If you can hear them on your LP and SACD, but not CD, unless your CDP has some issue, the odds are it is just not mixed the same. They are mixed differently, more often than not, especially a 25yr old CD, not to mention the record likely has some compression of the highest volume peaks that brings out the rest.
Is there any point in pointing out that the following statement shows a lack of understanding of the timing of multi-channel digitized audio (or any digitized system) for that matter? All that matters is the analog bandwidth, jitter, channel to channel timing jitter, and signal to noise ratio assuming the sampling rate is sufficient to capture the analog and allow reasonable analog filters. Sampling rate has literally no effect on timing beyond supporting the analog bandwidth. Absolutely none. It is a common misconception that it does, but it does not. No room for a full course on Nyquist here, but feel free to research.
Okay so you play your record, stylus traces the groove. Which by the way the smallest squiggles on the groove wall are so incredibly small they’re on the order of some large organic molecules. So incredible detail. No idea how many times you’d have to sample to trace an organic molecule but got a pretty good hunch its more than thousands per second, and probably up there in the millions. Orders of magnitude at any rate above any current or even proposed rates of digital.