Thanks Bob.
While I have been working on a keyboard I found an interesting "not quite" fault.
Firstly one of the CD4051 analog switch IC's had failed, replaced that keyboard works.
But as I have been going through it with the scope, to document things, I found something quite interesting. The other CD4051 IC (the keyboard uses two) is actually defective too. It is outputting a large voltage transient on the switching edges of the control (channel select) data. Once processed by the transistor amplifier, it is about 3V peak. The other transients due to small amounts of the capacitive coupling with the keys open are about 300mV.
It turns out though that this amplitude of this abnormal pulse is just a little below the trigger level for generating an output pulse from the transistor threshold detector. So the keyboard works normally. But if this problem/pulse was 700mV to 1.4V higher the keyboard would have malfunctioned. The new CD4051 does not have the issue.
It is being caused I think by failure of one of the Fets inside the CD4051, at a gate junction, that is allowing the control data bit switching transitions to leak into the the switched analog information. I have heard of these analog switches doing this sort of thing before.
It is interesting too to find a fault that is just about to, or hasn't quite happened yet. It is not unlike the movie Minority Report, with the department of Pre-Crime, where you find something about to go wrong and fix it before it happens.