Bad TI IC sockets
TI made some bad IC sockets back in late 1970's. They are easy to identify since they grip the IC pins along the edge of the pin and not the flat body of the pin. A typical high quality replacement would be a Mill-Max machined pin DIP socket.
DIP sockets are often problematic as instruments age. Gold-plated pins carrying low current in gold-plated sockets are the least problematic. Higher currents, such as occur in ECL logic, are more problematic.
For gold pin DIPs like many of the Tektronix custom ICs, one would choose a gold internal contact plating, and for other early non RoHS lead-tin plated ICs a similar internal contact plating is available from some suppliers.
Tek issued a service note (see below) that describes a transition from the original C93 socket design to a C95 design which was mechanically too fragile, and then a transition to an improved version of the C93 socket design.
For the 14 pin DIP at the left, a gold part number would be mill-max 110-93-314-41-001000, a non-RoHS lead-tin part number would be mill-max 110-99-314-41-001000. Both parts are non-RoHS since the solder-tail (PCB end) is also lead-tin plated.
Former Tek Bench Technician and Applications Engineer Jim Mauck says:
I noticed your article on bad Texas Instruments IC sockets. I literally replaced 100's of those sockets. I believe TI's part number was C95. They were used throughout Tek at the time but they do not always cause an issue. They didn't have a gas-tight contact with the IC pin and they would oxidize over time and develop a resistive connection. This wasn't necessarily a problem for low current applications but it was death to any ECL circuits. Unfortunately ECL was used extensively in logic analyzers. There were very high failure rates for the 7D01, DF1, and DF2 and eventually Tek had an unpublicized board exchange program for those instruments. Service technicians could replace the boards but there was no official modification kit or customer notification.
Failures due to these sockets were typically intermittent. A classic indicator of bad sockets was having the problem disappear if you wiggled all of the ICs in the socket. Doing that scraped off some of the oxidation and the instrument would work for another month or so until the contacts were again oxidized. The problem was exacerbated in some cases because Tek was using ICs with tinned leads intended to be soldered to the circuit board - not socketed. I never saw any problems in high current applications when the leads were gold plated.
maintenance note for 7D01, C93 to C95 to "new C93" change