In a previous post I discussed some of the common problems with DIY contact mics. I showed how to get better bass response. In this article, I will address one of the other main problems — hum.
The issue is that DIY contact mics tend to act as little antenneas that pick up electromagnetic radiation that is emitted by other electrical devices and by AC power generally, and which permeates virtually all of space. DIY contact microphones pick up this sound and convert it into audio, which can be heard as a humming or buzzing sound in the sound. The fundamental frequency of this hum is typically 50 or 60 Hz, depending on the AC power in the country where you live. However, it also typically has overtones at integer multiples of the fundamental, which makes it very difficult to filter out.
The solution is that every part of the microphone, including the sensor, the preamp, and all of the cables, need to be shielded. This means that everything needs to be completely enclosed in metal, and that metal needs to be electrically connected to the circuit ground. This will block out that pesky unwanted hum so that you can have nice clean hum-free audio. In this post I will show you the effect of shlding each part of the circuit.
I built a DIY contact mic using the Marshmallow DIY Kit. I started with everything completely unshielded, just out in the open, susceptible to hum. I plugged the output of the microphone into the input of my audio recording interface, and I plugged that into my computer. I recorded about a second of (ostensibly) silence, and then calculated the spectrum of the result. I repeated this 20 times and averaged the spectra all together. The result is shown in Figure 1.
Figure 1: A spectrogram showing what silnce sounds like when a DIY contact mic is unshielded. Note that this is not at all silent and is humming loudly ay 60 Hz.
In this plot, there is a huge spike at 60 Hz, and at each integer multiple of 60 Hz. This is the famous hum, and can be heard at 1:10 in the video above. It is quite loud and makes it impossible to get clean recordings of anything other than hum!
For the first step in trying to reduct the hum, I put the preamp but not the piezo disc down into a tin box. I soldered a wire onto the box, and another onto the lid, and connected those to the preamp ground. I repeated the experiment and the result is in Figure 2.
Figure 2: The new blue line depicts what silence sounds like when the preamp and nothing else is shielded.
There is still a huge spike at 60 Hz, and consequently a massive amount of hummung, but notice that the peak is around 6 dB lower, so that is at least an imporvement.
The next improvement was to encase the disc in aluminum foil. The brass part of the disc is already grounded, so the aluminum foil will be grounded too when it touches the brass. However, the inner crystaline part of the disc should not touch the foil, as that creates a short circuit. To prevent that, I covered the crystaline part of the disc in tape. Then I repeated the experiemnt and the result is in Figure 3.
Figure 3: The new green plot represents silence when the piezo disc is also properly shielded.
Again, there is still a lot of hum, but covering the disc in foil reduced the 60 Hz peak by another 7 dB. This is another good step in the right direction.
Up to this point I have been powering my circuit with a cheap wall adapter. It is very noisy and is the source of much of the humming. The Marshmallow DIY Preamp that I am using does have built-in filtering to try to reduce the power supply noise. However, filtering can only do so much. So for the next improvement I replaced the wall adapter with a lithium battery which has a very stable and quiet output. The result is in Figure 4.
Figure 4: The new red plot represents silence when the wall adapter has been replaced by a battery.
This resulted in a 13 dB reduction in the hum at 60 Hz, and perhaps more at the integer multiples of 60 Hz. This represents a significant improvement.
Up until this point, the piezo has been outside of the tin box, and connected to the preamp by alligator clip wires. These wires are not shielded and are also picking up some electromagnetic radiation, contributing to the hum. Normally one would use audio cable for this, which is already shielded. to simulate shielded cable, I put the whole disc and wires into the tin box with the preamp so that everything is completely shielded. The result is in Figure 5.
Figure 5: The new faint acqua-colred line represents silence when the disc and its cables are in the tin box with the preamp.
This plot can be seen by itself for clarity in Figure 6.
Figure 6: This is the same as Figure 5 but with all of the previous plots removed for clarity.
This gave the most drastic improvement yet, it reduced the hum by another 15 dB. This really highlights the importance of shielding every little thing and not leaving anything out in the open. The remaining 60 Hz spike is virtually inaudible and the result is a very quiet hum-free sound which can be heard at 3:20 in the video. The remaining (inaudible) hum is likely a real mechanical signal, not electromagnetic radiation. This is because my desk slightly vibrates at 60 Hz due to my refrigerator, HVAC system, and whatever else there is in my house that has motors that spin and vibrate at that frequency. You can never really get away from it totally.
To get a baseline, I disconnected my mic from my recording interface and reforded silence with the mic totally disconected. This tells me how much of the hum is due to the interface as opposed to the microphone. The result is in Figure 7.
Figure 7:The yellow line represents silence when the mic is completely disconnected from the audio interface. This is the noise floor of the recording interface.
Even here there is a tiny amount of hum. Note that the Acqua line also has a lot of low-frequencies present. This is most likely due to traffic driving by and other environmental sounds that are picked up very faintly by the microphone. Notice also that in gerneral the preamp is very quiet and does not add any measurable amount of noise to the exsisting the noise floor of the recording interface, at least not at unity gain.
Conclusion
Following these four steps reduced the hum by a total of 41 dB. This means that at the end, the 60 hz component had less that 1 percent of the amplitude than it did originally. This reduced the hum from being loud and audible to being completely inaudible.
