Sunday, March 08, 2009

Let me tell you what I've done. (part 2)

Hooboy, this is rapidly growing into a memoir. Fine. It's just going to take a lot more parts than I originally thought.

In Part 1 I covered my work history from my High School days until I got a job offer at Specialty Records in the Spring of 1992. In Part 2 I will cover my time at Specialty Records from my hiring through my years as a Statistical Process Control Coordinator.

The job that I was offered at Specialty Records in the Spring of 1992 was nothing spectacular: I would be working as a plater in the Plating department of CD Pre-Production.

I need to explain a little bit here, which is a problem from this point forward in my job list: you probably have no idea what I am talking about, and in order to make you understand, I have to explain several levels down. It's very difficult to do this in a conversation without running the risk of having the person you're explaining it to zone out, glaze over, lose interest. But explanations are important, or you might get the wrong impression of what these jobs are.

In CD Pre-production a "source" is provided by the recording studio to the manufacturer. At the time that I was working in that department, the sources were sent in on a magnetic medium known as a 1630 tape, though today sources may be CD-R's or (as is the standard now) digital files. This source was checked by our Pre-Mastering department and then sent on to CD Mastering. The Mastering department would transfer the information on the source onto a recordable master medium, using one of two methods. This is the point where the information taken from the tape is converted into the "pits" and "lands" that represent ones and zeroes on a compact disc. The form that the recorded master takes is a ghostly image of a CD on a semi-transparent material on a glass disc about twelve inches across and a quarter of an inch thick. Once this recording process is complete, a thin layer of metal is deposited onto the surface, adhering to the walls of the microscopic pits and lands that have been recorded. At this point, the Glass Master* is ready to be delivered to Plating.

In Plating the Glass Master is carefully prepped for the Plating - or electroforming - process. This process is really quite simple, and probably something you did in your high school Chemistry lab. The Glass Master is loaded into a workholder with a nickel "contact ring" around the outer edge. The workholder is mounted on a rotating shaft so the whole thing spins like a wheel seen face on. The workholder is then lowered into a solution of nickel sulfamate, positioned so the face is a fixed distance several inches from a basket of nickel pellets. An electropotential difference is set up across the solution, with the workholder and Master having one charge and the basket of nickel pellets having the opposite charge. Nickel ions from the solution migrate onto the molecules-thick layer of nickel on the face of the Glass Master and adhere; more nickel ions migrate out of the basket and dissolve into the solution. When the process is completed, you have "plated" the Glass Master with a layer of nickel a fraction of a millimeter thick. After carefully peeling this plated piece from the glass, you have a "Father" - the inverse image of the pit-and-land structure that was recorded at Mastering, as well as a blank piece of glass with some residual recording medium on it.

From this Father you now make a second-generation plated part - a "Mother". The trick here is you need to set up an electrochemical barrier on the surface of the Father before you make the next generation, or else you will have what is known as a "weld", or a "hubcap." From this Mother you make many inverse images, structurally identical to the Father, called "Stampers". These stampers are then trimmed to size and otherwise prepared for the Molding process, and then sent off to make CDs.

It was an interesting job. Turns out I'm allergic to nickel, as are many people; the allergy expressed itself as an outcrop of little red pimples on the underside of my forearms. I wore a lot of teal shirts so they wouldn't show the nickel sulfamate stains. Nickel sulfamate is now regarded with a lot more caution than it was in those long-ago days. Also, the edges of the plated parts were razor-sharp and less-than-razor-thin; everybody got cut at some point, but the trick was not to lacerate yourself too badly - or cut a tendon, or lose one or more fingers.

During my interview, the person who would be my boss's boss confided to me that the company was looking to hire college graduates with technical backgrounds for this job that barely required a high school diploma because they had something special in mind, something that would be coming down the road in a few months.

I started in May of 1992. I guess it was around December of 1992 that a posting went up for a new job in the Pre-Production department that required a technical background and a college degree. But what the heck was an "SPC Coordinator"?

I would find out. I put in for the job, and got it. I think my shift supervisor was angry that I had bid out of his department.

Statistical Process Control, or SPC, was a "new" way of looking at production processes that had first been developed in the early twentieth century. It involves data-based decision making and an advanced understanding of the nature of variation. All variation can be divided into two types: that which is a normal part of the system, and that which comes from outside the normal functioning of the system. Both should be minimized, but each must be dealt with in a specific way. The fundamental mistake of any operation is to treat the normal variation of the system - "Common Cause Variation" - like the variation coming from outside the normal operation of the system - "Assignable Cause Variation," and vice-versa. The goal of the SPC Coordinator would be to analyze the operations throughout Pre-Production, which included Pre-Mastering, Mastering, and Plating, apply SPC analysis techniques to these processes to identify Common Cause and Assignable Cause Variation, and to teach all employees of the Pre-Production department to use these techniques to improve the processes on their own.

Of course, this wasn't just something I picked up on my own. I went to Knoxville, Tennessee for a four-day seminar conducted by Dr. Donald J. Wheeler, covering not just the principles of Statistical Process Control but also the underlying Quality philosophy of W. Edwards Deming. I was accompanied by five other people from our company: the other person hired as an SPC Coordinator, the man who would be my supervisor, the head of CD Engineering, the head of CD QC, and the manager of all of Pre-Production.

He was the key figure here. He was the son of the President of the company and grandson of its founder, but that didn't mean he didn't earn his keep. He was a rough-and-tumble character, overbearing, demanding, imposing, loud, smart, all business but with an engineering background. He took me under his wing and taught me to stand up to him, so I could give as well as I got - while I would have several levels of management between me and him on the organizational chart, I was also technically on his staff, and he didn't want cringers or yes-men. He was also the visionary, the guy who was driving the Statistical Process Control movement in the company, as well as the one who was actively exploring new technologies even while Compact Discs were still fresh and new. He was also next in line to be President of the company. He was ambitious and determined. I always said he was a man who was driven to succeed at any cost, and he didn't care who he had to drag along into success with him.

So it was with that sort of support that I began my role as SPC Coordinator. My partner and I also had the advantage of having spent seven months in the trenches with the troops. We knew them and they knew us. We knew what they were going through and what their processes were. We had gotten up to our elbows in nickel sulfamate with them, and had worn bloody Band-Aids on our fingers under our gloves alongside them. That was another smart move by the guy in charge, one that - sadly - is not emulated in other contexts, where production workers are issued mandates from engineers and other "gods from above" who have neither experience in the processes nor a personal connection with the operators.

And it went well. We experienced that bizarre phenomenon whose name escapes me at the moment** that is reported by everyone who begins such a program: as soon as implementation starts, there is a disproportionate improvement in process yields, often before anything can actually be done to change the processes. There are several hypotheses to explain this, but I believe the leading one is that operators, realizing that someone finally cares enough about their processes to want to do something about them, independently start addressing all sorts of low-hanging-fruit issues that have been obvious all the while.

Eventually things settled down and we were able to start to distinguish Common Cause and Assignable Cause Variation, as well as process-to-process differences. My partner and I were given several assistants who we carefully groomed in the methods of SPC. After a year or so, after we had done one-on-one training with every member of the Pre-Production department, we designed our own SPC Seminar built around Don Wheeler's Understanding Statistical Process Control. Dozens of managers and engineers from throughout the facility went though our course, and came out with a basic introductory knowledge of SPC as well as a copy of Wheeler's book.

But things change. Things always change.

*Not to be confused with a "Recording Master", which we always assume is locked in a vault somewhere at the studio. This is sometimes not the case.
**As dee points out in the comments, this is the "Hawthorne Effect."

NEXT: Things change. Change is not always good.


marcoshark said...

DB, I am enjoying this. Your description of SPC is very clear and easy to understand. I don't think the new "gods from above" have a real clue as to your real talents.

dee said...

I think the phenomenon you're describing is the Hawthorne Effect. If I recall correctly from my Stat classes, it's that individuals alter their behavior when they know they're being studied, or in your case, a new process is being implemented, even though the intervention/new process may not have been designed to elicit a particular change in behavior. That's the simple explanation, but as I said I'm remembering from Stat class and those were a while ago...