JVC HR-S5902U band noise pattern from linear head?
 

I'm seeing banded noise patterns coming off the linear audio head - particularly bad with EP recordings. I've adjusted the head to make the audio as good as it can be but nothing puts a dent in that noise. I've tried grounding the head, no change. The ribbon cable is obvious very sensitive to interference, if I touch it with my finger, it makes a pronounced buzzing sound around 60hz with accompanying harmonics.

The spectrum very clearly shows the noise falls within a range of various frequency bands. Green lines were added for emphasis:
Attachment 18662

Here is a brief clip where the noise is very conspicuous. The audio program is very low, so the S/N ratio is terrible.
Recorded 1987 in EP Mode:
JVC HR S5902U Noise Patterm.mp4

With Hi-Fi recordings the noise isn't bad. On linear audio where the audio program further from the noise floor, it's a lot better, but visual inspect of the audio spectrum reveals the noise pattern is still there, just lower relative to.
Recorded 1994 in SP Mode:
Linear Audio Test 2.mp4

I also have a JVC HR-S5200U and the audio still has a terrible S/N ratio but the noise is more of an ordinary random hiss, not this nasty patterned stuff. I'd use that one on the old EP stuff, but the 5902 tracks the video signal better than the 5200 which has a lot of vertical jitter on the older deteriorated tapes. I can restored the audio but if there's a fix for capture, that would be better.

Is there most probable cause of something wrong with the 5902 itself, or is this typical behavior for low-signal audio on these machines?

Hard to say, but I suppose the way to check if it's an issue inherent to that model would be to do a deck swap from a similar model that doesn't have the issue which will also swap over the audio head itself. If the issue follows the deck, it's something in the original deck like a vibration or AC head that could probably just be swapped. If the issue follows the 5902 motherboard, it's probably something in the audio section there.

Other way to look at it would be to get another 5902 and see if it does the same thing or not.

This issue has been discussed occasionally on this forum. It's a shame decks with good picture can have such noisy linear audio. Normally the recording itself has little or no such patterned noise. It'll mostly or entirely emanate from the playback VCR.

Causes can include a noisy/unshielded VCR switch mode power supply (SMPS), unshielded ribbon cable from the A/C head (common in later model VCR's), poor shielding in other relevent areas in the VCR, noise emanating from the video head drum motor, and failing caps in various parts of the VCR.

For the average user this VCR born noise can be difficult or impossible to fix without the relevent skills and experience, and some are custom modifications, often the only way to extract the best linear audio performance from these slow tape speed formats.

Digitally reducing or removing these many tones (which is never ideal anyway and is more work) can be difficult as they may emanate from different sources in the VCR and so be unrelated harmonically. It can be tough just identifying the fundamental frequencies. As with so many things "prevention is often better than cure".

For the average user it's probably best to capture the linear audio separately on a VCR with lower linear audio noise and replace the noisy audio in your video editor.

Yeah, I'm tempted to re-capture the audio on my other VCR - knowing the sync will need to be wrestled to the ground. I'm going to edit the video and output just the audio from the final edit to see if I can use advanced noise reduction techniques to make is at least not annoying. I have stretches of "silence" for the reduction algorithm to sample from. Thanks for the response.

The sync issue may not be as bad as that. The two tracks should be pretty close as both are controlled by the control track on the tape. After capture and import, line up the starts of the two audio tracks in your editor, at least to the single picture frame accuracy, but better tighter than that. Then check the very end of the audio to see if they are still in sync. Again they should normally be pretty close.

My guess is that cleaner transfer will be a lot easier than trying to cleanly filter out all those many loud noise tones from the poorer transfer, pictured below as many spikes in a standard spectrum display.

If using frame TBC, sync should not be an issue.

Also gotta ask, what program did you use to visualize the noise spectrum there?