Think of nuclear bomb design as the direct inverse of landmine design. With the latter you want it to blow up under a wide range of circumstances, including if it is "tampered" with. You want sensitivity, within some wide range.
A significant concern with nuclear weapons is that they're small enough to steal.
You definitely do not want thieves or terrorists to be able to trigger a nuke.
So the trigger systems for nuclear weapons are encrypted and require a decryption key to be functional.
A key requirement for a successful (nuclear) detonation is nanosecond-level timing control of the explosion. Anything else will result in a fizzle with the conventional explosives just scattering the nuclear material in a small area.
It's possible that some nukes had deliberate self-destruct modes where the circuitry would react to tampering by triggering an asymmetric explosion, causing a fizzle, which is relatively harmless to city-sized targets.
We still don't have the whole story, maybe the whole story isn't even known.
But an anti-tamper destruct is a reasonable explanation for the K-129 incident.
These days it's easy enough for a bomb to fry it's electronics to brick it, but that was before electronics were so easy to fry.
Yes, what I mean is that the anti-tampering self-destruct mode could also be activated after the main device is triggered. That way, if the main device does not detonate, at least you know it will never detonate on its own (since all of the circuitry has been destroyed).