Open Reel Recorders: Erasure and Demagnetization (April 1981)

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting

Departments | Features | ADs | Equipment | Music/Recordings | History

The Mechanism Of Magnetic Tape Erasure

by: PETER VOGELGESANG [Manager, Advanced Recording Technology, Magnetic A/V Products, 3M Company, St. Paul, Minn.]

Certainly one of the greatest virtues of magnetic tape as a recording medium is the ease with which old information can be erased to make room for new information. Com pared to other kinds of recordable media, such as photographic film which is chemically processed and permanently written, magnetic tape can be used thousands of times through the faculty of erasure. Although the process of erasing tape is per formed automatically in recording systems and manually with the aid of bulk demagnetizers, the precise mechanism of era sure within a tape is not commonly understood. Perhaps this lack of understanding results from the apparent simplicity of the erasing process--a simplicity which does not challenge the interest of those who use the process. But tape erasure is not as simple as it appears, and perhaps the following description of the erasure mechanism will both interest and enlighten.

To understand tape erasure one must first understand the composition and behavior of the magnetic material used in tape.

The most commonly used material is the gamma form of iron oxide, or gamma Fe2O3. This material exists as tiny, needle-shaped particles having an average length of 0.4 micron, and is shown magnified 42,000 times in the photomicrograph of Fig.1. Although other magnetic materials such as chromium dioxide (CrO2), cobalt modified iron oxide, and metallic materials may have different chemical compositions and physical structures, they respond magnetically in a manner similar to gamma iron oxide.

Each iron oxide particle is a permanent magnet containing only a single magnetic domain. The single -domain structure of the particles dictates that each particle is forever a permanent magnet which cannot be demagnetized. If subjected to an external magnetic field of sufficient intensity, a particle will reverse the polarity of the field it generates, but the reversed field will not be greater or smaller in magnitude than the original field. Thus, each particle can be thought of as the magnetic equivalent of an electronic flip-flop which has two stable states. Just as a flip-flop can be switched into one of two saturated states, energy applied to a single-domain magnetic particle can cause it to reverse polarization virtually instantaneously but cannot cause it to generate a magnetic field of variable intensity within the particle.

The switching characteristic of a particle is illustrated graphically in Fig. 2. A magnetic field which varies sinusoidally in amplitude is applied to the particle so that the direction of the field is parallel to the long axis of the particle. (A different result is obtained if the applied field is at right angles to the axis.) Particles in most magnetic tapes are aligned along the length of the tape by immersing the tape in a longitudinal field while the magnetic coating is still fluid. Once dried, particles in the coating remain physically aligned in that direction.

As the applied field in Fig. 2 increases from zero, a level is reached (point a) where the coercivity of the particle is equaled, and at this point the field within the particle instantaneously reverses.

The particle will not again switch until the applied field has reversed and reached the equal but opposite magnitude. The hysteresis waveform of a single particle will be recognized by electronic engineers as similar to the graphs used to depict the behavior of electronic flip flops and Schmitt triggers.

Each area of a magnetic tape which is uniquely magnetized during recording contains thousands of magnetic particles throughout the width and depth of the recorded track. The magnetic field emanating from a recorded area is the sum of the fields produced by these individual particles. A magnetic tape is said to be saturated when all the particles are switched with the same magnetic polarity. A discrete area of tape thus saturated will produce the maximum external magnetic field of which the tape is capable.

In the erased condition, one-half of the magnetic particles are switched with one polarity while the remaining half retains the opposite polarity, and since the particles composing these halves are closely intermixed, their external fields cancel, producing zero external field from the tape.

Fig. 1 -- Photo micrograph of iron oxide particles used in magnetic tape, 42,000 X magnification. The particles magnetic properties are controlled largely by their shape, and their magnetic characteristics from stimulation fields which are parallel to the particles are different from ones which are perpendicular. The long dimension of the particles is aligned parallel to the recording direction in most tapes.

Fig. 2 -- Magnetization characteristics of a single -domain particle.

Varying levels of tape magnetization are produced during the recording pro cess by causing the particles to be switched to the opposite polarity in pro portions varying between the saturated (100% switched in one direction) and erased (50%-50%) states. It is seen, then, that the analog recording process in the strictest sense is really not analog at all but is a binary process in which the magnetic outputs of thousands of flip flop elements are summed in the pole pieces of the reproducing transducer.

It is also apparent from the foregoing that the magnetic transfer characteristic of a tape would not permit the recording of an analog signal if the coercivities of all particles in the tape were identical. A tape made with such magnetic material could exist only in alternately saturated states of two polarities, and would have a hysteresis curve identical to that shown in Fig. 2. (Note: In reality an analog signal could be recorded on a tape having a rectangular hysteresis curve because of the three-dimensional geometry of the magnetic field produced by a recording transducer [1]. In this instance, varying magnetization of a tape would be achieved by modulating the depth of magnetization within the thickness of the magnetic coating. The recording characteristics of such a tape, of course, would be quite different from conventional tapes.) Fortunately, the millions of particles that exist in even a small section of tape have coercivities which cover a broad range. The population of coercivities is similar to the distribution of a nor mal curve, as shown in Fig. 3. A value of coercivity assigned to a specific tape refers to the average coercivity of the oxide particles, and in most instances, this value will be the peak of the distribution curve.

Fig. 3 -- Distribution of coercivities in a large group of particles.

Fig. 4 -- Relationship of a magnetic material distribution curve to an erasure field.

Fig. 5 -- Measured distribution of a high-energy tape having a coercivity of 650 oersteds.

The task of erasing a magnetic tape can be defined as the necessity of establishing a condition whereby all external fields are zero as a result of a one-to-one ratio of oppositely polarized, closely intermixed magnetic particles [2]. Most explanations of a.c. erasure utilize the expedient of a family of hysteresis curves to show how the mechanism of erasure works [3]. An alternative explanation which may be more conducive to an understanding of the physical events which occur in the erasure process is presented with the help of Fig. 4. The uppermost curve represents the distribution of coercivities of magnetic particles used in a tape, where the greatest population occurs at 300 oersteds (the specified coercivity of the tape). The wave form below the distribution curve represents an applied a.c. field which varies sinusoidally and which diminishes gradually from a high magnitude to zero. This diminishing field is the kind of field that a tape will experience as a bulk degausser is slowly withdrawn from its vicinity or as a particular area of the tape moves away from the erase head of a recorder. The shaded half-cycles represent a field of negative polarity and for the purpose of illustration are folded over from the negative side of the zero line. The unshaded half-cycle portions represent the positive polarity.

Starting with half-cycle a, the peak applied field is of sufficient magnitude that all particles are switched in the positive direction, and at that point the total magnetic material is saturated positively.

Half-cycle b also saturates the tape, but with a negative polarity. Half-cycle c again reverses most of the magnetic particles, but leaves a small percentage in the negative polarity because the de creased magnitude of the field is too low to reach the highest coercivity particles.

The largest part of the particles are yet again reversed by half-cycle d, but this time a greater percentage are left in the positive polarity because of the diminishing erasure field. This process continues until the erasure field reaches zero. At this point note that approximately half of the particles remain switched positive and the other half are switched negative; in other words, the sum of the fields produced by all the particles is zero. Note also that the probability of arriving at a one-to-one ratio of polarities is greater as the number of half-cycles is increased (longer diminish time). This analysis shows why a tape must be withdrawn slowly from an erasure field to insure full erasure. It also shows why the field of a bulk degausser cannot be interrupted until the tape being erased is outside the influence of the field. To obtain complete erasure, the applied field must be greater in magnitude than the coercivity of the highest coercivity particles in the tape.

An erasure field which is merely equal to the specified coercivity of a tape will not totally erase it; the field will switch only those particles with coercivities in the lower half of the distribution curve, while those particles in the upper half will be unaffected.

Of increasing concern today is the erasure requirement of the so-called "high-energy" tapes which have coercivities considerably higher than gamma iron oxide. As a general rule, the width of the distribution curve of these high-coercivity materials increases in proportion to the increase in specified coercivity. Consequently, a tape which has a specified coercivity twice as great as another will require an erasure field of double the magnitude.

Fig. 6 -- Erasure curves for a gamma iron oxide tape using parallel and perpendicular erasing fields. The effective distribution of coercivities is broadened when the applied field is perpendicular to the direction of orientation, substantially increasing the field strength needed for erasure.

Fig. 7 -- Three sources of d.c. erasure noise. Any irregularities in a magnetic tape which cause the head gap to "see" varying magnetic material will generate noise when the tape is d.c.

erased and will also cause modulation noise from an a.c. erased tape when it is recorded with an analog signal.

The measured distribution of a 625-oersted magnetic tape is shown in Fig. 5, and it can be seen from this curve that a significant portion of the particles have a coercivity above 1,000 oersteds. The signal which remains after a tape has been erased to a level 60 dB below the maximum output level of a tape may be stored by only one -tenth of one percent of the magnetic material in the tape, and, of course, that small portion of the material will be at the uppermost end of the distribution curve. The tape represented by the curve of Fig. 5 requires an erasure field intensity of 2,000 oersteds to achieve this level of erasure.

The peak field requirement for a bulk degausser is even greater than that of an erase head, because in most bulk degausser configurations the direction of the erasing field is at right angles to the direction of tape orientation. (The exception, of course, is quadruplex videotape, where orientation is across the width of the tape rather than along its length.) Erasing with a perpendicular field effectively broadens the distribution of the magnetic particles. Figure 6 shows de magnetization curves for a gamma iron oxide tape in which the erasure field is first parallel with the direction of orientation and then perpendicular to it. Note that erasure starts taking place at a lower magnitude of erasure field for the perpendicular case, but that more than 200-oersted greater magnitude is needed to achieve maximum erasure equivalent to the parallel case.

Since bulk degaussers almost invariably use a magnetic field which penetrates the flanges of a reel or the wall of a cassette, the most intense component of the field is perpendicular to the tape.

Erase heads, on the other hand, have magnetic gaps which generate intense parallel as well as perpendicular erasure fields and consequently are somewhat more efficient than bulk degaussers.

Also, the erase head is in contact with the tape, whereas the bulk degausser must penetrate the thickness of the tape container. While a magnetic field experiences no difficulty in penetrating plastic or cardboard material, the physical separation caused by the container can greatly reduce field intensity. In selecting a degausser to erase a particular tape product, or in the design of degaussers and erase heads, the principal parameters which must be considered are:

1. The specified coercivity and distribution of the magnetic material involved.

2. The direction of the erasure field relative to the orientation of the tape.

3. Separation of the tape from the erase field (in the case of bulk degaussers the field must penetrate the full thickness of the tape pack).

4. The level of erasure which is required.

These parameters will be translated into an erase field intensity.

Most purchasers of bulk degaussers do not have the instruments needed to measure tape distribution or degausser field intensity and so must rely on the claims of manufacturers relative to the suitability of a specific product. Perhaps the best means of determining suitability is to erase a saturated recording with the degausser under consideration and then measure the magnitude of any remaining signal.

The discussion of tape erasure to this point has dealt only with a.c. erasing fields, which are commonly generated by erase heads used in recorders and by bulk degaussers. In terms of ridding a tape of a recorded signal, drawing the tape over the end of a permanent mag net will erase a signal as effectively as any other means. Permanent magnet erasure was used in early recording systems for the sake of simplicity, but it has been abandoned for years in favor of a.c. erasure. Only in a very few applications of magnetic recording is permanent magnet erasure still used.

D.C. erasure was abandoned to minimize the noise which is generated by tape erased in this manner. D.C magnetization of a tape will cause any irregularities in tape surface and thickness to generate fields which appear in the reproducing transducer as noise. Since the field generated by a given area of tape is the sum of the fields of all the particles within that area, it can be seen that localized concentrations or rarefactions of oxide will produce different sums when all particles are polarized in the same direction. The differences in these sums constitute a varying external magnetic field and a noise signal.

Great care is taken in the tape manufacturing process to produce uniformly dispersed oxide in the magnetic coating, but slight density irregularities are unavoidable. These irregularities, combined with slight thickness variations of the coating and slight surface irregularities, produce unacceptable noise when a tape is magnetically saturated in order to erase a signal. (See Fig. 7.) Strong evidence that d.c. erasure noise is caused by physical irregularities in tape is provided by the simple test of comparing the output noise waveforms from the same area of a tape on two successive passes where polarization of the d.c. erasure field has been reversed.

One would expect the noise waveforms to be identical but inverted, as indeed they are in Fig. 8.

Fig. 8 -- Noise waveforms generated as the result of reverse magnetization.

A correlation coefficient between two noise waveforms was determined by measuring the waveform amplitudes at 50 different sample points spaced at equal time intervals and then mathematically computing their similarity. When the tape was magnetized by passing it over a recording head excited with direct cur rent, correlation of the two waveforms was -0.89, where ideal correlation would be -0.9 (rather than -1.0) because of amplifier and system noise. This high degree of correlation is convincing evidence that noise output of a d.c. saturated tape results from permanent physical features of the tape.

The same density variations that cause d.c. noise also give rise to modulation noise, that is, noise which occurs in the presence of a recorded signal. Imagine, for example, that a tape has been a.c. erased and then recorded with a long -wavelength, high-level audio signal.

Insofar as the segment of tape which contains a half-wave of the signal is concerned, magnetization within this segment is no different than magnetization which might have occurred from d.c. erasure. Once again density variations produce external fields, but in this case the noise which is generated has a frequency relationship to the recorded signal and is termed modulation noise. D.C. erasure noise and modulation noise thus have substantially the same origin but are manifested in different ways. While d.c. erasure noise can be virtually eliminated by using a.c. erasure fields, modulation noise is minimized only by judicious design and manufacturing of magnetic tape.

In conclusion, the magnetic recording process is almost always preceded by an erasing process, either by bulk degaussing or by magnetic -head erasure.

Erasure is a fundamental step in achieving high -quality recordings. The simplicity with which erasure is performed belies the complexity of the process. It is hoped that the foregoing has created a little higher regard for this seemingly mundane procedure.


The author wishes to acknowledge the work of E.F. Wollack of the Data Re cording Products Division of 3M Company, whose research of the recording pro cess contributed significantly to the information presented.


1. Mee, C.D., The Physics of Magnetic Recording, North -Holland Publishing Co., 1964; John Wiley and Sons, Inc., distributor.

2. Pear, C.B., Jr., Magnetic Recording in Science and Industry, Reinhold, 1967.

3. Jorgenson, Finn, Handbook of Magnetic Recording, Tab Books, 1970.


Focus On Head Demagnetization


Until recently I accepted without question the widely propagated dictum that tape heads must be periodically demagnetized -- roughly after about 8 to 16 hours of use. My acceptance became embedded in concrete several years ago, but cracks have developed in this concrete as a result of several letters from "Tape Guide" readers and some inquiries I have made.

At this point perhaps a brief review would be appropriate of' the commonly given reasons for demagnetizing tape heads; in other words, The Gospel: The waveforms of most source material are asymmetrical, thus in effect contain a d.c. component, which tends to magnetize the heads. Distortion in the bias wave form has a similar effect. Magnetized heads act as erasing de vices, particularly as recorded wavelengths on the tape grow shorter, i.e. as frequency rises at a given tape speed. Further more, magnetized heads translate physical and magnetic irregularities in the tape into noise. Altogether, it is claimed, magnetized heads produce treble loss and noise.

Several months ago I received a challenge to this convention al wisdom from Ramon Valdes in Miami Lakes, Florida. He wrote: "Two years ago I did a test to see how important tape recorder head magnetization is. I took a TEAC 7010 GSL tape deck with automatic reverse and optimized it for 3M 206 tape; a test tape was then made with tones from 20 Hz to 15 kHz. I then took a 10 1/2 -inch reel of 206 tape and proceeded to record at 7 1/2 ips three hours at 20 Hz, three hours at 1 kHz, and 3 hours at 15 kHz in that sequence for 100 hours. I cleaned the heads every three hours but did not demagnetize them. After 100 hours, I played back the original test tape and also rerecorded the test tones. All parameters (distortion, signal strength) were the same as before the 100 hours of recording. I went another 100 hours before my patience ran out -- the results were the same."

My reply to Mr. Valdes was essentially as follows: "If I under stand correctly, at 7 1/2 ips you recorded and played tones of 20 - 1,000 and 15,000 Hz, each 3 hours at a time, for a total of 200 hours and after such use found no deterioration in the ability of the heads to record and reproduce a 15,000-Hz tone relative to 1,000 Hz.

"However, I have several problems with your findings. The first* is that you used a speed of 7 1/2 ips. A number of adverse magnetic tape phenomena, including erasure due to magnetized heads, are wavelength effects. That is, they become more severe as tape speed decreases. I wonder what you would have found if the same test were con ducted at a lower speed, say the cassette speed of 17/8 ips.

"Second, gradual magnetization of heads is attributed to the asymmetric nature of the typical audio signal. For your test, you probably used very pure sine waves devoid of significant asymmetry.

"Further, your letter says nothing about noise. In looking for head magnetization effects, one should look not only for deterioration of high -frequency response but also for increase in noise." Mr. Valdes shortly replied: "Your first paragraph is correct; that is the way I ran the test.

"Concerning your second paragraph, I only ran a 7 1/2-ips test since that is the (minimum) speed I consider useful for serious recording.

"Concerning the third paragraph, I did use very pure sine waves because that was then the only way I had to mea sure distortion and frequency response.

However, now I have a method of re cording precise square waves with different rise and fall times, resulting in asymmetrical waveforms. I can view them with a spectrum analyzer and photograph the screen. Let me know if you think making before and after comparisons, following 200 hours of deck operation, would be a valid test. Using a musical waveform would show whether the heads become magnetized, but only gross distortion would be noticed.

"Concerning your fourth paragraph, the noise increased after 45 hours, but it was due solely to tape erasure by the erase head. When the tape was bulk -erased, the noise disappeared." In turn I wrote: "It would be interesting to see if asymmetrical waveforms recorded over a long period do raise the level of magnetization of a tape head, and to compare the reproduced waveforms at the end and beginning of the test." As of the present writing, it is several months since my last letter to Mr. Valdes. This may well be too short a period to permit him to give an account of the results of running his tests with asymmetrical square waves. However, in view of continuing reader interest in the subject of head demagnetization over the years, I thought it best not to delay the present article by waiting further for this account.

About the same time that I received Mr. Valdes' first letter, another challenging letter came from Henry B. Ruh of Owen Valley Broadcasters, Inc., Ellettsville, Indiana: "... From my 15 years of broadcast and audio experience, let me state that if your deck needs to be de magnetized, you probably need a new deck! Back in the days when a permanent magnet was used for erasure .. . the continuous stream of one-way magnetized tape particles would over a period of time tend to magnetize items downstream from the erase magnet.

Thus, you had to demagnetize the heads, guides, capstans .... But when the cost of a high -power a.c. erase sys tem was reduced to a reasonable level and permanent magnet erase heads were eliminated, so was the problem of magnetized heads.

"Although I have checked hundreds of times on many different makes and models of tape machines, I have never found on a modern machine any residual magnetism which could in any way erase any portion of the signal on the tape....

"In conclusion, you can throw your head demagnetizers away and forget the problem. While in the service field, I en countered more magnetized tape heads caused by improper use of demagnetizers than I found blown fuses."

Comments By Several Authorities

After hearing from Messrs. Valdes and Ruh, I solicited views on the matter from several knowledgeable persons in the field of tape recording. While they do not necessarily constitute a representative sample, their positions lend authority and interest to their comments.

Delos A. Eilers, Technical Service Supervisor at 3M Co., St. Paul, Minnesota, wrote: "Head demagnetizing at regular intervals is a good preventive maintenance practice. What that interval should be is difficult to define.

"It is true that improper head erasure can cause more problems than if one didn't demagnetize the heads. If the de magnetizer isn't gradually pulled away from the heads, you won't be showing the head a gradually decaying magnetic field. You will leave the head magnetized. A slowly decaying field is essential for good demagnetization.

"The magnetization that a head can build up over a period of time is not from the earth's field but more from accidental exposure to magnetized particles or build-up of residual magnetization from nonsymmetrical bias or program signals.

Thus, we recommend that heads be demagnetized after every 10 to 15 hours of recording." The chief engineer of a well-known manufacturer of tape decks (who prefers not to be identified) indicated that the need for head demagnetization appears to vary from one deck to another. To illustrate, he cited his ownership of two fine decks, one requiring frequent de magnetization in order to prevent obvious treble loss, and the other apparently never requiring it. To play it safe, he periodically demagnetizes both.

Don Eger, an engineer with Crown International, Inc., gave the following re marks: "Our experience with tape re cording heads during the past two decades does not support occasional claims that tape head demagnetization is un necessary or harmful.

"The effect of using a head that has become magnetized is Loss of short wavelength material from the tape, as such material is erased by the magnetism of the head. If the tape is continually exposed to a magnetized head, the re cording will lose most of its high-frequency content.

"Occasional dissatisfaction with the demagnetizing process experienced by some recording enthusiasts may be a result of degaussing the head too quickly and/or improperly. If the degaussing coil is removed slowly from the head, it will demagnetize the head properly." Finally, we quote Henry C. Pollak, an electronics engineer who has spent much of his career designing profession al tape equipment and servicing consumer audio equipment, including tape decks. At one time he owned Western Radio Lab of Sunnyvale, California, an audio service shop. (It may be added that he is co-author, with me, of the book Elements of Tape Recorder Circuits, which was published in 1957.) . He stated:

"My experience with demagnetizing play heads has been that I have never found any benefit. It's like washing be fore eating. My mother said to do it, so I wouldn't think of not doing it, although now I'm certain that nothing adverse would happen if I neglected my upbringing. At Western Radio Lab, as a matter of course, we demagnetized all decks that we serviced. The amount of magnetization the play head receives from the tape oxide is extremely small. The narrowness of the hysteresis loop of the play head material is so small that the remanence of the head would be trivial.

And, anyway, the next playing of a loud passage would restore the head to its previous condition.

"I would worry more about the harm a head demagnetizer can do (if carelessly used): Bending flimsy tape guides, cracking plastic covers, or scratching head surfaces. Most demagnetizers are too bulky to do a proper job on the cassette heads. At Western Radio Lab, we covered the tips of the demagnetizer with mylar tape to prevent scratching the heads.

"There is an argument for preventing d.c. from flowing through any head--erase, record, or play. And, to minimize noise, it is important to have minimum distortion in the bias and erase wave form."


From the foregoing, it doesn't appear that a definitive case can be made for or against head demagnetization. However, some conclusions can be drawn:

1. Demagnetization of the heads is advisable as a precautionary tactic, for some decks might need it.

2. If used, a demagnetizer should be operated with proper care so as not to leave the heads in a more magnetized condition than previously. The device should be turned on while several feet from the heads, brought slowly to the heads, moved in slow circular fashion while near the heads, removed gradually, and turned off at a distance of several feet. The tape deck should have its power off during this process.

3. If the demagnetizer is used, care should be taken to avoid physical harm, such as scratching the tape heads with the tips of the demagnetizer, bending any of the tape guides, or damaging a sensitive record-level meter by subjecting it to an excessively strong magnetic field.

The "Tape Guide" will be pleased to hear further from readers who can sup ply authoritative information on the pros and cons and cautions of tape head de magnetization. Assuming receipt of such items, the subject will be continued.

(Audio magazine, April 1981)

Also see:

The Effectiveness of Bulk Erasing (Jan. 1988)

Tape Recorder Maintenance (Apr. 1982)

Beta Hi-Fi: Better Audio for Video (May 1983)


Top of Page   All Related Articles    Home

Updated: Friday, 2019-04-19 8:36 PST