ANALOG TAPE: An Overview of the Medium
© February 1997 by Eddie Ciletti

Please also see the recent articles link for an updated, three-part version on this topic that appeared in MIX 2000.

This is Part Two in a series about recording media. I thought it might be a good idea to review audioís historic obstacles to better understand why things sound they way they do. Part One begins with the phonograph record and appeared in the Feb 97 issue of EQ Magazine.  Click here for BIAS TIPS.

Discreet analog electronics delivers the best audio that money can buy.  However, as soon as we attempt storage, there is the potential for compromise. For this reason, mastering engineers typically suggest the use of half-inch analog tape and (optional) Dolby SR noise reduction for mix-down and archiving.


Unlike digital storage media, analog tape's frequency response is not flat.  Tape speed and track width will greatly affect the response at both ends of the audio spectrum.  For a perfect example, check out Figure One which is a frequency response chart that shows the performance variables of a Studer B67.  This is a  quarter-inch two track ó with new heads ó at its three speeds (from top to bottom): 30 IPS, 15 IPS and 7 1/2 IPS.  ( This format is equivalent to a two-inch sixteen track.)  Note that increased tape speeds extend high frequency response (well beyond that of 44.1 kHz digital) but also accelerate low-frequency roll-off.  If phatness is key, the extended low end of 15 IPS is for thee.

FIGURE ONE: How tape speed affects frequency response

MULTITRACK PHENOMENA: The skinny on the skinny

If half-inch stereo is the best of what analog has to offer, squeezing sixteen tracks into the same space becomes a major challenge.  Recording bass drum on a two-inch twenty-four track is not easy, so imagine how little headroom half-inch sixteen track has to offer!  From Left to right, Figure Two (below) starts with the ultra-phat all the way down to the micro-gnat, aka the four-track cassette.  Don't "dis" the skinny stuff.  Some folks appreciate low-frequency saturation, which, on narrow-format multitracks, is easier to get!

half-inch four-track or,
a half-inch section of two-inch
sixteen track
a half-inch section of two-inch
twenty-four track
half-inch eight track
sixteen track
4mm (.157inch)
Figure Two: Squeezing more tracks on tape

Remember how quickly the low frequency response rolls off at 30 IPS on quarter-inch (half track) stereo tape?  The same type of response is true for narrow-format machines at 15 IPS!  Head wear further exaggerates the bass response making the bumps bigger, the troughs deeper and the high frequencies droopier.  More important than electronic alignment is the condition of the head surface.  If your narrow format machine sounds tired, have the head relapped first.  The machine may not even need an alignment.


In addition, narrow format machines rarely provide the luxury of a read-after-write head. By contrast, two-inch twenty-four track machines can monitor playback while recording. This not only facilitates alignments, but also yields better performance because the record and playback heads are each optimized for their respective jobs. ( The typical one-inch sixteen track may have separate record and playback heads, but both are technically playback heads.)


On machines with a dedicated playback head, especially at 30 IPS, a record test is highly recommended. Instruments such as kick drum are particularly vulnerable to saturation at high speeds. While switching from input to repro, note how its low frequency character changes with dynamics.

When tape saturation is employed for effect, use a limiter to keep from exceeding the maximum level for the sound you want. (Start with a medium attack and a slightly faster release.) Application permitting, less trouble will be encountered if the kick resonance falls between 75 Hz and 85 Hz. In addition, avoid adding bottom to kick or bass instruments while recording. Let the tape and the head do their thang and make EQ adjustments on the playback side.


The warmth associated with analog tape has as much to do with what happens at high frequencies as it does at the opposite end of the spectrum. If youíre in the habit of pushing signals into the red, be particularly careful with instruments such as tambourine. On older machines equipped with VU meters, the average level will seem to be about "-5VU" but the peaks be off the scale and in the +10VU to +14 VU range. LED displays do not suffer from mechanical inertia.


With the exception of head wear and different tape formulations, tape machines donít go out of alignment unless something is wrong or someone is dangerous with a screwdriver. Pro machines were designed to be aligned so that users, taking advantage of this capability, can choose the operating level, tape formulation, speed and optional noise reduction on a per-project basis ó hence the generous array of adjustments.

Test tapes are full-track recordings with "no compensation for multitrack reproduction." The important reference tones are 1kHz and above. Accurate low frequency playback adjustments can be made after recording a frequency sweep very slowly from 500 Hz down to 20 Hz making recognizable stops along the way to note the peaks and dips. Youíll need this information in a moment.

Note: In addition to speed, be sure to choose the correct equalization (AES, IEC or NAB) for your machine. Also, demagnetize your tools and the headstack before playing the test tape.

Narrow format machines offer little in the way of record EQ adjustment and there is no low frequency playback compensation for worn heads. One trick Iíve used (on heads that are not yet toasted) is to find the nastiest low-frequency head bump ó note the level ó and set the 10 kHz reference tone (from the test tape) to that level. It is cheating, but otherwise, the noise reduction system will mistrack and exaggerate the response errors by a factor of two.

In addition to the head bump, it is also helpful to find the nearest trough. Machines with low-frequency playback EQ should be set so that these two anomalies are equally above and below 0VU. Then, pick the frequency in between that falls on 0VU, record it for at least 30 seconds and make a note on the box.


Hint: If youíre smart youíll start at number 8. Tip: If you accidentally turn an adjustment that does "nothing," put it back in place.

  1. Donít attempt a record alignment. Most sonic problems are likely to be related to head wear on the playback side.  Without a dedicated playback head, any attempts at record alignment will simply throw the machine more out of whack.
  2. Normal head wear can be corrected by resurfacing (lapping) while severe wear can damage the head beyond repair.  Replacement is expensive, so donít put it off. See the back pages of EQ for a lapper near you.
  3. Do not attempt a record calibration on machines not equipped with a dedicated playback head. You will be sorry! Donít be fooled by the record EQ adjustments, they donít do anything. In additionÖ
  4. Do not use bias to affect record EQ. Bias is optimized for minimal distortion. All narrow format machines require more "overbias" than pro machines.   2 ½ to 3 dB overbias (using 10 kHz @15 IPS) is typical for a Pro Machine, 4 ½ to 5 dB overbias is the norm for machines mit da skinny tracks.
  5. For the terminally curious, make a record test ó with the noise reduction off ó using an oscillator set to 1 kHz. Donít freak when all of the LEDís donít line up. A segment plus or minus means little. Use the oscillatorís level control to find an average level that satisfies most of the meters.
  6. On playback, confirm that the signal is at the same level.
  7. If the edge tracks are down at mid or high frequencies, clean the heads. If the improvement is negligible or inconsistent, itís lap time.
  8. Own an alignment tape and adjust playback only. Most 15 IPS machines use IEC EQ.
Note: Noise reduction is not bad ó it should be used on narrow format machines ó but it is level and frequency sensitive.


If youíve never done an alignment, it may be better to leave well enough alone. Otherwise read the manual and have someone show you not just the procedure, but also the pitfalls. It may even be necessary to pay a technician, but this is exactly what bias is about. Bias is a radio frequency that is like currency to the tapeís magnetic particles ó theyíll make music if you pay cash up front.

Adjusting Bias is like tuning an old fashioned AM radio. There is a generally accepted "window of correctness" where the signal comes through loud and clear. Underbias and the signal will be bright on top and mushy on the bottom. Excessive bias makes dull and dirty recordings.

Table One shows typical overbias values for reference purposes. (Contact the tape or the machine manufacturer if unsure.) The amount of overbias is not global ó it is not a magic number that works for all speeds and machines. The frequency used to make the adjustment and the amount of signal reduction (overbias) are speed specific. Change the speed and either the frequency or the amount must change as per Table One. Also note that at 7 ½ IPS, high-frequency tests must be made 10 dB below 0VU (20 dB below for cassette decks).

30 IPS 
15 IPS
7 ½ IPS
(-10 dB ref)
Pro (10kHz)
1.25 dB
2.5 dB
5 dB
Pro (2.5 dB over)
20 kHz 
10 kHz
5 kHz
Narrow (10 kHz)
2.25 dB
4.5 dB
9 dB
Narrow (4.5 dB over)
20 kHz
10 kHz
5 kHz 
Table One: Typical over-bias values for pro and narrow-format machines.

If, for example, the recommended bias is "2.5 dB over @ 10 kHz," put all tracks in record, adjust the bias to find the maximum output at this frequency, then increase the bias (usually clockwise) until the signal is reduced by the amount specified. This is what is meant by "overbias."

If you are not sure how much overbias is required, record a low frequency sine wave no higher than 40 Hz. A course adjustment will deliver maximum output. Listen to what happens as the bias is varied and adjust until the fuzz and harmonic distortion are minimized. (Congratulations! Youíve just used your ears as a distortion analyzer.) Finally, with a high frequency such as 10 kHz, record and make note of its output (A), then reduce the bias until maximum output is achieved (B). The difference in dB (B-A) is the amount of overbias. Increase the bias until the signal returns to the level noted in "A."

Recording low frequencies at 30 IPS on a two-inch 24-track (or narrower format machine) is a challenge both in terms of efficiency and headroom.  Optimizing the bias for the lowest distortion at 40 hz ensures that you get all the efficiency (maximum output), "cleanest warmth" and the "cleanest saturated punch" analog is famous for while minimizing the modulation noise. 

The purpose of going back to 20 kHz (or 10 kHz for 15 ips) after the "low-frequency tour" is to make sure the same amount of overbias is applied to all channels.  Doing so will minimize channel-to-channel HF idiosyncrasies -- such as phase and level -- so that all of the record EQ settings end up in the same relative place. 

If you biased for frequency response (rather than distortion and you pretended to not have access to the HF REC EQ adjustment), the resulting phase response from channel to channel would be way out.  (Not good for recording drum overheads.)  This is almost the case in narrow format machines that have no real HF REC adjustment. 

Biasing at 20kHz (for 30 IPS) is more precise than using 10 kHz.  For example: 3 dB of overbias at 20 kHz is only 1.5 dB overbias @ 10 kHz.   By being meticulous about the bias adjustment, you may discover track-to-track anomalies that could either be caused by aging components OR a head near the end of its life.

Figure One also makes reference to the fact that increased tape speed permits higher operating levels. Elevated levels are not recommended for narrow-format machines. Operating levels have been optimized by the manufacturer for use with a Noise Reduction (NR) system. NR is built-in out of necessity simply because smaller tracks yield less electrical output and therefore have a higher noise floor. Changing tape formulation and/or operating level will make your machine incompatible with other machines. Plus, the lack of a dedicated playback head makes alignments tedious.

Using NR improves headroom, so overdriving for "the sound" of tape saturation will likely yield more NR decode artifacts than actual tape compression. Narrow format decks typically have a global noise reduction on/off setting. Defeating NR on individual tracks is limited to the edge track designated for time-code purposes. (Some machines can be easily modified for individual control of noise reduction.)

Stick with standard formulations (Ampex 456, 3M 226 or AGFA/BASF 911??). These tapes are 1 1/2 mils thick and have 1 mil cousins 457, 227 and ??? High-output tapes such as Ampex 499, 3M 996 and AGFA/BASF ??? are physically more difficult to pull through the mechanism. The resulting tension increase tension will accelerate head wear.

The warmth for which analog tape is famous comes from the composite of its idiosyncrasies. There are obstacles, primarily noise, but also mechanical problems such as tension, flutter and tape path. These will be tackled in a future article. Iíll also be taking a closer look at why tapes shed and what you can do about it. Till then, be sure to play all your tapes all the way to the end and store them, tail out.

Part 1
The Phonograph
Part 2
Magnetic Tape
Part 3
The Mastering Process
Part 4
Part 5
You are Here