Steve,

Start here….

http://www.xilinx.com/support/techxclusives/DCoffset1-1-techX14.htm
http://www.xilinx.com/support/techxclusives/DCoffset2-1-techX15.htm

Thanks for the quick reply. I have read this article previously, and others like it. What I am trying to understand is how, during the recording process, can you curb this phenomenon ie: what devices cause this eq's maybe? dunno.
And also what is acceptable, ideal, and what precautions are to be taken when correcting it with respect to impact on audio.

..........? :d:

Reduce dynamic range: True.
Create subharmonic distortion: True... well, kinda.

Whenever you clip in the digital domain, the waveform is harshly lopped off to the maximum value that the digital data can represent. This results in a sort of square wave that generates a short burst of static upon playback.

A signal should ideally be able to travel an equal distance up and down before clipping - this provides maximum headroom. What DC offset does is translate the entire signal up or down, causing the average value of the signal based on time to fall above or below the middle. (This is shown pretty well in the diagrams in the links above).

This means that the signal can only go a shorter distance up (or down, for a negative DC offset) before clipping, and the "extra headroom" on the opposite side is wasted - Who wants that extra space to work with below when the tops are already clipping?

So, the assumption that the error occurs because of DSP and A/D conversion are not totally correct. It can happen there, but it can also happen in any piece of equipment that isn't calibrated properly. It's just a lot simpler to remove DC offset in the digital realm than it is in the analog realm (That's my opinion, of course)

Most equipment on the market will come with really minimal DC offset on the output stages - what you can do to check is get an oscilloscope and start looking at the signals. You shouldn't worry too much about it - at less-than-significant values the headroom you lose isn't that much, and it is easily corrected after all is said and done (assuming there hasn't been any clipping, of course ;) )

The only thing I can see affecting a signal which hasn't clipped but still has a severe DC offset would be a compressor with very low attack/release times - The compressor would move the signal closer to 0DC offset every time the threshold was passed and return the signal to its normal, biased state once the release kicks in - Could generate some weird harmonics that way.

cool!!!!!

I made that up. :)

I became more intrigued with this when I receive a project (on dat) from another studio.
The mix job was horrid, and I was asked to see what I could do to touch it up. Using cooledit the first thing I always do is correct for dc offset before I process anything. Normally on my own mixes I never see anything more than .005- .010%, usually less than .005. Upon anylizign this one I found l
differences of up to .050% and average differences of +- .045%. What the hell did he do? Supposedly this guy has an amek console and top notch outboard stuff. He uses adats for tracking and I think he also has 12" tape. I think he must use the tape for drums. They sounded like they were recorded on a badly calibrated machine. Either that or the cables from the mics were 7000 feet long.

Next question. Is there an optimum D.C. offset to shoot for on the premaster? I wonder how this translates to the speaker. With a severe offset problem would it create heat in the voice coil, and effect the dynamic sesitivity of the speaker in the same way as it does in the waveform.
I think it would.
Moreover, while considering the lo-fi world, would it be benificial to bias + or - to average the speaker differently on more dynamic mixes?

Next question. Is there an optimum D.C. offset to shoot for on the premaster? I wonder how this translates to the speaker. With a severe offset problem would it create heat in the voice coil, and effect the dynamic sesitivity of the speaker in the same way as it does in the waveform.

From my (admittedly rudimentary) knowledge of electronics, I'd say the extra power you're putting onto the speakers would be equivalent to the something like the percentage of DC offset. That is to say, with an offset of 0.05%, you'd be adding an extra 1/2000th the power needed to actually make the same noise without any offset. Does that make sense to anyone else?

Almost every piece of gear has something to block DC at it's input and output like a 'coupling' capacitor. Something has to be out of whack for it to pass DC. Like a leaky power supply or an opamp that has seriously drifted.

I think that may be, at least half the info I need. Am I correct in assuming then, that if I start ton encounter the issues in my studio, then I should start testing gear? I have read that with respect to crosstalk That once an emmitted rf sees conductive material of one tenth the wave length, then inductive crosstalk can occur. perhaps then, the first place to look for problems is in the console. If this is correct I may call this studio and see if the engineer is aware of the problem I encountered.
Also, what about rf signal induction to unbalanced patching? Perhaps crossing the wires over instead of lining them side by each would make inductive issues subside somewhat, due to misaligning the feilds.

Are we considering signals below the range of human hearing as technically DC?

But you're right, though. Oscillations like that one means that something could be out of whack with the potential to grow worse in the future. Should get all that equipment checked.

are we entering RFI and EMI? Oh.. that will be cool..

There is a very powerful FM station tansmiter less than 1 kilometer from my studio. There are days one can listen to a ghost signal, when recording high gained electric guitar.
When I move towards the north, the interference disapppears completly or minimizes a lot.
But it only happens with some single coiled guitars or with a bad grounded bass guitar.
With humbucking pickups, not problem at all. The solenoids arrangement already do the cancellment, right?

In addition to allof this, brings me to subharminic distortion, and the overall presence of frequencies lowr than say, 17hz. I have tested my own hearing and i cannot hear lower than this. When masterin should these ultra low frequencies be deleted? I think so.

would you guys dc offset before using mastering plug ins or after, or both? and why?

chris perra

Folks,

some of this has been said but let me repeat and reinforce some of the statement while adding some others:

0. Pure DC is a signal component at zero frequency, meaning that it will not change over time at all. Practically very low frequencies (below 5Hz) are also treated as DC because they have similar effects as pure DC although they are in theory perfectly fine AC signals.

1. DC offset is, for one, undesirable because it reduces the signal headroom. One side will hit the ceiling sooner than the other. Accordingly the optimum DC offset is zero.

2. Highpass Filters will eliminate the DC offset. They can usually be set to very low frequencies (e.g. 10Hz) so that they don't affect the regular signal. I for one pretty much cut off everything below 40Hz anyway (rumble doesn't do anything for my mixes) - this at the same time eliminates any potential DC offset from the signals.

3. Most electronic devices such as power amps and FX units have AC coupling at their inputs, they will block DC offset, meaning that whatever signal makes it into the unit does not have a DC component anymore. Also, the output of such a unit will usually not have a DC offset any longer. In power amps this filter is usually nothing else than a somewhat large capacitor and a resistor. Removing the offset digitally by means of a digital HPF is of course also possible and certainly cheaper if there is already a DSP at work in the system.

4. Amplified DC offset (which in reality doesn't really happen any more these days) is REALLY BAD for speakers because it sends a permanent DC current through the coils, thus consuming power and heating/burning them up.

5. DC offset well below 0.1% is not something that I would lose sleep over. The effect is neglible - in reality it is probably rather expensive to eliminate such small amounts completely.

6. Alecio's observation with single coil PU's has to do with a different effect. The long wires of the coil function as an antenna, picking up and demodulating (theoretically a rather complex procedure) the FM signal into something that is audible when the gain is high enough. This is not a real problem with Humbuckers because these are two single coils that are wound in different directions. The picked up signal is accordingly phase inverted in both pickups and cancels each other out.

Not trying to be a smartass here ... just hope it helps clarify things some more (... although a lot of it has been said before).

MisterBlue.

:) MisterBlue, I appreciate your direct response. I do have a few questions...

This statement:

2. Highpass Filters will eliminate the DC offset. They can usually be set to very low frequencies (e.g. 10Hz) so that they don't affect the regular signal. I for one pretty much cut off everything below 40Hz anyway (rumble doesn't do anything for my mixes) - this at the same time eliminates any potential DC offset from the signals.

What is the slope you use for the below 40Hz cut off? And do you apply this type of cut while tracking, at the 2-bus, or only in mastering?

Thanks,

--Rick

Rick,

I usually use a rather steep slope - mostly a 3rd order filter - giving me a -18dB/octave cut.

I don't apply any filtering during tracking. I like to keep my input chain as clean as possible, so no EQing in that stage... microphone or instrument, pre-amp, maybe a compressor but that's it before going into the converters. Any other processing is always done later during mixing so it can accordingly be reversed/adjusted if necessary.

I usually put the 40Hz HP-filter into the stereo sum (already during the mixing process). I found that it helps me to clean up my mixes significantly.

Hope this helps,

MisterBlue.

we've had a few threads discussing whether or not to lopp off these sub freqs. by applying these steep filters at high freqs, you could be doing more harm than good. There is plenty of good stuff down there and in some instances fundamental notes, not to mention the phase shift you create with this kind of filter.

like what was said early, some sub freq's can fool the program into thinking there is DC when there is not, just good ole sub freq's and by correcting this mistaken identity, you can throw things out of wack. most of the time, sub freq's are out of control because the music is mixed on speakers that can reproduce them, I will lower them but not cut them. These sub freq's are vital for depth and dynamics in a mix.

In response to Michaels reply, I agree that an hpf could do more harm than good. Personally I feel that 40hz is a little high if you were to apply this. Effectively managing low freq's can be tough, But some of the freq's you can't hear too well you can sure feel when played back on a good system. For the purposes of "quick mastering" I think the 30-40 hz hpf might not be a bad idea. I don't have the acoustics to accurately make strong decisions in this area. And "quick mastering" is the only service I offer to clients who will not be mass producing their product.
I feel that if some lows are a little out of control off the mix, when sent to a good mastering house they will at least be there, so that the engineer can do what he wants with them.

I would like to add that, if you track your sessions well, and stay away from the eq's in the 100hz and down range, you may find all this worry is unnecessary.

To Chris Perra:
You wrote:
would you guys dc offset before using mastering plug ins or after, or both? and why?

Even on a super quick job I always correct d.c. to zero before doing anything. I do this to maximize headroom.
After I finish, and it sounds good I check my work. I have never had to make additional corrections at this point, as changes are in the range of .01-.03%.
Just the same, double checking keeps me sane.

"If you have to do it twice, then you had time to do it right the first time."

Thanks everybody for this discussion. You have all helped me reinforce my knowledge of the subject. I am self taught, and I appreciate your advice and opinions greatly. :p:

LOL... Just like to point out that a DC offset of 50% (A really ludicrously high value) is a loss of 6dB of headroom.

1% is effectively a loss of of approximately 0.08dB of volume.

Hello all,
I am new to this forum,and just want to say thanks! to all of you sharing your knowledge.
Dudley.

would you guys consider usimg the dc offset option in wavelab?

or would you prefer to use eq's

chris perra