Track widths and track movement of that size will need to be measured if tape cartridge capacities are to increase into the 1- to 10-TB areas. Track widths will have to be less than 1 micron. But before tape drives can achieve this, one must be able to accurately measure movements in the order of 0.1 micron in the track position. Because, if you can't measure it, you don't know if you've achieved it.
Up to now, such precision was not needed. Track pitches were wide--as much as 500 microns in early 3490s. The tape could wander laterally half the track width and the signal could be read without servos because the track was so wide compared to the reader head. Guiding was simple.
Today, however, tracks are typically 20 microns wide in LTO drives. Researchers are presently working on drives with track widths in the 5- to 1-micron region. For these small widths, tape wander, properly called Lateral Tape Movement (LTM), must be less than 1 micron. This is because the track following head always lags the moving target and produces a position error signal. If the position error signal gets to be larger than about 1/10 the track width, the track cannot be read. However, it follows that the smaller the LTM, the smaller the error is. This allows more tracks can be placed on the tape.
[FIGURE 1 OMITTED]
With the advent of precision tape drive paths such as the porous air bearing deck, which yields about .7 microns of non-repeatable LTM, the need for precision measurement of tape position and wander has arrived. The question is, what methods are suitable for such measurement and how practical are they?
A rule of thumb is that LTM should be about the same as the width of a track. If tracks of 1 micron are desired, LTM should be 1 micron or less. To measure this, the measurement system should have an accuracy of 1/10th the measured object--or in this case .1 micron.
Measuring small in the computer hardware field is just part of challenge. The advent of nanotechnology and microscopic machines in other fields makes the capability to do this even more necessary. Measurement of this sort may be possible now on a limited basis. The challenge is to do it cheaply, conveniently, routinely and accurately. Therein lies the problem.
Four methods of measuring will be discussed here. First is the Fotonic probe, a method using fiber optics. Second is the position decoding of the pre-written servo track. Third is writing a high frequency track next to a low frequency track and measuring the Hf/Lf ratio. Fourth is to measure the movement of the tape surface with a laser vibrometer.
Method 1: A Fotonic Edge Probe
This instrument by MTI of Albany, NY, gives tape edge displacement by measuring the amount of light blocked by the tape. The fiber optic probe directs a curtain of light past the measurement target edge to a receiving bundle (Figure 1). The intensity of light received changes with the edge position. This translated directly into microns of tape edge motion via the appropriate calibration setup.
The calibration curve in Figure 2 is typical of the sensitivity one gets from the Fotonic probe. From this curve, one can see that the linear portion spans about 10 volts across 100 microns, or 10 microns/volt or 10 nm/mV. (MTI's own literature claims 2.5 nm/mV across 100 microns) This is what is needed for submicron track location.
This instrument provides the sensitivity needed, but what about the resolution and bandwidth? After all, tape movement can be rapid enough that the track following servo cannot keep up. These are typically frequencies above 800 Hz. What, then, is the accuracy of the Fotonic probe for high frequency movement?
Tests of accuracy were conducted using a notched rotating disk that gave mechanically stepped square wave input. The tests showed that, with the 10K low pass filter, the Fotonic probe faithfully reproduced the input pattern up to 1500 Hz. Beyond that, distortion of the step corners was evident.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
In practical terms of measuring tape edge quality or lateral tape motion in the 800 Hz to 1500 Hz region, a region of concern to tape drive developers, the MTI Fotonic probe would give the resolution and bandwidth needed to record the motion and waveforms. It is a very good choice; however, the MTI probe is expensive, and most labs can afford only one. If multiple measurements are simultaneously needed, an alternative is needed.
A low cost alternative is to use a common photo switch with appropriate noise reducing circuitry. This has been successfully used in various tape drive development labs including Carnegie Mellon University Mountain Engineering II has developed a similar device to the place where it matches the resolution and low noise of the more expensive Fotonic probe and is cheap enough to locate multiple devices along the tape path.
The disadvantage of this method is that it measures the edge of the tape, not the surface of the tape which carries the track. It follows then that if the edge of the tape was rough but the surface of the tape moved along smoothly due to good guiding, edge sensors would measure roughness that one might misinterpret as lateral tape motion.
The way around this is to record the edge a number of times at the same location--which is time consuming. To do this one must also have software to store many runs and record the exact position along the tape repeatedly (Figure 3). Superimposing these shows the repeatable LTM, which will be the edge profile and perhaps some reel once-around. However each pass will differ slightly. That is, it will have a non-repeatable portion. This non-repeatable portion is the true lateral motion of the tape surface. A histogram can be constructed from the position of the non-repeatable readings so that a statistical picture of the LTM spread may be seen.
Method 2: Position Decoding of a Written Servo Track
LTO and most other tape formats have a pre-written servo track. This servo track is, in a sense, an artificial edge positioned on the tape. Measuring this comes closer to measuring what is wanted: the absolute lateral position of the tape surface.
In open loop mode with the head fixed, the absolute position of the servo track is read. However, since it will measure the full LTM, both repeatable and non-repeatable, the multiple pass techniques of the previous section will have to be used to separate the repeatable and non-repeatable portions.
The advantage is that it does not require complex or expensive equipment. However, since servo tracks themselves must be written on special drives called servo-writers, there is a written-in error that is determined by the servo-writer's own accuracy in guiding the tape. Usually the servo-track error is small, a micron or less, and can be ignored in present day drives where the LTM is in the order of 20 microns.
However, as the industry approaches 1-micron tracks where the LTM must be 1 micron or less, the accuracy of the servo-writer becomes an important question. Measurements of servo track accuracy on different brands of tape show there is a difference in position accuracy that cannot be ignored The bottom line is that if one's guiding accuracy is limited to today's technology (20 microns), then this method is fine for measuring LTM. However, if one's guiding system yields 1 micron or less of LTM, then this method is marginal. Its results must be checked against other methods.
Method 3: Writing One's Own "Servo" Track
The problem with a pre-written servo track is interchange. If the servo track could be read back on the same machine it was written on, the repeatable LTM would be canceled out. All that would be left was the non-repeatable or true LTM.
One way to do this is to use an available tape deck, not necessarily of servo-writer quality, and write adjacent tracks--one of high frequency and one of low frequency. This method was used to write actual servo tracks in the 3590 era. With a fixed servo head straddling the two tracks, the ratio of low frequency to high frequency is measured. This ratio is proportional to tape's lateral position, a 50% ratio being on dead center. This gives a direct measurement of non-repeatable LTM. However because this gives the additive non-repeatable LTM of two runs, the result must be divided by two.
This method is fairly straightforward and can be used on tape drives available in the lab. The disadvantage is that calibration is very sensitive. It must be done with a head moved across the Hf/Lf margin with a micrometer stage. Head width, written track signal consistency, and other factors are variables, each of which adds complication.
Method 4: Using a Laser Surface Vibrometer
There are several instruments using lasers and the terms are sometimes confusing. Polytec of Germany, for example, makes non-contact surface measuring equipment for different applications. One is a Laser Surface Velocimeter, which is used in the steel industry to measure the velocity and length of sheets and bars as they are being rolled. It is meant for high-speed measurement. Lab tests on tape show that although it measures continuous velocity well, it does not have the resolution needed for instantaneous speed variation (ISV).
Laser vibrometers come in two kinds: in-plane and out-of-plane. In-plane is called a Laser Surface Vibrometer. Out-of-plane is called a Laser Doppler Vibrometer. Out-of-plane vibrometers measure movements along the axis of the laser beam and are common in the computer industry for measuring the vibration of actuator arms in disk drives. In-plane vibrometers measure movements perpendicular to the beam axis, similar to the laser surface velocimeter. The advantage of the in-plane vibrometer is that it can measure a continuous (DC) velocity and superimposed variable (AC) component, or instantaneous speed variation of a moving surface. This sensitivity makes it potentially useful for measuring speed variations of the tape and the quality of the speed control system.
If the laser head is rotated 90 degrees to the tape motion, there is potential that it can pick up the lateral motion of the tape surface. Integrating the lateral velocity produces the lateral displacement, or the actual LTM of the surface.
In the laboratory, this is challenging. For confidence in the results, the instrument should be calibrated on an object whose characteristics are known. Calibration on a shaker table with a moving tape target where the amplitude, frequency, and G-level are easily obtained by calculation showed the vibrometer to be very accurate. However, it took 30 averages of the sinusoidal tape surface motion because of the poor reflectivity of the tape. In a moving tape path, there isn't the luxury of averaging many passes to reduce the noise. Integrating the noisy velocity signal gave a noisy displacement signal that was not useful.
The conclusion is that on moving tape the results of a Laser Surface Vibrometer are not reliable. The tape surface is simply not suitable for reflected signal needed by this instrument. I have included the Laser Surface Vibrometer this discussion, not to recommend it, but to caution users because many tape laboratories have such equipment and may be tempted to push it beyond its limits of accuracy.
The Table summarizes ways to measure tape movement. Each has its advantages and disadvantages.
A combination of methods is best. Method 1 in combination with either method 2 or 3 would give a complete picture from very different measuring methods. In the laboratory, the first three methods gave LTM results less than .3 microns apart. All would be satisfactory for measuring LTM in the order of 1 micron. Method 4 is not recommended.
Method Characteristics Advantages Disadvantages 1. Optical edge This measures the This gives a full Time consuming, probe such as tape edge picture of the and not entirely an MTI Fotonic undulations motion of the representative of probe relative to a reference edge of movement of the fixed point. the tape, with surface of the As such, it repeatable and tape. measures both the non-repeatable Repeated runs edge raggedness components with the same of the tape and visible. tape on the same the lift-off of Calibration of deck are needed the tape from the the probe is to establish the reference or easy, is usually non-repeatable guide surface. supplied with the portion of the This means that probe, and can be signal. in subsequent verified with a Statistically, passes, it micrometer stage 30 runs are a measures both the setup. minimum to repeatable establish a portion (the tape Gaussian spread edge) and non- giving a picture repeatable of the LTM portion (the random lift-off) of the lateral motion. Repeated runs are needed. Synchronization of the data from run to run is done by decoding the LPOS information. 2. Servo track This uses the Uses a pre- Disadvantages of decode position written track repeated runs information already on the also apply here. encoded in the tape. Measuring The written-in servo track. A this position LTM of the servo- stationary head gives a closer writer, although reads the track. approximation to small, is The lateral the movement of included. position the body of the This requires information tape. that a tape have together with the Gives the full a servo track. LPOS is decoded LTM, both the Not all tapes and the data from repeatable and have one. all runs is non-repeatable synchronized. portions. 3. Write and Two adjacent This method is Calibration is read a reference tracks fast. It does not very sensitive. reference track are written on require repeated It must be done the tape, one passes but gives with a head moved high frequency the non- across the Hf/Lf and one low repeatable LTM of margin with a frequency, then a single pass micrometer stage. read back with a over time, for Head width, read head example 15 written track positioned seconds of a signal halfway between. single pass. As consistency, and The read head such, this other factors are should be small represents a variables, each compared to the fundamental of which adds track width for difference from complication. greatest methods 1 and 2, sensitivity. The both of which are ratio of Hf/Lf is a slice of data proportional to at one place the LTM. along the tape This method rather than a measures the slice across combined non- time, a vertical repeatable LTM of vs. a horizontal both the write slice as it were. pass and the The method of read pass. All writing a values are reference track therefore divided and then by two, and the decoding the result represents position of the the LTM of a track in single pass. subsequent read passes closely resembles the decoding of a pre-written servo track for input to the actuator. However this method is closer to measuring movement of the body of the tape. 4. Direct This measures the Gives direct Setup for optical tape surface measurement of measurement must measurement of motion rather velocity and be rigid. lateral surface than the edge displacement if Alignment of the motion with an motion. As such the S/N ratio is reflected beam instrument such it eliminates the high enough. off of moving as a Polytec edge raggedness tape is very in-plane Laser component as do sensitive. Surface methods 2 and 3. The reflected Vibrometer beam is too weak and noisy to give accurate results for the typical tape. This method is not recommended.
By Gary Collins, P.E.
Gary Collins, president of Collins Consulting (Boulder CO), is a professional mechanical engineer and has consulted for the storage industry for over 20 years. He can be reached at: firstname.lastname@example.org
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||measuring instruments|
|Publication:||Computer Technology Review|
|Date:||Mar 1, 2005|
|Previous Article:||High performance media for high performance drives.|
|Next Article:||Gartner: LTO now dominates DLT technology; Sales of SDLT drives decline for first time.|