• Please review our updated Terms and Rules here

When was audio storage first used for computing?

A

alank2

Veteran Member
Joined
Aug 3, 2016
Messages
2,427
Location
USA
I wonder if there is an interesting history for when audio was first used to store and load data. Back when I got my first computer, a NEC PC-6001 back in 1983, it was ready made to work with the portable cassette recorders of the day. What was the first computer to use audio for loading/saving data?
 
Very early. The Kansas City standard was devised back in 1975 and there had been several years of experiments with the technology by then. Scelbi dates their cassette board to 1974.

I believe the earliest example I have seen dates back to the late 60s but I am having trouble confirming that.
 
One I know of prior is Datapoint 2200 that uses cassette. Announced by CTC in June 1970 (with units shipping in 1971. What use audio means will probably cause disagreement on what machines qualify.

Next
 
Do you count mercury delay lines?

Also, I wonder if early (1960s) modems could have the sound output recorded and then played back and if that was ever used?
 
There were a lot of early attempts at audio storage of computer data, many of which were not robust enough to be useful. Don Lancanster's article that was reprinted in the TV Typewriter book laid out the problems. The KC standard was very similar to the Lancaster proposal. Better tapes and other improvements allowed for considerably higher data rates.
 
I can't speak to audio generally (delay lines!?) but useful in the above to note there was a distinction in cassette storage between digital data using compact cassette style media, versus data-as-audio-tones on actual audio tapes. Both use the standard compact cassette form factor, but they work differently. The former were/are called "digital cassettes" and imprinted the magnetic flux transitions to tape-- not audio signals. These emerged from the mid/late 1960s (e.g. the kind of thing in the article in Electronic Design linked above) and are sort of a miniaturized version of the mainframe era reel-to-reel magnetic tape storage-- lower cost than the big tape units, but the ones that worked well still needed sophisticated mechanisms and apparently specialized tape media formulations. The tapes can sometimes (often?) be distinguished by a notch at the top of the case that ensures a correct orientation in the drive. I believe Datapoint 2200 used digital cassettes.

Data storage as audio signals on consumer grade cassettes (using consumer grade tape decks) emerged as more of a desperate attempt in the early/mid 1970s to figure out some kind of cheap and easy mass storage for personal/home/homebrew/micro systems in the years before disk tech was affordable. (Digital cassette systems were also largely not cost-effective in this context.) Lancaster's article as noted (it's also in the inaugural Sept 1975 issue of Byte) talks through the proto-Kansas City standard. SCELBI's cassette interface (suggested above) may have been the first commercial interface standard of that kind, I'm not sure.
 
Last edited:
Makefile said:
Also, I wonder if early (1960s) modems could have the sound output recorded and then played back and if that was ever used?
Click to expand...

FWIW, I know that Sphere's early experiments in audio cassette data storage did exactly that by using modem chips designed for real time telephony and sending that to tape. My understanding is that this signal was difficult to recover reliably because consumer grade tape has all manner of variation in speed both between decks and wow/flutter on any given deck, and that was often outside the range of acceptable variation for the modem signal-- which was never intended to be tolerant of that kind of signal variability.
 
1944GPW said:
Philips-based cassette data peripherals were in existence back in the late 1960s as noted on PDF page 30 of this December 1969 issue along with a number of other cassette systems:
https://www.worldradiohistory.com/A.../1969/Electronic-Design-V17-N26-1969-1220.pdf
Click to expand...
That Sykes unit is interesting as it included a PCB position ("interface kit") in its internal electronics that enabled incorporating a bus-interface board specific to any of a handful of minicomputers. Three are seen in this ad, which specifically mentions the Varian 620i, Data General Nova, PDP-8, PDP-8/L, PDP-8/I and PDP-12. I know that the DG Nova interface quite literally was a cabled direct connection to the CPU back panel bus, which was a standard Nova-line method for adding peripherals in the early models.
 

Attachments

bzotto said:
I can't speak to audio generally (delay lines!?) but useful in the above to note there was a distinction in cassette storage between digital data using compact cassette style media, versus data-as-audio-tones on actual audio tapes. Both use the standard compact cassette form factor, but they work differently. The former were/are called "digital cassettes" and imprinted the magnetic flux transitions to tape-- not audio signals. These emerged from the mid/late 1960s (e.g. the kind of thing in the article in Electronic Design linked above) and are sort of a miniaturized version of the mainframe era reel-to-reel magnetic tape storage-- lower cost than the big tape units, but the ones that worked well still needed sophisticated mechanisms and apparently specialized tape media formulations. The tapes can sometimes (often?) be distinguished by a notch at the top of the case that ensures a correct orientation in the drive. I believe Datapoint 2200 used digital cassettes.
Click to expand...
While the notch and specialized mechanisms were true for some systems, the concept of storing data as "digital" flux transitions is used by most (all?) cassette recorders solely intended for computer usage, with a non-audio style interface. The perhaps most well known example is the various models used with Commodore 8-bit computers , but other examples are the ones for the Atari 8-bit computers, Spectravideo 318/328 computers, various Sharp 8-bit computers and whatnot.

A related tangent: There were also DIY projects to add this feature to regular audio cassette recorders, even if the computer produced and expected "audio" signals. In particular the DIY electronics magazine Elektor (available in various languages back in the days) had a DIY project specifically intended for the ZX81. It included a switch to switch between computer and audio mode, where the computer mode took the signal from the computer and converted it back to something closer to a perfect digital square wave, and then sent that to the rec/play head as flux transitions. Can't remember if this project in particular also did something for playback or not.
They also had other projects intended for those who didn't want to modify their tape recorders, like notch filters removing/reducing mains hum and whatnot.
I've never tried these projects as I had Commodore computers back in the 80's.
===========
I wonder how common/uncommon it would had been to use reel-to-reel recorders back in the days? I think I've never seen anyone use one of those with a computer (for storing data on the tape). At some point in time it must had been fairly cheap to buy used tapes when for example schools sold out their old tapes when switching over to cassettes, and a reel to reel tape recorder that only had mono sound would probably at least not had been more expensive than a cheap mono cassette recorder (and the sound quality would had been way better).
 
Cassette data storage became common in 1970s home computers, but earlier systems used audio tones on magnetic tape for data recording.
 
The introduction to the ECMA 34 standard says:

ECMA TC19 was set up by ECMA in January 1970 with aim to identifying and standardizing the physical properties and the relevant data format of a magnetic tape cassette for digital applications – below the performance range of existing magnetic tape standards – in order to ensure interchangeability.

It appeared to be a fact that the 3,81 mm tape cassette was already used and will be more used in computer applications.
The urgent need for a standard for this type of cassette was therefore recognized and-TC19’s first efforts were directed to setting up the present Standard ECMA-34, which was issued in September 1971.
Click to expand...

ECMA-34 - Ecma International

Data interchange on 3,81 mm magnetic tape cassette (63 ftpmm, phase encoded at 32 bpmm)
ecma-international.org ecma-international.org
 
ECMA 34 defines a streaming tape format. The notch on one side of the cassette is a giveaway. The format should be around 250 KB in practice with a theoretical almost 3 million bits available on a full length tape but about 1/3 of that would be devoted to gaps between blocks. If I did the math right, the 32 bpmm would be about 50% more than the 1500 bps audio format used by the Apple II.

Reel to reel and other audio tape designs were used for data storage. The problem is that the low frequency signals* used by the super cheap data storage audio formats were what noise reduction systems tried to remove. It was somewhat counterintuitive that the lowest quality tape and recorder worked the best, as long as the tape coating lacked any gaps. Specialized computer cassettes were supposed to make sure there were no dropouts in the much shorter length.

* Low frequency signals were the most likely to survive all the interference that excessively long unshielded cables would receive. The industry was littered with failed companies that had high speed tape designs that worked in the lab but not at the customer. The KC standard duplicating each bit four times to make sure one would be readable turned out to be excessive.
 
Last edited:
If we remove the "computing" part and leave it to simply "digital stream", if someone ever recorded a hellschreiber - that might be the very first application I would recognize as audio media storing a string of digital data.
 
krebizfan said:
Reel to reel and other audio tape designs were used for data storage. The problem is that the low frequency signals* used by the super cheap data storage audio formats were what noise reduction systems tried to remove.
Click to expand...
Noise reduction systems only affect higher frequencies, where the annoying background hiss of audio tape is most prominent. For example, Dolby B Noise Reduction only involves frequencies above 5 kHz, and thus won't cause any impairment upon data storage formats like Kansas City which use frequencies around 1 to 2.5 kHz.

Dolby also only affects the quieter portions of the recording, below the Dolby reference level (indicated by the double-D Dolby logo on the level meter of many cassette decks). Computer program data was usually recorded really "hot" on the tape (high in level) and thus Dolby would have no effect upon it.

In fact there were some music albums on cassette tape which contained a "bonus track" of computer program data at the end, and the whole tape was Dolby NR encoded, including the data.
 
krebizfan said:
ECMA 34 defines a streaming tape format. The notch on one side of the cassette is a giveaway. The format should be around 250 KB in practice with a theoretical almost 3 million bits available on a full length tape but about 1/3 of that would be devoted to gaps between blocks. If I did the math right, the 32 bpmm would be about 50% more than the 1500 bps audio format used by the Apple II.

Reel to reel and other audio tape designs were used for data storage. The problem is that the low frequency signals* used by the super cheap data storage audio formats were what noise reduction systems tried to remove. It was somewhat counterintuitive that the lowest quality tape and recorder worked the best, as long as the tape coating lacked any gaps. Specialized computer cassettes were supposed to make sure there were no dropouts in the much shorter length.

* Low frequency signals were the most likely to survive all the interference that excessively long unshielded cables would receive. The industry was littered with failed companies that had high speed tape designs that worked in the lab but not at the customer. The KC standard duplicating each bit four times to make sure one would be readable turned out to be excessive.
Click to expand...
I've never ever heard of unshielded cables being used for "analogue like" signals between a computer and a tape recorder. (Or rather I'm sure someone at some point in time did that by mistake, and also it would not matter for the signal from a speaker output of a tape recorder towards the computer as the output impedance would be low enough to drive a 4/8 ohm speaker, and thus wouldn't pick up interference within the audio spectrum).

The reason for the cheapest tape recorders tended to work the best with computers is simply that they had the least amount of amplifier stages that might introduce phase shift. I.E. for data storage purposes it's better if the frequency response looks like a sad mouth as long as there is little to no phase shift.
The reason for the cheaper tapes working better was that almost none of the cheap recorders supported anything else than type 1 tapes ("Ferro"/standard tapes, in contrast to type II (Chrome), type III (FerroChrome) and type IV (Metal)). In addition to the frequency response being wrong using the other tape types, it's likely that a recorder only intended for type 1 tapes wouldn't be able to erase existing recordings on the other tape types.
But this doesn't make the cheapest tapes better than the decent or the best type 1 tapes.
I bet that no one ever had any problems using Maxell UD or TDK AD in their cheap recorders.
(Nowadays it seems like a waste to have used such good tapes for data that could had been stored on say the cheaper Maxell UL/UR or TDK D tapes).
 
MiaM said:
... The reason for the cheaper tapes working better was that almost none of the cheap recorders supported anything else than type 1 tapes ("Ferro"/standard tapes, in contrast to type II (Chrome), type III (FerroChrome) and type IV (Metal)). In addition to the frequency response being wrong using the other tape types, it's likely that a recorder only intended for type 1 tapes wouldn't be able to erase existing recordings on the other tape types.
But this doesn't make the cheapest tapes better than the decent or the best type 1 tapes.
I bet that no one ever had any problems using Maxell UD or TDK AD in their cheap recorders.
(Nowadays it seems like a waste to have used such good tapes for data that could had been stored on say the cheaper Maxell UL/UR or TDK D tapes).
Click to expand...
I've have a couple of "Data Cassettes" from different manufacturers. Can you perhaps explain the differences among them?

Verbatim R 300 NH "Data Cassette"
Verbatim T 300H "Data Cassette"
Maxell CS-600HD "Back Coated High Density Data Cassette"

I also have this one; unsure what tape drive would use it:

Verbatim DC 30 "Data Cassette"
(rear label states "Do not record data on this side; Side B has a 1600 FCI pre-recorded clock track for tape speed control")
 
Sorry, I'm not familiar with those.
In general I would think that if the cassettes have that special notch they are "special" data cassettes, and the last one is obviously special.
There were also "data" cassettes that were regular type 1 audio cassettes, with a low enough dropout rate, and the selling argument was that they were short in order to not have to wind/rewind the tape so much to find a specific program.
(This reminds me that IIRC I used a lot of short promotion/demonstration tapes for data back in the VIC 20 days. I.E. tapes that would had been supplied with a cassette recorder back in the days when those were so rare that a customer might forget to buy tapes, or tapes with a few songs that were sent out to record shops, typically songs that wouldn't end up selling as well as the record companies would had hoped for. Or in some cases demonstration tapes for demonstrating specific tape models. While on this tangent, later on I did this with VHS tapes too. If only VHS tapes had been cheaper back then I would had had one more Activision promotion/advertisement tape (or still had had the content)).
 
pbirkel said:
Yes, they all have that slightly off-center full-thickness square notch on the edge.
Click to expand...
Ah, I see that ECMA-34 identifies that as "Asymmetrical Slot" in Figure 1.

"5.1.3 The back surface is provided with one slot slightly off-center (see Figure 1).
This slot makes it possible to distinguish between the sides A and B, both visually and mechanically."

Thank you @bsdphk for pointing out that standard.
 
You must log in or register to reply here.
Back
Top