Today we’re going to talk about signal to noise ratio when cutting plastic with a diamond stylus, because it comes up a lot but people don’t always know what to expect or how to check it.
SNR is the difference between your actual audio signal and the noise floor, for example hiss, scraping, rumble, and general surface noise that could be audible on a cut. A common misconception is that lathe cuts always have audible noise, like it’s part of the sound. We hear this a lot, especially from people who contact us wanting to get into cutting, stating that they have read things like this online. If you’re cutting properly with good calibration, the noise floor can be low enough that it’s not really audible at normal listening levels.
Below we’ve listed various noise issues to look out for and a basic idea on what causes them:
Stylus condition is one of the biggest factors. A worn or damaged stylus is often the source of a scraping, rough or hissy sound, that people hear and assume is just how lathe cuts sound. A stylus in good condition, cutting clean walls is crucial for your noise floor, and it’s also one of the easiest things to overlook because the wear can be gradual and hard to spot without proper magnification. Always ensure calibration is spot on, as not only stylus condition, but incorrect groove depth can also play a factor in this and also lead to stylus damage.
Ground loops are probably the most common source of electrical noise. This happens when multiple pieces of equipment in your cutting chain are earthed at different points, creating a loop that picks up hum at 50Hz or 60Hz plus harmonics. A persistent hum points to a ground loop and is easily identified as it’s an actual hum, rather than rumble.
Static is something that can cause clicks and pops. This is less common, but it’s worth noting that if you play your cut back and hear clicks and pops, try playing it back later or in another room, sometimes static buildup isn’t actually on the cut itself and is just present on playback, so always double check. Another static issue is swarf dropping and sticking to the blank, if it gets caught in a groove while getting cut, it will cause noise.
Blanks also play a part, material and flatness being the issue. If a blank isn’t flat, you get inconsistent groove depth as the stylus rides over the warp. Combined with vertical modulation, this can cause the groove to go shallow in some places and too deep in others. This likely shows up as noise that comes and goes, audible with the rotation of the disc rather than being consistent throughout. The material also plays a significant role, if having noise issues on one disc type, it is worth testing on another to rule that out.
Rumble often shows up in the very low end, quite often caused by a vacuum system, more rarely something a stylus will cause, but never say never. Always check your vacuum is decoupled where possible and not touching the lathe or its table. Rumble could also be present from platter bearings, but this is less likely, though worth noting.
To check your SNR, always cut silence and audio, just cutting an unmodulated (silent) groove can make things seem louder or quieter than they are in reality. Cut a good few seconds of silence and then audio at your normal cutting level. 0 VU is always a good place to check for calibration issues like this. The quieter you cut, the closer you are to the noise floor. We advise getting it inaudible when cutting at 0 VU, as when cutting louder it will be dead silent.
When everything’s dialled in properly, you shouldn’t be hearing significant noise on a cut. Don’t rely on meters to check this, always go from what you can hear.
Below is a real world example of a cut with an acceptable signal to noise ratio. You can hear the diamond noise for a few seconds before the music starts, this is from a slightly worn stylus, but it remains acceptable since it doesn't interfere with the music itself. We'd say this is a good benchmark to aim below. Also, for anyone wanting to diagnose diamond noises, we have a blog post HERE with a few examples.