CN103148876B - Single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel building method and coding/decoding method - Google Patents

Abstract

A kind of single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel building method and coding/decoding method.Described virtual incremental formula code channel is the signal of the pulse train composition utilizing the signal received from absolute type code channel to generate, and each pulse in described pulse train is that read head produced through a pitch, and its width is corresponding with the pitch of grating scale.The coding/decoding method that the present invention is based on virtual incremental formula code channel greatly reduces and that the produce probability of error in reading contaminated due to grating scale groove, and can realize Subdividing Processing.

Description

Single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel building method and coding/decoding method

Technical field

The invention belongs to optical metrology field, relate to a kind of single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel building method and coding/decoding method.

Background technology

Single-code channel absolute type displacement of the lines grating scale is a kind of grating scale belonging to international most advanced level, for high-grade precision machine tool.The a few countries such as Germany, the U.S., Japan are only had to grasp this technology in the world.Due to technical know-how reason, domesticly there is no like product, be still in development.

The reading of current single-code channel absolute type displacement of the lines grating scale is based on the vision signal of absolute code channel.There are two basic problems in this method: one is to do Subdividing Processing, affects the precision of grating scale; Two is that antijamming capability is poor.In other words, grating scale must have high-quality absolute type code channel, otherwise easily produces error in reading.

The basic skills solved the problem is on absolute type displacement of the lines grating scale, add physics increment type code channel at quarter, makes grating scale become dicode road absolute type displacement of the lines grating scale.But two physical code channel must be aimed at, this has very high requirement to processing technology, adds the manufacturing cost of grating scale.

Virtual incremental formula code channel technology provides another kind of approach for solving single-code channel absolute type displacement of the lines grating scale reading problems.It is with the difference increasing physics increment type code channel technology: it synthesizes a virtual increment type code channel, for replacing physics increment type code channel by signal processing method.There is no the report of virtual incremental formula code channel technology at present.

Summary of the invention

For the technical matters of above-mentioned absolute type displacement of the lines grating scale, the present invention proposes a kind of single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel building method and coding/decoding method, a virtual increment type code channel is synthesized by signal processing method, for replacing physics increment type code channel, the method greatly reduces error in reading.

For achieving the above object, the present invention adopts following technical scheme:

A kind of building method of virtual incremental formula code channel of single-code channel absolute type displacement of the lines grating scale, described virtual incremental formula code channel is with the signal that this signal generation pulse train forms after receiving absolute position signal from absolute type code channel, each pulse in described pulse train is that read head produced through a pitch, and its width is corresponding with a pitch of grating scale; Line style CCD pixel is divided into two large groups, each large component is L group, and the linear field of each group is 1/2nd pitch of grating scale, and 2L group in two large group is arranged alternately; When read head runs on absolute type code channel, often through a pitch, have the groove of Liang Ge group separately on grating scale in described pixel group at least thus produce a pulse, after these pulses being averaged, namely obtaining the output of virtual incremental formula code channel.

As the preferred embodiments of the present invention, when read head moves, the output level of each group is I _nk(t), wherein, n=1,2, k=1,2 ..., L, works as I _nktime corresponding character " 0 ", output low level, works as I _nkduring corresponding groove, export high level, the signal that described virtual incremental formula code channel exports is:

${\stackrel{\‾}{I}}_1\left(t\right)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L(I_{1k}\left(t\right)-I_0),{\stackrel{\‾}{I}}_2\left(t\right)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L(I_{2k}\left(t\right)-I_0),$

Wherein, differ half pitch, be respectively two outputs organized greatly, I ₀for CCD small group of pixels is by output level time character " 0 ", N is the number of groove in CCD visual field.

A kind of coding/decoding method of virtual incremental formula code channel of single-code channel absolute type displacement of the lines grating scale, the reference position of read head when first determining to power on, then, detect maximum point and the segmentation point of virtual incremental formula code channel, if what detect is maximum point, add 1 or subtract 1, as new decode value to previous decode value according to read head shift to right or left; If what detect is segmentation point, to add deduct segmentation value to current decode value, as new decode value, finally export decode value, wherein said segmentation point comprises equinox and quartile, equinox refers to the mid point of a pitch, and quartile refers to the point of a pitch quartern, and described segmentation value refers to the distance between current segmentation point and the previous virtual incremental formula code channel maximum point be adjacent.

As the preferred embodiments of the present invention, detect equinox with the minimum point of virtual incremental formula code channel output signal pulses, use zero-crossing examination quartile, zero crossing refers to the point that signal pulse is crossing with horizontal ordinate.

As the preferred embodiments of the present invention, described in when powering on read head reference position determine in accordance with the following methods:

(5.1) read head is made to move forward or backward from current location, detection signal or maximum point, if maximum point detected, then proceed to step (5.2);

(5.2) when read head is through a pitch, to two ends I ₁₁and I _2Loutput level does binary conversion treatment, 0 or 1 counts character array A and B respectively as the character read as a result;

(5.3) when array A is identical with the string code word length of specific coding with the length of B, read head stop motion, definitely decodes to A and B;

(5.4) if the absolute decode value difference given length of A and B, forward to (5.5); Otherwise forward (5.1) to and repeat above-mentioned steps, till the absolute decode value difference given length of A and B;

(5.5) Subdividing Processing, arrives t=0 the virtual code channel waveform of first maximum point make Subdividing Processing, the absolute decode value obtained by A or the B segmentation value that adds deduct exports as read head reference position value.

Compared with prior art, the present invention at least has the following advantages: virtual incremental formula code channel of the present invention utilizes the information obtained from the absolute type code channel of single-code channel absolute type displacement of the lines grating scale, the signal of the pulse train composition produced by signal processing method, each pulse is that read head produced through a pitch, its width is corresponding with the pitch of grating scale, so can Subdividing Processing be realized, simultaneously, virtual incremental formula code channel of the present invention is the average of multiple pulse, thus the probability of the error in reading produced because indivedual groove pollutes is greatly reduced.

Accompanying drawing explanation

Fig. 1 is line style CCD(of the present invention or CMOS) group pixels and virtual incremental formula code channel organigram, wherein, the lattice wherein in CCD represents one group of pixel, and the black rectangles on grating scale represents a groove.

Embodiment

Virtual incremental formula code channel utilizes the information obtained from the absolute type code channel of single-code channel absolute type displacement of the lines grating scale, the signal of the pulse train produced by signal processing method composition.It is mainly used in manufacturing and designing of single-code channel absolute type displacement of the lines grating scale, in order to replace physics increment type code channel, realize the Subdividing Processing of single-code channel absolute type displacement of the lines grating scale, a kind of new reading and coding/decoding method are provided simultaneously, overcome the deficiency that grating scale reading is easily made mistakes, improve reliability.

At present, the reading of single-code channel absolute type displacement of the lines grating scale is based on CCD(or CMOS) vision signal that exports, obtain absolute encoding by binary conversion treatment.Because the signal of process is bi-level digital signal, thus cannot segment.In addition, vision signal is subject to the impact of noise, grating scale manufacturing process.Micron-sized high-precision wire displacement grating scale is reached for precision, usually occurs error in reading.And virtual incremental formula code channel is the signal (signal of simulation or multi bit quantization) of pulse composition, each pulse is that read head produced through a pitch, and its width is corresponding with the pitch of grating scale, so can realize Subdividing Processing.And each pulse is the average of the multiple pulses detected from absolute type code channel, this is in theory exactly on average mean filter, and it makes pulse be tending towards the waveform expected, can noise decrease and indivedual contaminated interference of groove.Therefore, compare CCD(or CMOS) vision signal that exports, the number reading method based on virtual incremental formula code channel has higher anti-noise ability.

At present, people solve the method for single-code channel absolute type displacement of the lines grating scale segmentation problem is increase a physics increment type code channel on grating scale.Comparatively speaking, use the absolute type displacement of the lines grating scale of virtual incremental formula code channel technology only to need a physics absolute type code channel, eliminate physics increment type code channel, thus reduce the production cost of grating scale.

Below in conjunction with accompanying drawing, the inventive method is elaborated further:

1, virtual incremental formula code channel building method

By line style CCD(or CMOS) pixel be divided into two large groups, each group is divided into again L group, and the linear field (i.e. visual range) of every group is 1/2 of a pitch, as shown in Figure 1.Suppose each small group of pixels at the volley output level be I _nk(t), n=1,2, k=1,2 ..., L, and work as I _nkoutput low level I when corresponding character ' 0 ' ₀; Corresponding groove, namely exports higher level through character ' 1 ' time.

Suppose in CCD visual field, there is N number of character ' 1 ' again.When specific coding meets homogeneity condition, N>1, but may be different at the diverse location N of grating scale.So definition signal

${\stackrel{\‾}{I}}_1\left(t\right)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L(I_{1k}\left(t\right)-I_0),{\stackrel{\‾}{I}}_2\left(t\right)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L(I_{2k}\left(t\right)-I_0),$

Our explanation below the signal be made up of pulse train respectively, the corresponding pitch of each pulse, plays the effect same with physics increment type code channel.Therefore claim (or for the virtual incremental formula code channel of single-code channel absolute grating scale. differ half pitch, thus their difference half waveform.After translation, two train of impulses overlap.Below with for example illustrates.

being take time as the function of variable, is simulating signal.If export I after analogue to digital conversion _1k(t), so I _1kt () is quantized; Otherwise be not quantized.Here, do not need to carry out quantification treatment to signal.

2, virtual incremental formula code channel property analysis

(1) virtual incremental formula code channel is made up of train of impulses.When read head runs on absolute type code channel, if I _nkt () through character ' 1 ' (i.e. groove), then exports higher level; If through character 0 ' (without groove), export comparatively low level.Typical situation, two level differences are close to 1V.Therefore I _nkt () is forming a pulse through character ' 1 ' time.Due to CCD(or CMOS) there is multiple ' 1 ' (being ensured by the homogeneity of encoding) in visual field, thus when CCD moves across a pitch, there is multiple I _1kt () exports pulse, the average of these pulses just defines a pulse.In other words, when CCD moves, be made up of pulse one by one, the width of pulse is the time of CCD through a pitch, corresponding with pitch.When CCD movement velocity is very fast, narrowed width; When CCD movement velocity is slower, pulse width broadens.

(2) pulse waveform of virtual incremental formula code channel output is corresponding with read head position, therefore can realize Subdividing Processing by detecting pulse waveform.The equinox of trough (minimum point) the corresponding pitch such as between two crests (maximum point), the corresponding quartile of flex point.This Subdividing Processing also can be passed through with relation come.Because very big (little) value point with minimum (greatly) value point corresponding, namely maximum point be equinox; with joining be quartile, therefore by zero-crossing examination quartile.

(3) each waveform is stable, is mathematical expectation waveform, has stronger antijamming capability.This is because each waveform be the average of multiple waveform, namely to the average result that multiple groove detects.Detect the homogeneity that number of times depends on coding.The average of waveform plays denoising effect, can revise irregular waveform simultaneously.

In fact, the signal that we obtain is

${\tilde{I}}_{\mathrm{nk}}\left(t\right)=I_{\mathrm{nk}}\left(t\right)+{\mathrm{\ϵ}}_{\mathrm{nk}}\left(t\right),$ n=1,2，k=1,2,...，L

Wherein ε _nkto be average be zero Gaussian noise.After superposed average, obtain

${\tilde{I}}_n\left(t\right)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L(I_{\mathrm{nk}}\left(t\right)+{\mathrm{\ϵ}}_{\mathrm{nk}}\left(t\right)-I_0)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L(I_{\mathrm{nk}}\left(t\right)-I_0)+\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L{\mathrm{\ϵ}}_{\mathrm{nk}}\left(t\right)$

Notice

$\frac{1}{N}\mathrm{\Σ}{\mathrm{\ϵ}}_{\mathrm{nk}}\left(t\right)\→0$ As N → ∞

So when n is large, the interference of noise to virtual incremental formula code channel is very little.And i _nkt () is tending towards the expectation value of waveform, this illustrates when indivedual groove is subject to polluting, and the impact that virtual incremental formula code channel is subject to is less.

Mathematically, each I _nkt pulse that () exports represents a sample, therefore requires that specific coding has certain homogeneity, and in CCD visual field, the number of character ' 1 ' should be not less than 15.

The above analysis, we obtain following main conclusions: when absolute encoding is even, and the pulse waveform that virtual incremental code channel exports has good stability and anti-noise ability.

3, based on reading and the coding/decoding method of virtual incremental formula code channel

(1) Subdividing Processing

Step1, right with carry out filtering, do and polish process further;

Step2, to detect by first order derivative with extreme point, and to detect zero crossing. the corresponding pitch of adjacent two maximum points; minimum point or maximum point as equinox; zero crossing as quartile.

(2) when read head powers on, reference position is determined

Step1, make read head move to the left or to the right from current location, pass through detection signal maximum point, judge that whether read head is through a pitch;

Step2, when read head is through a pitch, to two ends output level I ₁₁(t) and I _1Lt () does binary conversion treatment, 0 or 1 count character array A and B respectively as the character read as a result;

Step3, when array A is identical with character string (absolute encoding code word) length with the length of B, read head stop motion, carries out absolute type decoding to A and B;

If the absolute type decode value difference given length (I of Step4 A and B ₁₁(t) and I _1Lt pitch number that () differs), forward Step5 to; Otherwise forward Step1 to, till the absolute decode value difference given length of A and B;

Step5, Subdividing Processing.By t=0 to first maximum point waveform makes Subdividing Processing, determines segmentation value.The absolute type decode value determined by A or B adds (or subtracting) segmentation value and exports as the decode value of read head reference position.

(3) coding/decoding method in read head motion

Step1, determine read head reference position;

Step2, the maximum point detecting virtual incremental formula code channel and segmentation point, if detect maximum point, then add 1 or subtract 1, as new decode value to previous decode value according to read head shift to right or left; If segmentation point detected, then to add deduct segmentation value to current decode value, as new decode value, described maximum point refers to the maximum point of the signal pulse of virtual incremental formula code channel, and segmentation point comprises equinox and quartile.

Step3, output decode value.

Claims (5)

1. a single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel building method, it is characterized in that: described virtual incremental formula code channel is with the signal that this signal generation pulse train forms after receiving absolute position signal from absolute type code channel, each pulse in described pulse train is that read head produced through a pitch, and its width is corresponding with a pitch of grating scale; Line style CCD pixel is divided into two large groups, each large component is L group, and the linear field of each group is 1/2nd pitch of grating scale, and 2L group in two large group is arranged alternately; When read head runs on absolute type code channel, often through a pitch, have the groove of the pixel of Liang Ge group separately on grating scale in each large group at least thus produce a pulse, after the pulse in each large group being averaged, namely obtaining the output of virtual incremental formula code channel.

2. building method as claimed in claim 1, it is characterized in that: when read head moves, the output level of each group is I _nk(t), wherein, n=1,2, k=1,2 ..., L, works as I _nktime corresponding character " 0 ", output low level, works as I _nkduring corresponding groove, export high level, the signal that described virtual incremental formula code channel exports is:

${\stackrel{\‾}{I}}_1\left(t\right)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L\left(I_{1k}\right(t)-I_0),{\stackrel{\‾}{I}}_2\left(t\right)=\frac{1}{N}{\mathrm{\Σ}}_{k=1}^L\left(I_{2k}\right(t)-I_0),$

Wherein, with differ half pitch, be respectively two outputs organized greatly, I ₀for CCD small group of pixels is by output level time character " 0 ", N is the number of groove in CCD visual field.

3. the coding/decoding method of a single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel, this single-code channel absolute type displacement of the lines grating scale virtual incremental formula code channel building method according to claim 2 and obtaining, it is characterized in that, the reference position of read head when first determining to power on, then, detect maximum point and the segmentation point of virtual incremental formula code channel, if what detect is maximum point, add 1 or subtract 1, as new decode value to previous decode value according to read head shift to right or left; If what detect is segmentation point, to add deduct segmentation value to current decode value, as new decode value, finally export decode value, wherein said segmentation point comprises equinox and quartile, equinox refers to the mid point of a pitch, and quartile refers to the point of a pitch quartern, and described segmentation value refers to the distance between current segmentation point and the previous virtual incremental formula code channel maximum point be adjacent.

4. coding/decoding method as claimed in claim 3, is characterized in that, detects equinox, use with the minimum point of virtual incremental formula code channel output signal pulses zero-crossing examination quartile, zero crossing refers to the point that signal pulse is crossing with horizontal ordinate.

5. coding/decoding method as claimed in claim 3, is characterized in that: when powering on, read head reference position is determined in accordance with the following methods:

(5.1) read head is made to move forward or backward from current location, detection signal or maximum point, if maximum point detected, then proceed to step (5.2);

(5.2) when read head is through a pitch, to two ends I ₁₁and I _2Loutput level does binary conversion treatment, 0 or 1 counts character array A and B respectively as the character read as a result;

(5.3) when array A is identical with the code word size of specific coding with the length of B, read head stop motion, definitely decodes to A and B;

(5.4) if the absolute decode value difference given length of A and B, forward to (5.5); Otherwise forward (5.1) to and repeat above-mentioned steps, till the absolute decode value difference given length of A and B;

(5.5) Subdividing Processing, arrives t=0 or the virtual code channel waveform of first maximum point make Subdividing Processing, the absolute decode value obtained by A or the B segmentation value that adds deduct exports as read head reference position value.