Which Was Better: Punch Cards, Paper Tape, or Audio Tapes?

[Pentad4k]

If you happen to see my other post, I've been reading about the early days of the computer market during the 1970s. The books I've read mention punch cards, paper tape, and audio tapes.

I was wondering which of the three was the most reliable loading programs? My first computer was a Commodore 64 with a tape drive in 1983. I never had any issues, but at school the Atari's, Tis', and Apple's had a terrible time loading tapes. The school switched to all disk drives fairly quickly because of the tape issues. At the time I was told Commodore's interface was much more reliable. I didn't know if the Altair (and others) had the same issues loading from audio tapes?

This may sound dumb, but was audio tape faster to load than punch cards and paper tape?


Lastly, how much information could be stored on a single punch card and paper tape? Did one punch card = X amount of bytes and X inches of paper tape = X amount of bytes?


Thanks for sharing your knowledge!

 

[krebizfan]

Speed is an easy question to answer. The standard paper tape moved at a rate of 10 characters per second. The lower end punch card reader did about 800 cards per minute. If each card has a full 80 characters, that gives a data rate of about 1000 characters per second. Audio tape varied a lot in terms of format and redundancy with the typical 1500 bps format* yielding 10K in a minute. Single density floppy data rate was 125000 bits per second.
1 minute of data is 600 characters for paper tape, up to 64000 bytes for punch cards, 10000 bytes for cassette, and 750,000 bytes for single density floppy drive.

CPU controlled cassette was the cheapest and could be fairly reliable as the later FSK techniques took hold. It also helped that tape quality improved during the early 80s so one was a lot less likely to encounter gaps in the magnetic coating. While a person listening to dictation might never notice a missing millisecond, that could well mean that one or two bits of data are gone and with some formats, the rest of the tape would also be incorrect instead of just that short snippet being wrong.

* variations on this format were used by the Apple II and TRS-80 Model III

 

[Pentad4k]

Krebizfan,

Thanks for the information! That is really interesting! May I ask a few follow up questions?


In Fire in the Valley they mention that Microsoft Basic was shipped on paper tape and was heavily pirated. Was it easy to copy one paper tape to another or did you have to load it into the Altair and save it to a new tape? Was it slower saving to paper tape than reading from it?

Was it easy to copy punch cards? If you wanted to backup your program could you copy the punch cards directly or did you have to load it into the computer and write to a new set of cards?

If you had to guess, do you know when paper tape replaced punch cards as the dominate format, and when cassette tape replace paper tape? I always wondered how long or quick that progression was.

Thanks again for sharing your information! I was always interested in know this, but never knew where to ask.

 

[bsd64]

...how much information could be stored on a ... paper tape?

The Data General Nova was a popular minicomputer in the '70s that could use a paper tape reader and paper tape punch.

The paper tape was 1" wide, 8 bits across to make up a frame, and 10 frames to the inch. The writer operated at max 63.3 frames/second, and had room for 300 feet of tape. The separate reader could hold 150 feet of tape and operated much faster at 300-400 frames/second.

For comparison, the paper card reader for the same machine could hold 12 bits/column, 80 columns per card, and could be read at anywhere from 100-1000 cards a minute.

The magnetic tape was quite a bit faster:


Stats are from the Data General Peripheral Programmers Guide 015-0000210-07. On a different note, the Colossus computer managed to get read around 5000 characters/sec from paper tape in the '40s ... amazing really.

 

[Robbbert]

Punch cards and paper tape become useless if they get wet or mouldy. Additionally if you're carrying a stack of punch cards and drop them, you might never get them back in order.

Magnetic tapes can get mangled, but that's far less likely to happen.

 

[Chuck(G)]

Depends on the era.

A deck of punched cards could be easily edited and are impervious to magnetic fields. Paper tape could not be easily edited.
As far as I'm aware, unit record (e.g. sorters, collators, reproducing punch, accounting machines) equipment did not exist for paper tape.
Even today, paper mark-sense cards are still the primary mode of retaining verifiable voting records. Magnetic tape data can be altered.
Initially, magnetic tape was used as a backing store for punched cards. It became the primary on-line storage before disks became available.

I'm not aware of any commercial operation that relied on audio tape for archival storage. Note that my mention of "magnetic tape" above refers to the half-inch magnetic saturating medium.

 

[cjs]

There is some (albeit rather limited) ability to edit paper tape. The ASCII DEL character is all-ones specifically so that you can overpunch an area of tape with it (all holes) to "delete" the data there. And the ASCII NUL character is all zeros so that you can write a bunch of NULs on tape to leave a "blank" space for later overpunching. I'm kinda feeling these facilities didn't get used much, though.

I suppose in theory one can also cut and splice paper tape, though I've never seen that done.

Cards can have serious issues in shipping. There's the classic story about the guy who sent a program to France on a deck of cards, only to be told by the recipients that it didn't work. So he made up another deck, checking it very carefully this time, but was told again that it didn't work. He finally flew over himself with a third deck and, going through customs, discovered that French customs had a habit of taking a few small samples of "bulk" material being imported: they plucked a couple of random cards out of his deck before letting it through.

 

[Chuck(G)]

Ever wonder why FORTRAN and other languages used columns 1-72 of an 80 column card long after the 704 had passed into obsolescence? 73-80 is used for sequence numbers. You could even program your UR patchboards to punch them into a deck.
That's what sorters (e.g. 082) were well suited to handle.

 

[krebizfan]

Anything likely to damage paper tape or punch cards would also cause problems for tape.

Audio tape did survive for a long time as storage for laboratory equipment. The problems of earlier iterations of audio tape had been largely resolved by then but almost nothing else could get by on the limited capacity. CPU timing audio cassette routines would never been chosen in the multi-user world since they would lock the processor to that task for several minutes. I have more of an appreciation of audio tape storage now that I know what caused the problems about 50 years ago.

Paper tape did have a problem of becoming difficult to reliable feed even with fairly short programs. Some of the introductory programs I wrote took several feet and risked tearing. A 36KB program would need a tape the length of a football field.

 

[Chuck(G)]

And old paper tapes, particularly the fanfold ones evenutally started developing cracks along the folds and tearing of sprocket holes-unless, of course, mylar tapes were used.
Mylar is pretty much forever stuff.

 

[g4ugm]

Well first of all the 1970's were not the early days of the computer market, that was the 1950's. Home computers started to appear in the 1970s so is that what you mean.

Both punched cards and paper tape long pre-date computers and were adopted as computer i/o devices because they were available. Herman Hollerith pioneered punched card use for data storage and manipulation from around 1880 onwards for use on the US census. However he took ideas from card-operated Jacquard looms and train ticket punches.

Herman Hollerith - Biography

Herman Hollerith was the inventor of the punched card system which predated electronic computers.

mathshistory.st-andrews.ac.uk

so the ideas go back further. In fact the cards in a loom are strung together are they cards or paper tape? I think the first use of an actual roll of paper with holes was for "Player Pianos" which again were developed in the 1880's and 90's. Punched tape as we know it was I believe developed for the telegraph system, perhaps earlier than cards..

so these have the great advantage that machines to handle data in these formats was widely available, so for example this IBM punch design dates from 1928...

IBM Type 11 Electric Keypunch – Time-Line Computer Archive

t-lcarchive.org

so commercial computers used these formats

.. now for your questions. I don't think many home users had cards. I didn't meet any. Audio cassette tape was the most common. Nearly every one had a cassette player. Interfaces were cheap to build. They were slow and not especially reliable. Usually they used standard modem audio tones and ran at 300 baud.

If you could find a surplus teleprinter with a tape punch and reader that was a little more reliable but the reader was mechanical and could wreck your tape easily. At 110 baud also slower. A good combination was an ASR33 to punch the tape, and a home brew optical tape reader using the sprocket holes to clock the tape.

Storage density, (I hope my maths is right)

A standard card was 80 characters. Cards usually came in boxes of 2000 so 160,000 bytes per box of cards, or 160K.
I believe a reel of paper tape was 340 yds, so round 12,000 inches which at paper tape 10 characters per inch is 120,000bytes or 120k
For audio tape at 300 baud, at 1 start, 8 data, 1 stop bit that is 30bytes/second, so on a c90 cassette tape around 160,000 bytes or 160k

so pretty similar practical capacites...
.. hope this helps

 

[cjs]

Usually [cassette tape interfaces] used standard modem audio tones and ran at 300 baud.

Actually not: Bell 103 modulation used 1270/1070 Hz mark/space for origination and 2225/2025 for the answering station. Most of the 1970s and 1980s home computer CMT interfaces used 1200/2400 Hz or similar for 0/space and 1/mark, probably because these frequencies are much easier to decode. (You can simply count "high" inputs or zero-crossings or whatever's convenient for you, and easily distinguish between the high and low frequencies because one is twice as fast a waveform as the other.)

The famous Kansas City standard used 4 cycles of 1200 or 8 cycles of 2400 per bit, which was indeed 300 bps, but it was probably more common even by the late '70s to use 1 cycle/2 cycles for 1200 bps; certainly by the early '80s it was.

 

[g4ugm]

The famous Kansas City standard used 4 cycles of 1200 or 8 cycles of 2400 per bit, which was indeed 300 bps, but it was probably more common even by the late '70s to use 1 cycle/2 cycles for 1200 bps; certainly by the early '80s it was.

I really should have checked that. Its funny how memories get confused.

 

[whartung]

Punch cards required a large amount of infrastructure. Those punch machines were not tiny.

But as a format, it was quite robust. It also allowed for "random access". If you were writing a program, you'd punch a card and put it in the right place for the deck. It was not just a data transfer medium, it was a composition medium.

Paper Tape was a storage medium. The punches were much smaller, but editing and changes required reading and writing the entire tape (I'm sure someone clever managed to tape pieces together, but that's uncommon). And as others have mentioned, it was pretty slow.

Cassette had the potential to be an excellent medium. Much like paper tape, it was simple storage, read it all in, write it all out (I can say that with some certainty, as I DID try to do some "random access" on a cassette -- it did not go well).

We had PETs in high school, and those tape machines were perfect. We never had any real issues using them for programs or data. Super reliable, even across the six machines that we had. We only had 8K of RAM so its hard to quibble about performance. We also had stacks of 5m or 15m cassettes just for this, so it worked really well. I also had reliable experience with the KIM-1. There, like others, just used normal 60M cassettes, and relied on the little counter that the cassette tape had to stick programs on the end. Mind, I never tried my tapes on someone else's machine.

The TRS-80 was notorious for the tape being fiddly. When I got my Atari, I got the tape drive, but since its experience wasn't anywhere near as good as what the PETs was, I was quickly dissatisfied and got the disk drive. Mind, having a lot more memory meant you were moving a lot more data, and the slowness of the cassette really reared its ugly head. In college, it took so long to load Telengard on the C64, we'd do it on a Friday and just leave the machine on all weekend.

I think data volumes are really what killed cassettes. Because tape CAN be reliable, but for some machines it just wasn't.

 

[Chuck(G)]

There were drives that recorded data on Philips cassettes using traditional (saturating NRZI) techniques. Very often 2-track, with one track for clock. I used to have a could of these drives; still have an old NCR cassette for one.
You can still occasionally find one of these drives in the used CNC PLC marketplace.
Priced pretty much out of reach for the average hobbyist.
My recollection is that early Wang WPs used these cassettes also as well as very early 2200s.

 

[Chuck(G)]

Cards were very durable and readers were fairly fast. I give you the CDC 405 with its vacuum feed.

Cards were loaded into the tray near the front and stacked in the rear tray. There's a very small stacker pocket in the rear for cards that fail a compare check. Cards to be fed were held on a vibrating shake table and separated by an air jet, passed through photoelectric read and check stations via vacuum capstans. A remarkable piece of equipment that saw service on many systems for years.

1,200 cards per minute.

 

[krebizfan]

Cassette could be improved by taking control of the mechanism instead of just using a standard tape deck. The dedicated streaming tape options or even the cheap micro design of Phi-Deck showed how much better quickly repositioning the tape could prove. For most of the smaller experimental tape companies trying to compete in the micro market, there just wasn't time to polish the design into a very reliable mechanism before floppy drives dropped to the same price point.

Paper tape could accept patches the standard way through a second paper tape. Some of the paper tape formats could set the address where a given block would loaded. This saved on space with an executable since any dead space could be left off the paper tape but it also permitted overwriting parts of the file.

The 70s and 80s included lots of proposed data storage concepts. One of the good things about the long lead times of magazines is that the company designing a new storage product would go out of business before the first ads reached potential buyers. Saved a lot on buying samples for testing.

 

[Agent Orange]

And old paper tapes, particularly the fanfold ones evenutally started developing cracks along the folds and tearing of sprocket holes-unless, of course, mylar tapes were used.
Mylar is pretty much forever stuff.

The Navy used Mylar tape with the Varian 620i on AN/ALM-106B M.A.T.E. ECM threat simulator because of good archiving characteristics when at sea. The 620i was 16-bit mini while the Navy version was a 4-bit using a 4004. The time frame was early 70's.

 

[Chuck(G)]

You used to see loops of mylar tape hanging on racks at CNC shops, since those tapes were used pretty much all the time. Regular paper tape would not have stood up to the wear.

 

[g4ugm]

Paper Tape was a storage medium. The punches were much smaller, but editing and changes required reading and writing the entire tape (I'm sure someone clever managed to tape pieces together, but that's uncommon). And as others have mentioned, it was pretty slow.

Whilst the readers on Teletypes were slow, many built high speed optical tape reader clocked off the the feed holes. There were also commercial high speed readers. The Ferranti TR5 could read 300 characters per second, so equivalent to 3000 baud. Many smaller card readers were slow, so we had a Honeywell L58 "mini" that read only 100 cards/minute so a maximum of 8000 cpm or 133 characters/second, slower than the optical tape reader.

I also feel its time to come back to "which was best" well at the start of the 1970s as a home computer person, there was little choice, "best" meant "available at a local surplus store at a price I could afford"...