H5: Huge Data Bus overshoot and undershoot

[ijor]

I've found that on my H5, the '373 latches sometimes produce huge Data Bus overshoot and undershoot. My H5 is populated with ACT373 chips, instead of the more common 74LS373 ones. These faster latches, with much faster edge rates, are probably the main cause of this (faster is not always better).

Here are a couple of scopes taken on the main CPU Data Bus. I tested different data bus signals at different locations and the results was always about the same. Not shown explicitly on the scope captures, but the worst case seems to happen just after the CAS leading edge, when the RAM toggles the latches input from high to low.

Note that I used the default longer probe ground lead. Ideally this should really be done with the short spring ground tip.

H5-D3-AtIo-Zoom.jpg (51.85 KiB) Viewed 2053 times

H5-D4AtIo-multiple.jpg (129.88 KiB) Viewed 2053 times

It would be interesting if other H5 owners that happen to have ACT373 chips could contribute similar scope captures. You can probe any CPU data bus signal (D0-15), on whatever location would be more convenient for you. Could be directly at the CPU, or at any other point directly connected to the main data bus. This would include almost any LSI chip except Shifter and the FDC.