than I am because USB audio endpoints are a bit of a faff on.

I've also developed a two new capsule modules that are just about to go into "beta" (both have been hanging around the 'shop for months because I got stalled by CMedia who refused to supply the software we needed for their later chips to "makers" like us. I'll stress though, that they're not required for the build and (assuming they work as designed) they can be easily added at a later date. 

* If anyone is interested in producing runs for the public to order, please feel free! Neither Matt nor I are set up for that. As few as 10 boards should be enough to represent a decent saving. The SMD ordering is fairly painless although there are a couple of caveats as the board requires impedance control. These things are better explained in a video of course. I'll do a full write up for everything as soon as the last couple of modules finally limp out of the 'shop. The project files (like a patent) contain everything you need, including the correct impedance controlled board code for JLC but people unfamiliar with the process - maybe hold off until  someone else has got some in stock. 

I can't promise when or even if a video will drop because that's entirely Matt's purview. This is a team effort and my part is purely the electronics. I set out to make a spiritual successor to V1 that could elevate "maker" microphones that rival some of the best out there while keeping them accessible and affordable.

I would imagine we'll start a new thread if Matt goes that way so any questions and development can go there, saving folk have to go through pages of me waffling. It's also useful for someone else to independently test the project files in case I missed something. First person to volunteer can have them before the Open Source files are published. 🙂

Hi, sorry for the interruption, but is there any board design (without a DAC) that does not use the THAT1512 or "exotic" FETs?

I've read a couple of pages, but sometimes it is hard to follow.

There are a few kicking around from earlier work. The most interesting one is based on the Padless Preamplifier (based heavily on a design by Douglas Self). You can find it on my github where the unofficial design repo. is.

It's an old design now but stands up surprisingly well. I developed that because a lot of people had issues sourcing the THAT151x series chips in some parts of the world. The Padless uses fairly unremarkable parts yet delivers a pleasant sound.

JEFTs aren't so much exotic (at least, the ones I've used here aren't) but the number of available parts is small and continues to shrink year on year as fewer are made in sufficient quantities to keep companies making them. More advanced JFETs will continue in high-end designs for commercial and military use but in JFET input operational amplifiers (like the OPAx134) series from Texas and the LSKx89s will persist until someone dreams up something better. Given the FETs haven't really changed all that much in over 100 years (the idea was proposed long before we could actually make one and decades before the bipolar junction) that's unlikely to happen but never say never.

The final V2 employs inexpensive parts but as they are SMD that does mean they are best bought direct and assembled from a board house like JLCPCB.

Alpha, which is based on Matt's original but with a couple of "creature comfort" improvements does us a THAT1512 but the THATs are based on a fairly popular layout for 8-pin INAs so it's not the only game in town.

If you really just want to use "common" and low-ish cost operational amplifiers, you can always make a discrete design from a pair of OPA2134s (leaving a single amplifier uncommitted which could, for example, act as an active filter. It really depends on what you want to do. I've probably gone through >40 iterations to settle on something I can finally call "good enough" (because it should outperform anything currently available to makers). Plus, as it will be fully Open Source, anyone with some experience can hack them around to do other things or (for example) employ an OPA input stage - even a single amplifier design - and drive an ADC and USB chip... or more. 

FREE PCBs! (And some updates.)

Santa is here already! I have a couple of spare Perks Alpha boards here, as demonstrated by @Muzammil17 so if anyone wants one, drop me a note and I'll drop one on the post. Sorry this is UK only and first come, first served. Support is limited as it's not an official design, just something I threw together for people who don't want the expense of a fully SMD design. 

Perks Alpha is very similar to Matt's original with the ability to narrow down its maximum output voltage so you don't accidentally blow your digitiser. This is through-the-hole (THT) design that uses Matt's original BOM with a couple of small additions like TL072 and some low-cost Schottky diodes. The silkscreen overlay makes it easier to construct and it uses the THAT and NMA0515 from the original.

This is not the V2 (which is out of prototyping, but having a few last-minute tweaks) and, perhaps obviously, requires Matthew to "bless" it with a new design. V2 is more modular and can accept a dual-head.

One of the new modules is the optional 48V inverter is complete but, required a specially designed capsule adaptor designed for donors like the BM800 from Neewer. I don't expect these will be ready before the new year since the need a new power unit to supply sufficient current to run JFETs like the 2N4416 and LSK170s which don't provide enough gain (gm aka |Yfs|) at the very low-voltages the V2 was designed for.

FETs are weedy little fellas (compared to their BJT cousins) and running the larger parts at very low currents reduces the gain further still. While there is sufficient gain at the amplifier, I felt it was better to deliver more signal to help drive other preamps.

Anyone desperate enough to get their hands on one can order them from JLC - and only requires a small, low-voltage electret capsule with internal FET (ideally). It will work with just about any small electret and offers a virtually flat frequency response across the entire audio range with a noise floor that's absolutely outstanding. Before the official release drops, support will be limited. The plus side is they are almost child's play to assemble.

Hello guys, it's been years since my last update on this topic (I think 2 or 3). I built the original v1 with all the original parts back in the day. So, I didn't got the time to read everything from all of you. I'm here to show you guys my set up. So here are some photos of my current set-up

You may not remember but I 3D printed a shell and bracket sistem. I did this for 2 different microphone with the same internals (one for me and another for a friend), same component, same internal isolation.
Those mic are great.
I connected mine using 3 meters of type-c cable and it's pretty good. Some samples atached

I'm very interested in you 3D prints actually.

Ok so quick recap on what i've done:
-V1 microphone with PCB reduced to the smallest i can 
-Added mute phisical switch
-Added 3.5mm jack output
-Added 43mm chinese Pop filters
-Added 4 vibration dampeners that divide the capsule from the body (they come from fpv drone parts) 

Every addon that I made is done scavenging parts from old electronics so i'm sorry but they are not a standard measure at all, for more context or measure I need to open it up again and I don't want to. Please refer to the CAD for more detail

For the 3D printed parts:
0,6mm nozle on a low end DIY printer, 0.2mm layer hight, 2 walls
-Printed in PLA for the most parts
-Green parts in PETG 
-Black capsule support in TPU 90A for better dampening

-Grey net voronoi part made with resin printer (any settings that works for you is good)

The models got some heat screw inserts in some parts, they are M3 screw ranging from 6 to 12mm. Heat inserts from Ruthex (M3S version, not ncluded in the CAD design but holes are big enought to take them)
The interal PCB is just wedged there in place, every other component is glued in place. Some clearence from the PCB is added to ensure a small brass net can be fitted to insulate from static noises as the original design suggest.
The top net is done using Meshmixer so here in Fusion is not as the printable models. 

Some screenshots of the project:

And for all the 3D file, you can find them on Cults3D from now on at this link:
https://cults3d.com/en/3d-model/gadget/diy-microphone-models-diy-perks

That's brilliant!

Hello everyone! I am new here and writing from Japan. I would like to start my DIY journey by making a USB-C Microphone. I am open to any advice, Also is it okay to order parts from Aliexpress, since most stores in link are not available in Japan.

Sure. Be careful with the capsule. That's the bit that matters most. There are some parts that are sold as clones which should work OK.

I suggest (since you're starting out) that you price your BOM first to see what your total cost is.

Matt and I are in the process of completing the V2 which is considerably better than the original and doesn't need much more than a USB-C cable and a plain electret capsule or capsule WITH a FET. I'm giving the new one a couple of minor tweaks at the moment, but it already exceeds the original in several important metrics.

Total cost is worth checking against having a V2 made up in China, but that will change according to import taxes and shipping. I use expedited shipping so that adds up for me, but I estimate the difference in cost is minor,

Posted by: @marcdraco

↑

Sure. Be careful with the capsule. That's the bit that matters most. There are some parts that are sold as clones which should work OK.

I suggest (since you're starting out) that you price your BOM first to see what your total cost is.

Matt and I are in the process of completing the V2 which is considerably better than the original and doesn't need much more than a USB-C cable and a plain electret capsule or capsule WITH a FET. I'm giving the new one a couple of minor tweaks at the moment, but it already exceeds the original in several important metrics.

Total cost is worth checking against having a V2 made up in China, but that will change according to import taxes and shipping. I use expedited shipping so that adds up for me, but I estimate the difference in cost is minor,

Thanks for the advice! I’m happy to wait for the V2 since it sounds very promising. Regarding the cost, I think it will probably still be cheaper than buying a branded one for several hundred $.

Do you happen to know any reliable sellers on AliExpress for the capsule or other parts?

Thanks again for your guidance!

Either version should outperform the "usual suspects". Done right, Matt's original with the veroboard layout far exceeds the performance you might expect from such a simple design. Copies of Verboard are made in China/India etc. but the SRPB boards are inferior and often will come apart as you're trying to solder them! I've done an Open Source version of Matt's circuit with a couple of minor tweaks as demonstrated by Muzzamill has demonstrated.

Ali is difficult to account for since there are so many sellers on there and we don't know if any are going to have a reliable source of <insert component here> so they might have a supply of THAT1512s (the 1510 is cheaper and similar in performance for electrets) or they may source them from someone else. We simply don't know until the parts show up.

Although the costs may be a little higher, there's a lot to be said for getting your board made up (minimum is two sadly) at JLCPCB. Over the several 100 PCBs I've had made or assembled at JLC over the last few years not a single one failed due to their error. That's impressive despite them being not as high-quality as some of European houses, the difference in cost is night and a blazing noonday sun. Also, we know that they generally supply the parts I specify. (NE5532s are a notable exception. JLC sources copies from all over the place and while they work, they're not original parts and may lack a quirk of the original.)

The capsule is (broadly) the most important part of the build which is part of why V2 has taken so long in development. We wanted to offer a variety of different solutions to cater to many tastes without sacrificing the original design, so it had to work with Matt's original vision and offer sufficient new features or performance to salve the cravings of musicians, singers etc.

A 48V capsule version is in the works but that will be a V3 upgrade and I can't even say when it will drop (other than to say that the individual sections are now fully functional). Gain is an issue on "unbiased" capsules (the ones that require a polarising voltage) as the required load resistor is massive - 1 to 10 Gb. You can "get away" with small resistors and even a bit of carbon from a pencil, but loss of signal amplitude is a serious problem due to the loading. I'd always assumed that the load resistor was required (and assumptions can get me into a lot of trouble) but in fact, with an electret capsule, the DC coupling allows the JFET to self-bias without extra faff. Anyway, this upgrade can be applied to the "Michelle" or "Nightingale" board at a later date since it's just a couple of extra PCBs (not hugely expensive ones), the killer is that 1G resistor.

While I can't promise anything, there IS another one in the works for the real purists using something called a NuTube which is a vaccum fluorescent (VFD) based valve that runs at safe voltages. (Valves typically require around 300 volts to work well and that's not something amateurs should go anywhere near. I avoid them myself out of respect; a simple slip could mean a deadly shock.) This might need a small Kickstarter as it's unlikely to attract anyone but the valve lads and most of us are pretty ancient now. 

Sorry the photo leaves a lot to be desired, but this is the current state of play re. V2. Well, this is the main board that works (pretty much) out of the box. It has a modest gain but that can be adjusted like Matt's original. Note the "missing" parts are for a dual-channel version which is only supported by the digital section and is not passed through to the USB due to limitations in the CMI108.

@marcdraco Been a while draco so glad to see the v2 is almost done!I have 2 questions mainly first im aware that the v2 is going to be usb c will it also have a option to be xlr? if so is they're any audio benifit to going the xlr way with how good usb c has gotten? lol also you said the dual channel version is a limitation due to the cm108 but how would a mic even benefit from being dual channel?

XLrrrr.. Ring My Bell

OK, so the confusion over XLR/USB comes from pre-digital era and yet exists in both.

XLR - is simply the connector and is a highly robust design that (unlike the traditional jack which came from the days of early telephony). The problem with Jack (be that 1.25", 3.5mm. 2.5mm, two, three or even four way) is that the "live" terminals connect before ground. With "high" voltage, low-impedance connections like a headset aren't affected, when you start to operated with low-level devices, the amplification causes serious noise issues when the device is plugged or removed.

The other thing with XLR (and a three-pin Jack) is that it can carry two signals and a ground (reference) and many of us think of this as being a stereo connection from the classic 3.5mm headphone jack, popular from the 1970s and is still used today for low-level consumer stuff.

Internally, USB (regardless of version) is similar to a "balanced" microphone cable but has an extra cable that carries a separate power supply (5V generally, but some devices with USB-C, called the PD standard) are configured to set themselves up with different voltages and drive current.

The term "balanced" is the crucial part here. You literally can think of the whole system as being a simple balance like a see-saw:

What matters, unlike my image, is that the impedance (the resistance to AC current) is balanced. A confusion arises in that many believe that the signals have to balanced - that is, equal and opposite - too. To quantify that, as the signal on "A" goes positive, the signal on "B" goes negative by the same magnitude. So if A=1, B=-1; A=-1, B=1.

In other words, the Sum of (A) + (-B) and vice versa is always 0.

Apparently not a whole lot of use but consider this:

If the two (A and B) cables are in close proximity - they're typically twisted quite tightly to ensure "balance" - then any interfering signal will reach enter each wire equally. That applies to AC and DC in fact - any signal that appears in equal amounts on both inputs is called "common mode" simply because it's common to the two wires.

So what we have is a situation where the NOISE (mains hum primarily and other low-frequency AC interference, R/F is a different issue) appear equally on both wires, hence the SUM A+B = 0 = no noise, in theory. In practice a very strong interfering signal can still punch through at some magnification, so it addition an outer screen works as a Faraday cage adds even more noise immunity.

We can see just how good even a simple "unbalanced" cable works because they are used in low-end wired microphones, including condensers and dynamic mics too. Dynamic mics are particularly prone to noise since they produce even weaker signals than condensers and yet they can still work quite well in consumer equipment - entirely as a result of the Faraday screen. That said, there are screens and there are screens - and some are better than others.

Passing a twisted pair through a screen (Faraday cage) provides a belt-and-braces approach to noise rejection.

The detail I missed out here is that to get a signal out of of balanced line, we need to deliberately invert one of the signal pair before it leaves the sending end and perform the sum (flipping some signs here to keep it simple):

(Signal + noise) - (-Signal + noise) = Signal + Signal + 0 Noise = 2 Signal;

In other words, we end up with twice the signal level and no noise.

Which is neat - but it's still only part of the picture. Imagine if you will, we only put a signal on one of those wires, we get this:

(Signal + noise) - (noise) = Signal + 0 Noise;

As a slight side-note: balanced connections are all over the place - even the data lines from your CPU to the RAM are balanced to prevent data corruption. It's a hugely powerful technique.

What matters here is the IMPEDANCE of the two wires with respect to the ground: that IS crucial. While data in USB only moves in one direction at a time the signal level on D+ and D- is always opposite. When D+ is logic 1, D- is 0 and vice versa but think of the data is sent using an electrical signal just the same way. Although the signals in USB are DC (that is, they change from 0 (off) to 5V (on) they are are still inverted with respect to each other so far as the USB receiver is concerned. Despite switch at very high frequency, this balanced setup ensures that USB is very resistant to noise.

I've said that the impedance of the connection matters - balanced audio that's nominally 50-300 ohms, but for USB it's 90. RAM has its own standard and UTP Ethernet another ... but in each case it's the impedance of the wires (as "seen" by the signals) that matters.

We can see why (it might not be obvious) from Ohm's law:

Current * resistance = a voltage.

Noise is a signal, but it's a signal that we don't want and there are two ways we can deal with that. 

1. We can "swamp" it by sending a much larger signal down the line.

2. We can use balanced cable.

Professional grade gear often does both. 

The low impedance (50-300) interconnect serves to reduce the voltage difference because any current impressed by interference is quite weedy, a small current multiplied by small impedance is small voltage, right?

It matters because our preamp is going to be boosting ANY signal it sees by a factor of 100 to 1000 and maybe even more.

If the impedance (not just the DC resistance) comes into play here because if we match the A and B cable impedance we can be sure that the noise signal that appears at the receiver terminals is the same. If it's equal then it all gets cancelled by the receiver while our relatively feeble signal passes through untouched.

XLR of the Opera

Matt's original could use a XLR connection (although there are issues here with phantom power I've not discussed) as can the V2, both use differential (balanced) inputs. IF you want to use an XLR, and you have the option to do that, you would simply wire the two centre conductors to the Drain and Source of the FET and the shield to the capsule's body.

I think what you're referring to here is a P48 Phantom connection as popularised by the OPA Alice by Scott Helmke which (as some wags have commented) is just an op amped version of the original Schope's Microfon circuit from the 1960s.

Schope's (designed by Jurgen Wuttke) circuit - famed and loved by mic builders the world over took an even earlier design paradigm based on transformers.

To understand what the deal is there we need to consider why we used 48V at all. That likely came from early telephony equipment since 50V is widely thought to be the safest maximum voltage to work on. No one wanted telephone operators getting shocks when they made up the connections back in the very early days.

When condenser mics became popular (they have multiple advantages over dynamic mics which are little more than a miniaturised speaker) they had the disadvantage of needing a polarising voltage to charge the capacitor.

Without diving too deeply into how condensers work (again, I've done it in the thread), they are just tiny, air-gapped capacitors that have to be charged before they can do anything. A DC voltage (sometimes as low as 20 volts or as high as several hundred in odd cases!) creates an electrostatic field that varies proportionally to the sound pressured being applied. The very best ones are so sensitive they produce a "self-noise" which is actually the sound of air molecules striking the diaphragm. This is self-noise will always be there but it's more than swamped out by differences in air pressure.

Back in the days before transistors the only way of getting that 48V up a cable and a balanced signal back was to use a pair of centre-tapped transformers. The DC polarising voltage was applied to the centre tap at the pre-amp and then "lifted" at the mic head to polarise the capsule. The A/C wiggle travelled back up the same wires but due in opposite phase (balance). A balanced input made with thermionic valves did the job of putting it all back together and although it sounds complicated, it's quite straightforward.

A lot of (sniffy) purists insist on "valve" sound or a "transformer" sound, insisting that the distortion caused in those transformers is pleasant and even desirable. Worse, what many won't admit, is that valves have a limited upper frequency response so those early condensers need a fair bit of boost above around 3-4K to give the audio a sparkle or presence that we know from life.

So Wuttke designed an a circuit that is as elegant as it is beautiful. He used PNP transistors in an Emitter Follower configuration so that he could obtain the voltage to operate a small impedance converter (the JFET in our case) with a balanced (differential) signal that was output via the transistor pair. I spent some time grappling it myself because it almost looks like it can't work - but when you see how it does... I won't draw the circuit again and there are lots of  copies of it on the web. If I ever manage to finish the book about this, there is a full description in there.

I did make a couple of JFET capsule adaptors (Varee is the best one, but that is being superseded by the later Songbirds) that are miniature versions of Wuttke's design and allow you to use a JLI2555 with something like a FocusRite Solo (or similar).

The Phantom Menace

So when we think XLR and balanced inputs the real question is are we thinking a impedance balanced signal (such as a dynamic microphone) OR a phantom power.

This is where it gets a bit tricky.

The input circuit has to have small capacitors to block the "phantom" voltage before it gets to the very sensitive input amplifier. This isn't usually a problem until someone trips over a wire and rips the plug out WHILE the power is applied.

When this happens the charge on the input capacitors has to discharge somewhere - and if we don't have some form of protection - that's going right into your sensitive, fragile and above all, expensive, differential amplifier. It could be a THAT1512 (or 1510) or any one of a number of these devices, but all of the suffer from the same problem. 48V rapidly discharging into a few ohms causes a reverse current to flow into the bases and will either weaken or destroy the chip entirely.

The solution is to use input protection diodes that short any loose phantom voltages back through the supplies - but that comes at the cost of extra noise.

The other reason we use low-impedance for microphones is that resistances generate noise. The larger the resistor, the greater the noise. This is only one of multiple noise sources but it's probably the most insidious (particularly as we have to have something like 1G to load the JFET in some cases).

To A-B or USB? That is the Question

It depends.

P48 is an option IF you want to connect a capsule to a "professional" mixing desk and while it's a good standard, it's primarily for running very long cables over some distance -- think 10s of metres.

USB can't do this because the digital signal degrades over the wire (regardless of interference) and errors creep in. Since USB doesn't have error checking that limits it to something like 5M total (hubs and the like add more decay for reasons I won't discuss here). TCP/IP could be used but that requires an entirely different design and then we're looking to radio for range without a wire.

For connection to a computer (or a USB-C capable audio recorder) a few meters of USB-C cable is more than enough.

The connection to the capsule has a simplified version of a phantom power in the V2 and can go as low as 1,8V (this is an option I'm working on) to run consumer capsules with their own JFETs and 10V for a greatly improved (albeit more costly) capsule adaptor for ones like Matt's original.

Unbiased capsules come with their own issues, primarily that while the JFET can tolerate fair amount of DC, it's nowhere near the +48V biasing voltage so the capsule has to be isolated with a small capacitor (typically a nano Farad or so). 

TL;DR

If you hate reading as much as I do you've probably skipped to this bit. The rest will finally appear on the Artificial Ignorance and search engines so there's that. 😉

USB is a far cleaner way of transmitting sound over short distance and since it's also balanced to preserve signal integrity, it's almost immune to noise. You can use an XLR to feed some microphone cable to the capsule adaptor IF you're making a head like Matt's beautiful design (I still love the Steam Punk look) but if not, you can put the whole thing in a slotted tube and make a "hypercardioid" (shotgun mic). Another alternative is a suspended retro design from the 1940s radio. I'm providing the electronics and Matt can do what he does best and make it beautiful.

What I didn't "advertise" at this point as it's still undergoing testing is that the digitisation isn't done by the CMI108 (which is found in the little red Behringer box) but by a very high quality differential amp (better performance than the THAT) and a 32-bit (capable) digitiser.

So the V2 has raw digitial OUT and IN (passthru). I've worked on some test code with Earl Philhower to get a Pi Pico to read and write the digital signal which is I2S. This gives us remarkable performance, while not "ultimate" yet since it's still 16 bits out of the CMI108 which is sent to the computer at the moment. A fully 32-bit, 192K stereo is also capable from these output but at the moment we don't have the USB code for the Pico. A later (planned) module will likely replace the Pico with an STM32F7 series or even a Teensy 4.0. 

However, and Matt is the master at this side of things, we can even add other boards like the ADAU1701 to get effects.

And this is where the dual head comes in.

Although the CMI108 is mono only, CMedia only sends the "left" channel - the right is crippled in their firmware. 

But this doesn't stop us from combining the signal from two separate heads mounted back to back OR (better) a single head with dual outputs. By mixing these we can balance the rejection of sounds coming from the rear more effectively than a simple pressure gradient and that means we can have a switch to set the pattern from cardioid to figure or even omni. The 108's job is to convert I2S to an audio signal that we can read on a computer.

And better yet, we can tap from the digital signal and have things like a level meter, Fourier analysis and more. The sky really is the limit here. Some things I can do, some things Matt can and since this is all Open Hardware, anyone else can improve it.

I've dumped the noise-prone NMA0515 for a much quieter (inductorless) charge pump which supplies more than enough go-juice to drive several amplifiers. You'll note this one doesn't have those huge unipolar capacitors for example because using SMD parts I can access far better components that are usually reserved for professional grade mics from people like Rode.

To give you an idea, the NMA0515 has a maximum drive current of 33 mA total (say 16 mA per power rail). The new beastie is capable of 250 mA at +/- 5V. 

So yeah, this thing is a true successor to Matt's original improving every corner of the design. Of course, it's taken a lot longer to get here but I can assure you it's worth it. 

I hope that covers everything (if briefly) because like Matt, I keep my promises: we promised you an amazing experience.

I'll get a demo up soon but the last thing you want to hear is my voice, so I've organised an acoustic guitar session with some jazz musicians I know. If I can get the new capsule adaptors back in time, I should be able to show an unbiased capsule too. All the main electronics is done but I've been working on a more sensitive FET head using a superior JFET (a matched pair in fact).

Once I can get the dual channel digital section working over USB, we'll have the capability to do first order Ambisonics - although that's a way away for now, but it's in the roadmap.

There really are far too many improvements to list here - but the overall board (assembled) cost around the same as the original BOM but you won't have to chop up an existing part to make it work or fiddle with USB connections because they can create issues of their own.

@marcdraco It's me again after 6 months of checking this thread every once in a while. Just wondering if the V2 is in a state now where I can just get the BOM and gerbers and make the circuit?

@Wizard_pope: firstly, may I apologise for the slow reply. I was offered some free food by the family and they cook better than I do. PB&J sandwiches get a little boring after a few months. Levity aside, I did mean to post this yesterday, but came back to do some routing and forgot, so that's on me. 😳 😳 😳 

(There are still two parts Wizard, the main board is in good shape (although it will get periodic updates as time and budget allow). I hate this "process of continuous improvement" malarkey myself but there's also a time when you just "stick a pin in it" and say, that's it. It's good enough now.

Short answer, yes, you can get the design files and it works a treat with small "low-cost" capsules. I've found that different ones work better (or worse) by adjusting the load resistors on the input, but the big change for the unofficial/official V2 is I've dropped the supply from +/- 15V (30V total) to just 1,8V to really bring the noise down to imperceptible levels. 

This means (unfortunately) that we don't see the best response from more "powerful" JFETs like the the 2N4416, LSK170, etc. It's possible to swap out the load resistors but at such low voltage, it's not really practical to get the gm (the JFET gain) up to a reasonable level. Better JFETs seem to be specced to run at 10V - something born out from the datasheets and other's experience.

To "convert" the existing design into something more like Matt's you'd need remove the load resistors and power the head separately.

All of the details are in the KiCAD for people to tinker with - indeed, I hope someone else will develop the digital parts. You can take a feed from the I2S signals on Michelle and (although I haven't tried yet) you can even insert a digital processor such as a ADAU1701 into the digital chain and add effects, digital gain, VU or even Fourier level meters. My coding skills are a bit rusty so I'll need to complete the electronics work before I can get stuck in to that. Ultimately, I'd prefer to dump the CM108 completely and run the digital (I2S) to a different processor and then out to USB audio - this way we could get full, 32-bit 192K studio-grade recording. The CM108 is the bottleneck and it's hard coded onto the chip by CMedia (same with the lack of a right channel).

However (not all is lost)... 

The other half of this build was always a FET carrier (capsule adaptor) with a balanced output.

I've been chasing a decent quality and decently priced way to get a 48V phantom but I've come to the conclusion that the P24 standard suits us better and I've already done a tiny 5 -> 24V boost board capable of powering most phantom mics. (I've also done a charge pump capable of +60V which is about as far as we can reasonably go with that due to losses, etc.)

Internally, every P48 mic I've been able to find schematics for seem to regulate their main power to 12V (presumably due to the ease of using a 12V6 zenner. 10V is slightly better as it's what the manufacturers seem to target and it's where the best data for noise, etc. can be determined. It's not clear if a TL431/432 will add or remove any noise and the only way to be sure is to build some here and test them: proof of the pudding...

I can upload these but it might be better to have a separate thread so people don't get lost in this one looking for the various bits. A pinned post would let me keep everything current (GitHub is an issue right now... I, had an issue with an old phone and my authenticater so I've temporarily locked myself out. (The shame, OH, the shame I feel!)

Both these boards are separate but can easily be incorporated into the main pre-amp (Michelle) - the decision to wait or use the current ones is entirely up to you. Self-noise from this one is undetectable right now so it could be adapted to ribbon or dynamic mics too.

I'm finalising some versions of the capsule adaptors today (hopefully and cat allowing, he's still young and demands a lot of attention, I'm jiggling a laser pointer and typing with my other hand rn in fact. I do need two hands for KiCAD though).

If I can help in the meantime, just hit me up. Alpha is an easy build for people who only want to use through-hole parts (and rather amusingly, the Michelle makes for a very good digitiser too - I did an earlier version for exactly that that doesn't have the instrumentation amplifiers. 

Hi everyone! I hope youre all do doing great. im having a couple of problems woth my build and any help would be greatly appreciated. First of all thank you to all of the people who have sent help on this forum before, i was reading through all 39 pages and learnt almost as much as from the original video itself. As for my problems, they may be because of my inexperience but i dont have any idea how to fix these. The main problem im having is the preamp circuit: im finding it very difficult to solder up as this is one of the first soldering projects im doing, and ive already messed up a whole board. if it hadnt been for the great bitsbox's prices and speed i probably would have dropped this project already. are there any prebuilt amps which would work fine for this build? On top of this, ive seen many people talking about how the original preamp circuit has some mistakes in it but i havent been able to find out what those mistakes are. Also, ive been hearing rumours of a V2 or alpha board, so if someone could drop a schematic (that a non professsional can understand) then that would be amazing.

Another problem im having is with the overall amplification of the sound. Ive done some research and some places have told me that i dont need the first JFET transistor nor the actyual preamp and that i can just buy a cheap preamp from somewhere else and itll have similar performance. this has also really confused me so any help would be appreciated. 

The final problem ive been having is understanding the whole usb interface dongle thing. I bought one from ezcap and it looks simialr to the one form the video, but im struggling to understand how im supposed to conect it into the circuit because, in the video, Matt uses a USB breakout board which, forgive me if im wrong, was a female connector , whereas the usb a i have is a male connector. Would it be possible to bypass the whole interface and instead use normal audio cable if my PC has line in input?

Sory if any of the questions i asked were stupid or have already been explained before, ive been trawling through this whole thread for weeks trying to figure this out but i havent been able to. any help at all would be greatly appreciated. Thank you and happy holidays!

also, in case you need to know this, most of the parts i have are same as from the video (capsule, transistor, all the preamp stuff) but ill be 3d designing my own casing to go with my desk aesthetic

Afternoon, it's your local neighbourhood cat's butler here to answer your questions

Posted by: @qwertuy

↑

Hi everyone! I hope youre all do doing great. im having a couple of problems woth my build and any help would be greatly appreciated. First of all thank you to all of the people who have sent help on this forum before, i was reading through all 39 pages and learnt almost as much as from the original video itself. As for my problems, they may be because of my inexperience but i dont have any idea how to fix these. The main problem im having is the preamp circuit: im finding it very difficult to solder up as this is one of the first soldering projects im doing, and ive already messed up a whole board. if it hadnt been for the great bitsbox's prices and speed i probably would have dropped this project already. are there any prebuilt amps which would work fine for this build? On top of this, ive seen many people talking about how the original preamp circuit has some mistakes in it but i havent been able to find out what those mistakes are. Also, ive been hearing rumours of a V2 or alpha board, so if someone could drop a schematic (that a non professsional can understand) then that would be amazing.

I once wrote, "experts are not born, they are hewn from the bedrock of endeavour and the granite of experience". Perhaps a little grandiose but it's a fact nontheless. 

I've made a few mistakes in the thread but rather than quietly fix them, they might as well stay there to prove that I'm big enough to admit when I drop the ball or the entire shipping container of balls ...

Perks Alpha

Here's the schematic of the improved version of Matt's original. The main changes are the inclusion of a capacitor multiplier (or which is a fancy name for a simple low-pass filter with high-current capability) on the input to the NMA0515 voltage inverter/booster. 

The other two major changes are from a dual TL072 (any JFET amp will work here but the TL072 is cheap, easy to find and unlike the common NE5532, doesn't gobble up so much juice there's nothing left for anything else).  I didn't want to break the original and in fact, I'd hope to keep everything as was with just the addition of a digitiser on board. Alas 'twas not to be (digital noise on the USB power rails just beat me).

The notes aren't entirely up to date but it give you an idea of what's going on and what the additions are. If you're in the UK, I still have a couple of bare boards already for Santa Claus to deliver - you won't even pay the postage.

The "errors" aren't so much mistakes as things that could have been done better but I have a lot more time on my hands than Matt does so I've been able to dedicate that to improving this thing and boy, what a journey that's been! I could have developed a better electric car in this time. 😉 The biggest issue was that a simple "whoops" when testing the thing on full volume and you could easily blow the doors off the digitiser which gets costly after you've done it a couple of times.

My friend @Muzammil17 very kindly made and tested one for us and he got fairly decent results. Support is limited though since I only made this one up for people who can't access the better boards.

Perks Alpha can be connected to a Line Input on most PCs if you want to eschew the digitiser.

Posted by: @qwertuy

↑

Another problem im having is with the overall amplification of the sound. Ive done some research and some places have told me that i dont need the first JFET transistor nor the actyual preamp and that i can just buy a cheap preamp from somewhere else and itll have similar performance. this has also really confused me so any help would be appreciated. 

This is where the V2 comes in (more on that in a moment because it also answers your final question.

Some of this bit might be a bit technical but I've tried to keep it fairly simple. Feel free to skip or ask for clarification.

You absolutely DO need the 2N4416 JFET (or something very similar) for "FET-less" electret capsule unless you use a JFET op-amp (on INA) because a normal amplifier with a typical input impedance of 10k - 100K is going to overload the capsule (you won't be able to hear a thing).

Understanding this problem isn't easy - we're usually talking in terms of impedance. But impedance (a "complex" form of resistance to allow for the use of imaginary numbers and which also includes resistance) is really another way of explaining the numbers in terms of voltage and current using good ol' Ohm's law.

I won't bother with all the fiddling quantum effects at work here but you can think of it in terms of the capsule being a small voltage source in series with a VERY large resistor, perhaps 10 gigaohms or more. If you start with a ridiculously large resistance and precious little voltage you get ... (drum roll) a very small amount of current.

Even if we put the capsule in the path of a 787 as it took off (and before it got blown away) it would still be unable to produce much in the way of current. (It has to do with the way capacitors hold charge in an electric field, and the capacitor in the microphone is just a few pF - around 40-50 pF typically for these (tiny in other words).

A tiny amount of current through a modestly sized resistor ends up with a fairly feeble amount of voltage (Ohm's law) which isn't anywhere close to enough for a conventional bipolar amplifier to work. Bipolar transistors by nature of their design, pull a fair amount of current from the preceding circuits, far more than a condenser mic can produce.

And that's where JFETs (specifically JFETs) come into their own because the gate (= to the base on a normal transistor) doesn't draw current beyond a few picoamperes of leakage - it looks like a huge resistor. You'll see from the symbol that a JFET (the n-type that we're using) has an arrow pointing in, towards the bar. That's to indicate the presence of a diode junction - but it's back to front! Some tricky quantum effects cause this to appear like a tiny capacitor (even ordinary silcon diodes do this). For our purposes since a capacitor blocks current, we can assume that it's like a massive resistance. In practise the input impedance of a JFET is in the 100s of megaohms to many gigaohms. MOSFETs are similar but the gate there is an insulating layer so that gate really IS a tiny capacitor, not a diode junction pretending to be on. But MOSFETs are noisy blighters so that leaves JFETs a the tool of choice for low-noise, high-impedance sensors like microphone capsules.

I suspect the comments saying that you only need a simple pre-amp are probably thinking of standard electret capsules - which come with a JFET fitted into the case. This accounts for the majority of electret mic capsules made every year by a huge margin but it also means you're stuck with the specifications of the FET fitted to your part of choice and since you can't see it, you're stuck with whatever the datasheet says.

Noise isn't so much of an issue as JFETs are pretty quiet anyway, but how much signal you can get out of them, that's a whole different can of worms. There's no adjustment possible and even hacks like the Linkwitz mod (for a capsule that's no longer in production and in short supply too) are just that, hacks. (Linkwitz is a good hack for what it's intended for but it's limited to mostly high-level recording for live music, etc.)

The OTHER type of electret capsule like the JLI2555 and similar ones, doesn't have a JFET on board, but it does have a permanently charged plate. 

A WHAT? 

Weird right. It's a bit like capacitor that's pre-charged but doesn't lose charge even when shorted out. They work by pushing and pulling electrons from into and out from the circuit they're connected to but in such a way that they never lose or gain any. If that makes your head melt, don't worry, did mine too.

Finally there's the original type of condenser mic capsule, the unbiased form. These are generally considered to be the best (to my mind the jury hasn't heard all the evidence on this). A lot of that reputation likely comes from the very, very high end condenser mics that occupy the >£1000 mark typically. Each of these require a polarising voltage - around 40 -> 60V but it can be more or less - to charge one of the plates in much the same way as the electret material holds charge. The difference is that when the polarising voltage is removed, the charge dissipates and you're just left with a hunk of metal and foil. 

Unbiased capsules (there are three or four major designs) have been perfected over many decades and do sound musically brilliant but that sort of technology doesn't come cheap and to get the best from it requires even more esoteric design and exotic parts putting the cost alone outside of the reach of most of us.

This is where well-designed electrets - and JLI produce some of the best - fill in a huge hole between "good enough for a telephone" and "good enough for a platinum-record selling recording artist).

I'm really skirting here - some of this is covered in more detail in earlier the thread but that's the broad strokes.

Posted by: @qwertuy

↑

The final problem ive been having is understanding the whole usb interface dongle thing. I bought one from ezcap and it looks simialr to the one form the video, but im struggling to understand how im supposed to conect it into the circuit because, in the video, Matt uses a USB breakout board which, forgive me if im wrong, was a female connector , whereas the usb a i have is a male connector. Would it be possible to bypass the whole interface and instead use normal audio cable if my PC has line in input?

Sory if any of the questions i asked were stupid or have already been explained before, ive been trawling through this whole thread for weeks trying to figure this out but i havent been able to. any help at all would be greatly appreciated. Thank you and happy holidays!

Songbird "Michelle" et al.

Although I stopped filling this thread with every update (I failed more often than I succeeded) there were several things I wanted to address for a true successor to Matt's original.

1. Better specifications.

2. As few "moving parts" as possible.

3. Similar total cost.

I spent ages trying to figure out why Matt's, made up on Veroboard could sound so good (and it does) and no matter what I seemed to try, I couldn't rid myself of this infernal 1KHz whine.

This, it turns out wasn't where I was looking but hiding right under my nose. I have quite a large beak so it had plenty of room there. 😉

The noise wasn't coming from my electronics, but from the USB data impressing a voltage onto the wires connecting the whole thing up. It's a massive problem for anyone who uses USB to power sensitive audio devices because no matter what we do, the very design of the USB cable means we've got high-speed digital signals running alongside a ground and power cable. That's a huge "NO!" insofar as signal integrity goes but USB doesn't care because the lines are balanced and already digital so hard luck folks!

So began a search for a system that could meet our needs and I came up with .... zip, zilch, nada, diddly-squat nothing... 

Nothing that is both easily available AND low cost could come close to the performance of the original, they're just too noisy. Many devices (like the CM108B) are designed for things like headsets so they are meant for near-field phonics (that's a mic sat right in front of your mouth) and don't do well with far field at all. The 108 has an internal amplifier but it's noisy, lacks proper drivers (it works as a standard audio device) and a lot of the specifications aren't published by C-Media.

What it DOES have however is an I2S (Inter-IC sound) input and since it's still in production we should be able to get our sticky fingers on them for some time to come. 

The thing with I2S is that, being digital, it's noise free (within reason) so if we could just hook up an I2S microphone to that, we'd be laughing! The best known is the INMP441 which is found on low-cost digital microphone boards and has a 24-bit interface. Perfect right?

Weeeellll - almost.

The INMP441 is also a near-field unit and (worse) it's got an omnidirectional pattern so it doesn't reject sounds coming from the rear very well - vs. a conventional "cardioid" pattern mic like the JLI2555. It's a very small device too so lacks a low-frequency response.

So - long story short - I dropped in a 32-bit I2S digitiser from Texas Instruments and a good quality INA (instrumentation amplifier like the THAT1512) and we're off. The V2 has a number of other little tricks; circuits I'd developed along the way which make this build a LOT easier to modify for people who want to tinker with the Open Source hardware. The +/-15V of the NMA0515 has been replaced with a much smaller, but more powerful, inductorless chip that provides up to 250mA at +/-5V - meaning we can drop those huge capacitors needed to smooth out the main supply.

V2 is more modular and can be customised to suit a variety of different scenarios. The current version (pre-final) only drives the JFET at 1,8 volts as a way of reducing the noise from USB, etc. and is easily tailored to a variety of electret mic capsules - the ones WITH JFETs on board. These FETs are intended to run at low voltage - 1,5 to 3V and even with the medium gain resistor on the build, it can still overload the recorder without breaking a sweat. V2 has a number of ways to adjust the gain - it's got a fixed setting from the factory but you can remove the resistor with a soldering iron and use a resistor chain on a rotary switch as Matt did. V2 is also capable of stereo or dual capsules but the CMI108 only outputs a single channel, even though a stereo signal is sent to the digital "taps".

How much soldering is involved? In the simplest case, you would only have to solder the wires from a two-pin JST to the back of your capsule and then the socket to the board. So four joints in total. Result? Actually not that bad and the most cost-effective solution. You can get small - 10mm - cardioid capsules from JLI and Primo with decent JFETs on board so if you wanted to go that way you can.

Beyond these basics, we need a bit more voltage since of the generally available "good quality" JFETs from companies like Interfet and TI, work better at 10V. Indeed if you glance at the spec sheets, you'll see that most of the specs are listed at 10V and some fixed current - 2 - >5 mA. Parts manufacturers are very much of the "but your actual mileage may vary" crowd and often give examples of how well their part works under conditions that aren't practical (or that we can't actually achieve in a working circuit because each example uses a different setup to demonstrate a given metric). Crafty buggers.

And what'll really spoil your day if you're designing with FETs - the blasted things have vast tolerances for important metrics - 10x or more worse than even low-cost, jellybean bipolar parts.

As as result of all of this, I've done a number of different capsule adaptors (JFET carriers) that perform a few different tasks to make this a little easier. Soldering an transistor point-to-point at Matt did for the original is fraught with gotchas from snipping the wrong leg to snapping a leg off. All very messy, so I've made some boards that can carry a variety of different devices from a 2N3819 (a jellybean JFET in a TO-92 epoxy case) through to SOT-23 and SC-59 SMD parts. This makes it a little easier to get a JFET that suits your needs but in those cases, you're still limited to the "simple" degenerated common-source design I've used on Michelle although it does mean you can pick a nicer capsule - again the JLI2555 is probably the best in this sector.

Finally, the current development is for a MUCH better capsule adaptor that descended from the early experiments into the Schoepes design. The OPA Alice microphone is an example of this circuit (simplified down somewhat) and replaces all the active amplification and impedance matching with a single (dual channel) JFET operational amplifier, the OPA2134.

Alice is intended for 48V Phantom - abbreviated to P48. In this example we're send a POSITIVE (set to 48V) voltage up each of the two signal lines and extract a balanced signal at the other end. Matt's original dodged that one rather cleverly using a split rail but the effect is broadly the same.

Thing is that 48V is a real bind to have to generate from a piffling 5V USB power supply but it turns out that 24V isn't and there's a competing standard (nowhere near as well known as P48 called: shocker, P24. There's even P12 but I found that's too tight to regulate well AND get a 10V supply from). 

Regardless of HOW we do it, we have to power the impedance converter at the capsule (the JFET) AND any support electronics that might be needed to make that work. P48 has an practical upper limit of just 2 - 4mA before the voltage drops to silly levels as a result of the way the power is sent down the cable.

P24 uses much smaller current limiting resistors meaning we can get more power at the to that electronics. And the real shocker is that every mic I've managed to find schematic for, uses a 12V Zener regulator - most of that voltage is being thrown away, esp. when the mic is using an electret capsule which doesn't require that biasing voltage! 

24 volt "phantom" isn't a perfect solution either. For short runs of cable, it's debatable if we even get any benefit from a balanced cable at all and in a custom setup like this. In fact, we could just say, "who cares" and use our own separate power and signal lines. This eases design considerably but it does limit the distance from capsule to pre-amp to less than a metre or two. Balanced lines by their nature squash most of the external interference from low-frequency sources like mains - which is the biggest problem for stage and studio mics. Even in a family home, the interference from the mains is significant if we don't employ some method to keep it out. 

The number of design choices we can make on this is staggering when you consider (in real terms) how simple the task is on the face of it.

TL;DR

So current state of play is V2 is ready IF you're content with using a simple capsule from JLI/Primo designed for a low-voltage setup. They do make a few cardioids in this format although I don't know if anyone has tested them against each other. (I guess not.)

V2 is getting a P24 boost circuit bolted on (for no better reason that it's cheaper to get the whole thing made up in one shot rather than lots of little boards) but that WILL be an option. I really will write up the whole thing in far more detail so everyone knows which bits do what and what you can change, why etc. but Matt will hopefully cover most of that in a future video.

Putting P24* on this thing opens up a whole new range of possibilities - which I've narrowed down to three or four candidate circuits based on the following JFETS:

2SK170 (Toshiba) or LSK170a - (Interfet). The other variants will work but not as well.

2N4416 

LSK389a: (a dual-channel, monolithic JFET) for lower-distortion.

JFE2140: Texas Instruments similar to the LSK389.

Several others will work with minor adjustments but I'm trying to keep the cost AND complexity down so the less we have to tweak and change at home the better.

I haven't stopped there either. For those who don't mind having a big chunky box (vs. the Steampunk of the original, which I personally love) we can push into even more exotic territory like unbiased, dual diaphragm capsules. These aren't part of the DIY Perks "official" canon but the Michelle will work with pretty much any balanced input you throw at her (just remove the power source resistors and you're off).

*P24 is almost ready but the switched-mode chip generates a little bit too much noise due to the low-loading, so I'm running some tests to see how much active smoothing is needed to tame that. This is the focus of my current work. I expect to have that done early in the new year. I'm putting routing the designs on a single test board to see which (or what combination) one works out best.

@marcdraco first of all thanks very much for your reply and i see that youve replied to many other people and helped them with their issues so thanks very much for that. im not oging to lie to you, i feel like i may have thrown myself a little in the deep end here because i dont understand half of the terminology you used 🙃 but from what i understood the difference between the original and the new alpha version one is:

a different board (the alpha one) which i suppose requires most of the components from the original and maybe a few more

a digitiser? im not going to lie i dont know what this is but ill do my research and try to figure it out

but from what ive seen the rest of the procvess should be the same, as in soldering the transistor (2n4416) to the capsule and connecting that all onto standard audio cable, but the only difference is the circuitry in the middle. also im very sorry but i still havent quite understood how the whole usb interface thing will be changed.

once again im very sorry for repeatedly bothering you, dont feel compelled to respond too quickly, and send love to your cat please!

thank you very much

p.s an alpha board from santa would be great, but i feel the need to pay for at least the postage

since i have line in on my pc i can just use that i think? also just to clarify the digitiser from the original video is the amazon usb dongle thing isnt it?

also, i feel like i should maybe give a parts list s you know what im working with. I currently have the jli 2555 (form the original video) connected too a 2n4416, connected to a custom lightweight cable connected to normal audio cable. i stared the preeamp so i have all he parts form the original preamp circuit from the video. finally i have the excap digitiser thing (which i dont think ill need now though)

You're very welcome to an Alpha, consider it (if you'll excuse the programmer's pun) a Beta tester, so you don't pay anything. You're helping the community out by trialling it.

Walter sends his best wishes for Christmas too. He's currently trying to play catch the laser pointer while I try to reply.

Alpha works (we've seen @Muzzamill get his working with almost no help from me) but some things on there aren't as good as they should be - mostly that I dropped the 5V regulator so the actual adjustment for the output voltage can go higher than it should. This isn't a major issue as it's intended as a trimmed adjustment that you set up before you connect up to a digitiser.

And yes, the digitiser is the bit from Amazon in the long black block that Matt opened up. A digitiser converts the audio signal coming from the pre-amp (which is what Matt designed and I've added to here) and converts it into a digital stream over USB that your computer converts into ... WAV, MP3 what have you. Or is just broadcast over your feed if you're doing a video for example.

Any modern PC likely has at the very least a microphone input and most will have some sort of Line input. The difference is crucial though - the mic input is designed for very low-level signals of perhaps 100 thousandths of a volt. The Line input expects much larger signals.

The problem I found with the original is that the THAT1512 (the chip that does most of the heavy lifting) is far too powerful with a +15/-15V supply. Internally, most digitisers work at 5V or less, perhaps as low as 1,8V. You don't need to understand semiconductor theory to understand that if you put too much voltage into a circuit, something is gonna break: permanently.

A good digitiser (one that the designers thought ahead should have some protection on the inputs (diodes) but most don't. Alpha adds both protection diodes and a adjustable hard limiter that will stop the THAT from driving the digitiser with too much voltage swing. It won't stop you from "over volting" the the inputs but it will help provided it's trimmed out first.

There's an position on the board for a trimmer which is a little variable resistor with a screw terminal for fine adjustment. When you're powered up, you take a voltage reading from the pad near the NMA0515 and adjust that to somewhere around 3,3 to 5V. It's a crude solution but it works.

 Understanding how this works is a little difficult, but it relies on the properties of diodes - if the positive end of the diode (the anode - the part of the arrow) goes above around two tenths of a volt more than the voltage at the cathode (the bar) the diode becomes (in effect) a short circuit and conducts current away from the point. The same is true of the lower diode (but in reverse) so that's a little more tricky. The other half of the protection circuit (not shown here) pushes the centre point of the signal up away from the ground reference so you can get a full swing as expected by the receiving device: your computer or a digitiser.

This is really a hack although it's a legitimate one. THAT1512s are highly regarded in professional audio where they are found in very high-end mixing desks but in this case, the designer has integrated everything as a coherent whole. Downside of this is it means a LOT more work on things like the power which is almost certainly going to be coming from a mains operated power unit.

Also, don't worry about not knowing stuff - everyone is a blank canvas. Google and AI make it a bit easier these days. I learned from books, tutors and blowing a lot of stuff up - at one point in my tender years, something went with such a bang it nearly took me with it. That taught me to respect mains electricity and stay the hell out of old television sets as next time I could have been a smoking pile of ash on the floor. Which would save on the cremation  so there's that. 😉

When I got to do training this stayed with me and I'm extremely careful to keep things safe when I design for community projects. Blowing something up is bad, injuring or killing someone is far, far worse. To get a suggestions on how to stay safe and still enjoy the hobby, Big Clive and Electroboom are both excellent channels where experts demonstrate and discuss this stuff in more detail. Electroboom even shocks himself deliberately just to show in some cases! Both men are as obsessive as I am when it comes to electrical safety though so they're worth a look - both are often hilarious too.

I'm trying to finish the book I'm writing on electronics which will be ready middle of next year detailing some simple design procedures without most of the horrible mathematical proofs to get you started in the hobby without needing to go into the weird details of mathematical proofs (as you'll see on Wikipedia).

V2

So the V2 (after a couple of years of faff and failed experiments that I really should have tested at home rather than make up PCBs but I love routing, it's quite relaxing) I gave up on the THAT as too much trouble when combined with a simple power supply and 5V from the USB.

The adaptor Matt used is a simple way to convert the USB-A plug to a USB-C socket. It doesn't have to be USB-C in fact, USB- micro B socket would be just as good and is very slightly cheaper. The good thing about USB and why I kept it is that it's a modern standard and you can plug stuff in either way up. No more of this "is it in the right way round... oh no... flip it over ... darn it still won't go in..." rigmarole that comes with USB-A.

The other thing that V2 does is to throw away all messy soldering with adaptors, chopping wires off and resoldering them (which can be tricky) and combines everything into one fairly painless board that requires minimal soldering.

It has all the features of the original and is (mostly) compatible with the original capsule design although it will likely work better when I've sorted out the "phantom" power. Recall that's how we get power to the condenser capsule, or, in this case to the 2N4416 that's fixed to your capsule.

Now there's A LOT of jargon in here, hence I'm writing a book to cover it with instructions that hopefully anyone with sufficient interest can understand. Since that's a little way away, don't hesitate to ask if you don't know.

<rant mode>

You may feel stupid but NO ONE here will laugh at you - because that's an instant ban if they do. We all have to learn and the best way to learn is to ask if you don't understand and if you still don't get it, ask again. I only bark at stupid people who refuse to learn because they think it's beneath them - I even co-wrote a book about Vani Hari, an undeniably attractive (and wealthy) American food blogger for this very reason. Hari, who styled herself as The Food Babe, spread all sorts of poor quality and in some cases dangerous information about the American food supply. I think she meant well but being well meaning can be more dangerous than just outright lying. Over time she morphed herself into a disinformation machine, destroying (for example) Subway's sandwiches for using a raising agent that she couldn't pronounce and telling people that all manner of perfectly natural and safe compounds could cause cancer. This sort of "buy my product because the alternative is a slow death from cancer" was the last straw and the book grew out of that. Much of this came from the simple problem that people like her (many of us in fact) don't understand the difference between hazard and risk. To simplify, an alligator is hazard but your risk of being bitten by one on the beach at Scarborough is effectively zero. Part of the WHO (yeah, the World Health Organisation) operates in France and is called the IARC. It produces what are called Monograms which are papers discussing the HAZARD of exposure to things from radioactive materials to simple sugars. Each one gets a hazard rating from 1 (does cause cancer) to 5 - can't cause cancer.

Now pretty much everyone knows that exposure to ionising radiation is a serious cancer hazard - note hazard. But your actual risk of developing cancer is determined by the dose of radiation you receive, how often and where. So it's nowhere near as simple as saying radiation = cancer, but that's the equation that Hari draws.

We all know smoking is bad but most of us (particularly old farts like me) knows someone who smoked a lot of cigarettes and lived to a ripe old age. Risk is a very, very difficult thing to quantify and this detail matters. As I said, I co-wrote a 300+ page book on this so you can imagine how strongly I feel about this stuff. Science and facts (provable facts, not anecdotes) matter. RFK Jnr. and DOGE in the USA has ensured the deaths of thousands of people in the developing world and now measles, pertussis and other diseases are running rampant in across the US. This sort of anti-medicine conspiracy has proliferated on the Internet with people like Dr. John Campbell (who is a Brit) giving medical "advice" on vaccines even though his doctorate is in TEACHING. He's actually a nurse, so lacks the expertise to back up his claims. But I digress...

</rant mode>

thank you so much for your help! im currently in school and trying to get into some electronics as a hobbyist so i apprecate the support! regarding the build though, is it fine if i just clarify a few things:

1. does the alpha just completely replace the preamp and, if so, does that mean i need to buy a few more electronic components from bitsbox such as the diode?

2. im sorry but i didnt really understand, will i be able to bypass the digitiser and just uise line in?

3. will it take a lot of troubleshooting and adjustment (usually im fine with this as i am a bit of a pc enthusiast but im not sure how it would go in a topic which i havent ever stepped foot in before)

once again thanks for the support and help 😊 

Hey, don't sweat it - I know I keep saying this but we ALL start out as a blank canvas. Some end up as a Van Gough, others a Picasso and some Tracy Emin... 😉

1. Alpha contains the whole circuit from Matt's original with the extra protection and noise reduction (which might or might not help). 

2. You can connect the output from this to either a Mic input or a Line In but it's important to have a meter - Aldi usually have a few in for about a tenner and they're not bad little meters. The meter will allow you to set the correct level so you don't blow something up but it's also giving you enough output signal for the system to be able to "hear" it. Mic inputs are far more sensitive and they also have a small biasing voltage so you have to deal with that potentially.

3. It's actually quite simple. Matt's design just works out of the box (so long as you have everything fitted right), mine takes a little bit of setting up and needs a couple of extra bits (if you know what you're doing, you can bypass those too). I can help you do this if you just want to convert it back into something more like the original.

The only important things are to make sure you put your THAT in the right way up. Pin 1 is top left, near the little notch. It might also have a divet up there - but ask one of us if you're not sure. I'll have a rummage in spares and see if can get you some of the other optional parts to save you having to get everything. Some of the capacitors from Matt's BOM (bill of materials - shopping list if you will) might not fit because I'm cheap. LOL.

There's also a couple of missing bits on the silkscreen but we can sort those little bits out when you have your board.

Michelle is a superior design (not surprising after all the development it's had!) but to use that you'd have to lay out a fair chunk of change as they really have to be assembled in China (or any board house that offers the service).  I aimed for and got pretty close to the cost of the original but I overshot a little. This was unavoidable if we were to get to the sort of advanced board I was aiming for. Even now, Michelle is "less than perfect" primarily because I kept it as simple as practical. I know some of the more experienced lads will kick up at some of my design choices but they were deliberate in order to keep the cost DOWN. I didn't skimp on the INAs (the bit that replaces the THAT in this application) and they're quite expensive so most boards will only need ONE. The CMI108B replaces the earlier CM108 and that could have been slightly better but that would mean extra parts since it requires a timing crystal. For amateur (and most uses actually) the CM108B is good enough. We're just stuck because there's a dearth of parts that convert I2S into a USB-capable signal. Perhaps the big lads don't want us all doing what I'm doing here. 😉

In case you're confused, the difference is this:

The CM108 requires an external quartz crystal oscillator running at 12.288MHz but the CM108B internally uses a "phase locked loop" to derive a 12.288MHz timing signal from the USB's clock. 

On the plus side this reduces the cost of the BOM (which matters in mass production). The downside is that the PLL can only track the USB clock and if that wobbles a bit, then the timing goes off a little. I remain in some doubt this will be audible for a 16-bit sample at 48KHz (which is better than CD quality) but there will be people who kick up. If I can get the digital side working with the right firmware (which exists but it's not mature) then I can put a better crystal on the board and push this thing to 96KHz per channel at 32 bit. This is a potential later development that someone else will hopefully do. My next job after finishing this version is the book! There I can give everyone more details on the design process and offer some basic training to get everyone started in electronics.

thank you so much, I've been worrying about how I'm going to finish this project for a while now 😁. fron what I've understood, all I need to do is connect up the electronic components, which are mostly all similar to the ones from the original build with a few taken out and a few extra added, and get a meter (I'm assuming you meant a multimeter, in which case I already have one), set the output using that, connect it to line in or mic in, and it should be good to go. I just had a little question, I was wondering how does the board receive power then? do I connect it via USB as well or is there something ik overlooking? 

also just as a side note I love the work you're doing on the v2 and im looking forward to seeing a subsequent video (hopefully) from Matt about it soon