DIY Microphone

This is a microphone that I have custom made. And if you’re wondering how good it sounds, well you’re hearing it right now. I’d say that it sounds really good considering that it is homemade. The circuit isn’t even that complex, although there certanly is a lot of room for improvement, as you will see later in this video. Really, the secret to the audio quality all starts with this piece right here, the microphone capsule. So, without further ado, let me show you how I made this.

Let’s start with perhaps the most basic form of a microphone that other YouTube videos will show you. Here is this tiny electret capsule. This electret microphone is a type of condenser microphone, as opposed to dynamic microphones. Electrets are different from other condenser types because they are permanently polarized internally. What does this mean? Well let’s take a closer look at what goes into an electret microphone exactly. Inside each microphone, is a sort of suspended capacitor. When you talk or make noise, the sound waves come and move the plates of this capacitor, thus changing the overall capacitance. We can convert this into electric signals by looking at one of the more basic capacitor equations: Q = C * V (charge equals capacitance multiplied by voltage).

The electret is permanently polarized, meaning that it has a constant charge, Q. That leaves us with a changing capacitance and voltage. When the sound waves push the plates closer together, the capacitance increases. This forces the voltage to decrease in order to keep the charge constant. The inverse is also true. When the plates vibrate away, the capacitance decrease and the voltage increases. And so yes, we will get a voltage representation of our sound waves across the pins of the capsule. But wait, we can’t use the signal yet, because the signal is both very high-impedance and very low voltage. In order to make it more usable, we have to create an impedance converter.

Most of the time, this so called impedance converter is in the form of a JFET. This kind of transistor is useful because of its high-impedance input, which will match the high-impedance output of our capsule. Likewise, the output of the JFET is low-impedance. This makes our output signal much easier to work with. Sometimes, the JFET already comes with the capsule, like in the case of our cheap capsule, so you won’t have to worry about getting one.

What you will have to do, however, is add power to our JFET impedance converter so that it may convert our audio signal. This is often done through a somewhat large resistor, like a 10k for example attached like shown. I will be using 9 volts for this example. Now this will work, and we will get our audio signal from the JFET, but we still have this rather inconvenient DC offset from our 9 volt power source. This can be easily solved by adding a capacitor on the output to filter this DC out. The value of the capacitor depends on how low of a frequency you want to filter out, but I’ll stick with a 10uF capacitor for now.

And as you can see on my oscilloscope, we are able to successfully detect and measure our audio signal. It is still much too weak to use yet. That’s where we can use a microphone pre-amplifier. This specific preamplifier is essentially just an op-amp circuit that will raise our audio signal up to line levels so that the computer’s ADC will be able to read it properly. I will be using this NE5532 audio op-amp for this purpose. I will use a 10k and a 51k resistor in order to get a 5 times amplification. You will also have to add a virtual ground since the op-amp is limited to our single supply. To do that, you can simply add a voltage divider to the non-inverting input. And now the op-amp will amplifiy the audio signal aroung this virtual ground point. And, well, this is how it sounds. So, it’s probably not good enough to be used in my future videos.

We can actually improve upon this idea and improve its performance even further. The first thing that we can do is introduce a higher quality electret capsule. JLI electronics is perhaps one of the better distributors of these capsules. For my project, I’ve selected the JLI-2590A because it has the required JFET built in. You can see it as this FET impedance converter in the schematic. If you want to follow along, but get a microphone without this impedance converter built-in, simply purchase an external JFET and attach it like this diagram shows. The only external components I used for my specific capsule were these 10k and 2k2 resistors, which I followed from the datasheet’s recommendations.

Now, let’s investigate the next stage, which is the preamplifier. This time around, I will be using the TL071 op-amp instead. The special thing about this op-amp is that it also has JFET inputs. This time, I also used the TLE2426 as my voltage divider. This should make the voltage division a little easier and more precise. On the output, I used a 10uF capacitor. And attached it to a single ended audio jack. Before I attached the capsule, I used my function generator to make sure that everything was working… and yes, we can clearly hear both the sine and square waves. After attching the capsule, this is how it sounds.

With the confidence I gained from making this prototype, I decided to start work on a PCB board. This time around, however, I planned to make it fully integrated with XLR. To do that, we need to supply another output and also receive the 48 volt phantom power from the audio interface. Well what’s in the second output? Basically, XLR uses two outputs in order to reduce noise. The outputs are identical except that one is the opposite of the other. Then, when the signals reach the audio interface they are both put into a differential amplifier. This will remove the difference between the two and restore our original audio signal. Why is this important? Well, let’s imagine that there is a significant amount of noise on the line from interference in the air. This interference will affect both outputs the exact same. Then, when the outputs reach the differential amplifier, the interference will cancel itself out, leaving the original audio signal intact.

This XLR setup is rather useful, and luckily, there is an easy way that we can implement it. I upgraded the TL071 to a TL072. It’s the same thing, but it has two op-amps on-board. All I did was feed the original audio signal from the first op-amp into the second. The second op-amp was configured as an inverting amplifier, so it would invert the signal as required by XLR. Make sure to use another decoupling capacitor here as well. All that’s left is to receive the phantom power. Here is how phantom power is wired internally in audio interfaces. The 48 volts in split to both lines through 6k8 resistors.

On our side of the circuit, we can place our own resistors and combine the phantom power again. I used two 1k’s. It’s up to you at this point for regulation, but I opted for a 12 volt zener diode to act as my regulator. Make sure to place capacitors here as well to stablize the voltage. And that should complete the circuit, and after assembling it… it didn’t work as expected. There was some serious issues with audio quality in the recording. After some debugging, I found a few issues with this circuit. First, there is a 32 kHz oscillation in the power supplied from the zener diode. At first I thought that I could fix this with a couple of capacitors, but to no avail. Afterwards, I found another issue. The voltage across the zener was only seven volts. This is likely what is causing my other issues. I believe that this is because I am drawing too much current through the phantom power. And as you know, more current means more voltage drop.

Placing some larger decoupling capacitors along the zener and the virtual ground definetly helped a great deal, but circuit still isn’t perfect. If you are wondering how it sounds at this point, let’s do a comparison. So far, aside from the intro, I’ve been using my normal microphone: The AT2020. And here is what the DIY microphone sounds like. It’s far from professional and the biggest problem is the background noise, but I’d say it’s good for a first attempt. This is what I think may need improvement in the design.

The captured audio sounds good in and of itself, but somewhere in the pre-amp, a significant amount of noise is being injected. I have a few ideas of what the source might be, which I may test for a second version. First, the phantom power isn’t providing enough voltage to keep the zener at the 12 volts which I hoped for. Second, these TL072 op-amps are pretty noisy, so they may be contributing some of the parastic noise. Either way, I’m still happy with the result, since it definetly has really good potential for improvement.

To top off the whole project, I 3d printed this microphone body, so that I can easily hold it, like I did back in the intro. It comes in three pieces: the main body with an xlr attachment at the bottom. The middle piece which attaches onto the main body and also holds the 3rd piece, which is this structure which holds the electronics in place. I would’ve liked to attach a mesh on the outside, but that was unsuccessful at the time of recording this video. Like I said, this whole project has potential for improvement.

That should just about cover it for this microphone. It’s surprisingly simple and affordable if you’d like to give it a try youself. If you’ve enjoyed the video up to this point, please consider subscribing so that you can see my future videos. Also check out my buymeacoffee page to support this channel so that I can keep making videos. Thanks for watching, and see you in the next one!