This is the first post in a new series on this blog called "What's that horrifying thing on your desk" Where I explain horrifying things that are on or near my desk. Our first eldritch companion is this hand built audio attenuator. I promise that I can explain how it is and why it is, but before we get to that, we have to explain how does computer make sound.
In general, a computer has some idea for what sound it would like to make. That idea can be as simple as BEEP! if the output is a 1980s style PC SPEAKER, or as complex as you want to get with modern surround sound audio setup. My specific computer sends its ideas of what sounds it wants me to hear through this USB audio card
The problem is that if I just plug my headphones into my sound card, it's way too loud. I don't know what it is about my ears, but my pain threshold is much lower for loud sounds than others. I just find the defaults that are offered are painfully loud. In the sketch above more electrons means more magnetism which means more movement of the permanent magnet diaphragm which means more sound. One solution to this is to give the electrons something to do that isn't in the headphones at all. That's the idea of an attenuator. It's the sort of thing that you can buy of the shelf and indeed I did, and this is the one that I used for several years.
The problem that I had with the off the shelf attenuator is that my left stereo channel was ever so slightly louder than my right stereo channel. Unfortunately, this sort of thing is really to be expected. The way that attenuators work is that a resistor (wire but uses up electrons) is wound around a doughnut shaped core that's set up so that a metal wiper can touch the top of every wind. You connect the input to the wiper, and the output to one end of the long wound resistor, and then you can use any length of the resistor between the end and what the wiper can touch to use up the energy of electrons before they make it to the speaker. The off the shelf attenuator that I bought is just two of those doughnut wound resistors and wipers put on the same axis and connected to the same knob. The attenuator needs to have two properties to make it convenient to use for headphone audio. The first is that at full quiet it should use up all the electrons, that is, it should be close to an open circuit. The second is that the resistance shouldn't be even (linear) because the human experience of sound isn't even but exponential. This means that most of the resistance is at the very quiet end of the attenuator. This is also where most of the manufacturing variability is. And this is what drove me to build that dead bug looking thing you saw earlier. When I put the knob for the attenuator where it was comfortable for me to listen to, the only thing keeping the left channel and the right channel at a similar volume was the manufacturer's ability to keep the resistances even at the most difficult point for them to do so. As a consequence I wound up with one channel annoyingly louder than the other.
The two boxy components are just audio attenuators, I've just attached one to each channel so that they can be adjusted independently. The two thin round banded components are fixed resistors, and cheap ones at that. Normally they would be wildly inappropriate for an audio application. Two specific things mitigate these disadvantages. First, I know that they will only ever operate in a climate controlled environment. Their relatively poor thermal stability won't ever be a problem for me. Second, I have a box of dozens of them and a multimeter, I kept trying pairs of them until I found two that were really well matched. This is something I can do, but wouldn't if I were selling them. The main point of the resistors is to soak up some of the attenuation load so that the real attenuators are operating in the center of their range where they are most sensitive. If they're not quite the same, the variable resistors can correct it out.
Unbalanced stereo channels showing risidual voltage on the multimeter.
Tuned atttenuation of the left and right channels to match well enough to be below the noise floor.