CD volume levels

I'm a bit confused by various recording levels I've found on CD's. While most fall in a narrow range of mastering/pressing levels, every so often, I'll run across a CD that was manufactured with a really hot (loud) level, while others are rather anemic (soft). I'm not talking about compression or limiting on the master tape, just the absolute mastering level on the CD. While I understand the physical limitations of cutting vinyl (a combination of science, art, with a little voodoo throw in for good measure), why would there be such a variation in the digital domain? Are there trade-offs with different levels when pressing CD's, just as on the analog vinyl?
While I am no expert on this subject, I read an article that said that the earlier CDs were recorded "soft", and did not make use of the full dynamic range that CD offered. Later,I think in the '90s, they discovered that they could get better dynamic range, and better sound, by recording them "hot". So, if you check when these CDs that you are talking about were made, you may find a correlation.
I cannot speak to the technical side of this topic either but can confirm that on my system I can definitley hear differences in the overall volume of the music recorded. I always assumed it was a matter of how the music was recorded. I have also noticed that on less than revealing audio systems, the difference in overall volume levels is not as apparent.
One way to quantify the differences is to playback on a CDR with level meters.
Hey Fatparrot...a really hot CD is probably compressed (or at the very least, limited) quite a bit to be as hot as it is.

When analog signals were run hot to magnetic tape, the tape would still accept the signal (up to a point), but it would compress them naturally. This is the analog tape saturation/compression that many people love (especially for rock music). It was still compression, but people seemed to think that it was more natural sounding. I think the volume level of LP's was limited because loud mastering volume would cause the needle to jump from the groove (I think...I don't know for sure).

Digital signals can only be so hot...they distort when they reach their peak volume. There is no natural analog-like compression that takes place...instead the signal breaks up and distorts badly. I assume most modern, louder CDs are using that 0db point as a maximum...making sure that the peaks of the recording don't hit that distortion point. A very dynamic, uncompressed recording will have a lower average volume because they need to allow that headroom to avoid distortion. For example...the average level of MFSL CDs is usually lower than standard mass market versions because they don't seem to use as much (if any) compression on the master...they lower the level a bit and allow the full dynamic range of the recording to show. They allow you control the volume with your amp & preamp (as it should be). Most modern, mass-market CDs will usually try be as loud as possible (at the request of the record labels). I believe the idea is that people think a louder CD "sounds better" and the louder volume level will also sound "stronger" on the radio. Many mastering engineers think this trend has gone way too the point that it hurts the music (you could probably do a web search on this topic and find some info).

ALL CD's have the same maximum volume level. If a CD is much louder than others, it was probably compressed quite a bit while recording, mixing, mastering, or all three. This is especially true if it's louder than most modern CDs, because their average levels are already very high. A loud CD's average volume is much higher because the signal has been compressed (or limited) to a point that it has very few dynamic peaks. This allows them to raise the overall volume because they don't have any dynamic peaks left to breach the threshold and cause distortion.

I don't know if I did a very good job of explaining this, but I hope you get the general idea. If you're interested, you might be able to find some on-line interviews with some mastering engineers...I often see them mention this topic in interviews. Some refuse to master CDs to that overly-compressed point, but I've read that others (the ones that do most of the big labels' mastering) tell the clients, "Okay...this is as loud as I can make it without compromising the signal. If I make it any louder we'll lose dynamics, etc. It's your call....what do you want me to do?". He went on to say that most of the label reps tell him to master it louder. Oh well...

Your explanation makes a lot of sense. A good example is the the track Private Investigation on Love over Gold album by Dire Strait -- recording level is low but it has tremendous dynamic range. The same track on the latest compilation Sultans of Swing has a much higher recording level at the expense of reduced dynamic range. BTW, One thing that I like about the earlier Dire Strait albums on CD is the presence of tape hiss. It sure adds another dimension.

P.S. Do you guys think that all those 16Bit (storage and playback) "Digitally Remastered" re-issues are nothing more than boosted up recording level? I mean CD recording is 1bit PCM after all. What did they miss out originally?
I think a lot of the digital remastering is hit or miss. It definitely has a lot to do with marketing, but some of it is sincere and done well. Unfortunately, labels can have something cleaned up, boost the treble, and compressed and many people will tend to prefer it to the older version when they compare the two. The compression can actually make a recording sound like it has more details too. It doesn't only squash the loud brings up the quiter signals, both musical and unmusical (sibilance, reverb, string noise, etc). That makes it hard to tell whether a remastered version has more detail because of original masters and modern 24 bit remastering equipment, or simply because the signal has been compressed and altered. It all depends on how poor the original release was, and how well the remastering was handled. There are definitely many remasters that were done by talented people, using original master tapes, and good equipment. I suppose your ears will tell you which are good and which are bad. Excessive compression and loss of dynamics *usually* makes a recording more grating and harder to listen to for long periods of time. The ones that you end up preferring and listening to for hours and hours are probably the ones that were mastered properly.

As far as Dire Straits releases go...I'm not sure how well the originals were done. If I remember correctly, wasn't their first album cut a little short on one song?? A premature fade, or something? The tape hiss issue could mean that it was digitally removed on the new version (which would mean that some musical signal was probably removed with it), or it could mean that they used a different set of master tapes...quiter originals versus a copy. THere are a lot of possibilities, but I do think you're probably right about the newer remaster being compressed more than the older's definitely a trend.
"Hotter" CD's in the pop jenre all have to do with marketing. When artists hear their song on the radio, and it is quiter than the next guy, they get upset. Most pop cd's have been mastered with TREMENDOUS amounts of compression and peak limiting. The reason being to have the disc play "as loud as possible". Fletcher/Munsen were the first psycoacousticians to create an average frequency responce vs. spl graph of human hearing. At louder absolute levels, our ear/brain mechanism tends to smooth out and become more "flat". At lower listening levels, we become less sensitive to bass frequencies and high frequencies. That's why you see "loudness" buttons on some older gear. It boosts the bass and high frequencies to make the level appear louder without turning up the volume knob. What I'm getting at is the same piece of music, played louder, will tend to sound "better" than the quiet version.
Now, if you look at most pop CD's released today on a digital meter (DAT deck, CDR etc) you'll notice that, if the mastering engineer did what he was told, the meters will be pegged at 0dBfs and jiggle up and down by no more than 3 db. At this level of dynamic range, a 16 bit recording has the mathematical dynamic resolution of a 4 bit PCM recording system. What the hell is the point? Classical orchestral discs (if mastered properly with minimal if no limiting or compression) for example sound quieter because, even though the musical PEAKS reach 0dBfs (1111111111111111), the avarage level (mezzo forte) lies at about -10dBfs. The quiet parts (Pianissimo) can be at the -30 to -50dBfs point. This is a dynamic range of 50dB! That's what truly tests a 16 bit system! But even in the classical field, we engineers are feeling the pressure to make as loud a CD as possible. That's why it's not unheard of to use a peak limiter to bring the average level of the CD up a couple to 3dB. Even a little bit of limiting is audible though, and those of us with a conscience avoid it like the plague.
Hey Alcides...thanks for all of the technical info and insight. I find this stuff fascinating. I was referring to rock/pop CDs in my rambling above, but I must admit...there have been a few jazz remasters that have made me wonder.
No problem Phild. I do want to add one thing though. There are very few, very talented mastering engineers that really know their rooms/monitors/and especially how to use their gear, that can manage to make a heavily compressed and limited record still sound immensly involving. If my record would have to be squashed inorder to compete in the marketplace, I'd be able to sleep at night knowing it came from their capable hands
Would you mind listing them or e-mailing me with their names?? I may be looking into mastering a rock band recording sometime soon. I'd be especially interested in any in, or near, Chicago. Thanks again!