TSF/TSFi neighbor data scope test Jul/24/03

Setup:

Slot 1: TSF-1 + TSFi-7 playing both neighbor Hi/Lo into
Slot 8: TSF-2 + TSFi-5 receiving both hi/lo neighbor data

The input memory wire map leading to neighbor output bits 0-27 are assumed to be:

TSF crate backplane test points:

CH3 (blue) = TSF-2 J2 D9 Neighbor_in27
CH4 (red) = TSF-2 J2 D3 Neighbor_in3

Playing all 28 nominal neighbor bits 0-27 together, while setting all other non-neighbor bits to zero:

Playing all wire bits in input memory together:

Noticing the large cross talk on neighbor_in3, a binary search test from Jaime narrowed it down to the wire "#" in the map table above which is solely responsible for the negative cross-talk spike on IN3 as playing "#" on/off drives the spike on/off. Unfortunately when Eunil powered off the crate to take out the ZPD, the large negative spike problem went away to prevent further debugging. Interestingly, this may well be correlated with another problem: in the map above, the neighbor bit 21 from column (bit) 0; row (word) 11 is not actually sending the neighbor data on line 21, Instead, playing wire "#" next to it actually drives the neighbor bit 21.

Further looking at the cross talk: turn off the IN3 bit playing and turn on the wire "#" playing to watch it cross talk to IN3:

Playing all other bits but turning off IN27 playing to watch cross talk on IN27:

Although crosstalk is rather heavy, if the neighbor data receiving actually clocks in the data during the last clock30 of the clock4 cycle as in original design, it should not matter still.