Retrotechtacular: 1990s CD Mastering Fit For A King

Before it was transformed into an ephemeral stream of ones and zeroes, music used to have a physical form of some kind. From wax cylinders to vinyl discs to tapes of various sizes in different housings and eventually to compact discs, each new medium was marketed as a technological leap over the previous formats, each of which justified incrementally more money to acquire.

But that’s the thing — each purchase resulted in you obtaining a physical item, which had an extensive manufacturing and distribution process behind it. And few artists demanded more manufacturing effort than Michael Jackson in his heyday, as revealed by this in-depth look at the CD manufacturing process for The King of Pop’s release of the HIStory double-disc set in 1995.

The video was produced as sort of a love letter to Michael from the staff and management of the Sony Music disc manufacturing plant in Pittman, New Jersey. The process is shown starting with the arrival of masters to the plant, strangely in the form of U-matic videocassettes; the 3/4″ continuous loop tape was normally used for analog video, but could also be used for recording digital audio. The digital audio is then sent for glass mastering, which is where the actual pits are created on a large glass disc under cleanroom conditions. In fact, much of the production process bears a strong similarity to semiconductor manufacturing, from the need for cleanrooms — although under less stringent conditions than in a fab — to the use of plasma etching, vapor deposition, and metal plating operations.

Once the master stampers are made, things really ramp up in replication. There the stamper discs go into injection molding machines, where hot polycarbonate is forced against the surface under pressure. The copies are aluminized, spin-coated with UV-cure lacquer, and sent on down the line to testing, screen printing, and packaging. Sony hired 40 extra full-time workers, who appear to have handled all the tedious manual tasks like assembling the jewel cases, to handle the extra load of this release.

As cheesy as this thank-you video may be, it was likely produced with good reason. This was a time when a Michael Jackson release was essentially a guarantee of full employment for a large team of workers. The team was able to produce something like 50,000 copies a day, and given that HIStory sold over 20 million copies, that’s a lot of workdays for the good folks at Pittman.

48 thoughts on “Retrotechtacular: 1990s CD Mastering Fit For A King

  1. I did a lot of work with both u-matic and 1” tape, and the audio quality of the Sony u-Magic’s was amazingly good. I didn’t know they used it as the analog mastering medium- that’s cool.

    It was not endless loop though.

    1. I think it was PCM sound: pulse code modulation. Every scan line contained data in the spot where normally the actual video data is going. It also had smte time code in certain versions. Sony even made a pcm device goimg with their portable betamax recorders. The defacto mastering format for a couple of years.

    2. At some point in the future, somebody will be handed a bunch of these tapes and be tasked with reading them.

      There were millions of U-matic machines built, but relatively few PCM-to-video converters, so they’re going to end up writing some software that looks at the image captured video, reads the little black and white bits, and decodes the audio from it.

      It’s going to be the binary equivalent of playing an old Edison wax cylinder via video camera.

      1. That is, if the master tapes aren’t deliberately destroyed by the big record labels.

        The copyrights will expire eventually so they can’t make money off of the licensing, it costs money to keep the original tapes, and they won’t just hand them over to potential competitors, so what do they do? Burn them.

    3. PAL U-Matic recorders were used because the amount of 16 bit digital audio that could be recorded came out to exactly 44.1K samples per second. NTSC format would have allowed a higher data rate but at some odd number. Less data per frame but at 29.97 frames per second VS 25 was more data per second.

      Using the higher rate would have made the run time of an audio CD less and for various reasons the companies developing the CD wanted a Compact Disc to hold 74 minutes, so 44.1K was good enough.

      Some people will promote various other reasons why for 44.1K but they’re all BS. It was 100% due to the video recorder chosen. If somehow it had come out to exactly 44.2K or just 44K then that’s what would have been used.

      1. These early digital recording systems encoded the digital audio as a video picture that could be recorded on a video tape.
        PAL umatic tapes were used for PCMF1, a 44.1kHz domestic digital audio format (originally designed to work with Betamax recorder) in Europe, the US etc used an NTSC version of PCMF1, both versions of the digital interface could replay the other.
        Electronic editing of PCMF1 on unmarked was possible, however the edit did not always work musically due to the fact it was a video edit so +or- 1/50th sec.
        Professional PCM1610 & the later compatible PCM1630 system used 30 frame per second B&W umatic recorders, not the 29.97fps NTSC machines used for colour video recording.
        1960 etc used a separate digital editing processor that enabled proper musical editing of the audio.
        The 70 min iCD limit was due to the longest tape cassettes available, later an 80 min tape was produce but was les robust due to using thinner tape.

      2. Both PAL and NTSC have roughly the same potential data rate, about 14,400-14,500 usable scan lines/sec, with similar luminance bandwidth.

        If you figure that you could encode 180 bits per line in the luma channel with reasonable s/n margins that gives you a usable bitrate of about 2MB/s. The bandwidth of 2 channels of 44.1K, 16 bit audio is 1.4Mb, so that’s a reasonable reserve for framing and (probably) some kind of RLL encoding, though not as much space for error checking as they probably would have liked.

        My assumption for why they used U-matic in the early days is that some Sony engineer was told go figure out a digital audio tape storage system but no, you can’t have a budget for a dedicated transport. U-matic was ubiquitous, already a Sony product, and had enough bandwidth to store the data, if it were packed correctly.

    4. An endless loop covering a whole music album would be what ? a HUGE box with thousand of pulleys in it ?
      I know the bloggers here hate fact-checking their articles, but do they even understand what they write ?

      Why is it SO difficult to put up editorial lines and improve the quality – quantity is well enough – of the articles here, Mr Williams ?

      1. 8 track cartridges were loops containing a whole album; they relied on a tape with a dry lubricant backing to be able to pull the tape from the center of a spiral of tape. See the wikipedia article for a photo.

      2. That’s literally how they made duplication masters for cassette tapes back in the day.

        All four tracks (two of them in reverse) were copied to a big endless loop of 1″ tape, and this was then copied over and over, at many times normal speed, onto to a long, long, long, reel of 4mm tape that would later be cut and loaded into individual cassettes.

        The master tape didn’t use rollers, it sat in a big tray, on edge. It naturally formed, big, meandering loops, and it’s own bending stiffness was enough to support it.

        It was a cool process to make cassettes at an industrial scale. There are several good videos on YouTube.

    5. Interesting – we used DLT tapes I think (certainly for blue ray, and I think for CD before that?), though eventually Aspera (a UDP-based file transfer protocol) replaced it.

      1. DLT tape came later, after U-Matic, DAT and CD-R (the last two being horrible as masters, but explain that to musicians wanting CD’s pressed from their demo recordings).

  2. I had access to a CD burner in…1994? As I recall, the CD-Rs were like $50/ea, the burner itself was probably 4’x2’x2′ and was connected to the computer via SCSI and only wrote in 1x. The computer had to have two hard drives RAID0 so you could put the image on it before sending to the CD-R. We had a success rate of maybe 50%?

    1. The good ol days right? I remember when me and my friend got burners (late 90s), of course he had his earlier than me because computer nerd and all, but it was a great time.

      I just remember burning everything to a CD to save it for posterity.

      1. Indeed the good old days, likely 95 or 96, CD-Rs about £10 a pop, maybe 75% success rate. My partnership was I had the burner, friends supplied me copy-protected games on CDs, I cracked them & burned them, sold them to other friends, friends got a cut of the profits from their game :)

    2. I remember buying my first CD-R blank in the early ’90s – it was almost $100 for a single item. Really cool looking though; shiny gold on one side and iridescent emerald green on the other.

      I used it to archive a stack of 9-track magtapes, which I figured back then would not be a thing for much longer.

        1. Nope. The drive didn’t have buffer memory and the operating system wasn’t pre-emptive multitasking, and the hard drive was PIO instead of DMA, so at worst moving the mouse cursor during the 1+ hour burn could cause a sudden sustained CPU load and halt all data transfer, then a buffer underrun and a coaster.

    3. Yes, the joys of starting a burn… leaving the computer _alone_ (because any sort of other system activity upset the burner application)… and if you were lucky, you had a burned disc with your data.

      The burner we used was a 8× read, 4× write (did not re-write). SCSI interface. The machine we ran it on was a Dual Pentium PRO 200MHz with SCSI disks and Windows NT Workstation 4.0.

      “Buffer underrun” was a common failure mode. We were never without drink coasters.

      1. I still have the very first CD-R I burned. Gold backed with dark blue dye so it appears dark green. I haven’t checked it in some years to see if it’s still readable. Probably is. It was even readable on the old Mitsumi 1x CD-ROM, the model where you push on the front and the entire inside pops out then lifts a clamshell top. Supposedly that drive wasn’t capable of reading CD-R. Probably couldn’t read the aluminum backed, light blue dye discs.

        BURN-Proof and other buffer underrun prevention systems were very much welcomed, along with the much larger RAM buffers in the drives. That all became less of an issue when PCs got fast enough to throw data at the burner quickly enough, along with multitasking getting better.

        What would reliably coaster discs is a subwoofer in the cabinet the PC tower was sitting on. A friend griped about how his computer burned coasters so often, until I was at his place when he was trying to burn a disc while playing some thumpa-thumpa music. I pointed out that the problem was likely vibrations from the speaker exceeding the tracking correction capability of the burner. D’oh! From then on he didn’t play tunes while burning tunes, audiobooks, or data. ;)

        1. Oh yeah, it was sensitive alright, but there were tricks to it. My first burner was a 2x Hewlett-Packard unit, and I couldn’t afford high-performance SCSI HDDs, I just had my simple little UDMA-33 PATA disk.

          System tuning was everything. Defrag the drive first. Disable the screensaver. Slay all background tasks. Move the keyboard to another desk for seismic reasons. Walk like there’s a quiche in the oven. Don’t slam any doors.

          For the next 35 minutes, be absolutely perfect.

      2. That was the fault of Windows memory management, not the burners. If you edited system.ini and set MaxFileCache to a sensible value you could play DOOM while you were burning a CD.

        1. I was going to say I burned plenty of CDs while doing various tasks in the mid to late 90s although using Linux and my success rate was high.

          Later media quality decreased and errors increased.

        2. Yep, it must have had something to do with Windows, because people with 50MHz Amigas were burning CDs too, with no greater hassle than the windows users with 4x the clock speed and RAM.

    1. In an earlier post I ccommented on working around cleanrooms. The video is most likely correct for the 90’s. Modern cleanrooms you wear what is called a bunny suit. Full body suit with a built in hood but yeah the masks are right. If you look at the hands in the video you will see the requisite cotton gloves with latex gloves on top pulled up over the sleeves. I’m not sure what class this cleanroom is but the aparell is also class dependent.

        1. It was normal for the workers to wear the mask below the nose. It does not meet protocol but no one seemed to call them out on it. In the cleanrooms you are also not supposed to wear deoderant or cologne/perfume but quite a few would pass on that as well.

    1. 110V plug is most definitely retro… the rest of us moved to 230V 50Hz years ago. :-P

      DVD is retro in the sense that devices to play it are becoming harder to come by. Most laptops eschew any kind of optical storage device, and while you can get Blu-ray players for your lounge room, I think they are on borrowed time because rightly or wrongly, we’re being shepherded towards streaming whether “we” want it or not.

    2. First of all, DVDs are absolutely a retro tech at this point. Even if you totally ignore streaming (hint: you shouldn’t, since that’s how people consume video now), it’s still a generation behind Blu-Ray, which itself was released to the public over 15 years ago.

      That said, music isn’t generally released on DVDs. This is a video about CDs, a technology so antiquated that apparently you didn’t even know existed until this point.

  3. Re: Betamax audio – circa 1980s I was mildly an audio dork. I recorded my vinyl to cassettes for daily use to preserve the originals, and I spent hours switching back & forth wearing headphones to test the fidelity of my copies. With a decent cassette deck and quality tapes, it was excellent. There was some noise, but that was a trivial compared to the dust and scratches the disks would have accumulated with constant use.

    In the 1990s I discovered that VHS audio was far beyond cassette quality. Not as convenient, but I made “master” copies of my albums and could make cassettes (& later CDs) for portability.

    Good times.

    1. Same. But both VHS and Beta Hi-Fi suffered from a “purring” noise during high-frequency sounds (like a bell wringing). This was never mentioned in discussions of it. My guess is that it was caused by switching between the two audio heads on the spinning head drum.

      1. It’s claimed that the issue was caused by alignment problems rather than the system itself, but having a discontinuous signal does tend to cause aliasing artifacts with high frequency sounds.

        The line frequency of a PAL signal is 15625 Hz so any signal above about 10 kHz would probably get modulated audibly.

    2. NEC made a PCM audio adapter for the V70 and V71 Betamax VCRs to store many hours of digital audio onto a video tape. Dunno what sampling rate was used, 44.1K or 48K.

      NEC was one of the few companies that went with the higher priced Betamax license. The lower cost option required purchasing many of the parts from Sony. NEC didn’t use a single Sony made part in their Beta VCRs. The V70 had around 40 inputs, outputs, indicators, and controls covering the front and rear. It had a light and a mirror that flipped down to visually check the tape remaining through the lightly tinted tape slot flap (for those not trusting of digital counters), and a NiCd backup battery for the clock and program settings so a short power outage wouldn’t make it miss a recording.

      OTOH, the NEC Betamovie camcorder was simply a Sony Betamovie with the Sony logo swapped for the NEC one.

  4. The good old days.. My first CD-R was shipped by HP, and was purchased at Best Buy.
    The date on the driver (On floppy) was one week before the purchase.
    And, try to buy blanks at a good price..
    By the 1990s, I was authoring CDs, both picture files and cute games/programs as shareware
    at the local computer swap meets..
    Then we (In the business) found out about “Bit Rot”.
    So the CD was actually not a “Lifetime Backup Medium”.
    I kinda miss those (Busy) days.

    1. Yeah, when I first got my burner, blanks were in the $5-10 range, but falling like a rock. They seemed to stabilize around the magic $1/disc range and hover there for a loooong time, and became a sort of currency — I remember buying Wendy’s for a friend and him reimbursing me in CD-R media.

      Then came another precipitous tumble to “Free after rebate” loss-leader status. That must’ve been around 2003 because I was active on the coupon sites and I remember pulling some crazy retail hijinks.

      Here’s a scenario: Circuit City and Staples have the same spindle of Verbatim white-label blanks on sale. Circuit City has it for $25 – $15 rebate = $10 AR. Staples has it for $15 – $7 rebate = $8 AR. What do you do?

      Don’t go to Staples, you chump! That’s where the unwashed masses are going. Go to Circuit City, pricematch the Staples price, you’re now paying $15 but getting a $15 rebate, free (except for sales tax) after rebate. Done and done.

      HOWEVER, remember those teeming masses thronging Staples? They’re a problem. Because Staples is going to sell out. And the moment that happens, Circuit City will no longer honor the pricematch. (And yes, they do call and check throughout the morning, so Staples can’t have 5 units in stock as sacrificial bait to cost Circuit City a bunch of money.) So, you’ve got to get to Circuit City before Staples sells out.

      And the problem is that the sales go live on Sunday morning, and Circuit City doesn’t open until 10 on Sunday morning. Staples opens at 9.

      Dun dun dunnnn.

      So, somebody has to volunteer as a ringer. Be in line outside the door at Staples before they open. (Nothing crazy like Black Friday lines, but on weeks with good sales, there’d be maybe a dozen.) Get in there and make a beeline for the CD-Rs, get two or three units in-cart, cover them up with other stuff, and push that cart around the store looking interested in other things. Maybe head over to the furniture section and check out some chairs, you’ve got a little over an hour to kill.

      Now when the rest of the crew visits Circuit City at 10:01, and the manager calls Staples to see if the pricematch is valid, well look at that! Computer says there’s a couple units still in stock! Figured those would’ve sold out by now but guess not, huh.

      Clear out Circuit City, make sure all the receipts look good, get back home and dial the ringer’s pager, send an all-clear code (we liked 4444, because shift-4 is dollar signs, we’re in the money!). The ringer calmly puts the items back on the shelf and leaves Staples, and some Staples customer later in the day is flabberasted to find the hot item inexplicably still in stock after the morning rush. (Get home and whoever made out best in the sale owes the ringer a free spindle for their trouble.)

      We were teenagers with more time than money. OF COURSE we pulled stunts like this. Every weekend, if conditions were right. Hey, I didn’t make the rules, I just play by ’em.

  5. “Hi Michael I’m Janette, this is Mike…” uh uh. no way. did they hire them for that?

    And how did they get young Jonah Hill to do a cameo at 7:03?

  6. Awesome, thank you for sharing this!
    I nominate it for Best “How it’s made” Soundtrack.

    Sadly, I couldn’t help thinking “look at all those folk who lost their jobs to the interwebs.” If I’d’ve seen this twenty years ago, I might’ve bought more of my music.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.