Sometime in the late '80s I ran a software group that built systems that integrated equipment with embedded micros and multi-user computer systems (DEC PDP-11 and VAX at that time). We had an 11/83 (rack-mount BA23 plus another Q-bus chassis) that we used for a lot of development purposes (including cross assembly for the micros). It had big disks (for the time), 9-track tape drives for backup, lots of serial terminals, a dialup modem or two, and ethernet to talk to our little MicroVAX 3100. The 11/83 left DEC with an 18 MHz clock, and it was the fastest PDP-11 CPU DEC offered at the time. Memory was maxxed out, too. But I needed more horsepower.
Late one night, I popped the CPU board (DEC M8190) and spotted the 18MHz crystal to the right of the DCJ11 CPU. I went over to the hardware lab and pawed through the crystals until I found a 24 MHz crystal, and soldered it in place. The system came up at this clock rate but crashed as soon as the operating system tried to load. I went back and found a 22.1184 MHz crystal. That one worked, and even passed all the diagnostics I could think of to run (CPU, floating point, cache, RAM, KDA50 disk controller, etc.). It seemed fine, so I left it that way. I never told my boss, or any of the programmers who worked for me (although at least one commented that the system was running very well). And I certainly never told the field service people who came to do preventive maintenance on the system.
I figured if it died then I would simply solder the 18 MHz crystal back in (and clean up the flux) before calling field service to swap the board. I later found out that the gate arrays for bus arbitration, etc. were the limiting factor on that board. That's probably why the console seemed to come up but the system wouldn't run RSX-11M-PLUS at 24 MHz.
Late one night, I popped the CPU board (DEC M8190) and spotted the 18MHz crystal to the right of the DCJ11 CPU. I went over to the hardware lab and pawed through the crystals until I found a 24 MHz crystal, and soldered it in place. The system came up at this clock rate but crashed as soon as the operating system tried to load. I went back and found a 22.1184 MHz crystal. That one worked, and even passed all the diagnostics I could think of to run (CPU, floating point, cache, RAM, KDA50 disk controller, etc.). It seemed fine, so I left it that way. I never told my boss, or any of the programmers who worked for me (although at least one commented that the system was running very well). And I certainly never told the field service people who came to do preventive maintenance on the system.
I figured if it died then I would simply solder the 18 MHz crystal back in (and clean up the flux) before calling field service to swap the board. I later found out that the gate arrays for bus arbitration, etc. were the limiting factor on that board. That's probably why the console seemed to come up but the system wouldn't run RSX-11M-PLUS at 24 MHz.
Last edited: