The maximum number of pad touchdowns are limited to three.

An example of this (three touchdown) worst care scenario is: touchdowns

In the above example:
  • If the (4 pin/needle/site quad) probecard is scanning from left to right and encounters the "retest die" pictured above on the left (which will be on the probecard's right), it will shift over one more die to the right (to put a different probecard needle/site over the die) for retesting.

  • When the (4 pin probecard) quad head goes on to shift four die to the right (to test the next group of 4 die) and encounters the "retest die" pictured above on the right (which will be on the probecard's left), it will shift one die to the left for retesting -- hence the "retest die" pictured above on the left will experience (the maximum of) three touchdowns.
Die Retest Plan File Entries
Die retesting was controlled via the planfile [test_switches] entries:
RE_TEST [bin] [turn_on_consec] [cum_confirm_turn_off]
[bin] Is the bin number that kicks off a retest

[turn_on_consec] Is the number of times that this bin has appeared on the same site consecutively. Once this limit is reached, re-test keeps reoccurring.

[cum_confirm_turn_off] Is the number of times retesting (with a different tester site) confirms the failing bin has really failed. When this (retest, and decide it really was a failing bin) limit is reached, retesting is turned off for the this bin for the rest of the wafer (since the die are genuinely bad).

RE_TEST_LIMITS [no_of_retests] [min_percent_lowered]
[no_of_retests] After this number of retests ...
[min_percent_lowered] ... we should have at least this (integer) percentage of die bincodes lowered (due to retesting). - if not, turn off ALL retesting for the rest of this wafer (since retesting is wasting test time)
Note that:
if [no_of_retests] = 0 then this limit check is disabled
if [min_percent_lowered] = 0 then we turn off retesting after [no_of_retests]