Return to EQ index
All information in these pages is copyright (c) 1989-2003 by Roger Nichols. All rights reserved. Permission for personal reference only, and may not be reproduced by any method without written permission.
My Dog Ate My Computer
By Roger Nichols
I’m late with my column this month, so I tried to use an updated excuse for my tardiness. In high school the favorite excuse was always "My dog ate my homework." Or the one I used was "I put it on your desk as I was leaving yesterday. You didn’t get it? Gee, that was my only copy."
With the advent of computers the ‘that was my only copy’ ploy no longer works. But ‘my dog ate my computer’ seems like a good one.
Nope! The editor is not going for it. He said if I didn’t turn it in by the end of today, he was going to flunk me so… here goes.
Surround stuff
As I am writing this, Elliot Scheiner and I are printing the last of the surround mixes for a Steely Dan DVD that will be out this summer. What I think is different about this DVD is that the focus is primarily on the quality of the music portion with the video taking a back seat. The DVD will have a stereo PCM channel, a Dolby Digital AC-3 surround channel, and a DTS surround channel. If there is any room left on the disc, it will be filled with video.
The speaker setup for the surround mixes was as follows: The front right and front left speaker were 6 feet 6 inches apart (tweeter to tweeter) and 6 feet 6 inches from the mix position. The center front speaker was half way between these two, but still 6 feet 6 inches from the mix position, which made it a little farther back from the console bridge. Oh, the left and right speakers were on stands behind the console, not sitting on the meter bridge.
The rear speakers were a mirror image of the front. 6 feet 6 inches apart, 6 feet 6 inches from the mix position. Herein lies the rub. This is the accepted positioning for the nearfield monitors when mixing 5.1 surround, but I don’t know anyone with that setup for listening at home.
The front speaker setup is fine. It doesn’t matter to me if the playback speakers are a little higher, or a little farther apart, just as long as the stereo spread sounds reasonable. The rear speakers are a different thing altogether.
If you are facing forward while mixing in this configuration, you can get a pretty good balance of all the instruments in all 5 speakers, but if you turn your head just a little either way, the rear speaker information seem too loud. I think this is because the rear speakers are too close together and the sound just hits the back of your head, or the back of your ears, and you don’t get much sonic information. If you move the rear speakers farther apart or if you move them up higher, the definition improves tremendously. This wider rear speaker positioning also seems to be the preferred setup in most homes with surround systems, including mine. I have noticed that at home I have to turn down the rear speaker level for most 5.1 surround music discs, as compared to the levels I set for surround movie sound tracks. Film soundtracks are mixed in a different acoustic environment with the speaker positions more closely matching those in a theater setting. The rear speakers are higher and farther apart.
So what does all of this have to do with the price of corn? Nothing. But I think that the standard speaker setup for surround mixing should be modified so that the end user doesn’t have to change the level of his rear speakers to hear the proper balances on music surround product. So there!
It’s Over Between Us
I get a lot of questions about over lights on DAT machines, Pro Tools meters and digital processor boxes. Here is everything you ever wanted to know about overs, but were afraid to ask.
Over lights can not really measure anything that is over, because you can never get any signal louder than all of the bits used up.
The over lights have to guess at whether the signal would have gone over. Counting how many samples are at full level does this. If just one or two samples are at full level, the possibility exists that the signal just barely reached full level and then went back down again with absolutely nothing wrong.
If you have a full level signal for more that two samples, then the only possible explanation is that the signal was still going up but there aren't enough bits to record it. The result is a chopped off top (or bottom) to the waveform. This is 100% distortion. The reason it is 100% is because none of the original detail is there, only 100% artifacts. If you isolated this piece of the signal you would only get a full level dc component and no modulation.
Professional digital meters in high-end hardware have settings for the number of samples allowed at full scale before the over lights come on. The Sony DMU-30 was the standard by which all others were measured. All of the CD mastering facilities used this exact same meter or detecting overs. This is the same meter as the meter section built in to a Sony 1630 which was the machine that produced the CD master that was sent to the CD manufacturing plant. The standard adopted was that 4 samples at full level or more was unacceptable.
In the early days of CD manufacturing the plant would not accept a CD master that contained overs. They would bounce the master and have the mastering facility make a new one. If early CD players were hit with an over, they would produce a loud pop or click. The severity of the pop varied depending on the CD player. Overs were illegal.
If the master sent to the mastering facility was full of overs, the mastering engineer could turn down the final output by 0,1 dB and even though the flat tops were still there, the over lights would never come on because the level never used up all of the bits. When the rap records and hip-hop craze flooded the market, every artist wanted their CD to be louder than the competition. Limiters from 1994 were not fast enough to crunch the audio without making it sound like a bad FM station. But you could raise the level 6dB or more and let the 16 bits fill up until they clipped the signal. Just like a peak limiter set to zero attack time with infinite compression ratio. The results were that the over lights came on and stayed on until the entire CD was played.
The mastering guys just brought the final level down 0.1dB (now you can bring it down 0.01dB) and the over lights would extinguish. Any digital device such as a digital console or digital EQ will allow you to lower the level by 0.1 or 0.2 dB. The tc Finalyzer allows you to set the maximum "never exceed" output level in .01dB increments.
Current CD players do not click or pop when presented with this square signal and the chopping off of the waveform actually adds to the punch of the kick or snare beat. If the square top waveforms are unacceptable to the artist, engineer or mastering engineer, a hi-pass filter in the chain will remove the flat tops and smooth things out.
Back to counting over samples.
I won’t go into binary counting here. For further information you can search the Internet, or cut off all but one of your fingers.
Approximately 6dB before clipping the most significant bit goes on. Some older meters turn on the over light when this condition is met. This means that the over light will warn you 6dB before you clip the recorded signal. This seemed like a fair place to set the over lights for a consumer machine. Keep them out of trouble by showing the over lights early. The first Sony PCM-2500 DAT machine was this way, which was based on the first consumer machine. If you want to be closer to the actual over level, you have to look at more bits. If you look at two bits you are about 3dB away. If you look at 15 bits (the 16th bit is the sign bit telling the converters whether the sample is positive or negative) you can give an indication of when the bits are all used up. Pro Tools counts all of the bits before the meters show over.
It takes some additional hardware or software to actually count the number of samples that have reached the "over" status. The meter must have a memory of what the last bunch of samples was (or just whether they were over). If you have the meter set to light up the over light when six samples in a row were over, then the meter must remember the last six sample conditions. The Sony DMU-30 meter, the meters in the Apogee AD-8000, and the Mytek digital meter are some of the hardware meters that count over samples before lighting the over light.
So, now you understand why you can record something on a Fostex D-10 DAT machine with no overs, and when you play it back on a Sony PCM-2500 the over lights are on ALL of the time. The levels on the tape are no different, the sound will be no different, it’s just those pesky over lights.
Most of the newer DAT machines have very accurate meters and even display digital headroom (how many dB away from over the loudest peak is). Sony and Fostex machines have this feature. If the headroom display says 3.1 dB, then you can turn your mix up 3.1 dB more before you run out of bits. On most DAT machines the over lights do not lock on as they do on stand-alone converters and outboard digital meters. After you have played through a mix you can look at the headroom display. If it says 0.0 dB, then you probably went over somewhere along the line. Play the mix again and watch the meters. I ALWAYS make a pass to watch meters before I print the final mix.
The bottom line is that one or two overs here and there are not going to ruin your master. If you can’t hear it, then it is probably OK. It will probably get fixed in mastering anyway.
That is it for this month.
Roger, WILCO, over, and out!