Reedholm addressed hot switching damage to node zero relays with new matrix designs, by minimizing damage to node 0 switches which bear the brunt of hot switching from sneak paths. The node 0 triple pole, single throw relays (force, sense, and guard contacts) have been replaced with independent single pole relays for the force and sense lines. Resistance added to the node 0 sense path prevents welding of the sense relay, and provides a dis-charge path if there is voltage present. The force relay is not allowed to close until voltage is <1VI, which is below the arcing voltage. Thus, even if there is switch bounce, the voltage is too low to transfer material, which leads to welding, or sticking, of contacts.
Getting rid of possible sneak paths can be a difficult problem when a system has been doing production testing for any length of time. Hundreds, if not thousands, of test lists can be affected. Stuck relays and scrambling are often hightly intermittent, and can bring testing to a halt with different symptoms from wafer to wafer and week to week.
Stuck Node 0 (ground) relays problems result in lots of problems for customers. It halts testing, forces reprobing of wafers, and generally cuts down on production test time. Few customers have the resources to do what really ought to be done – fix test conditions or test structures to eliminate the root cause. Part of this is because for many of the products, wafer lots are run years apart.
Without hot switching, dry reed relays in Reedholm systems have lifetimes of 10^9 operations, with end of life defined by contact resistance increasing to several times the initial resistance of 100mΩ. Even if operated continuously, which never happens, a relay would last more than 20 years. That is why we say relays do not wear out. That is also why we say that relay failures are due to hot switching.
One step in the quest to eradicate hot switching involved changing the power control logic (PCL) board to prevent turn-on of the 120V supplies when AC power is applied by requiring software initialization first. Another source of uncontrolled hot switching are sneak paths that charge unassigned pins, only to have those paths discharged to ground between tests. More details can be found in SN-136 on our web site.
Getting rid of possible sneak paths can be a needle in the haystack problem when a system has been doing production testing for any length of time. Stuck relays and scrambling are highly intermittent events that can bring testing to a halt, even though problem tests vary from wafer to wafer or week to week.
If you have not found the time or resources to change test conditions, but are faced with stuck node 0 relays, you might want to consider changing to matrix modules that have circuitry and software that lowers the chance of a node zero force relay being closed with voltage on a pin. Migrating to the new PAM-16N and CPM-N does not eliminate scrambling or all hot switching, but does minimize damage to node 0 switches, which are the most likely to face hot switching.
Customers that have migrated to the improved matrix card report significantly lower failure rates (19 to 2) and some have seen no stuck pins. One reason is that resistance added to the node 0 sense path prevents welding of the sense relay and provides a discharge path if there is voltage. The force relay is not allowed to close until voltage is less than 1V, which is below the arcing voltage. Thus, even if there is switch bounce, the voltage is too low to transfer material, which leads to welding, or sticking, of contacts.
Please contact us at firstname.lastname@example.org if you would like to discuss your system updates. We look forward to hearing from you.