Paper Tape

Now, those of us who actually used paper tape readers are somewhat older than your typical texter today. During my time with ICL (International Computers Ltd, a UK company with New Zealand offices in the capital, Wellington at that point), around 1972, I helped support a paper tape reader on a System 4 computer at the NZ Post Office, on The Terrace I think. I worked with some really great guys who (in retrospect) were very tolerant of a naive and arrogant youngster. I learnt a huge amount about computers and working in a professional environment, and left only because of a disagreement over a training trip to Australia. I returned to University in Christchurch to get married and continue my ‘degree interruptus’.


Paper Tape reader unit, showing stepper motor drive and feed roller. This is only the read head and guides, not the whole unit.

The System 4 computers were a very different architecture to ICL’s mainstream 1900 series. The former were byte-based, whereas the latter were 24-bit word based (from my failing memory). I tried, reasonably successfully, to avoid the 1900 series. The System 4 was eventually phased out, another example of what certainly seemed to be a superior technology being ousted by a more popular one.

To get to the point, the paper tape reader at the Post Office was ‘A Beast’, both in terms of trying to keep it working, but especially in terms of performance. This thing could read around 1400 characters per second and if you don’t think that’s impressive, build something that literally shoots paper tape one metre horizontally before it even has time to droop.

Paper tape is about an inch wide and comes on spools up to 8 inches or so across. It has a sprocket hole down the centre (roughly), and 9 hole positions across the tape, for one byte per position, plus parity. There was longitudinal parity too and I think also a check sum - these things were seriously error-prone, especially at high speeds. (looking elsewhere on the Web, I see 5 holes per tape was the rule. When I get a minute, I’ll go find the unit shown here and have a closer look).

To read a tape, it was first loaded on the reader supply spool, with the tape threaded over some spring-loaded take-up arms, then over the sprocket feed mechanism, though some more take-up arms, then onto the take up spool. The two spools were servo-driven from the take-up arms, like a magnetic tape drive. That worked OK on slower paper tape readers, but from memory again the one at the Post Office was so fast the take-up spool couldn’t keep up, so it had a big hopper where the tape could drop - hopefully without tangling. That might also have been an option for where the tape was to be re-used. Regardless, the controlling computer issued a read request, which set the tape moving past the light-sensitive heads. As each character went past it was buffered and when a block was complete it was error-tested and sent on to the computer, which did what it needed to do. The tape could accelerate and decelerate extraordinarily quickly, much faster that the eye could follow. Roughly speaking, it had to be able to stop in a single character distance (which isn’t much at 1400 cps).

Apart from tapes breaking, the main issue was getting the light-sensitive diodes reading all the bits of a character (which were across the tape) at the same instant. We had test tapes to feed through that had all holes punched and used an oscilloscope to set the ‘read instant’ for each diode based on the light from the timing hole (the sprocket hole I think). You did each channel separately - rather time consuming. The main cause of mis-alignment was the tape skewing as it went past various guide posts, or a failure to accelerate quickly enough to be up-to-speed before the first character got to the read head.

The tape reader in the photos here is, I suspect, a much later item. It uses a stepper motor drive, whereas the ones I worked on were all servo-driven. I don’t know when stepper motors came into common use, but it was after this.

As magnetic tape drives became more affordable and data entry equipment could create magnetic tapes directly (such as the ICL Key Edit equipment that was my main responsibility - I looked after the small unit (8 stations I think) at Commercial Union Insurance and the much larger (16 or 32 station) unit at the Reserve Bank), paper tape ceased to be the main source of input and as a result, paper tape readers could operate much more slowly since they were then used only for the initial ‘read’ of the data, which was immediately transferred to magnetic tape.

The tape head shown here is quite ‘low tech’, albeit brand new - it was a spare part that never got used, I am told. It certainly looks pristine - real ones had (often major) wear marks on the capstans and the head itself. This unit is only a reader, but most tape readers I saw were also paired with a tape punch, which worked very similarly but obviously punched holes in the tape using solonoids instead of using photo diodes to read the holes.