I've removed advertising from most of this site and will eventually clean up the few pages where it remains.
While not terribly expensive to maintain, this does cost me something. If I don't get enough donations to cover that expense, I will be shutting the site down in early 2020.
If you found something useful today, please consider a small donation.
The first four parts described the basic of magnetic tape theory, and the very basic QIC tape format - which was nothing more than an extension of the first reel-reel tape recorder designs.
This section will talk about the evolution of tape backup systems for small computers, and discuss the fundamentals of rotating head tape recorders. In the next segment I'll try to go into more detail on the difference between each of the current types of data backup systems.
The QIC tape format brought tape backups into the small machines. Prior to that you had the large tape transports you see in old movies with 10" reels of 1/2" tape. These were just extension of the reel-reel design with rather sophisticated tape handling mechanisms for high-speed movement and extreme acceleration movements - using air-bearings and vacuum chambers. They were not cheap, but disk drives and their removable packs were very expensive too.
In these system, instead of placing the bits in a serial stream and then winding that stream across a tape as it changed directions as a cartridge tape or reel-to-reel audio does, the bits were placed side-by-side, using heads with 5, 7 or 9 tracks. That meant the bits were in parallel. If you have seen pictures of old paper tape used in teletypes of that era, this was the magnetic equivalent. Thus a byte occupied 1 magnetic domain in linear tape length, but the full width of the tape. This would make it 8 times faster in transferring a byte than it would if the data was in a serial fashion.
But back to the small computers we all know, love, and sometimes hate. As small computer speeds increased and the amount of data increased, we needed to move data to the tape faster or we'd never get any backups finished before it was time to put the machine back in a production mode.
Remember that QIC tapes increased their speed by making the bits smaller and also by moving the tape faster and faster. As with anything the faster you go the harder it is to control.
Well let's digress a bit and see how we got to where we are today by seeing that there are other ways to make the tape/head speed increase other than by moving the tape faster and faster.
In the 1950's television was coming into it's own. However the only method of storing programs was via a kinescope, which was a modified movie camera which photographed a TV tube. It was costly, time consuming as it had to be developed, and there was a large degradation in quality.
There was a concerted push in the industry to be able to record pictures on magnetic film, just as there had been the drive to record audio on magnetic tape a few years earlier, so that performers didn't have to perform once live for the East coast radio audience and then repeat their show for the West coast audience a few hours later.
Audio required a bandwidth of 20 kHz - or KCS - Kilo-Cycles/Second as it was called then. The first 'rule-of-thumb' for audio tape recorders was 1" of tape for KC of bandwidth. If you needed to record to 30kHz - the tape was run at 30IPS - which was the first standard. As things improved 15IPS became the standard quality mode and 20kHz was obtainable even at 7.5IPS. 20kHz audio response was finally obtainable in cassette machines running a 1 7/8 inches per second - at the sacrifice of other sonic components that only the most critical listeners [like myself being an audio recording engineer] would notice.
[Analog recording is dependent on every step in the process. The real advantage of digital is that we only make a blueprint of the original signal - but no matter how battered or beat up the blueprint gets - if it's still readable we can recreate the original signal].
Figuring the 20kHz upper limit at 7.5IPS and multiply that figure by 100 and you see that when the tape is moving at 750IPS we only get a 2MHz bandwidth - not quite enough for video. To get to 4MHz we'd have to move tape at 1500IPS. Anything over 200 IPS is difficult as we have so many physical limitations as I have mentioned before.
The key is the head/tape speed relationship. We could hold the tape still and move the head at 1500IPS and that would give the same effect. That would be harder than moving the tape - but we can split the difference and move both the head and the tape to increase the speed difference between them and stay within manageable physical limits.
In 1956 the engineers at Ampex introduced the first broadcast video recorder - the Mark IV - and Ampex made history. [Ray Dolby - well known for the Dolby Noise Reduction system that bears his name was part of the original group which brought forth the Mark IV].
They developed what is known as the Quadruplex system. They mounted a 2" diameter wheel on the end of a motor shaft. Then they placed four heads 90 degrees apart on edge of the wheel. The capstan spun at 14,400 RPM and the heads ran vertically against a horizontally moving 2" wide band of tape. They would switch one head on as it entered the tape while simultaneously turn off the head that was just leaving the tape area. The did this 4 time for each revolution. [Though the concept it simple the execution of it was quite complex because of the timing problems.]
Take 14,440RPM and divide it by 60 to get revolutions per second, and that gives us 240RPS. We don't write across the full-width of the tape as there is room needed for a control track, guard bands, etc, but it we could you would see that 240 RPS times 2" per head pass times 4 head passes per revolution would yield 1920IPS. That's more than we would need to get the signal we wish to record. [The actual speed of writing on Quad machine is 1508 inches/second.
The next transition was the helical scan where the tape drum and the tape were not at 90 degrees from each other to closer to 30 degrees and we only used two heads. At this angle that the head would travel longitudinally on the tape 3 times further than the height of the tape, and would form the side of a triangle IF the tape were stretched out straight, and not wrapped around the drum.
Since the tape is moving, the path is curved into a shape as if were wrapped around a tube - a helix - which gave rise to the term helical recording. This first appeared in reel-reel commercial units, and then Sony introduced the first tape cartridge / cassette to use this format - the U-Matic - which led to the data cassettes of today.
Only after Beta emerged as the leader after the early video cartridge/cassette systems failed, the Cartrivision, Great Time Machine, and the V-cord, did the helical recording enter the broad consumer market. VHS entered the market the year before the first small consumer computes - TRS80, Commodore Pet, and Apple II made their appearance.
Tape quality improved as did head technology, and the ability to write smaller and smaller domains meant that the tape/head speed kept decreasing. The Quad machines started at 1508IPS, U-Matic dropped that to 404IPS, Beta at 272IPS and VHS as 228 IPS.
Only after the PC gained a foothold and Seagate introduced the $2500 5MB HD did the search for cheaper data storage begin in earnest. No one wanted to have to back that amount of data up on 360K floppies - the high-density standard at that time. There were some early attempts at using VHS tapes - as at $15 each they were much cheaper than the QIC cartridge - also just emerging.
In the video world Technicolor introduced an 8MM video cassette system only to disappear. The 8mm format re-appeared with Kodak and was finally brought to the fore by Sony. The 8MM tape writing speed was even lower at 148IPS.
It was the Sony format that Exabyte adopted and it was the first successful adaption of a video form-factor to data storage. The Exabyte s were targeted to the workstation market - such as Sun. The first HW data compression units in wide use for tape backup were add-on devices added onto the back of the Exabytes.
In the meantime the DAT [Digital Audio Tape] recorder was under development. It had severe teething pains and had several long delays getting to market. The CD was being introduced at this time and the CD manufacturers were extremely worried. The felt they had to get at least 5% of the music market before the DAT was introduced in order to survive because the world had moved from disk to 4-track, then 8-track and cassettes dominated the market.
The consumer had grown so accustomed to audio cassette tapes that typical comments about CDs were "it will never sell because it won't record". Developmental delays and pending legislation because of record company worries [will nothing ever change] delayed the introduction long enough where the CD became the fastest growing consumer market item the world had ever seen - until the DBS satellite dishes and DVD players.
That left the DAT to the professional audio industry and a few people who adapted it to data recording. The first DAT devices recorded only 1.3GB on what were essentially audio tapes. Because of problems with people trying to use audio tapes instead of data tapes - MRS - Media Recognition System - was adopted and the data DATs were renamed DDS as we know it today. Although they look the same the DDS is a far cry from the original DAT. The original audio design assumed a tape would be inserted and played from beginning to end, and never subjected to the back and forth motion seen so often in data. That was one of the weak point in using audio DAT tapes.
The head drum diameters shrank from the almost 3" in beta down to 1.5" in 8MM, and the drum diameter in the DATs is in the vicinity of 1". However while most of the video systems used 1800 RPM drums some of the DAT systems rotate the heads at 4000 RPM.
With HW compression, high head speed, finer magnetic domains, and overall technological advances, the 300MB+ min speed of DAT today make the past look awfully slow, but even those pale in light of the 8MM TAPE RAID arrays that can reach up to 5 times those speeds. I'll save that for a future article.
If you found something useful today, please consider a small donation.
Got something to add? Send me email.
More Articles by Bill Vermillion © 2011-03-09 Bill Vermillion
C++ is just an abomination. Everything is wrong with it in every way. So I really tried to avoid using that as much as I could and do everything in C at Netscape. (Jamie Zawinski)