Revolutions Per Minute

How Records are Made, Part 3: Quality Control

The telephone rings. “We have a few pallets for you, can we deliver tomorrow morning?”

The next morning, an 18-ton truck with a hydraulic tail lift arrives. It is July and hot enough to fry an egg with no additional heat source. Several pallets are unloaded into the yard, each full of several stacked boxes, all wrapped up together. Each box contains several smaller boxes and each one of these contains 20 vinyl records. People race back and forth bringing the boxes into a climate controlled warehouse before they get too hot. The last pallet is short, only a single row of boxes. One box in the corner looks like someone dropped a washing machine on it, which they probably did. The damage is noted. Some destroyed record sleeves are visible through the torn cardboard. Some quick photographs are shot for reference, documents are signed, the truck leaves.

This is a lot of several thousand, hopefully identical records, which have just arrived from the pressing plant, shrink wrapped and almost ready to sell. The test pressings were already approved a few weeks ago. Everything sounded great. So what could possibly go wrong now?

I could write an entire book detailing all the things that could still go wrong. In fact, I am in the process of doing just that, but this is another story. First and all too common, shipping damage. That little washing machine (or was it elevator motors?) incident…Phew, it only hit the edge here! Only two of the smaller boxes with visible damage. That’s just 40 records to write off, if we are lucky.

Then we start unpacking. Is this actually the correct artwork? Usually it is, but every now and again…Is it properly printed? This is a can of worms for the graphics design department to deal with. Just like the audio people and the test pressings they approved, the designers received a print proof, weeks in advance of the actual printing. Printing, cutting, folding and gluing record sleeves is full of technical challenges. It does happen occasionally that records are rejected solely because of sleeve defects. These may even originate prior to printing, such as missing or incorrect information on the original recording.

Time to dig deeper. We have several thousand shrink-wrapped records. For obvious reasons, it would be extremely impractical to open, inspect and listen to each and every one of them. Could we just open one and assume all the others would be the same?

As we saw in the two previous columns in Copper #92 and #93, if the mastering and plating have been successful and the test pressings were satisfactory, then we are fortunate to have a set of good mothers and at least one set of good stampers. However, for a run of several thousand records, several sets of stampers will be needed. These need to be grown on the mothers, centered, sanded, formed and mounted on the press molds accurately. With each new set of stampers, major issues could appear. But, even within the span of a single set of stampers, they may get worn or damaged through repeated heating/cooling cycles and high compressive stresses. Defects could still exist in only a small number of records out of 1000 or more, pressed with a single set of stampers. How do we weed them out?

My approach is that the first and foremost course of action is preventative. The better pressing plants are capable of manufacturing a very consistent product and have internal quality control procedures, which in the vast majority of cases, effectively weed out defective records and recycle them long before anything is shipped out. The potentially higher cost of a serious pressing plant is certainly money well-spent, as it prevents much higher and unforeseen expenses down the line and ensures you will receive a product you can actually sell.

But, still, there is no way around proper quality-control procedures, carried out by suitably qualified engineers, at the receiving end. Several records will have to be opened and inspected. These cannot be sold. So how many records do we need to “sacrifice”?

The way to go is a statistical selection of samples, initially only a small number, which will indicate if further samples need to be taken. To begin with, we would take a few samples per pallet, randomly selected, from different boxes on each pallet. It is vital to note down which records came from which box, for our statistical analysis.

These records are our initial samples. They are opened and carefully inspected. Are these actually the correct records? Are the inscriptions on the lead-out area matching the ones on the lacquer masters? This indicates that the correct lacquer disks, mothers and stampers have been used. Are the labels correct and on the correct side? Are they in good shape? Are the edges properly trimmed? Are the center holes well-formed and correctly dimensioned? Are all samples identical in these respects?

Any marks, craters, scratches, blisters, finger prints, dust, hair, insects, plastic or paper residues, or bubble gum visible on the record surfaces?

The next step is to inspect the records on a special turntable with a measurement microscope fitted, for proper centering, flatness and groove structure integrity.

After that comes the critical listening session. The records are placed on a reference turntable, tonearm and cartridge system, calibrated for level and frequency response. The reproducing system must be able to resolve detail and must not introduce any noises of its own. Ideally, the listening session will take place in an acoustically treated room with a very low noise floor of its own, using high quality amplification and accurate full-range monitor loudspeakers. This description is close to the ideal mastering environment and indeed, this part of the quality control assessment often takes place in the mastering facility which cut the masters for that record. This environment does not really resemble the average home listening conditions. It is meant to be far more resolving and revealing to the extent of not necessarily being pleasant to listen to.

The reference turntable system.

It is meant to bring out every little noise and defect, exposing a much “uglier” side of the recording, which would normally remain well concealed on the average home system. If a recording can be made to really sound good on such a revealing system, it will also sound good on any less revealing system as well. Perhaps different, but still free from technical defects. A mastering engineer needs to be able to hear everything a home listener is going to hear, plus a lot more that the home listener may not get to hear. For accurate loudspeakers to really perform, they need an acoustically accurate room to work in. For a low noise signal path to make sense, it needs a low noise room to not mask it with extraneous noises. The mastering environment is a laboratory, not an entertainment space.

It is, in effect, an audio microscope.

In this audio microscope, the records are compared to the test pressings and archived reference cuts, to establish whether any audible differences exist. The same is then repeated through headphones, which provide an entirely different listening perspective. While most world-class mastering engineers only trust their loudspeakers for making decisions regarding dynamics, frequency balance and overall sound character, headphones certainly have their place in quality control assessment. No changes are made to the sound at this stage, we are simply checking if there could be anything there that shouldn’t be there.

Clicks and pops, for instance, shouldn’t be there and are not there on high quality pressings. They are a defect, not an inherent feature of the medium. If such noises are heard, then something needs to be fixed.

While on the calibrated reproducer, the records are also measured objectively by means of laboratory instruments, such as level meters and spectrum analyzers.

If all sample records prove near-identical to each other and to the test pressings, all is good and the records can be sold. If all sample records are near-identical to each other but vastly different from the test pressings and defective, then most likely all records have the same defect and that 18-ton truck comes back to collect them, to be returned as rejects, along with a proper engineering report detailing and justifying the reasons for rejecting.

Sometimes, however, life is not that simple.

While most sample records will be near-identical to each other and to the test pressings, one or two may have a strange defect. In that case, more samples are taken from the same pallet. The aim is to locate the extent of the defect occurrence. It may prove to only affect 3-4 records, usually the ones right next to the original defective sample. These are weeded out and destroyed.

Special attention is required when signs of shipping damage exist. In the case of that washing machine incident (true story), although only about 40 records had obvious damage, another 20 next to them were badly warped, which could only be noticed after opening them one by one and removing the records from the seemingly intact sleeves. This was probably because the washing machine (or submarine parts) which fell on the records when the truck swerved to avoid a suicidal driver, actually stayed there, leaning against the records for another three days of bumpy roads, diesel hum and country music, bending the neighboring records out of shape…!

Both record and sleeve are scrutinized for defects—all the more painstaking when they are as elaborate as this release!

Do all record labels take the quality control of their products as seriously? Considering the stories of major label, high profile artist releases, which made it all the way to the retail stores with one side by an entirely different artist and album, or with songs appearing twice in a row, it is clear that some will not even bother checking the test pressings, let alone do any further quality control on the final product. These are usually the ones who keep the mediocre pressing plants in business.

In the last few years, I have been observing a consistent trend towards significantly higher sales figures for high quality recordings, properly mastered, plated and pressed, with decent quality control procedures throughout, compared to many mediocre products which sit unsold, taking up shelf space.

This is encouraging, as it suggests that it is the record buying public ultimately having the final say on product quality. By choosing to purchase the high quality product instead of the mediocre one and developing a preference towards record labels and artists consistently offering higher quality products, not only are you keeping them in business, but you are making a statement regarding quality standards for commercial releases. This has been a good incentive for record labels and artists to starting to pay more attention to recording and manufacturing quality, seeking out the qualified professionals and producing better records, with all it takes to do so.

To conclude our “How Records are Made” series— it is not easy to make a really good record. It takes skill, experience and patience. But, it can be done and the result can provide a most rewarding listening experience. Seek out some outstanding examples and discover for yourself just how good this 130-year old format can sound!

[Photographs courtesy of Magnetic Fidelity.]