 |
| Home | The Simulator | Telnet Clients | PDP-11 History | Links of Interest | Code Archives |
|
Early Computing at Wofford Original Witch Witch Reincarnation Access the WITCH Wofford Witch Documents Image Gallery PDP-11 Collection (w/documentation) Projects Message Board Videos |
 |
A DH11 Installation Story John E. McNamara, former DEC employee and the designer of the DH11 multiplexer, tells the story of his first experience installing a DH11 for a client This story is reproduced here with John E. McNamara's permission, from an email correspondence with W. Ashley Carder dated December 7, 2005 In March 1973, a mere 32 years ago (!), DEC had a major customer in Pittsburgh called Online Systems. I believe they had ten PDP-10's front-ended by about 30 PDP11/40's. They ordered the first two DH11's and insisted that the Project Engineer had to come to Pittsburgh and install them. I had installed one in a BA11-F (I guess) and was sitting cross-legged on the floor about to install the second one that was temporarily lying on the floor. A fellow previously introduced to me as their VP of Software Development came by and inquired, "How many of these do you have in the field?" I replied, "Two - that one in the BA11 and that one there on the floor." He looked distinctly pale and said nothing more. John McNamara talks about DH11 Diagnostics, Jan. 4, 2006 Hi Ashley, A brief word about diagnostics. At one point there was grave concern that many of the PDP-11 diagnostics did not detect all possible faults. A "fault insertion" system was developed in which various points in an option were faulted to ground or otherwise inhibited, and diagnostics were run to see if the fault could be detected. As I recall, for some options, only 65% of the faults were found. To ensure that all faults were found, the Field Service folks insisted that the DH11 diagnostics had to run each line individually at all formats (5, 6, 7, 8 bits per character, 1, 1.5, 2 stop bits, and at all speeds (50 through 9600 bps). The coverage was excellent - well over 90%. (Unibus faults were difficult to isolate.) Unfortunately, the full suite of tests, DZDHA through DZDHK took over an hour, principally because the DZDHF Single Line Data test took over 50 minutes! After there were a few thousand DH11's in the field, the Field Service folks showed up at my doorway complaining that the DH11 was one of the "top 10 trouble options" as measured by how many man-hours were spent repairing them. I thought I had been very conservative and careful in my design, and was very disturbed until I realized that the very long diagnostic run times lengthened every service call considerably. About the same time, the PDP-10 group was developing the KL10 processor, which had an 11/40 and DH11 in the front end. Since they wanted the Mean Time To Repair for the whole system to be less than an hour, the existing DH11 diagnostics were completely unacceptable. They developed two new diagnostics, I think they were DZDHL and DZDHM, that handled the data line testing much more rationally, had the same fault detection capabilities as DZDHA through DZDHK, and ran in a minute or two. So, when the Field Service folks came to complain, I suggested that they test the capabilities of the new diagnostics, and if acceptable, use them instead of DZDHA through DZDHK. Of course, the new diagnostics were just as good, so I never heard any more complaints. In summary, if you can find DZDHL and DZDHM rather than DZDHA through DZDHK, you can save a lot of time ;-) Happy New Year to you and may there be bits flowing through your DH by this time next year! -John |