METHOD FOR CONTROLLING THE EMISSION FREQUENCY OF A LASER

Abstract

A method for controlling the emission frequency of a laser comprises: recording a first spectrum by passing a laser light emitted by the laser through a sample onto a detector, the detector being connected to a multichannel analyzer which assigns pulses detected by the detector to a channel; determining a first channel to which the maximum of a first signal in the first spectrum has been assigned; determining a second channel to which the maximum of a second signal in the first spectrum has been assigned; recording a second spectrum in analog fashion like the first spectrum; determining whether the maximum of the first signal in the second spectrum has been assigned to the first channel and whether the maximum of the second signal in the second spectrum has been assigned to the second channel; adjusting the operating temperature of the laser in the event of deviations determined in the previous step.

Claims

1. A Method for controlling the emission frequency of a laser, comprising the following steps: a) recording a first spectrum by passing a laser light emitted by the laser through a sample onto a detector, the detector being connected to a multichannel analyzer which assigns pulses detected by the detector to a channel, b) determining a first channel to which the maximum of a first signal in the first spectrum has been assigned, c) determining a second channel to which the maximum of a second signal in the first spectrum has been assigned, d) recording a second spectrum in the same way as the first spectrum, e) determining whether the maximum of the first signal in the second spectrum was assigned to the first channel and whether the maximum of the second signal in the second spectrum was assigned to the second channel, f) adaptation of a working temperature of the laser with deviations determined in step e) such that the maximum of the first signal in the second spectrum is assigned to the first channel and/or that the maximum of the second signal in the second spectrum is assigned to the second channel.

2. The method according to claim 1, wherein, in the case of deviations determined in step e), a laser parameter is additionally adapted which causes an extension or compression of a spectrum recorded with the aid of the laser.

3. The method according to claim 1, wherein the following is carried out for determining in step f): determining a first value for the first channel and for a predeterminable number of higher order channels adjacent the first channel, the first value representing a signal intensity assigned to these channels, determining a second value for the first channel and for a predeterminable number of lower order channels adjacent the first channel, the second value representing a signal intensity assigned to these channels, and forming a first ratio between the first value and the second value.

4. The method according to claim Error! Reference source not found, wherein the first ratio is used to adjust a temperature control parameter of the laser.

5. The method according to claim 1, wherein the following is carried out for determining in step e): determining a third value for the second channel and for a predeterminable number of higher order channels adjacent the second channel, the third value representing a signal intensity assigned to these channels, determining a fourth value for the second channel and for a predeterminable number of lower order channels adjacent the second channel, the fourth value representing a signal intensity assigned to these channels, and forming a second ratio between the third value and the fourth value.

6. The method according to claim Error! Reference source not found, wherein the second ratio is used to adjust a laser parameter which causes an elongation or compression of a spectrum recorded with the aid of the laser.

7. The method according to claim Error! Reference source not found, wherein the number of adjacent lower order channels corresponds to the number of adjacent higher order channels.

8. The method according to claim 1, wherein the following is carried out for determining in step e): determining a third channel to which the maximum of the first signal in the second spectrum has been assigned, and/or determining a fourth channel to which the maximum of the second signal in the second spectrum has been assigned, and determining whether the third channel corresponds to the first channel and/or whether the fourth channel corresponds to the second channel.

9. The method according to claim Error! Reference source not found, wherein the determination of the third channel and/or the fourth channel is carried out by adapting a curve to measured values of the second spectrum, wherein a maximum of the curve corresponding to the maximum of the first signal is used for the determination of the third channel and/or a maximum of the curve corresponding to the maximum of the second signal is used for the determination of the fourth channel.

10. The method according to claim 2, wherein the following is carried out for determining in step f): determining a first value for the first channel and for a predeterminable number of higher order channels adjacent the first channel, the first value representing a signal intensity assigned to these channels, determining a second value for the first channel and for a predeterminable number of lower order channels adjacent the first channel, the second value representing a signal intensity assigned to these channels, and forming a first ratio between the first value and the second value.

11. The method according to claim 10 Error! Reference source not found, wherein the first ratio is used to adjust a temperature control parameter of the laser.

12. The method according to claim 2, wherein the following is carried out for determining in step e): determining a third channel to which the maximum of the first signal in the second spectrum has been assigned, and/or determining a fourth channel to which the maximum of the second signal in the second spectrum has been assigned, and determining whether the third channel corresponds to the first channel and/or whether the fourth channel corresponds to the second channel.

13. The method according to claim 12, wherein the determination of the third channel and/or the fourth channel is carried out by adapting a curve to measured values of the second spectrum, wherein a maximum of the curve corresponding to the maximum of the first signal is used for the determination of the third channel and/or a maximum of the curve corresponding to the maximum of the second signal is used for the determination of the fourth channel.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Further advantages and details of the present invention are explained in more detail by means of exemplary embodiments and Figures.

[0050] FIG. 1 shows a simulated absorption spectrum of carbon dioxide gas in the spectral range from 2299 cm.sup.?1 to 2295 cm.sup.?1.

[0051] FIG. 2 shows a selected spectral range of the .sup.12CO.sub.2 band and the .sup.13CO.sub.2 band in an IR absorption spectrum.

[0052] FIG. 3 shows an example of the procedure for determining a possible deviation of the position of the two absorption bands shown from a position in a previously recorded spectrum.

[0053] FIG. 4 shows a graphical representation of an exemplary embodiment of the dependence of the TempSet parameter on the ratio AB of the flank integrals of a first absorption band.

[0054] FIG. 5 shows a graphical representation of an exemplary embodiment of the dependence of the parameter Imax on the ratio C/D of the flank integrals of a second absorption band.

[0055] FIG. 6 shows a graphical representation of typical curves of flank ratios of two absorption bands as a function of time under real measuring conditions using an example of the procedure for controlling the emission frequency of a laser.

DETAILED DESCRIPTION

[0056] FIG. 1 shows a simulated absorption spectrum (using the Lambert-Beer law) of carbon dioxide gas, wherein .sup.12CO.sub.2 bands are represented by a solid line and .sup.13CO.sub.2 bands by a dashed line.

[0057] Typically, the spectral range to be measured is tuned from a start frequency to an end frequency. If the frequency is generated by a fast temperature change of the laser, e.g. by a current flow, then there is an initial and final temperature, which are related to an initial and final frequency. However, the dependence between the applied heating voltage and the emission frequency is not linear so that either a calibration (e.g. with a germanium etalon) must be carried out for each measurement, or at least two absorption lines must be clearly identified whose frequency positions are known.

[0058] The frequency positions of typical gas absorption lines are well known and can be looked up in the HITRAN database (available online at http://hitran.org/) with uncertainties of less than one thousandth of a wavenumber.

[0059] The changes in the properties of the laser or the laser environment (e.g. temperature, pressure, humidity, current flow, voltage) discussed above often lead to a systematic drift to higher or lower frequencies. Frequently, the emission frequency of the laser oscillates also around the adjusted frequency value. The example of a rotational vibration absorption spectrum of carbon dioxide in the gas phase, which is recorded experimentally with a narrow-band infrared laser (IR laser), illustrates the associated technical challenge. Let us assume that in this example only the temperature changes the laser properties of a quantum cascade laser (as a light source).

[0060] The absorption lines shown in FIG. 1 are based on data taken from the HITRAN database. The line width of the absorption lines is 0.19 cm.sup.?1 and results from the natural line width, the natural widening and the widening due to the measuring apparatus.

[0061] In the following, an exemplary measurement and an exact control of the laser frequency are presented.

[0062] The aim of the measurement is to determine the concentration of .sup.12CO.sub.2 and .sup.13CO.sub.2 in a sample. The selected spectral range extends from 2298 cm.sup.?1 to 2296.3 cm.sup.?1 and is shown in FIG. 2. The simulated .sup.13CO.sub.2 absorption band and the simulated .sup.12CO.sub.2 absorption band are marked with a crosshatch and a corresponding label. The points connected with an adjusted curve represent measured values of a sample measurement.

[0063] In order to tune the laser spectrally, a current ramp is applied, which generates a constant current increase, so that the temperature and thus the frequency increase. This was described in the following publication: Nelson, D.; Shorter, J.; McManus, J.; Zahniser, M.: Sub-part-per-billion detection of nitric oxide in air using a thermoelectrically cooled mid-infrared quantum cascade laser spectrometer. Applied Physics B 2002, 75, 343-350.

[0064] Then an absorption measurement can be carried out at freely selected (here almost equidistant) time intervals, which is assigned to a channel by means of a multichannel analyzer. The current ramp can have different shapes: It can ascend, ascend and then descend, or even oscillate. The measured channels must then be assigned to the real frequencies (in wavenumbers).

[0065] For this purpose, the absolute positions of the absorption peak positions are known, for example, from the HITRAN database. So one can now stretch or compress the measured absorption signal depending on the channel position so that it matches the peak positions from the HITRAN database. However, this would mean considerable post-processing of each individual spectrum before it could be averaged. Such an approach would, however, be possible in principle, but very costly.

[0066] If, however, the laser parameters shift so far due to a systematic temperature drift that, for example, the baseline of the spectrum changes, the measurements can no longer be averaged by post-processing each individual spectrum. This is because the measurement was no longer carried out under identical conditions. Averaging would lead to erroneous results.

[0067] For this reason, the measuring conditions and the frequency and temperature range must be kept constant or their deviations kept to a minimum. In order to achieve this, the procedure described below was developed as an implementation example of a procedure for controlling the emission frequency of the laser.

[0068] The examined spectral range shows three absorption bands. One .sup.12CO.sub.2 absorption band at 2297.58 cm.sup.?1 (B1), one .sup.13CO.sub.2 absorption band at 2297.19 cm.sup.?1 (B2) and one .sup.12CO.sub.2 absorption band at 2296.45 cm.sup.?1 (B3) (see FIG. 2).

[0069] In order to find the correct laser working point, the temperature of the laser is set so that it emits light at a frequency of about 2298 cm.sup.?1 without a current ramp. Then the current ramp is started and the measuring points are recorded when the current ramp is applied. Assuming the current ramp would pass through the spectral range by 1.7 cm.sup.?1 within 1 ms (from high to low wavenumbers), then 20 to 20000 measurement points, i.e. 20 to 20000 individual absorption measurements, could easily be carried out in this time if the electronics are fast enough to process these absorption measurements.

[0070] In the present case, several hundred absorption measurements were carried out at different current values and thus different emission frequencies of the laser. The measured absorption spectrum shown in FIG. 2 is the result of the sequence of the measuring points. First, however, a spectrum is obtained in which the signal intensities are not yet assigned to the corresponding wavenumbers, but only to channels. Each measurement at a new current value corresponds to a new channel.

[0071] The position of B1 is assigned to the channel where the maximum of the first band (from left) is identified, the position of B2 is assigned to the channel where the maximum of the second band (from left) is identified, the position of B3 is assigned to the channel where the maximum of the third band (from left) is identified.

[0072] The linearity or non-linearity of the frequency positions can be determined as a function of the channels by assigning the three bands to the individual channels and a one-time etalon calibration. This is typically constant for a constant working point of the laser (also considering the corresponding tuning range of the laser at this working point). Thus, the current ramp, the initial temperature, the initial frequency, the non-linear dependence of the frequency on the channels for a defined operating point of the laser is characterized. Together with the settings for the operation of the laser, this defines the optimal operating point of the laser.

[0073] One now proceeds as follows for the further measurements: If (under ideal conditions) there are no changes in the environmental properties or the operating point of the laser, there is no change in the laser frequency. Unfortunately, this is never the case under real conditions. Since the laser temperature is controlled by a temperature controller, it is always controlled to a property (e.g. the temperature of the temperature sensor close to the laser) that does not correspond exactly to the laser property. In prior art solutions, the temperature of the laser is measured by a temperature element that is close to the laser and has good thermal contact. If the laser heats up, this is slightly delayed by the temperature element and measured in a slightly altered (attenuated) form. This is due to the temperature distribution in the components. This means that the temperature gradient leads to a heat flow that produces a temperature change in the temperature element to which it is then controlled.

[0074] Unfortunately, such a control always lags the real temperature and can lead to systematic temperature deviations with rising and falling outside temperatures. This is often the case, especially during longer laser operation, as the laser can continuously heat up due to the current or heat input into the laser. Even after switching on the laser, it takes up to one hour until the laser is in an acceptable equilibrium.

[0075] According to an exemplary embodiment of the procedure claimed here, however, the position of the absorption peaks is directly tracked, as shown in FIG. 3.

[0076] FIG. 1 shows an exemplary frequency stabilization based on the position of the CO.sub.2 absorption bands. The hatched areas are compared symmetrically to the peak maximum of the individual absorption bands on one side of the respective absorption band (channels with a higher ordinal number) with the other side (channels with a lower ordinal number) of the same absorption band by forming a ratio. A change in the area ratio indicates a frequency shift.

[0077] If the temperature of the laser changes, the emission frequency of the laser changes. Consequently, the positions of the absorption bands shift. If the non-linearity or the dependence of the laser changes with current flow (for example as a reaction to heating), then the relative distance between the absorption bands shifts.

[0078] The operating point of the laser is set according to the described exemplary embodiment so that, for example, a peak maximum can be found for channel 18 and channel 36 in FIG. 3. If now the signal of the channels from 14 to 18 is integrated as parameter A and the signal of the channels from 18 to 22 as parameter B and the ratio of the integrals I1=AB is formed, then an increase of I1 indicates a shift of the peak to lower channels and a decrease of I1 indicates a shift of the peak to higher channels.

[0079] One proceeds in the same way with the second peak. If now the signal of the channels from 32 to 36 is integrated as parameter C and the signal of the channels from 36 to 40 as parameter D and the ratio of the integrals I2=C/D is formed, then an increase of I2 indicates a shift of the peak to lower channels and a decrease of I2 indicates a shift of the peak to higher channels.

[0080] The integration of the flanks of the absorption bands is a possibility to calculate the shift of the maxima of the absorption bands from the ratios. It has the advantage that the signal-to-noise ratio increases during integration and that the shift of the maxima of the absorption bands can be determined well even with noisy spectra.

[0081] It is possible to include different flank components in the integration, which do not necessarily have to be symmetrically arranged. Flanks up to 90% of the maximum, especially up to 80%, especially up to 70%, especially up to 60%, especially up to 50%, especially up to 40%, especially up to 30%, especially up to 20%, especially up to 10% can be included. The more spectrally isolated the bands are, i.e. the less overlap there is with other bands, the larger the edge fraction included in the integration of the edges can become. There are also other possibilities to determine whether there has been a shift in the maxima of the absorption bands, e.g. by evaluating the inflection points and the maximum, etc.

[0082] A shift of both peaks can occur due to a changed initial temperature or frequency. This is specified in the regulation of the exemplary embodiment described here with the value TempSet (shift of the spectrum). If the relative distance between the peak positions is changed, the non-linearity between the frequency and the channels has changed. This is indicated by the value Imax in the control of the execution example described here. The current flow is increased from zero to Imax during a complete ramp. A change of Imax corresponds to a compression or extension of the spectrum.

[0083] A change of one or both peak positions is corrected directly after the measurement of a spectrum by variation of TempSet and Imax. Since the absorption spectrum directly reflects the current operating point of the laser, the operating point can be optimized directly in a feedback control.

[0084] The Temp Set value is a value that defines the working point of the laser. The change of the TempSet value ?TempSet is described in the following table and results from the flank integrals A and B of the considered absorption band.

[0085] The value Imax is also a value that defines the working point of the laser. The change of the Imax value ?Imax is described in the following table and results from the flank integrals C and D of the considered absorption band.

[0086] The dependence of the gradients ? and ? are determined during the setting process of the optimal operating point. The value ? can, for example, be determined by displaying the TempSet parameter over the ratio AB as shown in FIG. 4. The value ? can, for example, be determined by displaying the parameter Imax over the ratio C/D as shown in FIG. 5. The parameters TempSet and Imax are measured independently of each other at the optimum points of the other values.

[0087] The following three regulations are presented as examples.

Example 1

[0088] A linear equation is used to correct the operating point and the compression of the spectrum. The two controlled variables are decoupled from each other.

TABLE-US-00001 Displacement of the entire spectrum Spectrum compression [00001]$?.Math..Math.\mathrm{Temp\_Set}=\left(??\left(\frac{A}{B}-1\right)\right)*\mathrm{factor}.Math..Math.\mathrm{LNR}$ [00002]$\mathrm{?I\_}.Math.\max =\left(?\left(\frac{C}{D}-1\right)\right)*\mathrm{factor}.Math..Math.\mathrm{LNR}$ Temp_Set.sub.new = Temp_Set.sub.old + ?Temp_Set I_max.sub.new = I_max.sub.old + ?I_max ?: Rise of TempSet vs. A/B ?: Rise of I max vs. C/D A, B: left or right area of peak; sum of C, D: left or right areas of peak; sum of channels left or right of the defined channels left or right of the defined central channel. central channel LNR: Correction factor -> LNR: Correction factor -> Determination Determination via standard deviation of via standard deviation of A/B and C/D A/B and C/D with active control; with active control; typical values: 0.01; typical values: 0.01; 0.05; 0.1 0.05; 0.1 Temp_Set: Value which is given to the I_max: Value which is given to the control loop of the electronics. control loop of the electronics.

[0089] In the control listed here, the positions of two absorption peaks (P1 and P2) are controlled via a feedback control. The TempSet setting should adjust the position of peak P1 so that the maximum is in channel K1. The gradient ? now describes the change of the value TempSet depending on the edge ratio A/B. The ratio A/B describes the integrals of the left and right edges of peak P1. If the peak is shifted, the ratio A/B changes. In the example shown in the table, peak P1 is positioned at the desired channel when the ratio is 1. If the ratio deviates from 1, a correction value results (A/B?1). This is multiplied by the gradient ? and thus serves for the exact setting of the TempSet value required for the desired working temperature of the laser.

[0090] The Imax setting is to set the position of peak P2 so that the maximum is in channel K2. The gradient ? now describes the change of the value Imax depending on the edge ratio C/D. The ratio C/D describes the integrals of the left and right flank of peak P2. If the peak is shifted, the ratio C/D changes. In the example shown in the table, Peak P2 is positioned at the desired channel when the ratio is 1. If the ratio deviates from 1, a correction value results (C/D?1). This is multiplied by the gradient ? and thus serves for the exact setting of the Imax value required for the desired working temperature of the laser.

[0091] Since, however, an additional temperature control very often regulates the laser temperature, it often makes sense in practice to adapt the control steps to the feedback control and to select a value smaller than 1. Furthermore, it must be taken into account that the gradients ? and ? have a statistical error, which must be included in the size of the control steps. This error is quantified in a calibration step. For this purpose, the standard deviation of the dispersion of the ratio A/B is determined as a function of the factor LNR with active control. Both influencing factors mentioned above are taken into account by the LNR factor. Ideally, LNR would be one.

[0092] The control of the emission frequency of a laser shown as an example above makes it possible to obtain very stable positions of absorption bands in a spectrum. This is shown as an example in FIG. 6, which shows the progression of the flank ratios A/B (upper line) and C/D (lower line) as a function of time for a measurement in a clinical environment with fluctuating temperature conditions. On average, the edge ratios are very constant, since the control of the emission frequency of the laser described above was applied. Without such control of the emission frequency, the edge ratios would change massively, corresponding to a shift in the absorption bands considered in the spectrum. It does not matter whether the regulated conditions are at one or deviate from it. To stabilize the frequency, only a constant mean value of the ratios with small deviations (small standard deviation) is important to successfully control the emission frequency.

Example 2

[0093] A linear equation is used to correct the operating point and the compression of the spectrum. The two control variables are coupled.

TABLE-US-00002 spectral shift Spectrum compression [00003]$?.Math..Math.\mathrm{Temp\_Set}=\left({?}^{}?\left(\frac{A}{B}-1\right)+{?}^{.Math.}?\left(\mathrm{TempSet}?\left(I_{\max }\right)\right)\right)*\mathrm{factor}.Math..Math.\mathrm{LNR}$ [00004]$?.Math..Math.\mathrm{I\_max}=({?}^{.Math.}?\left(\frac{A}{B}-1\right)+{?}^{.Math.}?\left(I_{\max }?\left(\mathrm{TempSet}\right)\right)*\mathrm{factor}.Math..Math.\mathrm{LNR}$ Temp_Set.sub.new = Temp_Set.sub.old + ?Temp_Set I_max.sub.new = I_max.sub.old + ?I_.sub.max ?: Rise of TempSet vs. A/B ?: Rise of I_max vs. C/D ?: dependency on TempSet when changing Imax ?: Dependence on I_max when changing TempSet A, B: left or right area of peak; sum of channels left C, D: left or right surfaces of peak; sum of channels or right of the defined central channel. left or right of the defined central channel LNR: Correction factor -> LNR: Correction factor -> Determination via standard Determination via standard deviation of A/B and C/D deviation of A/B and C/D with active control; typical with active control; typical values: 0.01; 0.05; 0.1 values: 0.01; 0.05; 0.1 Temp_Set: Value which is given to the control loop I_max: Value which is given to the control loop of of the electronics. the electronics.

[0094] In the regulation listed here, the positions of two absorption peaks (P1 and P2) are regulated via a feedback regulation. The TempSet setting should adjust the position of peak P1 so that the maximum is in channel K1. The gradient ? now describes the change of the value TempSet depending on the edge ratio A/B. ? describes the change of the TempSet value when Imax is changed. The ratio A/B describes the integrals of the left and right edges of peak P1. If the peak is shifted, the ratio A/B changes. In the example shown in the table, peak P1 is positioned at the desired channel if the ratio is 1, for example. If the ratio deviates from e.g. 1, a correction value results (A/B?1). This is multiplied by the gradient ?. Since Imax is set in the same or a subsequent control step, this position shift would also influence the position of P1. To counteract this, an additional term ? is introduced, which takes this dependence on Imax into account.

[0095] The Imax setting is to set the position of peak P2 so that the maximum is in channel K2. The gradient ? now describes the change of the value Imax depending on the edge ratio C/D. ? describes the change of the value Imax depending on TempSet. The ratio C/D describes the integrals of the left and right flank of peak P2. If the peak is shifted, the ratio C/D changes. In the example shown in the table, peak P2 is positioned at the desired channel if the ratio is e.g. 1. If the ratio deviates from e.g. 1, a correction value results (C/D?1). This is multiplied by the gradient ?. Since TempSet is set in the same or a subsequent control step, this position shift would also influence the position of P2. To counteract this, an additional term ? is introduced which defines this dependency on the TempSet.

[0096] The correction of the individual values for TempSet and Imax can be carried out simultaneously (parallel) or successively (sequentially). For a parallel version of the control, the complete set of curves must be recorded for the TempSet(Imax) and Imax(TempSet) variables. In the case of a sequential execution (correction of TempSet followed by Imax) a case distinction can be made: If the relative error is ?TempSet/TempSet??Imax/Imax, i.e. if the influence of the change on the position of peak P1 is greater than on peak P2, then ? can be set to zero, whereby the value for TempSet is changed so that peak P1 has its maximum in channel K1. P2 is corrected according to the above formula, but the shift of peak P2 due to the change of TempSet is taken into account. If the relative error is ?Imax/Imax??TemSet/TemSet, i.e. if the influence of the change on the position of peak P2 is greater than on peak P1, ? can be set to zero, whereby the value for Imax is changed so that peak P2 has its maximum in channel K2. P1 is corrected according to the above formula, but the shift of peak P1 due to the change of Imax is taken into account. It is not possible to overcontrol the regulation.

[0097] Since, however, an additional temperature control very often regulates the laser temperature, it often makes sense in practice to adapt the control steps to the feedback control and to select a value smaller than 1. Furthermore, it must be taken into account that the gradients ? and ? have a statistical error, which must be included in the size of the control steps. Furthermore, it must be taken into account that the gradients ?, ?, ? and ? have a statistical error that must be included in the size of the control steps. This error is quantified in a calibration step. For this purpose, the standard deviation of the dispersion of the ratio A/B is determined as a function of the factor LNR with active control. Both influencing factors mentioned above are taken into account by the LNR factor. Ideally, LNR would be one.

Example 3

[0098] Intervals are specified in which the feedback loop is regulated.

TABLE-US-00003 A/B > (a.sub.i * peak deviation.sub.AB + 1) C/D > (c.sub.j * peak deviation.sub.CD + 1) B/A > (a.sub.i * peak deviation.sub.AB + 1) D/C > (c.sub.j * peak deviation.sub.CD + 1) Temp_Set.sub.new = Temp_Set.sub.old ? b.sub.i I_max = I_max ? d.sub.j A, B: left or right area of peak; sum of C, D: left or right area of peak; sum of channels left or right of the defined central channels left or right of the specified channel. central channel a.sub.i, b.sub.i: preset parameters for interval i c.sub.j, d.sub.j: default parameters for interval j peak deviation.sub.AB: preset parameter peak deviation.sub.CD: preset parameter Temp_Set: Value which is given to the I_max: Value which is given to the control control loop of the electronics. loop of the electronics. For each interval i different parameters for For each interval j, different parameters a.sub.i and b.sub.i can be used can be used for a.sub.j and b.sub.j.

[0099] The third control is independent of a predetermined dependency of TempSet on A/B or of Imax on C/D.

[0100] In this variant, two peak positions P1 and P2 are assigned to channels K1 and K2. For the deviation of the values of the ratios A/B and C/D by a preset value, e.g. 1, intervals can be defined for which TempSet is changed by a preset value. This interval division is represented by the value peak deviation.sub.AB*?. An interval (peak deviation.sub.AB*?.sub.1) is sufficient for small changes to the TempSet value. For example, if there is a deviation from A/B from the value 1, which is greater than the value peak deviation.sub.AB*?.sub.1, the value TempSet changes by the amount (+b.sub.1). The value for peak deviation is very small and can be 0.002. If, for example, ?.sub.1 has the value one, the value TempSet is changed by b.sub.1 (e.g. 0.0005) if there is a deviation of 0.002. Further intervals can be defined for larger deviations. However, the intervals may not have to be equidistant. So for the intervals peak deviation.sub.AB*?.sub.2 (e.g. value two), peak deviation.sub.AB*?.sub.3 (e.g. value three) and peak deviation.sub.AB*?.sub.4 (e.g. value five) larger b2, b3, b4 values for the changes of TempSet can be selected. Within an interval (values between peak deviation.sub.AB*?.sub.1 and peak deviation.sub.AB*?.sub.i-1) TempSet is always corrected by a fixed amount (b.sub.i) for this interval. If the value lies in a different interval, TempSet is corrected by a different amount b.sub.i. For each interval there is a defined b.sub.i.

[0101] The regulation is carried out iteratively. This means that in a first control step, the ratio A/B is determined and compared against the specified intervals. Depending on the interval in which the ratio A/B lies, the value for TempSet is corrected by the corresponding value b.sub.i. In a subsequent step, this procedure is repeated and Temp Set corrected again if necessary. By performing these individual steps one after the other, a control loop is created which corrects the peak position P1 per loop pass with the aim of keeping the ratio A/B within the smallest interval and thus peak position P1 at channel K1.

[0102] With this method, deviations to lower or higher temperatures can be introduced with different control steps. This is useful if the cooling of the laser is close to its limit and, for example, cooling takes longer than heating.

[0103] The deviation B/A shows a peak deviation in the other direction than A/B. Accordingly, TempSet is controlled in the opposite direction (TempSet.sub.new=TempSet.sub.old-b.sub.i), analogous to the method described above.

[0104] The second peak P2 is controlled on channel K2. The procedure is analogous to the TempSet control. However, the intervals from when control is started (c.sub.1, c.sub.2, c.sub.3, c.sub.4, etc.) and how much is controlled (d.sub.1, d.sub.2, d.sub.3, d.sub.4, etc.), as well as the peak deviation.sub.CD are determined separately for Imax from the ratios C/D and D/C.