Should this "pop"?
YES... That IS a "pop"
If I saw that, yes, I'd expect to hear a "pop" of some kind on that channel, because there's a "sharp" amplitude change, disrupting the continuity of the waveform. Isolated sharp angles like this in a soundwave generally generate a very short burst of broadband noise, whose harmonic content is naturally dependent on the shape and length, as can be seen in this spectrogram of a 440 Hz sine wave with a noisy synthetic "pop" ("synth pop"! ;)).
The nature of soundwaves means any change in amplitude in the time domain will generate additional frequency content in the frequency domain (and vice versa) as we are effectively applying amplitude modulation, and the more abrupt the change, the "brighter" the resulting spectral content. When we build wave shapes by adding sine waves (additive synthesis), we know that in order to create a sharp angle, we must use high frequency harmonics to get the corner as sharp as we can. So when we deconstruct that wave shape via FFT, we know that it will reflect this rich high frequency content.
A very useful wave shape in subtractive synthesis is the sawtooth wave, and the reason it's so useful is that it's the wave shape with the richest harmonic content, because it contains these, albeit periodic, abrupt changes in amplitude which could individually be considered "pops" or "clicks".
In reality, as we add constituent waves to build our "pointy" wave shape, the corner becomes increasingly more defined, but as the sample rate limits the amount we can add, they never really become ideal corners. There will always be what we call "ringing" in the analogue signal, as can be seen in this generated sawtooth wave with approximated analogue wave display (using the RX default sample interpolation order of 33).
Which brings me to the next problem with "pops" and "clicks". Due to the abrupt change in amplitude, as with the sawtooth wave, a relatively high "true peak" can occur, particularly at higher frequencies. The true peak is the measurement (
dBTP) of where the resulting analogue signal is likely to peak.
For example, let's say you purposely create some distortion and think, yeah, it looks fine, the highest peak is around
But as you can see in the waveform stats, the true peak has a value of over
3 dBTP! That's around a
4 dB increase. If we use the 33rd order interpolation, we get this better approximation of the analogue signal.
This will cause more distortion in the resulting analogue signal and could even damage equipment (Although that's not very likely to happen for short bursts like this).
All this results in a relatively intense "pop" or "click" as the intense discontinuous impulse displaces your speaker cone with a fidelity and frequency response dependent on the specs of the monitor/speaker. It disrupts the continuity as if you had given the cone a hard, sharp flick.
So that is why this waveform would make an audible "pop" that may sound different through different output devices. You can see the pop in your spectrogram, though. That doesn't lie. Although it does look like it's not very "sharp"(remember? High frequency content = sharp angle). If there are other similar-looking events, I'd fix them too, with simple interpolation or patching, because they may be a bit smoother than this one, so have a high frequency roll-off, making it a "dull" "pop" that could be masked by other sounds. As for the cause, it could be a sample clock sync "hiccup", possibly a CPU or memory performance issue, as this type of anomalous distortion of the signal is usually the result of a large, approaching instantaneous, step in amplitude.
I hope this is the kind of answer you were looking for.