On 08/22/2015 10:07 AM, Jay Jaeger wrote:
One other type: spring arm, where the tape reels do all the tape moving, and a speed sensor, rather than a traditional capstan. The tension arm is part of the servo system to which the reel motors respond. In this case the reel motors and tension arm are anything but independent.
Historically, the two are very different in application. If you've got a COBOL handy on your tape system, try running the 1974 Navy Audit Tests, once the set of benchmarks by which CODASYL compliance of a vendor's COBOL was judged. (Al, do you have a copy of those on bitsavers?)
Very short tape records were written and read back in one of the tests (I forget which one; it's been 40 years after all). Vacuum-column tape drives made the most itneresting noises as they went through their start-top tape motion. You could actually play melodies on some of them, simply by varying the block length.
Vacuum-column drives, if you will, are inertia-trading devices. The reel motors are powerful in a 300 ips drive; they have to be. Practically speaking, it's not realistic to expect the system composed of the mass of the tape reels and the motors behind them to go from 0 to 300 ips in a fraction of a second--nor, to do it accurately. So, the reel motors handle the approximate movement of the high-inertia reel system and only maintain a few feet of loose tape, held in the vacuum columns, said short length massing almost nothing. The capstan(s) can then manage the quick tape movements quite nicely. The nature of the capstan mechanism varied between manufacturers--some used a pinch-roller sort of affair; the others used perforated capstans whose selection of nothing, vacuum or positive pressure could be managed quite nicely by a voice-coil valve.
The point is that during the 50s and much of the 60s, diskless/drumless systems were not uncommon (anyone remember S/360 TOS?) and tapes were used as working storage. Sometime during the 70s, with the proliferation of disk storage, tapes became relegated to archival or offline storage, not working storage. Pick up a copy of Flores or Knuth on sorting and you can see how important tapes were for handling and manipulating large amounts of information.
Comes the minicomputer and you begin to see spring-arm and direct servo drive units aimed toward the archival use of tapes--that is, a tape held information to be copied to a disk, so start-stop on a dime wasn't necessary. If one overshot a record, the drive/formatter only needed to bring the tape to a halt and read backwards until the gap before the desired block was reached, then back up another block or so and get a running start at the target.
This change in role probably is what governed the brain-dead treatment of tape by later operating systems such as UNIX and PRIMOS, to name a couple. Until the advent of cheap cartridge tapes, almost all microcomputer OSes were utterly ignorant of the existence of tapes.
--Chuck
