|
New Scanning Method for LIDA Scale Tape
Ever increasing demands regarding the measuring accuracy of linear and rotary encoders mean that often the design of these units must be reconsidered. Often small changes are sufficient for a more efficient performance, but sometimes entirely new developments are required. A new scanning method has been developed for HEIDENHAIN's LIDA steel scales and has been introduced in the LIDA 105 measuring system. This new design, referred to as single-field scanning, has smaller grating periods of 40 µm, but still has large mounting tolerances and remains customer-friendly. The big advantage is that it is less sensitive to ever-present scale waviness and also to contamination. As a result of this, the scanning signals are of very high quality and can be interpolated to a measuring step of 0.1 µm.
Features of Incremental Measuring Systems
Besides measuring accuracy, measuring length and signal periods, the other main features of incremental measuring systems are mechanical dimensions and electrical properties of the output signals. There is a close relationship between these features and the type of scanning process used (usually photoelectric). What is especially important for the user is the resulting measuring accuracy. Therefore, the manufacturer must offer a suitable and system-specific range of scanning methods.
Accuracy
The accuracy of a measuring system depends primarily on the quality of the scale tape and the scale graduation. This in turn depends on the accuracy and width of the lines, and the quality of the line edges, on the density and the size of imperfections and on the homogeneity of optical characteristics such as degrees of reflection, transmission and absorption. Besides this, the topography of the scale also plays an important role: roughness, scale waviness and flatness must satisfy small tolerance limits. It is important to remember that when the scale is in use, it is possible it will become contaminated. Of course, it is also important that the encoder be mounted precisely.
Impact of the Scanning Method
How does the scanning method affect the precision of the encoder? We must differentiate between two effects. On the one hand, the sine and cosine signals emitted by an incremental measuring system are not optimal, even if the scale is perfect. When these signals are then interpolated, interpolation deviations often result – the typical incorrect signal period pattern. On small measuring paths, these deviations are responsible for the measuring inaccuracies and are therefore very disruptive in many applications. The second effect concerns the differences in sensitivity of various scanning processes to work tolerances and to contamination on the scale tape. It should be mentioned that contamination on a scale tape can lead to important errors.
Incremental Linear Encoder LIDA
In 1977, HEIDENHAIN introduced a new LIDA encoder for long measuring lengths. This encoder has a graduation of reflective gold lines and alternate light absorbing spaces on a steel scale. Scanning is photoelectric and measuring lengths of up to 30 m can be scanned with high accuracy. The grating period is 100 µm and, after interpolation, measuring steps of 1 µm are possible. LIDA encoders are used in big and small measuring machines alike, in PCB drilling machines and as sealed units in large machine tools. More recently, it has also been used increasingly in linear servo drives.
The New LIDA 105
In recent years, more and more has been expected from linear encoders. A redesigning of the LIDA became necessary and the result of this was the LIDA 105. This LIDA 105 has the now well known distance-coded reference marks. After switch-on, the absolute position can be read from the reference marks after a maximum traverse of 80 mm. Measuring accuracy is improved in this encoder – a new scanning process allows the typical errors on scale encoders to be compensated.
Figure 1: Photoelectric scanning method used on reflective scales until now.
Photoelectric Scanning, Old and New Demands
The scanning method of the old LIDA encoder was reliable and had a large scanning gap. Figure 1 shows the old scanning process: light from a light source passes through a condenser lens, then through the four scanning fields of the scanning reticle and onto the incremental scale graduation. Here it is reflected, and after passing back through the scanning fields, it is directed onto the solar cells. Four scanning signals are generated here, which are phase-shifted by 90 degrees, and from which the typical zero-symmetrical output signals of the encoder are formed. The high quality of the signals allows 100-fold interpolation. The work tolerances of the steel scale are important here – they are significantly larger than in glass scales.
In order to improve the measuring accuracy of the LIDA 105, and in particular, the interpolation accuracy, the first aim was to reduce the grating period from 100 µm to 40 µm, without changing the size of the scanning gap of 0.8 mm, or the customer-friendly mounting tolerances. Secondly, it was important that sensitivity to scale waviness and to possible contamination be reduced.
Influence of Scale Waviness
Because the steel scales used in these encoders are so thin, sometimes the smoothness of the graduated surface is not equal to that of optical model encoders. Local sloping imperfections of up to 0.3 degrees are possible. In this case, the reflecting light rays are deflected slightly, and conventional scanning results in a somewhat displaced shadow pattern of the scale graduation being produced on the scanning reticle. Depending on the degree of sloping, the recorded position is slightly distorted and the measuring values are inaccurate. It has been shown that a so-called "neutral center of rotation" exists for every scanning process (see Figure 2). According to the definition, either the scanning head or the scale can be moved (very slightly) around this point without the position value changing. If the effect of uneven surfaces on measuring accuracy is to be kept to a minimum, then this center of rotation must be located on the level of the scale graduation. In the conventional scanning process in Figure 1, this point is however, located on the level of the index grating. This unfavorable location may be explained by the disruptive interaction between the displaced shadow pattern of the scale graduation and the amplitude grating of the scanning reticle (transparent and opaque lines). This can be avoided if the index grating is a phase grating, as opposed to an amplitude grating. All of the light reflected by the scale then passes through the scanning reticle without any losses or distortion, regardless of whether the surface of the scale is completely flat or not. Here the neutral center of rotation is located in the scale graduation and uneven scale surfaces have little or no effect.
Figure 2: The comparison of two scanning methods with the neutral center of rotation located in two different areas; on the left, the center of rotation is in the index grating and on the right, it is the scale.
How can a phase grating be used, however, to scan the amplitude grating of a measuring scale? The answer lies in the ability of a grating to produce an image of itself (see Figure 3). In 1836, Lord Talbot discovered that parallel light rays passing through an amplitude grating do not only result in a shadow image line pattern (intensity modulation). Interference effects produce other intensity-modulated areas at certain distances – the so-called Talbot distances. Further examination of the Talbot effect shows that between these areas of high intensity modulation, a phase modulation of the optical wave is produced at approximately half Talbot distances. From this phase modulation, interference causes another intensity modulation, again at half Talbot distances. It is precisely this effect that can be used, i.e., the phase modulation is produced by the phase grating. The amplitude scale grating can be scanned using the intensity modulation at half Talbot distances. Using a phase grating has another advantage -- the scanning gap and the scanning gap tolerances can be increased considerably.
Figure 3: Self-imaging ability of gratings: Both amplitude and phase gratings can produce line patterns at certain distances with alternate high (yellow) and low (gray) intensity.
Figure 4 shows the result of an experiment in which a LIDA measuring scale with a grating period of 40 µm is laid over a step with a height of 200 µm. With the conventional scanning process, the scanning head of the LIDA prototype produces increasing interpolation deviations around the area of the step, as can be seen from the width of the measuring curve. Grating deviation with an amplitude of approximately 2 µm are superimposed.
Figure 4: Influence of scale waviness on measuring accuracy. Scanning gap in LIDA prototype between 150 and 350 µm, scanning gap in LIDA 105 between 700 and 900 µm.
The new LIDA 105 shows significantly smaller interpolation deviations in the entire area. A small grating deviation of approximately 1 µm can be seen in the area of the step. When comparing both systems, the difference in size of the scanning gap must be taken into consideration. As we saw in Figure 3, this is due to the different imaging abilities of amplitude and phase gratings. If the neutral center of rotation on the LIDA 105 was located in the index grating and not in the scale grating, error deviations of between 8 and 10 µm would result. Optimal location of the neutral center of rotation is therefore especially important when the scanning gap is large.
Figure 5: Index grating of the LIDA 105 as seen under a scanning electron microscope.
Reducing Sensitivity to Contamination
Exposed LIDA measuring systems are often subjected to heavy contamination. The dimensions of a contaminated area are usually similar to the dimensions of a scanning field. This scale contamination has a different effect on each of the four fields in the typical four-field scanning method. As a result, the zero symmetry and the relationship between the amplitudes of the two output signals are distorted. This results in relatively large interpolation deviations.
The interpolation errors are less pronounced when all scanning signals are affected to the same extent by contamination. Therefore, it would be better if all the scanning signals came from a single scanning field or at least many small scanning fields spread across the scanning plane. Single-field scanning is known to be particularly reliable, and even when contamination is present, the scanning signals in single-field scanning are of high quality.
Some single-field scanning was used in the past. In the majority of these cases, a phase grating on the scale was necessary. This meant, however, that the neutral center of rotation was almost on the index grating and this is not beneficial with tape scales. Other scanning conditions must also be taken into consideration, for example low cost, small dimensions and availability of appropriate technologies, etc. As none of the existing single-field scanning processes fulfilled all these criteria, a new single-field scanning process for amplitude scale grating was developed. The basis of this system is a two-dimensional phase scanning reticle, which does two tasks at once (see Figure 5). On the one hand, it must have the same effect as the above described one-dimensional phase index grating and on the other hand, it must divide up the phase-shifted scanning signals, which are superimposed on the scale, and send them to the various photovoltaic cells. Because of its two-dimensional structure, the light beams can be diverted in various directions so that they land on a different position of the focal plane (see Figure 6). The four photo cells are to be found on these positions.
To demonstrate the difference between sensitivity of four-field and single-field scanning to contamination, we stuck five pieces of light absorbing adhesive tape, of different shapes and sizes, onto the incremental graduation of a LIDA measuring system with grating period of 40 µm (see Figure 7). Both encoders were testing approximately the same size areas of the scale graduation (in the same line and measuring direction). It must be noted that the height of the lines scanned corresponds to only 63 percent of the total height, i.e., the pieces of adhesive tape are covering a significant part of the scanning field.
Figure 8 shows the deviations of the LIDA prototype with four-field scanning. The effect of the five pieces of adhesive tape is clearly to be seen -- the high-frequency interpolation deviations were of the nature of ± 6 µm. What is particularly interesting is the double and four-fold effect that the pieces of tape cause. If the pieces of tape enter into the area that is being scanned, then the deviations become decidedly more pronounced.
The Lissajous curves also testify to the disturbance caused by the pieces of adhesive tape (see Figure 7). These curves are round and centered in ideal scanning conditions. In this case, the signals are off the specification.
Figure 6: Optical layout of the LIDA 105.
The deviations on the new LIDA 105 were then measured using the same "contaminated" scale (see Figure 9). Where there was no tape, the interpolation error was between ±0.1 and +0.15 µm, which is five times less pronounced than with the four-field scanning. Where there was tape, the advantage of this single-field scanning was even greater. The effect of the "contamination" was scarcely to be seen. Only where the tape was running in the same direction as the lines (second from the right in Figure 7) did the error become noticeable -- approximately +0.5 µm, but this was still 10 times less than the corresponding maximum in the four-field scanning.
Figure 7: Simulated contamination on a 40-µm-LIDA scale.
Summary
The new scanning method of the LIDA 105 offers significantly improved measuring accuracy. In this new LIDA 105, the grating periods were reduced to 40 µm while the size of the scanning gap and the favorable mounting tolerances were kept the same. The new scanning process takes the features of exposed linear encoders into consideration, and combines the more advantageous features of several different scanning methods. An optimal location is found for the neutral center of rotation and scanning is single-field through a two-dimensional index grating. Firstly, this scanning method has the same effect as a one-dimensional phase index grating with desired interference effects. Secondly, this method divides and sends the light beams to appropriate photo cells. The new scanning method is less sensitive to scale waviness and scale contamination. With the improved scanning accuracy, measuring steps of 0.1 µm are now possible.
Figure 8: Sensitivity of a LIDA prototype with four-field scanning to contamination. Top: measuring deviation along the "contaminated" scale (see Fig.7). Bottom: Lissajous curves of the scanning signals on the marked scale position.
Figure 9: Sensitivity of the LIDA 105 with single-field scanning to contamination. Top: measuring deviation along the "contaminated " scale (see Fig.7). Bottom: Lissajous curves of the scanning signals on the marked scale positions. |
Website Optimized by: Cherryoneweb.com
