I've gotten into that heap of old keyboards down in a corner of the basement, and looked at 'em with meter and 'scope. Certain ones use positive-going strobes, and others use negative-going strobes. Now, these are primarily LSI encoded, so the logic is not easy to discern, but it is apparent that some use positive strobes, simply because they have pulldowns on the rows rather than pullups. The majority, however, do use pulldowns, and, as one would expect, when they're scanning, they have short negative-going strobes.

It's just puzzled me to no end, that some would use what one would expect, i.e. negative-going column strobes, while others use positive ones. I can understand why the negative-going strobes are more common, first, of course, because when the same function was done with TTL SSI/MSI logic, that was easier, but when the designer has control over all aspects of the design, I truly wonder why some designers would use the positive-going strobes. Historically, when done in TTL, the low-going strobes made more sense because of noise/crosstalk risk. With more-or-less balanced outputs, it's not as big a factor, but I have seen CMOS circuits behave very strangely on an extender card, when a hand was passed nearby. I can only assume that this is a contributing factor.

The power consumption may be a factor, I guess, though back in the '70's, when these were designed, it was not a major concern, as they often used PMOS, i.e. they required three supplies, despite the fact that their outputs were TTL levels only, and they certainly could have done whatever they wanted. The CMOS 74C922 and 923 keypad encoder spec's specifically state that they "see" a resistance less than 50 k-ohms as a key closure. Perhaps that sheds some light.

I haven't spent a lot of time on this, but it is a mystery and I'll be looking at it further.

RE