from Bob Feretich - Mar 1, 2006

We made very good progress yesterday working toward a valid reset state for the TAU.

• Fixed bugs in the reset logic tree.
• Fixed bugs specifically effecting "WTM" and "WRITE mode" resets.
• Discovered a very convenient scope setup that is making troubleshooting reset problems a snap.
• Place scope in storage mode
• Attach Channel A to the reset switch logic (before pulse generation)
• Set scope to trigger on Channel A
• Set sweep to 10 uSec/div. Some reset pulses are 1.5 uSec wide. The scope will not store two channels below 10 uSec/div.
• Probe troubled logic using Channel B
• Use the TAU CE Panel "Reset" switch to stimulate the logic.
• Most common bug is a clipped (.3V instead of 6V) reset pulse caused by a leaky transistor. I will send a separate e-mail discussing transistor problems.
• We are seeing a very high density of failures (bugs per pound of cards used in the logic function) in the TAU. If this continues it could become a major problem. Some failures seem to be related to card handling and inspection. The SMS logic cards used in the TAU seem different than the ones I have seen used in the CPU logic We often have no spares of the TAU card types. That means we need to troubleshoot each card if we suspect a problem and repair bad ones before we can proceed. This has slowed the TAU debug process.
  

• The "Go" latch is not being reset.
• I think that the RDD counter logic is not being reset, but I don't know how much of it is supposed to be reset.
• Then we can start debugging manual initiation of simple (non-data moving) TAU operations. I suspect that we will see similar failure modes on these operations and that the above scope setup (but moving the Channel A probe to the appropriate switch) will work well for debug.
  
  

• The Tape Channel Analyzer is working well in responding to simple selection handshakes. 
• We now think that we don't need to replace the 26LS32s that are sensing response signals. Ron Crane has devised a way to compensate for their current draw by adding one resistor to each signal. No unwrapping will be needed.
  
• Ron Crane believes that the -N response signals will not need 26LS32 threshold adjustments. 
• I would like one more debug session using the Analyzer before I make the modifications to it. (I'm busy this weekend and can't work on it anyway.) I will plan to add the threshold adjustment logic and compensation resistors to the +P response line sense logic late next week. In the mean time, I am learning PIC18 architecture and assembly language.