Listening for Tracking Error
I’ve put this discussion as a side piece to my blogs as it’s rather in-depth. But for me this topic really is essential to understand system performance, and enable better design too. As audiophiles I believe we need to know what tracking error sounds like. And we need to understand the limitations of ‘transparency thinking’ as well as the serious flaws of assessment with frequency response.
We need to turn our thinking away from the frequency domain and into the time domain. But be careful here, I don’t mean musical ‘timing’ - this is is about the real form and progression of a signal as it actually is - a positive or negative value that changes over time.
But let’s start with frequency response. Analysing a signal in terms of it’s power response across a frequency spectrum is a mathematically convenient viewpoint for rudimentary engineering that should only be used to check basic things like gain performance of an amplifier and speaker roll-off. Any more than this and it’s hampered by two significant limitations. First, assessing any component in high fidelity terms with a slow moving single tonal point does not simultaneously load the test component across its operating range as normal music does. And this means that real faults are not exposed. Secondly, designing good audio equipment relying heavily on frequency domain Fast Fourier Transforms (FFT) does not work. There are many different FFT algorithms all of which throw away data and build approximations - in essence they aggregate steady state data. This tends to lead designers to use things like error correction algorithms and high levels of amplifier feedback that will deliver apparently better FFT measurements. Unfortunately this results in designs that suffers from time domain faults that FFT will not reveal. I think people who like things like NOS DACs and simple amps, and don’t like things like MQA intuitively understand this problem.
And what are we thinking with this method when we listen? Sure, we can listen for good bandwidth and frequency response in a system and maybe that helps show up problems like lumpy bass caused by room acoustics. But beware, this then leads to incorrect assumptions, the classic being blaming bad midrange and treble on frequency response faults – and then try to compensate with dull sounding equipment. The biggest schoolboy error in hifi.
So lets come on to transparency thinking. I described a very useful analysis philosophy using transparency in my blog number 8. This is really a great thing to use as we start to learn to assess a system more effectively, and avoid genuine lack of transparency pitfalls. But what does transparency change sound like? Well first, we must remember that when we improve transparency we let more of the bad through as well as the good. So if we change a component for a genuinely more transparent one you should notice more information. It might not be better yet but there should be more than you’ve heard before with a few new things surprising you in your recordings. But here’s where the second worst schoolboy error comes in, one i’ve seen so many people do – they say “yes there is more detail but it also sounds brighter, or has less bass - It’s got things I don’t want, so I’ll stick with my old one thank you”. But the start of real enlightenment is to accept there’s more transparency (more information) and use that now as a tool to find more faults in other parts of the system! As a minimum, adjust the system; reposition speakers, adjust your mains loom and try your mains conditioners in a different configuration, and adjust supports.
So now to tracking error. As you will have read from my blog number 9, tracking error can be equated to the problem of a cartridge not properly tracking the shape of a record grove. When mistracking, the tip of the stylus is not in the position it should be, at that moment in time. It’s not moving as it should be, it’s not following the curves on the surface of the groove faithfully. It’s in contact with the surface of the groove some of the time, but at other times it leaves the surface – and this gets worse the more complex the shape of the groove (the complexity of the recording). Notice the element of time here and that some of the time the stylus is in the correct position and other times it is not. It’s in essence bouncing along the groove and of course as soon as it leaves the surface it’s output can no longer be the output that is defined by the recording.
Every piece of hifi exhibits this same problem. Take an amplifier and consider how it responds to different supports. Improve the arrangement of footers and shelves and it sounds better. Putting aside the discussion about what’s happening there physically, what’s actually happening to the left and right audio output when we get an improvement? Simple, it’s tracking better.
So, we should then start to understand what tracking error faults sound like - and remember from our stylus example, it always gets worse the more complex the music. And we need to keep in mind that we want to increase transparency to find the tracking errors, then reduce the causes of tracking error in order to significantly improve performance. Here are the main examples:
- An inability to play complex music. The system sounds better with simple music and we tend to stick with it, thinking that rock, pop, electronica etc is all badly recorded rubbish (most of it is brilliant).
- An inability to play at louder volumes. We have a few favourite albums and sometimes like to pump it up a bit. But we are aware that the sound gets hard and shouty and has considerable listener fatigue. Other people in your house are not keen on the experience at all.
- Vocals, percussion and lead guitar sound particularly harsh and bright.
- Imaging is not wide and deep. With more complex music the sound drops onto the face of the speakers.
- One-note bass or bass resonance at certain pitches. Can be room acoustics but can also be feedback into the system causing gross tracking error when excited at particular frequencies.
A personal note. I learned a great deal about time domain signal theory from my beloved friend the late Dr Gareth Humphreys-Jones. I owe him a considerable debt.