In a room you’re not allowed to go in, inside a machine you’re not allowed to look at, a 30-milliwatt laser beam that you’d barely be able to see anyway is burning infinitesimally small indentations into a spinning plate of glass. The indentations aren’t into the glass, exactly–as the laser switches on and off, the light is actually exposing tiny dots on an extremely thin layer of clear, light-sensitive coating. And they’re not dots, really, they’re more like flattened Good & Plentys of varying lengths.
The interesting thing about this process–the making of a “stamper” for pressing multiple copies of an audio compact disc (or of a video laser disc or even a CD-ROM information-storage disc)–is the combination of speed and precision with which it is accomplished. The light-sensitive coating is a bit more than one micron thick, a micron being one millionth of a meter. The elongated dots that the laser beam exposes vary in length from about 0.8 to roughly 3 microns. There may be half a billion or more of these dots on the surface of a CD; the laser, turning off and on, off and on, inscribes them into the coating at the rate of a million or so dots per second.
The machine is owned by the Digital Audio Disc Corporation; it and its brothers and sisters are housed in one part of a 300,000-square-foot facility in Terre Haute, Indiana. DADC is owned by Sony, the Japanese electronics conglomerate that bought CBS Records in 1989. Terre Haute has long been a center of music manufacturing and distribution. Before being converted and expanded for CD manufacture, this building housed CBS’s main record-pressing plant. (That business–what was left of it–went south to Georgia in 1986.) Not coincidentally, Columbia House, CBS’s huge record club (now mostly a tape and CD club), has been located right across the street since 1956.
The plant sits on the outskirts of Terre Haute, a blue-collar town of about 60,000; it’s a vast, low-slung affair, set back from Fruitridge Avenue by a large parking lot. There’s nothing on the outside to suggest it, but this is one of the two or three largest CD factories in the world, the source of about 30 percent of all the CDs bought in America last year. Inside, a bit more than 700 people work in four-day shifts of 12 hours each; the plant runs 24 hours a day, seven days a week, producing 120 million compact discs a year, nearly 250 per minute.
It wasn’t always running at this rate, of course. When DADC opened, in 1984, its capacity was about 10,000 discs per day; but the company’s growth has paralleled the growth nationally and worldwide of the compact disc. While the tape cassette supplanted the record album in both dollar sales and absolute numbers sometime before 1980, the industry remained focused on the vinyl LP as the flagship provider of recorded music. But after the introduction of the compact disc–1982 in Japan and Europe, 1983 in the U.S.–sales grew almost exponentially: from 6 million discs in 1984 to 100 million in 1987 to more than 300 million in 1990. The long playing record, which now accounts for less than a tenth of the industry’s total sales, is for most practical purposes extinct.
Inside the DADC building there’s a noticeable dichotomy between the extreme high-tech nature of the machines and the low-key manner of the company. It is a factory, after all: Most employees, from management on down, wear blue company vests. The lobby, the cafeteria, the office of president James Frische–all are functional and nice, in that order. Russ Kunz, the head of manufacturing, who directly oversees about 75 percent of the plant’s employees, uses for an office a small open corner of a plain work room just off the manufacturing floor, a space he shares with secretaries and technicians.
The hierarchy is simple: Frische was recently promoted and spends only one day a week at the plant; day-to-day operation is now the responsibility of executive vice president Mike Moran, an extremely affable guy with salt-and-pepper hair and a casual wit. The plant’s manufacturing end is handled by Kunz, who like his two deputies–who oversee replication and packaging respectively–is in his early 30s. One of those deputies is an electrical engineer named Mike Mitchell, a 31-year-old Northfield native who came to DADC seven years ago after a spell with Martin-Marietta. Mitchell’s friendliness and his deep understanding of the plant’s immensely complicated processes combine to make him the designated tour guide for most visitors.
Terre Haute is no one’s idea of paradise, but everyone’s adjusted. Moran’s a fairly major corporate figure in town, and gets his picture in the paper often enough to be teased about it. Kunz hates the bustle of a metropolis, and proudly shows off the gorgeous two-story, 3,500-square-foot house he’s just bought–at a price roughly one-sixth what a similar structure would cost in Lakeview. Mitchell, the youngest of the bunch, misses Chicago, where he grew up, but likes the people in Terre Haute and likes his job so much he’d never think of leaving. And there are side benefits, too: the CBS store (now the Sony store) sells CDs for $3 a throw, and for management–and dozens of technicians, surpervisors, operators, and engineers–there are those training trips to Japan (or even Salzburg, Austria, home of another Sony plant), sometimes for weeks at a time.
“Making a record or a tape,” says Mitchell, “is easy–you can do it in your garage.” To make records, all you need is a record-pressing machine. It’s not something that most people have in their basements, but it’s not a hugely complex affair. Relatively serious tape duplication isn’t that hard, either: the big plants do it not by “real time” duplication–i.e., having a thousand or so tape machines all duplicating at the same time–but through an ingenious process called “sprinting,” which strips an endless spool of virgin tape against a master at very high speed. The magnetic impulses are transferred directly, which eliminates the problems that normally attend high-speed duplication. (As those with home high-speed dubbing decks know, it exacts a woeful price on frequency response, giving the recorded tape that “distant” sound.)
But making a compact disc is of an altogether different order. A CD, in a way, has very little to do with music–it’s a high-tech information storage and retrieval system much more analogous to computer discs than to records or cassette tapes. Indeed, compact discs would be even more like computer discs if sound weren’t so demanding in terms of storage. For the purposes of CD-ROM–the ROM stands for “read-only memory,” which means you can access the information but not change it–a CD-sized disc can store the Encyclopaedia Britannica. By contrast a music CD, physically identical to a CD-ROM, holds a maximum of about 70 minutes of music.
Almost all the complications of CD manufacturing spring from the need to get that much information onto a disc of manageable size. (One of the prime goals Sony and Philips set in developing the CD was that it be small.) As everyone’s heard by now, “digital” means that the information on a CD is conveyed through a binary computer language of 1s and 0s. That language could just as well be stored in a very long list written out in longhand; programmed back into a computer properly, the information could once again produce music. The problem, of course, is that it would take an awful long time to write out the billions of numbers that would eventually produce, say, Born to Run. (And one suspects that the same process undertaken for, say, Barry Manilow Live would be even longer.)
What they do at DADC, essentially, is cram those billions of bits of information onto a little piece of plastic about four and a half inches across. It is not a simple process.
Anyone can get a CD made at DADC, but if you’re a big record label it’s easier. (A large part of the plant’s volume is for discs on CBS’s Columbia and Epic labels, but these it handles like any other order.) If you can produce your own master tapes and you buy in big quantities, you’ll probably score DADC’s friendliest price, which is less than a dollar per disc, including the pricey plastic “jewel case.” With the rise of digital audio tape as the recording medium of choice for even smaller labels and one-man operations, DADC is seeing a lot of submissions on DAT; the information’s already digitized and that makes for easy and clean transfers. But in theory the company will take anything, on any kind of tape: if you want to make a CD of your kid playing piano, or of your address to the Rotary Club, either of them recorded on a $3 Maxell cassette, DADC will take it. For smaller quantity orders, of course, there are correspondingly higher per-item charges, and the company won’t waive the mastering fees as it does for large orders. Still, they’ll do it. “You can get work out of us for about $5,000,” Mitchell says.
The tapes submitted to the company are kept in a nondescript storeroom crowded with a couple of shelves and every kind of tape machine you can think of–cassette decks, DAT decks, video decks, strange European models. What’s coming up on the manufacturing agenda waits casually on the shelves: a new Pat Metheny album on high-quality tape from Warner next to submissions from small labels; high-tech tapes for CD-ROMs (these and video lasser discs are also made here) cheek-by-jowl with vanity discs from Pepsi and Disney World.
The first thing DADC does with this material is “premastering,” usually a three-step process that assembles all the information to be embedded on a disc. First the tape is listened to for any sound problems. Second, it’s run through a machine called a 1630 processor, which translates the sound into digital numbers, then subjects them to a hugely complicated process called the Cross-Interleave Reed-Solomon Code, which provides the basis of the CD’s formidable error-correction capability. (More on this later.) The third step is adding something called a subcode.
The premastering is done in a warm and friendly all-wood room. Mounted in the wall is a massive speaker system, in front of which is a complex array of screens, keyboards, and a tape editor. The premastering supervisor is John Macdonald, a young audio engineer with a long ponytail. He’s working on a submission from a private party–three DATs that will make up a 20-song CD sampler of alternative bands. Before him is a detailed sheet that the producer has submitted; it includes a song-sequence list, the times for each track, and notes of unusual sounds and effects so Macdonald won’t mistake them for flaws in the tape. “Every tape we do is listened to 100 percent,” Macdonald says. “While we listen, we draw up notes: if we encounter any kinds of problems or questionable situations with the tape that aren’t already approved by a client–clicks, buzzes–then we’ll get in touch with them and give them the opportunity to do something about it.”
The problematic sounds range from distortion on a guitar to the kind of stuff that bands think is neat but that sound engineers just roll their eyes at.
“This client, believe it or not,” Macdonald says, “wants the CD to start with the sound of a guitar plugging in.”
He rewinds the tape: you can hear a hum from an amp and a loud “pop!” before a guitar strum leads off the song.
He shakes his head. “I would just make it start with the music, but they have a specific note that said leave the sound of the guitar plugging in.”
Since this sampler album has been submitted on DAT, the music is already digitized. Macdonald’s next step is to subject the digital info to an encoding process that will help CD players correct errors. This is a unique aspect of CDs; on a record or a tape, what gets played is literally what gets played, if you get the point. But since a CD is just a readout of binary information, there’s the possibility of compensating, in part, for potential errors. It’s lucky there is, because the tiny scale on which CD information is conveyed makes the technology extremely susceptible to screwups.
Time out for a little CD background.
If you’ve heard or read anything about digital sound, you’ve probably heard that there is–in theory and for all intents and purposes in practice–no loss of sound quality in recording or reproduction. In the analog world–records and conventional tape–there always is. “Every time you duplicate it you’re going to lose something,” says Mitchell at lunch one day. “It’s like if I had five glasses here, and this one was full of water and I poured it all into the second glass, and then into the next one, one by one, by the time I get down here I’ve lost a little of the water, to dropping it and evaporation, and the residue in the glasses.”
To put it another way: You might think a xerox machine makes good copies–until you make a xerox of the xerox, and then a xerox of that. With each step a little bit of quality is lost, and eventually the small imperfections add up.
In analog recording there are many steps. Barry Manilow sings into a microphone, the diaphragm shakes in relation to a magnet, and the shakes are converted into electrical signals. That’s one. The signals go through a tape head, which magnetizes particles on a strip of plastic. That’s two. Another tape head converts the magnetic patterns on the plastic back into electrical signals. That’s three. The signals shake a stylus, which makes indentations on a master disc. That’s four, and we haven’t even got the sound onto vinyl yet, let alone onto your turntable and out of your speakers.
Digital recording also involves many steps, but they’re not as destructive. Digital recording doesn’t represent sounds by physical analogy–by shakes or wobbles or magnetic patterns or whatever. Instead it sort of reconstructs sounds. A digital recorder takes periodic readings of Barry Manilow’s voice and stores them in the form of 16-digit binary numbers. This process has imperfections of its own, of course. Using an 18- or 20-digit number would give you more precision, for one thing. Also, because the reading is periodic it can only approximate the continuous wave created by Barry’s warble–it gives you a series of dots, in effect, that your ear has to connect. But it gives you 44,100 dots per second, and your ear does connect them without noticing that anything is missing. (Movies work the same way, of course, and even TV screens do something similar.) The big advantage is that at most steps of the recording and duplicating process the only thing that’s being recorded is numbers, so there’s no loss of quality from one step to the next.
Unless you get the numbers screwed up.
Which brings us back to the Cross-Interleave Reed-Solomon Code–the CD’s error-correction system. A compact disc is similar to a vinyl record in this respect: they both consist of a single spiral of information which is read by a needle or a laser. (A CD, incidentally, is read from the inside out, the opposite of a record.)
That spiral track of information is very small on a CD. A traditional album side, 20 minutes long at 33 rpm, has a spiral that winds around the record about 650 times; stretched out, the groove would be about the length of a football field. A CD groove, by contrast, is about half a micron wide. (Human hair is about 40 times as thick.) It would stretch out to about three miles.
The problem with that tininess is the number of things that can go wrong on that scale. The most minuscule particle embedded on those dots, the most microscopic flaw in the disc itself, any scratches on the outside of it–all of these things can disrupt the information flow, by eliminating not just a few but thousands and thousands of “bits.”
Such disruptions are inevitable, and in fact appear by the dozens on any CD: the CD has to compensate. That’s what the error encoding does. It actually does two things: it gives every 16-bit word (actually it’s two 8-bit words, now; to make things easier, they’re split in half) something called an “error code.” That’s the Reed-Solomon part. And it rearranges the data into a nonsequential order–that’s the Cross-Interleave.
The error code, attached to each 8-digit number, is simply another number generated by a strange mathematical formula; think of it as sort of a social security number, another way of naming the same thing. As your CD player courses over the disc, a microprocessor inside constantly checks the names against the social security numbers; if it finds a name that has no match, it knows something’s wrong. In a lot of cases, the microprocessor can infer from the error code what exactly is wrong and summarily fix it.
(Math junkies interested in the codes or anything else about CDs are advised to check out Ken Pohlmann’s Principles of Digital Audio.)
Sometimes, though, the problem is more complicated. This is where the Cross-Interleave comes in. When a CD player encounters a large mistake–like, say, a void where there should be information because of a flaw on the surface of the disc–its circuitry tries to reconstruct the sound that should be there, based on interpolations of what goes before and after. This would be impossible if the numbers were stored in sequence on the spiral of information, because if one were damaged chances are the surrounding ones would be damaged as well. So the Cross-Interleave code breaks the data up into “frames” and within each frame scatters the numbers around a bit, in effect putting one in a shoebox, the next under the flowerpot, the next beneath the bed. The CD player, which knows the distribution pattern, skips around each frame and reassembles the data in its proper order. So if, say, number 384 is lost to a speck of dust, numbers 383 and 385 are probably still accessible, and by interpolating between the two–sound can’t change that much in a forty-four-thousandth of a second–the processor can make a passable repair.
Once the original tape has been subjected to all this encoding, John Macdonald adds something called the subcode–basically, it’s the timing fuction that allows the CD player to tell you how long the song is, how many seconds have elapsed, and so forth. This process is somewhat analogous to stripping a ruler alongside the music–a ruler that measures out time in thirtieths of a second. The time coding also makes possible the random-access feature of a CD player–it can find the beginning of any song on a disc within a second or two.
This information, too, gets sent along into the digital stream–according, that is, to the client’s wishes, for the codes are readily manipulable. For the CD version of Led Zeppelin II, the subcode programmer left no time at all between the songs “Heartbreaker” and “Living Loving Maid,” preserving the album’s beloved instantaneous segue. Prince’s Lovesexy album seems to have left out the song-division subcode entirely: actually, it’s simply programmed to tell you that the album consists of one song, 45 minutes and 3 seconds long. More recently, Minneapolis’s Gear Daddies “hid” a tune on their most recent album, Billy’s Live Bait; because of the way the subcode was programmed, your CD player thinks “Zamboni” is part of the last cut. And then there was a John Cage album whose title Macdonald can’t quite recall. “He said to put in a subcode every ten minutes whether there was a break in the music or not.”
After premastering, the music ceases to exist in the factory. From here on in there is only data, and the physical process of embedding it onto a disc.
The mastering process is the most sensitive part of the sequence. Here is the room that you’re not allowed in. DADC obsesses about dust and microscopic particles; workers change into plant-only shoes when they enter each day, and visitors must wear blue paper booties to reduce the impact of outside debris. The mastering room, when the actual “stamper” is being made, is the most sensitive area of all. The mastering room is called a “class 1,000” room, meaning that there are about 1,000 particles a micron wide floating about in any million-part sample of air. This, they say, is very clean. Additionally, most of the machines are under special hoods whose air is filtered to a level of class 10–or no more than ten particles in a million parts of air.
Another big concern in the mastering rooms is vibration; the laser-cutting machines are anchored to foundations a dozen feet underground, and isolated by air pockets from the rest of the building.
The few people who are allowed inside wear protective clothing. Female employees can’t wear makeup.
“It’s not pleasant,” grins Mitchell. “You have to spend 12 hours in those outfits and you come out looking like you just woke up.”
But it’s necessary. “On the CD’s surface,” says Mitchell, “you’re talking about a beam spot of about 1.5 microns; it doesn’t take anything very big to just wipe out its signal. In this area, everything is exposed and just waiting to accept contamination.”
In the mastering section–you can watch suited employees moving about through orange-tinted glass–the first step is taking a piece of high-tolerance glass and cleaning it through a variety of chemical and ultrasonic processes, making sure the surface is exactly flat and perfectly clean.
The glass is treated with a binder to make sure every last bit of that light-sensitive coating–it’s called “photo-resist”–sticks. Then the photo-resist is put on in liquid form, cured with nitrogen, and left to stabilize for at least four hours. When it’s ready, the disc is put on the machine that actually fires the laser. It’s linked to a tape deck containing the premastered, subcoded, equalized, Cross-Interleave Reed-Solomon-encoded tape. An operator hits the play switch, and the “cutting” begins.
The laser beam doesn’t really turn on and off a million times per second; instead it’s interrupted by a lens a million times per second, in accordance with the 1s and 0s on the tape. After the cutting is done, the disc is run through a developer, and the exposed areas are washed away. (That’s why the glass was treated with a binder before the photo-resist was applied–you can’t have bits of the photo-resist falling off of their own accord.) The washed-away areas leave indentations about one micron in depth. These are the elongated dots, called pits.
The pits are the key to the way your CD player reads the surface of the disc. Popular perception is that the surface of the disc is a series of “bumps,” which the CD player reads as 1s or 0s. Actually the surface area of the pits–and the surface area of the corresponding “lands”–are both read by the player as one or more 0s, depending on their length. But the transitions from pit to land are the key; these the player reads–via a change in the light reflected back up into the player–as 1s.
The next step is chemical plating–Mitchell says the precise chemical is a proprietary secret–which gives the design etched onto the glass enough strength to undergo a nickel electroplating bath. After that bath, the nickel plate is physically pulled off the glass by a technician. “It’s hairy the first time or two you do it; you’re thinking, ‘This took over five hours to make,'” says Mitchell. “But you go around the edge with a little knife and then get your fingers underneath it, and it comes right off.” Now you have a mirror image of the disc imbedded in nickel–a stamper.
But before you start stamping you might as well make a few more. “They’re digital,” says Mitchell, “so it doesn’t matter how many times we transfer it. So we make another metal part from our first-generation copy, and make another number of metal parts–mirror images again–from it. So if we have a Michael Jackson, we can take 20 of these stampers in just two more steps, put them on 20 machines, and get them out of here and onto the shelves.”
The tour now moves on to the actual manufacturing room. It’s another clean room, but one with a less stringent standard–it’s technically a class 10,000, but it’s actually at about 3,000 parts per million–and visitors are allowed, if they’re properly suited up.
Required attire is the company’s one-piece jumpsuit: you slide into the suit, pull on boots over your shoes, put on large plastic goggles, and pull a white paper hood over your head. Then you have to walk into an air lock, on the floor of which is a sheet of tacky blue paper that removes the debris from your feet. This chamber looks like something out of 2001: the walls are covered with dozens of air nozzles, large, industrial-strength versions of the air blowers above your head in an airplane. Once the door closes on the lock, you’re bombarded with an “air shower” that blows off any dust on your suit and quickly turns over the air in the room.
Only then are you allowed to proceed into one of the cavernous manufacturing rooms, which echo constantly with low booms, silky hisses of injection, buzzers and horns. These rooms are dominated by long, low stamping machines, maybe four feet high and 20 or more feet long, depending on the model. Affixed to each one like fancy barnacles are dozens of meters, gauges, switches, lights, lamps, tubes, wires, pipes–all riding along the undercarriage, curling over the top, disappearing into the ceiling. Atop the machines are long tubes filled with little clear pellets–these are polycarbonate, a high-grade plastic that makes up most of the CD’s bulk. The pellets are fed into the machine and a large screw compresses them; the pressure melts the stuff down into something approaching a liquid, which is then injected very quickly into a mold, where the stamper is pressed against it. The disc is immediately cooled through cold-water “jackets” and the center hole is punched out. Then the disc is lifted with robot arms–they grab it with a little cluster of vacuum suctions in the less-sensitive center area–and stacked on a spindle.
The polycarbonate disc is not yet playable–the CD player needs a reflective coating for the laser to bounce off. But the polycarbonate is more than just a medium for this coating; it plays a key part in both the mechanics of a CD and its resilience. The laser that is beamed onto a CD by a player is actually, in relative terms, quite large–it’s much bigger than the pits it’s supposed to be reading. But the clear plastic acts like a lens; when the beam hits the outside of the CD, the coating focuses it down to one thousandth of the size it was when it came out of the laser.
This is why so many of the marks on the outside of a CD–from fingerprints and smudges to cat hairs and even some gouges in the plastic–don’t affect the CD player: the flaws are “out of focus” to the important part of the laser and don’t affect its workings.
The reflective coating is applied in a “vacuum evaporator,” a huge dishwasherlike contraption where the discs are placed along with a few pellets of pure aluminum. A series of pumps removes almost all the air inside. Then “we hit it with a couple thousand amps,” says Mitchell, “and the aluminum turns into a liquid. We hit it a little harder, and voom! the aluminum just vaporizes, and ends up on the surface of the discs.”
(The vacuum evaporator is on its way out. DADC is beginning to use a new machine that combines stamping and vacuum evaporation; in this one, the stamped plastic discs are grabbed by a robot and put in a little mini-evaporator–a “stutterer,” DADC calls it. The process takes about three seconds.)
At this point the disc is playable–and, indeed, some discs from this lot have been played already. As each lot is put through a stamper, a sample from the front of the batch is immediately plated and inspected for everything from surface flaws to minute problems with pit geometry.
The extensive checking early on in a run is meant to catch major problems before a thousand discs have been stamped–which it does, most of the time. About once a month, Mitchell sighs, someone does the wrong thing, and a stack of worthless CDs is the result. “Just the other day we had someone screw up and print 88 lots [nearly 9,000 CDs] on the wrong disc. We had to eat ’em.”
When Mitchell brought up the company’s quality-control level–he says it’s higher than 95 percent, where it should be–I asked him about the issue of CD deterioration, one of a number of concerns techno buffs have brought up since the advent of the CD age. Many of these concerns–digital sound’s “coldness” and such–have faded in large part as engineers and producers have learned to work with the new technology. But reports of “laser rot”–oxidation of the CD’s aluminum plating–persist, and some audiophiles get seriously concerned. Mitchell isn’t dismissive of the concern, but he says DADC discs won’t have the problem. “A lot of people [in the business] didn’t take that seriously at the beginning,” Mitchell concedes. “But we know what it is that makes CDs degrade: we put a lot of time into it and studied it hard and we found out what causes that to happen. We have absolutely no concern here.”
The plant has five environmental chambers, Mitchell says, that subject the discs to all sorts of maladies: “We can put discs in there for 80 days at high heat and high humidity and they come out just fine. It’s just a question of what to do about it.”
What exactly DADC does do about it isn’t clear; Mitchell hints that the company may be doing something its competitors aren’t: “I don’t know if they’ve figured out the mechanism yet. On the surface it looks like one thing, but when you get through that and you’re still having the same types of problems, you have to look elsewhere,” he says elliptically.
In the labeling section, a sophisticated loading and electronic silk-screening setup can churn out one- and four-color labels at the rate of one per second; here, for the first time, workers can see exactly what CDs the plant has been producing. (Keeping the secret as the discs move through the factory helps to minimize pilferage; so does the cut-rate employee disc store.) Today Christian pop star Amy Grant is having her new album printed simultaneously, if not congruously, with an older catalog item, the Beastie Boys’ 1986 debut, Licensed to Ill. Also strewn about the room are giant racks stacked with Earth Wind & Fire’s I Am, the Rolling Stones’ Black and Blue, a goofy record called Television’s Greatest Hits, Tesla’s Five Man Acoustical Jam, and the new Bob Dylan three-CD Bootleg Series retrospective.
These last two albums won’t even be available on good old vinyl, at least in the U.S. At DADC the original charter of the building–pressing phonograph records–is hardly remembered. Kunz gave all of his LPs away; Mitchell hardly plays his. From the engineers’ point of view, the records’ usefulness is just a function of their capabilities, which are so obviously subpar, and their future, which was limited from the outset. “Remember,” says Mitchell, “that as time went on they printed about 750 million copies of Abbey Road. If you compare when it was first done, when you had a virgin record of the very first master, and one made ten years later, there’s going to be a big difference. You just keep losing the music.”
For information on Terre Haute, see the Visitors’ Guide in this issue.
Art accompanying story in printed newspaper (not available in this archive): photos/Mike Tappin.