Not many contact microphones sound fantastic, and I wasn’t really sure why that was. They all look the same but what exactly makes one sound better then the other? Being in the foley and sound design business I wanted mine to sound as full range and flat as possible. While I was at it I thought this might be a good time to document this.

First and foremost, what is a contact microphone and how does it work? It’s not a standard microphone that sends a voltage to an amplifier in response to air pressure. It uses compression caused by vibrations and deformation to a piezoelectric ceramic plate to generate a current. In other words, anything that resonates or deforms the piezo disc can be recorded. For example, speaking very close to it or picking up surface vibrations by making contact, as the name suggests. What it does very well is reject the background ambiance or reverb of a room, allowing you to isolate a sound or instrument without the requirement for a quiet environment.

 Speaking of reverb, contact microphones are also utilized in plate reverbs, which is a type of specialization effect used in music production. These effects are now digitally recreated with software, but back in the day, mix studio basements were filled with massive metal plates manufactured from various materials mounted in large cabinets. On one end, they’d connect a speaker, and on the other, they’d use one or more contact microphones to record the plate’s resulting scattered vibrations.

By reversing the action of a microphone, a piezo disc can also be used to generate sound. Yes, these can be loaded to create a small speaker. Higher voltages are required for audible results, but it is possible. Their main application is as buzzers for user feedback, therefore they are often tuned to a resonant frequency to buzz at a certain tone and require you to use a matching AC current to get the loudest possible result. 

Typically, microphones are made of a thin plastic sheet that vibrates like a piece of paper, causing a coil and magnetic element to generate an AC voltage, which means alternating current, up and down, like an audio signal. It works similar to a speaker but in reverse. Air pressure microphones are implemented in various applications and are recorded by many types of devices so usually they are made fairly easy to drive and produce mic level signals  between -30 and – 60 db  at a voltage roughly at 2.5 millivolts.

Contact microphones on the other hand are not really designed for recording high end audio. They are more used as detectors or buzzers so they require some more attention if you want a usable sound. The main difference is that they are not easy to drive, don’t have a good natural flat frequency response, produce a highly dynamic exponential signal as to the kinetic source and have a much higher impedance (Z) level. 

Impedance

A term that causes a lot of confusion amongst buyers of electronics. Thinking that greater spec values always produce better results. Not exactly! Let’s look at the definition first.

High impedance is typically an undesirable feature in the commercial microphone market because it introduces some inherent problems. The only time manufactures are happy to use that spec is when it needs to be low cost! Yes, those piezo discs cost only a few cents to manufacture. What’s the reason? High impedance (Hi-Z) mics produce louder signals and thus require little to no extra parts to reach line level.

 So what are the drawbacks? Using a Hi-Z signal in a cable is more prone to noticeable RF interference. Doing so basically creates a line that will act like an antenna, so not the best option.

 Running a long cable with Hi-Z (>1000 ohm) can cause a natural low pass effect (fewer high frequencies) due to the capacitance of the cable interacting with the impedance. It is not surprising then, that a lo-Z output is the industry standard when using professional microphones. Signals with Hi Z almost always require a DI box to convert the to lo-Z (mic level) in order to travel longer distances (ie. guitars) to avoid this effect. Think of your typical low end consumer karaoke microphone and amp. I’m sure you can’t remember it for having a great sound with rich low end.

 Not to say that hi-Z is inherently bad audio in any way, the contrary is actually true as they record at a lower noise floor ratio. Just note that lo-Z mics  are more flexible to use at any length  of application and are less likely to overload the amp. 

 The reverse is true for speakers and mic preamps, this is where some confusion arises when you look up these concepts. The two look to do the same but are built in reverse and thus have opposite impedances. Mic amps use hi-Z, headphone and speaker amps lo-Z. 

Impedance: The electrical deep dive

Okay, so for those who are still unsure about what impedance is, I decided to dive a little deeper and clarify what it actually does; feel free to skip this part. 

In an audio circuit, the voltage (also known as electrical pressure, potential energy) typically holds the actual audio signal. The more of it there is, the louder the signal can be and the greater the ability to drive ie. a speaker. The current (also known as flow, or amp) is caused by the polarity of the voltage. It represents the rate at which electrons pass through a given point. With audio both voltage and current are variable and depend on what the audio signal does. 

Without enough current the voltage can not deliver its full potential and the more the resistance the lower the current for the same voltage.

To visualize how resistance works, think of a water hose that has some obstruction in it. On one side the pressure and flow is high due to a water tank pushing the water, on the other side the water will be weak and slow. The circuit will eventually take longer to completely drain the stored electrons. As a result of the conversion some of the voltage in the resistance will partly turn into heat. 

Current  = Voltage / resistance

2 Amps = 6V / 2 Ohm

6 Amps = 6V / 1 Ohm

Lo-Z type speakers will cause the audio to be louder as the coil is lighter but typically less accurate. Hi-Z speakers will produce a more quiet signal as the element is more resistant to move, it requires more power (current and voltage) to get it moving. What differentiates a lo- or Hi- Z element is typically the amount of coil wire used to create the electromagnet that moves the speaker’s membrane. The stronger the magnet the quicker and more precisely it is able to place the membrane at its position according to the signal. Typically that should produce a more defined low end as a weaker coil struggles to keep up with the electrical signal and has a harder time fighting the physical air pressure. On an electrical level, the more coil is used the more the impedance will be and thus more power is required. 

 Impedance also behaves differently at different frequencies. Yes! If you sweep a sinus through a microphone or record some pink noise, you’ll get a very bumpy curve that corresponds to the efficiency of how the coil reacts to different electrical frequencies. Manufacturers must frequently compensate for this with an acoustical design or some form of corrective eq, as this can significantly color the sound.

So back to our piezo, we now understand that we can change the behavior of the mic by supplying different impedances, essentially drawing currents from the mic and thereby getting different results in pickup frequencies. This process of supplying enough voltage and current is -once again- called impedance matching and is done within the audio industry to good taste and often personal preference.  

Impedance: Matching the load

In order to fully take the microphone’s impedance load and not introduce coloration due to a lack of current draw, the pre amp input impedance should be at least matched. Typically one has an amp that is actually 4 to 10 times the impedance of the microphone; this is just a safety feature and is not better in any way. This is just to make sure that when you run parallel or serial setups with multiple devices the amp is able to keep up. 

So why is there a signal drop if you don’t use an amp with a high enough impedance? There’s the ‘load,’ which is transferred from the microphone to the amp and forms a circuit like we saw earlier. When current flows from hi-Z (high resistance, low current) to lo-Z (low resistance, high current), electrons are abruptly released from a small hose into a large hose, to keep the water analogy. The current has insufficient electrons passing through to maintain the same pressure. It’s the same reason why your home’s water supply never transitions into larger diameters, as this could cause air pockets and pressure loss. Current always has to get ‘slower’ or remain the same due to the resistance to make sure there is enough at the next device. Doing the opposite would drain the current faster than the source is able to provide, plummeting the voltage and signal. Mismatching impedance or the ‘Loading down’ of your signal can be a creative choice into altering your sound but it introduces things like reflections (echo’s) where the signal passes back in the opposite direction, summing the signal with standing waves and altering the sound in an unpredictable way. This is more noticeable when you have long cables, like with a few kilometers, so nothing big to worry about in our use case!

Variable Impedance control

Now that we have all this theory down, it’s time to get creative! Introducing the variable Impedance ‘mic activator’, a device that works like a tonal shaper. The greater the impedance, the more body or beef you can achieve, and vice versa. As previously stated, impedance operates at a frequency level and is not linear to the output. As a result, changing it can significantly alter the shape of the frequency response in a unique way. A mic activator or ‘lifter,’ such as the cloudlifter Z is a popular device for accomplishing this. Simply turning a knob changes the power balance between the device and preamp and does this by adding parallel impedance to the circuit while also optionally amplifying the signal. This results in increasing the low end, giving more bass.

 Balanced cable

Now is the time to introduce the next pillar in this project, the balanced cable also known or referred to as an  ‘External Line Return’ or XLR in short!

In any case, having a balanced cable is always better for all of your audio applications because the design of the cable produces very little to no extra noise; let me clarify with our use case.

When using our contact microphone, we should ideally convert the result to lo-Z with a DI box before running the signal through a cable. That is a good setup but without a balanced cable, our signal is still likely to be affected by all types of RF humming caused by casting devices like Wifi or cell phone networks, your local radio station close by, etc. These are almost everywhere these days so the chance you encounter them is quite high, just think of your speakers producing that annoying ticking sound when your phone has an incoming connection. If you care anything about your signal, a balanced solution is primordial, and so is the quality of the cable! I would say the cable is more important than the DI box converting the signal. With a balanced cable you can run any signal through a longer cable and be relatively sure that there is nothing that will affect it, even higher impedence signals. Why? The XLR has a simple but amazing design, and I’m not talking about cosmetics.

The XLR works by sending two identical signals from the same source through the cable on a seperate line, both identical but phase inverted (the up signal is then pointing down and vice versa). If interference hits the cable it is likely to cause the same noise on both lines in the same manner and phase. 

The magic of the cable happens when the phase is flipped back by the preamp or recorder  to align both original signals in phase, essentially creating a double in amplitude. When this happens, the interference is canceled out because they are now out of phase and both signals are summed together! Ah, so elegant and pure physics! This is also why many balanced cables are usually twirled quite hard so that the interference hits the cable lines in a uniform way. 

So, can we conclude the theory now? Not yet! This is where things get exciting. We can actually take the concept from the balanced cable a step further. Recording the audio signals with two piezo elements! Why? If you had paid attention you would have noticed nothing was in place to flip the phase of our piezo at the source, you would need some circuit component to do that, or not?

 As you recall, the cable’s design cancels out the difference in signal. When we record with two elements, all of the self noise produced by each piezo is only half the total amplitude and possibly not in phase with the other, so we will get a thicker bed of noise but not as loud unless it happens to be totally in phase. As for electromagnetic interference, it will be canceled out, allowing the common signal from both piezo discs to shine like a diamond!

 As long as you correctly apply vibrations to both elements and face both piezo’s opposite to one another by physically flipping them 180 degrees (out of phase), your signal will come through perfectly.

Building the contact microphone

Ok let’s get to work. This is where we discuss the design of your microphone and the materials you’ll use as bonding components. The microphone should be able to vibrate as a single unit, but not be so rigid that the piezo cannot deform. That’s the difficult part!

In the audible range, a round disk also has more resonance frequencies. As we saw most of them are even tuned to a frequency. As a result, cutting the brass in a non-uniform manner, such as a triangle, would aid in shifting those out of the spectrum. Please note that if your triangle shape is off, the frequency spectrum will change slightly and part of the spectrum will get lost by the cancellation so make sure they are identical and make some extra to test for a fuller sound.

The addition of a second chemical component causes the resin to harden. The hardness is pretty high depending on how you mix the components, usually still a little flexible, so it does not really break off quickly.
This is the point where you can let your creativity flow on how to seal it. I tried putting it in a pocket of duct tape, but my pocket wasn’t as watertight on every mic, so my cardboard became a big mess after a while. Yes, resin is as liquid as water at first, so make sure it’s tightly sealed, multiple times. Also prepare more then one mic as you don’t need much epoxy for one, or else it will be a waste. 

We don’t want both contact mics to touch each other’s ceramics, but we do want them to share vibrations. Both bonding chemicals are non conductive so it should work. But for safety I wrap both ceramics in a liquid electrical ceiler as I will be pressing them together. 

Because the voltage is generated by the deformation of the ceramic, I wouldn’t always recommend totally gluing them shut unless you expect to record really loud vibrations. When both ceramics are glued together, the flexibility of the piezo is reduced, picking up less signal. So, in the end, I made 2 prototypes, I kept 1 hollow in the center and one completely filled with epoxy.

After doing some test recordings the epoxy ones were less responsive, had less bass, but proved super useful on very loud sources. Their property kept signals under control in a 24 bit clippable environment. Contact mic recordings are very dynamic because you are touching the element all of the time, especially if you are dealing with metal surfaces. Remember that they produce an exponential signal compared to the kinetic source. What does that mean? Well, let’s say you hit your microphone with a pencil twice as hard as before, you will find that the volume meter will peak 4 times as high, not twice! So prepare your ears with some cushions as things might give you ear ringing after a while.  

Testing the contact microphones

For my first test I had planned to record some kitchen knives. I had made 4 contact microphones, one hollow pair with a round form, and one triangular pair filled with epoxy. Since I had some mics lying around I decided to add 2 more small condenser microphones to compare them to.

Having two of each identical I could record in stereo. I stuck the epoxy filled mics on my knife using some blu tack gum. They were less responsive and thus more reluctant to handle noise with the cables. I configured them in a mid/side stereo configuration as there is no spread possible. The hollow microphones I stuck to my countertop as room pickups, These I put far apart and configured them in a left and right stereo configuration.Then as a reference I put up two Oktava MK 012 super cardioid microphones as room ambience and reference signal.

Yes that is a lot, and you would think that riding the gain on this would be quite the task. You are right if I would have used a common 24 bit environment for recording. Remember the exponential signal to source ratio we mentioned. Just slightly touching your knife could cause your signal to clip the recorder! 24 bit is a relatively medium sized digital ‘band’ to record on, just fine for most applications as most audio for CD (is this even still a standard? I haven’t seen one in years) is even 16 bit. The bit depth for 24 is 16,777,216 levels of sound, plenty you would think. If we translate this into signal amplitude we would end up with 144.5 dB of dynamic range. To put this into context the maximum physical sound pressure at sea level is 210 db. So what does this mean for our use case? Well 24bit is hard to manage and would require much gain riding, recordings rejected for lack- or the clipping of signal.

The last pieces of the puzzle

Now enters 32 bit float. Not to turn this into a bit depth math class, but simply remember that this encoding stores the signal in a different way (obviously larger on disk) where the dynamic range is so large you can describe a whopping 1500 dB, more than you ever need, enough to record the ending of the world. You may be unaware but internally your DAW works most likely at 32 bit. Although standard on a computer, having this feature in a field recorder is quite uncommon and for most applications not considered for the amount of data that it stores. Luckily for us some are being made that are not costing us an arm and leg, like the Zoom F6 field recorder.  

Not describing audio in levels but in decimals, 32 bit float has no limit set to what the maximum amplitude is. It might look red on the meter, but the file itself is perfectly described. Well, a small side note that the accuracy decreases the more you go up above the suggested limit as more rounding errors occur, so it is still good practice to stay within the suggested range.  

With this format you don’t have to gain stage no more, you can focus on the actual mic handling and not worry about the recorder as you can always  pull back the audio level in your DAW or Audio editor before converting to 24 or 16 bit, which is still the target bit depth for consumer audio. 

I set up my recorder in poly wav, a format that is not supported by all DAW’s but by most audio editors. It is a format that has multiple tracks and keeps things nicely together in one file. 

Having frequencies well above hearing level doesn’t seem very interesting if you keep them there, but when pitching a sample down by an octave (12 notes) you are essentially shifting the frequency spectrum half way down. An octave up  is nothing more than doubling the speed. So when we turn a squeaky knife sound down an octave we end up with an immense metal object that still has some top-end up and over 10 khz, sounding natural and not as muffled. When I look at my recording file I see a signal up to 100khz and more. That would give me a 3 octave range to end up at 12.5 khz which I consider the minimum to get a full range audio spectrum with some high frequency action, almost the frequency range of a mp3 file (16khz) which most people still experience as ‘full range’. To compare with 44.1 khz, that is just 1 oktaves to end up with 11 khz. 

Remember that as a rule of thumb to capture full fidelity audio one should always capture at double the bandwidth of the highest frequency one would want to capture, hence the 44.1khz. That is the smallest rate (without going into details on the actual odd number) for the 22 khz of the human hearing. So if you want a true 96khz signal (studio quality) according to the Nyquist-Shannon theorem one should record at 196khz; just to say that 196khz is actually not as crazy high as one would think. 

– Tristan Morelle.

© 2022 Omnion Media

Thank you for reading my endeavors, and excuse me if I made a mistake somewhere. Again, I’m not an engineer or electronics guy, just a sound designer who wanted his mic to sound as good as possible.

Please leave a comment if you enjoyed the article.