Steve_Graham@ub.cc.umich.edu (12/06/90)
[Since there was a request, I have solicited the following two articles from
the resident recording engineers. Enjoy. -tjk]
Here we go. Please keep in mind that this is all from memory--Not Responsible!
Magnetic tape has a non-linear transfer function; the change in output is not
directly proportional to the change in input. For low level signals a
relatively large change in input level yields a relatively small change in
output level. But there is a portion of the dynamic range of the tape that
is relatively linear. Bias is a signal that is added to the tape to bring
the total signal level up into that linear range.
If there was no bias and we tried to record an audio signal, or a sine wave,
we would find that the waveform was badly distorted. You would have something
akin to "crossover distortion" as the waveform crossed the zero axis. If you
add a DC biasing signal you can bring the total signal up to where the sine-
wave (or other signal) is quite undistorted, but in doing that you throw away
more than half of the tape's dynamic range (since you are biased constantly
in either the "positive" or "negative" direction) and in addition you find
that the biasing signal adds noise to the tape. That is because any
irregularities in the tape's surface, dropouts, modulate the strength of
the dc bias on or from the tape. (At least that is part of the reason.)
If we add an ultrasonic AC signal instead of DC to bias the tape we can use
the entire undistorted dynamic range of the tape. The audio rides on top of
the AC bias, and both the audio signal can modulate the tape in both the
"positive" and "negative" directions. AC bias causes less noise than DC
bias because the AC signal "averages out" to zero (over time) so that
"modulation noise" is less noticeable. (This is my own, possibly flawed
interpretation!)
AC bias has another advantage: it helps the record head do its job. Tape has
a natural resistance to changes in its magnetic state. I believe this is called
"coercivity", and is related to the aforementioned non-linear transfer curve.
When the tape passes the record head the bias signal excites the tape and
makes it ready to accept the recording. The actual recording doesn't take
place at the head gap itself, but rather at a point where the tape is
leaving the gap, and the location of this point varies slightly according
to the strength of the bias signal. This point is important because it
affects the "stereo phasing" of a stereo head and may affect how the azimuth
is adjusted.
AC bias does have one drawback, and that is that the bias signal can tend to
erase the signal we want recorded. (In fact most machines send a stronger
version of the bias signal to the erase head to erase the tape.) This
phenomenon of "self-erasure" is most pronounced at high frequencies (or
short wavelengths, if you prefer) and, again, varies with the strength of
the bias signal.
So the bias signal does these things:
1. Greatly reduces distortion
2. Maximises the dynamic range available from the tape
3. Affects the frequency response because of self-erasure
4. Affects the overall loudness because of (a) greatly reduced distortion and
(b) self erasure. (These are the mechanisms I'm aware of, at least)
5. Affects the "modulation noise" and level stability of the tape
The explanation for this point is as follows:
As you increase bias the level coming from the tape will increase to a point
(due to lowered distortion and perhaps greater efficiency in the record head)
and then begin to decrease again due to self erasure and possibly tape
saturation. The area at the top of this curve is relatively flat, that is,
small changes in bias will have less effect on output level when the bias
is near the "maximum output" point than when there is too little or too much
bias. Therefore dropouts, which affect the strength of the bias recorded on
the tape, will have smaller effect on the signal output level, and will
cause less modulation noise since they are causing less modulation of the
signal, when the bias is adjusted for maximum output.
There are many methods of adjusting bias. The old-fashioned way is
simply to put in a midrange tone (1 kHz at 15 ips, for example) and adjust
for maximum output. To make this more precise you can back off until the
output level drops .5 dB, then increase through the peak to where the level
again drops .5dB, and set your trimpot half way between those two points.
But I should back off here to say that before adjusting bias you should have
made playback adjustments, for which you will need a test tape.
The sequence of alignments is something like this:
1. Playback rough level
2. Playback azimuth
3. Playback frequency response (treble) and recheck azimuth
4. Playback fine level
5. Record bias
6. Record head azimuth
7. Record rough level
8. Record EQ (treble)
9. Playback EQ (bass) if available, set by recording tones rather that test tape
10. Record level fine, and recheck azimuth and overall response.
Since I am asked about setting bias we will assume other adjustments have been
dealt with. (Though tapes have varying characteristics, often resetting just
the bias will provide near optimum results if you switch tapes, and have
previously performed all the adjustments properly. As you might expect, the
bias setting is more critical at low speeds.)
A commonly used method nowadays is to input 10 kHz and adjust the bias until the
level drops a specified number of dB, beyond the peak. This is useful because
it is easily repeatable, but you need to know the required setting for the tape
and speed you are using, and it varies according to the dimensions of the head
gap.
Another quick way to set bias with three head machines is to put in a low
frequency signal (10 to 40 Hz), and listen to the sound from tape passed through
a filter to remove the fundamental. You will hear thumps and noise, and
as you approach the optimum spot there will be a null where the noise and
distortion are minimized. This is the approach recommended for those who
use dbx and other similar types of noise reduction.
If you are using a two headed deck that has previously been set up correctly
and you want to change to another tape, you can try just setting bias
so the frequency response is correct. This can be done by ear, by recording
fm interstation noise at a level at least 20 dB below 0vu and comparing
the sound off the tape with the sound going to the tape. The volume must
be the same in both cases, however. If you do this with a cassette deck
be sure to turn off the Dolby or other N.r. circuits first, as slight level
differences will affect their performance, and will also have to be adjusted
out, after the bias is set. (We are talking quick and nasty methods here)
--Steve