DECtape, originally called "Microtape", was a magnetic tape data storage medium used with many Digital Equipment Corporation computers, including the PDP-6, PDP-8, LINC-8, PDP-10, PDP-11, PDP-12, and the PDP-15. On DEC's 32-bit systems, VAX/VMS support for it was implemented but did not become an official part of the product lineup. DECtapes were 3/4 inch wide, and formatted into blocks of data that could each be read or written individually. Each tape stored 184K 12-bit PDP-8 words or 144K 18-bit words. Block size was 128 12-bit words (for the 12-bit machines), or 256 18-bit words for the other machines (16, 18, 32, or 36 bit systems). From a programming point of view, DECtape behaved like a very slow disk drive.
Contents |
DECtape had its origin in the LINCtape tape system, which was originally designed by Wesley Clark at the MIT Lincoln Laboratory as an integral part of the LINC computer. The design of the LINC, including LINCtape, was in the public domain, and LINCtape drives were manufactured by several companies, including Digital. DECtape used the same transport mechanism as LINCtape, but the tape was run in the opposite direction, thus the supply and takeup reels were reversed. Mechanical dimensions, speeds, and signal characteristics were identical, and at least one system, the PDP-12 (with the TC12-F option), was capable of using either LINCtape or DECtape on the same transport.
On the PDP-12, the DECtape drives were tightly integrated into the LINC CPU instruction set. There were simple LINC instructions, single instructions, for reading and writing tape blocks.
While LINCtape was designed to support high-speed bidirectional block search, it only supported actual data read and write operations in the forward direction. DECtape used a significantly different mark track format to provide for the possibility of read and write operations in either direction. Some but not all DECtape controllers supported reverse read. This bidirectional data transfer capability is the subject of U.S. Patent 3,387,293.
In turn, LINCtape's origin can be found in the magnetic tape system for the historic Lincoln Laboratory TX-2 computer. Best and Stockebrand's 1958 paper "A Computer-Integrated Rapid-Access Magnetic Tape System with Fixed Address" is the direct ancestor of LINCtape, including the use of two redundant sets of five tracks and a direct drive tape transport.
DECtape was designed to be reliable and durable enough to be used as the main storage medium for a computer's operating system. It was possible, though slow, to use a DECtape drive to run a small OS such as OS/8 or OS/12. The system would be configured to put temporary swap files on a second DECtape drive, so as to not slow down access to the main drive holding the system programs.
Upon its introduction, DECtape was considered a major improvement over hand-loaded paper tapes, which could not be used to support swap files essential for practical timesharing. Early hard disk and drum drives were very expensive, limited in capacity, and notoriously unreliable, so the DECtape was a breakthrough in supporting the first timesharing systems on DEC computers. The legendary PDP-1 at MIT, where early computer hacker culture developed, adopted multiple DECtape drives to support a primitive software sharing community. The hard disk system (when it was working) was considered a "temporary" file storage device used for speed, not to be trusted to hold files for long-term storage. Computer users would keep their own personal work files on DECtapes, as well as software to be shared with others.
The design of DECtape and its controllers was quite different from any other type of tape drive or controller. Physically, DECtape used dual-redundancy to keep the error rate low. Each bit was written twice across the width of the tape, using Manchester encoding (PE). During read, the two read heads for each bit were wired in series, so the resultant output was the sum of the two bit amplitudes. This meant a "drop-out" on one channel could be tolerated; even a hole punched through the tape with a 1/4 inch hole punch would not cause the read to fail. Another reason for DECtape's unusually high reliability was the use of laminated tape: the magnetic oxide was sandwiched between two layers of mylar, rather than being on the surface as was common in other magnetic tape types. This allowed the tape to survive many thousands of passes over the tape heads without wearing away the oxide layer, which would otherwise have occurred in heavy use on timesharing systems.
The fundamental durability and reliability of DECtape was underscored when the design of the tape reel mounting hubs was changed in the early 1970s. A metal hub with a retaining spring was replaced by a lower cost single-piece plastic hub with 6 flexible arms in a "starfish" or "flower" shape. When a defective batch of these new design hubs was shipped on new DECtape drives, these hubs would loosen over time. As a result, DECtape reels would fall off the drives, usually when being spun at full speed, as in an end-to-end seek. The reel of tape would fall onto the floor and roll in a straight line or circle, often unspooling and tangling the tape as it went. In spite of this horrifying spectacle, desperate users would carefully untangle that tape and wind it laboriously back onto the tape reel, then re-install it onto the hub, with a paper shim to hold the reel more tightly. The data on the mangled DECtape could often be recovered completely and copied to another tape, provided that the original tape had only been creased multiple times, and not stretched or broken. DEC quickly issued an Engineering Change Order (ECO) to replace the defective hubs, to resolve the problem.
Eventually, a heavily-used or abused DECtape would start to become unreliable. The operating system was usually programmed to keep retrying a failed read operation, which often would succeed after multiple attempts. Experienced DECtape users learned to notice the characteristic "shoe-shining" motion of a failing DECtape as it was passed repeatedly back and forth over the tape heads, and would retire the tape from further use.
DECtape II was introduced around 1978 and had a similar block structure, but used a much smaller 0.150" tape[1] (the same width as an audio compact cassette). The tape was packaged in a special, pre-formatted DC150 miniature cartridge consisting of a clear plastic cover mounted on a textured metal base. Cartridge dimensions were 2 3/8" × 3 3/16" × 1/2". The TU58 DECtape II drive had an RS232 serial interface, allowing it to be used with the ordinary serial ports that were very common on Digital's contemporary processors.
Because of its low cost, the TU58 was fitted to several different systems (including the VT103, PDP-11/24 and /44 and the VAX-11/730 and /750) as a DEC-standard device for software product distribution, and for loading diagnostic programs and microcode. The first version of the TU58 imposed very severe timing constraints on the unbuffered UARTs then being used by Digital, but a later firmware revision eased the flow-control problems. The RT11 single-user operating system could be bootstrapped from a TU58, but the relatively slow access time of the tape drive made use of the system challenging to an impatient user.
Like its predecessor DECtape, and like the faster RX01 floppies used on the VAX-11/780, a DECtape II cartridge had a capacity of about 256 kilobytes.
The TU58 was also used with other computers, such as the Automatix Autovision machine vision system and AI32 robot controller. TU58 driver software is available for modern PCs.
Early production TU58s suffered from some reliability and data interchangeability problems, which were eventually resolved. However, rapid advances in low-cost floppy disk technology, which had an inherent speed advantage, soon outflanked the DECtape II and rendered it obsolete.
|
||||||||||||||||||||||||||||||||||||