Category: Tech History


Vinyl Records

Vinyl has become trendy again, and record pressing plants are pumping out as many new records as the plants can produce.  Some plants are even expanding.

Vinyl records have a mythology around them promulgated by audiophiles.  It is said that they are analog (they are), and thus more accurately reproduce the original audio than the digital “stair steps” (they don’t), and that, somehow music heard via vinyl is “purer” than digital music.  Almost exactly the opposite is true.

I hate to break it to you, but vinyl is a terrible medium for reproducing audio, and its various deficiencies require countermeasures that significantly change the audio.  Tom Scholz, the leader/recording engineer for the rock group Boston, supposedly tried to get the first Boston album recalled when he heard what his mixes sounded like on vinyl.  Tom Scholz’s experience aside, many of the countermeasures make changes to the audio that audiences can find pleasing.

These countermeasures were implemented in the process of “mastering”.  Originally, mastering was just creating a master disk, from which the pressing plates for the vinyl records would be made.  The mastering setup was simply a cutting lathe that created the sound groove in a metal plate.

One of the physical properties of a vinyl record is that the width of the groove is determined by the volume of bass frequencies.  When music started being recorded with electric bass, mastering engineers found they could often only get five or ten minutes of audio per side of a long-playing record, instead of the normal 15-20 minutes, because the grooves were too wide.  This resulted in them adding devices to the mastering setup to do compression and limiting on the bass frequencies.  The same measures are required for classical music with lots of tympani and/or low brass, and jazz with a prominent bass part.

Another issue with vinyl is that it does not reproduce high frequencies well, and midrange frequencies tend to be prominent.  Mastering engineers added equalization to their mastering setups to partially compensate, and recording engineers would often boost high frequencies in their mixes to help them be audible on the record.  Even with these measures, high frequencies on records gradually disappear toward the top of our hearing range.

The dynamic range of vinyl–the range in loudness from the quiet background hiss of the record to the loudest sound it can produce–is much smaller than that of our ears.  On vinyl it is about 70-80 db, while our ears have a range of about 120 dB.  Every 3 dB represents a doubling in loudness, so the extra range can be pretty important.  For music that goes from being quiet to very loud, it can exceed vinyl’s limits, so the quiet parts are buried in the background hiss.  To deal with this issue, vinyl mastering engineers compress the entire mix (as well as adding extra compression and limiting for the bass frequencies), which reduces the dynamic range.  This technique is used on all types of music, but it is most important on classical recordings because they often have wider dynamic ranges.

There are other, more arcane, measures taken in mastering, but many listeners find the ones I’ve described add a quality pleasing to the ear.  Overall compression makes it easier to hear all the parts, bass compression often makes the bass sound better, and the rolling off of high frequencies results in a sound many describe as “smooth” or “warm”.

At least part of the blame for the vinyl mythology has to do with a shortcut record companies took.  When Compact Discs first came out, the record companies believed that they didn’t need to do any mastering for digital because digital didn’t have vinyl’s limitations.  They sent the master tapes to CD manufacturers with no mastering, and the CDs that were produced did not sound anywhere near as good as vinyl.  They didn’t have any compression (or only what the recording engineer used), and because the high frequencies were boosted for vinyl, they sounded “harsh” or “tinny”.

These problems were caused by a lack of mastering, not, as audiophiles believed, an inherent flaw in digital audio technology.  It took a few years for the record companies and engineers to figure out that, in order to sound good, a similar mastering process was required for digital media.  CDs manufactured in the early 1980s often have these sonic problems, while later “remastered” versions mostly sound better (to my ears) than the vinyl, or at least more similar to the original master tape.

Today, great tools exist for mastering digital recordings, and pretty much every digital recording, whatever medium, gets mastered.  Mastering engineers have built on the vinyl techniques to create a large bag of tricks that make recordings sound better to listeners.  Over time, the ears of audiences have adjusted to being able to hear high frequencies without cringing, so they accept recordings where you can hear what the cymbals really sound like.  As a friend of mine who is a mastering engineer said to me yesterday, even an mp3, if it has a reasonable bit rate, will sound much closer to the original than vinyl will.

If you love the sound of vinyl, please enjoy it with my blessing.  Apart from the sonic aspects, I find the 15-20 minute album side a more satisfying chunk to listen to than a 3-minute mp3.  Just let go of the idea you are hearing what the recording engineer heard when he was mixing.

Now that I’ve rained pretty hard on the vinyl parade, do I have an alternative?  Is there a different technology that I think will serve listeners even better?  Stay tuned for Giving Good Audio for Music Part II: 24-bit Audio.

The First Shuttle Landing

STS-1 Landing at Edwards AFB

As we prepare for America’s very last shuttle mission, I thought I would share the story of my small role in the first shuttle space mission, STS-1.  (No, I didn’t get the word order wrong.  There were previous shuttle missions piggybacked on a modified 747 that did not go into space.)  Anyway, be warned.  This post is a little long.

In 1980, I arrived as an engineer at Edwards Air Force Base, working for Kentron International, the engineering services contractor for the base.  In college, I had wanted to study computer science, but at the time, almost no schools offered a degree in computer science.  I ended up studying electrical engineering with a computer science “area of specialization”.

In my interview for the job at Edwards, I talked about programming microprocessors, a skill I was sure they would be interested in.  The guy I interviewed with did not see things quite the same way.  It turned out they did almost no microprocessor work, doing most of their designs as large circuit boards covered with hundreds of logic chips.  The guy explained to me that microprocessors were a “fad”, which would quickly pass.  (!)  I got the job because I convinced him I could do circuit design as well.

About the time I finished my first microprocessor-based project for them (I never did any circuit design there), Edwards got the news that a large system they had ordered years before would be a year or two late in arriving.  This news caused more than a little panic, because the system was required for them to participate in the orbital portion of space shuttle missions.  Sure, the shuttle would still land at Edwards, but if they lost the orbital portion, it was not only humiliating; they also would lose a substantial amount of funds, much of which they had already spent preparing for the shuttle landing.  And they were rightly very proud of the accuracy of their two RCA AN/FPS-16 radars (16-foot diameter dishes) made during the golden age of radar in the early 1960s.  They calibrated these radars by bouncing a signal off the surface of the moon.

Previous space missions had been done “unplugged”, at least the tracking part.  Each site that did tracking of the spacecraft would watch the horizon at the point the spacecraft was expected to appear, and when it was supposed pop into view, the radar operators would madly search to find it.  When they found it, they would lock on with the radar, and it was automatic from that point.  The tracking data was recorded on a tape drive, and it was processed later.

This was to be the first space mission to use “continuous track”.  Live data would be sent over phone lines from a site tracking the shuttle to the next site it would pass over, and that site would slave their radars to the data to locate the shuttle before locking on.  That site would then send live data to the next site, and so on, providing continuous tracking data.  All the live data also went back to Houston so they could immediately see where the shuttle was.  The system that was going to be late did the slaving and data transmission that allowed them to do continuous track, as well as lots of other stuff.  They could do without the other stuff, but they needed the slaving and data transmission.

There was a very short time available to create a replacement.  In order to participate in the shuttle mission, a site had to succeed in a test of the continuous track system.  The test was to track a dead satellite continuously around the world, and the test was scheduled about 90 days after Edwards found out the system was going to be late.  My fellow engineers, who designed with logic chips, estimated they could do a replacement in nine months or so.

Being young and foolish, I spoke up and suggested a way we could do it in the time allotted.  There was an off-the-shelf computer system intended for industrial applications that could meet the requirements.  There were plug-in circuit boards from several vendors to do the different things we needed.  The other engineers were smart enough to understand that if someone failed at this task, he would almost certainly be fired because of the political weight of the issue.  I was naiive about the politics, and there was a consensus in our group that I should be the one to get the assignment.

I ordered the parts, put them together, wrote some assembly language code and started doing tests.  Everything worked, except sending and receiving data across the phone line.  We used an unusual mode of data transmission (synchronous, rather than the normal asynchronous), but I couldn’t get it to work no matter what I did.  As I countinued to beat on it without success, everyone got more and more nervous.

After a couple of weeks of this, my boss hired a consultant to come in and help me.  They did want him to get it up and running, but they also wanted him to tell them whether they should fire me right away.

The consultant and I got along well, and he eventually identified the problem in a place I had not thought to look.  It turned out that the plug-in board  I bought to do the data transmission had a design flaw that made it work fine for asynchronous data, but not work for synchronous.  I cut a few traces on the circuit board with an X-acto knife and soldered on a few wires to correct the problem, and everything was running just as it should.  The consultant gave a very positive report on me, and later tried to hire me.

Edwards participated in the test with the dead satellite, with me at the radar all night as the test continued, “just in case”.  A couple of sites failed the test (not Edwards), and they did another test a week or two later, again with me standing by in the radar all night.

About ten days before the shuttle landed in April of 1981, my boss told me I needed to attend the base commander’s staff meeting.  Once again, being young and foolish, I thought maybe the commander would thank me for all the all-nighters I put in to get the project done.  For the entire staff meeting (an hour and a half) no one even glanced at me.  When the meeting was over, the commander peered at me and asked, “You Gloster?”  I nodded.  He said, “I need you to be at the radar for the whole shuttle mission, understand?”  I wasn’t quite sure what to say, so I nodded.  He said, “That is all”, picked up his notes and walked out of the conference room.

When I got back to the office, I asked my boss, “He can’t really do that can he?  I’m a civilian, not an airman, and not even a civil servant.”  My boss said, “Don’t make an issue out of it.  Just do it.  I’ll give you some comp time later.”

So, as America watched Cape Kennedy prepare for the launch, I was holed up in the tiny cinderblock building that housed one of the FPS-16 radars, where I remained for three days with my sleeping bag, as radar operators went in and out for their shifts.  Nothing went wrong.  If it had, I had a spare unit I had built, but I’m not sure I could have done anything other than swapping the unit out.

Two days later, when the shuttle landed, I was a little bleary.  Sleeping a couple of nights on a concrete floor behind humming racks of equipment while you are excited and a little worried doesn’t really give you quality rest.  But the radars are on a hill overlooking the lakebed, and watching the landing from there gave me the best view of anyone.  Getting the best seat in the house for that historic event made all the project’s late nights, political undercurrents and difficulties worth it.

Edwards AFB