JPS6029199B2 - radiation generator - Google Patents

Description

Detailed Description of the Invention The present invention relates to a radiation generation device equipped with a linear electron accelerator that generates radiation such as X-rays or electron beams, and stabilizes the setting parameters of the device by detecting radiation intensity. This is what I decided to do.

An example of this type of device is shown in FIG.

In the example shown in Figure 1, a device is used that is equipped with a deflecting electromagnet that deflects the electron beam, and the deflection angle is 90 degrees. The same applies to 110o, 2700, and others. In FIG. 1, numeral 1 indicates an electron beam accelerating section.

This acceleration section includes an electron gun that generates an electron beam, an acceleration tube that has an acceleration function, and a microwave source for acceleration.
For this purpose, it has a high-pressure pulse modulator, a microwave transmission system vacuum holding mechanism, and a cooling mechanism. It may also include an electromagnetic lens for converging the electron beam generated from the electron gun, a steering coil for correcting the beam trajectory, a quadrupole electromagnet for converging the accelerated electron beam, and the like. Reference numeral 2 denotes a deflecting electromagnet, which has the function of deflecting the electron beam at an appropriate angle.

3 is a radiation generation mechanism that generates radiation such as X-rays or electron beams using a deflected electron beam. Normally, when generating X-rays, a heavy metal target is used, and when generating electron beams, a scattering mechanism is used. Use foil.

4 indicates the trajectory of the electron beam.

A is the radiation generated by the radiation generating mechanism 3, and 5 is a flattening filter, which is used when the radiation A is an X-ray and is removed when it is an electron beam. Reference numeral 6 denotes a radiation detector that measures the intensity of radiation A, and a transmission type or fingertip type ion chamber that generates an ionizing current by radiation is usually used. 7 is a radiation detector similar to radiation detector 6, but it is for detecting the intensity of different areas within the irradiation field by radiation A, and two or more detectors are installed like 7a and 7b. Ru.

8 is an intensifier that intensifies the signal of the radiation detector 6, 9 is a control mechanism for stabilizing the radiation intensity, 10a and 10b are intensifiers similar to intensifier 8, and 11 is an intensifier 10a. ,10b
This is a differential multiplier that differentially multiplies the output of the

12 is a control mechanism for keeping the radiation intensity distribution of the irradiation field constant using the output of the differential intensifier 11, and controls the accelerating section 1.

Reference numerals 13a and 13b are field limiting mechanisms, which form a pair of 13a and 13b, and there is also another pair in the right angle direction.

(Not shown) There is also a multi-division limiting mechanism that can configure an irradiation field other than a rectangle. The irradiation field limiting mechanisms 13a and 13b mainly have the function of defining the irradiation field in the case of X-rays, and use 14 electron beam irradiation tubes (electron beam applicators) in the case of electron beams. Next, the operation of the conventional device will be explained.

The electron beam accelerated to a predetermined energy in the accelerator 1 is deflected by a deflecting electromagnet 2 which generates a predetermined magnetic field in accordance with the energy, and guided to a radiation generating mechanism 3, where it is emitted as an X-ray, an electron beam, etc. of radiation A is generated. The radiation A is irradiated within an irradiation field defined by the irradiation field limiting mechanisms 13a, 13b or the electron beam irradiation tube. In the case of X-rays, the distribution of radiation intensity is flattened by the flattening filter 5, but in the case of electron beams, the flattening filter 5 has a mechanism to be removed from within the irradiation field. The intensity of the irradiated radiation is detected by a radiation detector 6, and an ionizing current proportional to the intensity is generated.

This is converted into current-voltage by the intensifier 8 and further amplified. Normally, this device is operated in pulses, and the radiation intensity depends on the pulse repetition frequency and the duration of the pulse, so the output of the intensifier 8 displays the radiation intensity and at the same time sends a signal to the radiation intensity control mechanism 9. , has a mechanism for controlling the pulse repetition frequency or pulse duration. However, since the electron beam is deflected by a deflecting electromagnet, the accelerated electron beam energy may deviate from a predetermined value, or
If the trajectory deviates, the incident angle and the incident position of the electron beam on the radiation generating mechanism 3 usually deviate.

For this reason, the distribution of radiation intensity within the irradiation field deviates from the ideal distribution B to a distribution C, in which the flatness of the radiation intensity distribution is lost, as shown in FIG. C is an example of a case where flatness is lost. However, since the device is required to maintain flatness such as the flat distribution B, the following control mechanism is usually used. That is, radiation detectors 7a and 7b are installed to detect the radiation intensity in different areas within the field of view, and measure the radiation intensity in each area. When the electron beam energy changes, the beam trajectory when deflected by the deflecting electromagnet changes, and due to the trajectory change specific to the system of the deflecting electromagnet used, the angle of incidence and/or the position of incidence on the radiation generating mechanism 3 changes. C in Figure 1 because the central axis of the directivity of the radiation changes.
This results in non-uniformity in strength as shown in . Therefore, the ionization current of each of the radiation detectors 7a, 7b increases or decreases, thereby causing a difference in the outputs of the radiation detection devices 7a, 7b. These are multiplied by multipliers 10a and 10b, respectively, and the difference between the two, including the sign, is detected or multiplied by differential multiplier 11. Therefore, the output of the adder booster 11 is sufficient to 4
- Operate the control mechanism 9 so that the radiation intensity distribution decreases (±0) to restore the original flat radiation intensity distribution. At this time, if the intensity flatness of the electron beam energy is to be maintained, one or more of the accelerating microwave frequency, accelerating microwave power, accelerating beam current, etc. is controlled as a control target.
In devices where the accelerated electron beam trajectory fluctuates or is controllable, a steering coil for trajectory correction is installed in the output section of the acceleration tube, and the current value of this steering coil is controlled to adjust the trajectory of the electron beam. may be modified. At this time, the incident position of the beam on the deflecting electromagnet 2 changes, and it is possible to correct for the loss of flatness. When this control mechanism is used, it is often operated in parallel with a control mechanism for correcting the electron beam energy. Second
The figure shows the basic relationship of its operation. D in Figure 2. is the detected value by the radiation detector 6, and D, , D2 are the detected values by 7a and 7b, respectively. x is an amount that sets the electron beam energy to be controlled, and is normally set to xm when the device is maintained in an optimal state for accelerating with a predetermined electron beam energy. When the mechanism for adjusting the radiation intensity, which can be set arbitrarily, is kept unchanged, Do normally obtains the maximum value Dm at Xm. At this time, D,=
D2 and flatness was obtained. When X changes from Xm, Do decreases from Dm and, at the same time, D,≠D2 and flatness is lost, so X is controlled so as to return to D,=D2. Conventional radiation generators using linear electron accelerators have the configuration and operation described above, so when non-uniformity of radiation intensity in the irradiation field occurs, as shown in C in Figure 1, several control operations are necessary. Because they are parallel, the result is complicated and they interfere with each other, making it difficult to obtain stable output and a flat radiation intensity distribution.
In devices and electron beams in which D2 exhibits similar characteristics to Do, there are drawbacks such as the fact that the electron beam energy cannot be controlled by the output of the radiation. This invention was made in order to eliminate the above-mentioned drawbacks of the conventional device, and it performs maximum value control of normalized radiation intensity, and performs maximum value control without using a differentiation circuit.
The object of the present invention is to provide a control device that can stabilize electron beam energy and radiation output.

First, the principle of the present invention will be explained.

In other words, as shown in Figure 2, if the radiation intensity adjustment mechanism is kept unchanged, if the parameters are kept in the optimal state to obtain a predetermined electron beam energy (the setting parameters represented by X in Figure 2) By using the performance that the radiation intensity Po detected by the radiation detector 6 becomes the maximum value Dm, and the radiation intensity distribution becomes flat as D, = D2. A control method is used so that Do becomes Dm.

At this time, in order to make Do irrelevant to the amount set by the radiation intensity adjustment mechanism (represented by y), DN=Do/
Normalization by y is performed so that DN becomes the maximum value DNm. FIG. 3 shows the basic configuration for achieving this. In FIG. 3, the output of the radiation detector 6 is transferred to an intensifier 8.
The normalization circuit 16 normalizes the output.

Normalization is performed by the amount by which the radiation intensity is set by the radiation intensity adjustment mechanism 15, and DN=Do/y as already shown. This is differentiated by a differentiating circuit 17 to generate a disc N/dt, but this is done by a gate signal generated by an 18 timing circuit 18, and this gate signal is generated at intervals of Δt. Therefore, dDN/dt occurs at intervals of Δt and is maintained for Δt. 19 is the output ±phantom D of the differentiating circuit 17
This is an intensifier that increases N/dt, and its output is sent to the absolute value circuit 2.
Enter 0.

The absolute value circuit 20 outputs an amount corresponding to the l-mountain DN/dtl, and the control amount is determined. On the other hand, the output of the intensifier 19 is input to the control parameter change rate discrimination circuit 22,
If >0, maintain the changing direction of X and set the dish N/dt<
If it is 0, it has the function of reversing the direction in which X changes and generates a signal corresponding to it. When the output of the absolute value circuit 20 becomes rainbow DN/dt=0, the gate of the control discrimination circuit 23 is closed and the output of the control parameter change rate discrimination circuit 22 is stopped from being transmitted to the control signal generation circuit 25. As described above, the control signal generation circuit 25 generates the direction in which X should change and the control amount when DN/dt=0, and maintains the current value of X when N/dt=0. It is working and the DN is at its maximum value. However, at the beginning of irradiation, the mountain DN/dt
= 0, so no control signal is generated, so the initial condition setting circuit 24 sets an appropriate control amount for the absolute value circuit 2 during the first Δt, and also determines the direction of change of the control parameter change rate discrimination circuit 22. Set. Simultaneously with the start, the timing circuit 18 sends a signal for generating a gate signal to the timing circuit 18. As described above, the control signal generation circuit 25 controls the electron beam energy setting parameters of the accelerator, thereby controlling Xm in FIG. 2 to obtain stable electron beam energy and radiation output. It is something. Here, the circuit shown in FIG. 3 uses a differentiating circuit 17, but since a differentiating circuit using an intensifier is likely to generate noise, there is a risk that the device is likely to malfunction. The present invention achieves maximum value control of normalized radiation intensity without using a differentiation circuit.

An embodiment of the present invention will be described below with reference to the drawings. In FIG. 4, 26 is an irradiation start signal gate generation circuit, 27 is a timing pulse generation circuit, 28a, 28b, and 28c are sampling/hold circuits, respectively, and 29a, 29b, 29
c is sampling/holding circuit 28a, 28b, 28
This is an amplifier that configures the output of c by impedance conversion, and includes sampling/holding circuits 28a, 28b,
28c. Multiplier 29a
, 29b, 29c have a high input impedance so that the outputs of the sampling/holding circuits 28a, 28b, 28c can be held for a sufficiently long time, and their output sides have a low output impedance. 30 is a differential multiplier, 31 is an absolute value circuit for configuring the absolute value of the output of the multiplier 29c, and 32 is a flip buffer.

The output is initialized by the output gate of the irradiation start signal gate generation circuit 26, and when the output of the intensifier 29c is positive, the output is not inverted and the output of the intensifier 29c becomes negative. When the output is reversed, the output is inverted. Reference numeral 23 denotes a parameter control unit to be controlled to determine the electron beam energy of the acceleration unit.

Flipf. The output of the circuit 32 determines the direction of increase or decrease in the quantity to be controlled, and the absolute value circuit 31
The control amount is determined by the output of . Next, the operation of the apparatus according to the present invention will be explained.

Radiation such as X-rays or electron beams is generated by the radiation generation mechanism 3, its output intensity is detected by the radiation detector 6, and the output intensity is increased by the intensifier 8. At the same time, the detected values generated in pulses are averaged. do. This output can display the radiation intensity and the cumulative amount of radiation. On the other hand, the output of the intensifier 8 is normalized by an amount proportional to the radiation intensity adjustment value set by the radiation intensity adjustment mechanism 15.

Let DN be a normalized value or a value proportional to it. 2
7, a timing pulse generation circuit 27 generates three types of pulses with different timings during a period Δt.

During the pulse, it has a value that is sufficiently small with respect to Δt and long enough that the inputs of the sampling and holding circuits 28a, 28b, and 28c respond to the sampling and holding circuits 28a, 28b, and 28c. When the first pulse S, occurs after a certain time, this causes Δt. After a delay, the next pulse S2 occurs at time t2. Further delayed by Δt2, the third pulse S3 arrives at time t.
3, and then S, occurs again after Δt3. Therefore, one cycle is △t: △ et al. + △t2 + △ et al. These pulses S, S2, and S3 are applied to sampling and holding circuits 28a, 28b, and 28c, respectively, and during the gates opened by each pulse, sampling and holding circuits 28a, 28b, and 28c output the values of their inputs. Communicate to the side. Sampling hold circuit 28a, 2
In order to maintain the output values of 8b and 28c for a sufficiently long time, high input impedance multipliers 29a and 29b are installed on their output sides.
, 29c are installed and maintain this value during the period Δt.
The outputs of the multipliers 29a, 29b, 29c are respectively connected to the D side DN2.
, DN3. The output of DN・, DN2 is the differential amplifier 3
0 allows the difference between the two to be detected. It is assumed that the output of the differential amplifier 30 is Dd, and Dd is, for example, Dd=DN2-DN. Dd is sampled and held again by the sampling and holding circuit 28c, and the output is held by the amplifier 29c having a high input impedance. Multiplier 29c
The output of is DN3. Therefore, these relationships will be explained using FIG. 5.

Now, if the period of △t is at the n-th period after the start of irradiation, the pulses S, , S2, and S3 are tn・, tn2, respectively.
, tn3, and in the next (n+1)th cycle, it occurs at the time t(n river, t(n+,)2, t(n river 3).Now, the controlled object X changes. It is assumed that the normalized radiation intensity DN gradually increases in the n-th period in Fig. 5, and as a result, X
changes and there is almost no change in DN (n+
1) It is the th period.

Due to the pulse S, generated at time tn, the intensifier 29
The output of a is obtained by reading the value of DN at that time, DN, at time t.
(It is held until time t(river,).Next, due to the pulse S2 generated at time tn2, the output of the multiplier 29b reads the value DN2 of DN at that time and is held until time t(river,)2. .Therefore, the differential intensifier 30 has Dd=
Since DN2-DN, is output, when DN is increasing, the value of DN2 at the time of t(n-,)2 is stored, so it becomes negative during △t, and DN2 is output by t rudder. When is updated, DN2>DN, and Dd changes to a positive value. In this way, at time tn3, the value of Dd changes to pulse S3.
This appears as the output DN3 of the multiplier 29c. t
Until n3, the output of the multiplier 29c was positive because the value of Dd at time t(n-,)3 was stored and DN was increasing. Since DN is closer to the maximum value, DN2 and DN・
Since the difference between
The value of 3 has become smaller than before. Due to the value of DN3 up to time tn3, the flip-flop 32 changes the value of the controlled object X in a certain direction because DN3 is positive.
Further, the control amount is determined by the absolute value circuit 31, and the control section 33 has been issuing control drive signals until time tn3. At time tn3, DN3 is newly generated, but since it is still a positive value, the flip-flop 32 does not change the direction of change in X, and continues to control until t (sail) 3 with the control amount determined by the absolute value circuit 31. Section 33 generates a control drive signal. (n+
1) DN,, DN2, DN in the same way for the th period.
3 occurs, but it is close enough to the maximum value, so
When DN, detected at the interval of △t, and DN2 become almost equal values, Dd at the time of t(n river 3 is almost Dd = 0, so DN3 = 0, and t(n river 3 and thereafter) When the control amount becomes 0, the control section 33 stops the drive signal that changes X and holds the current value.

If the state does not change after this, even if the period Δt is repeated, no drive signal is generated to change X at DN3-0, so the optimum state is maintained. If for some reason DN
When DN3 starts to change, DN3 becomes DN3≠0, that is, DN3<0, and the direction of change in X indicated by the flip-flop 32, which has been set in the previous state, is reversed and x is changed. As a result, if DN increases and DN3 becomes positive, X is controlled until DN3=0 again. However, as a result of this, if DN3 is still negative, the direction of change in x indicated by flip-flop 32 is reversed again, and a drive signal is generated that changes x in the direction in which DN increases. In this way, control is achieved in which the DN reaches its maximum value. The above explanation is about the operation when DN starts to change once after starting irradiation. If DN does not change at the starting point, that is, if X does not change, it is not known whether the value of X at that time is the maximum value. Therefore, an irradiation start signal gate generation circuit 26 is required to set the initial conditions. In FIG. 6, irradiation is started at time .alpha., and the pulse period described above is repeated from time .alpha., which is Δt after .alpha. That is, t, . , t,2, and L3, pulses S, , S2, and S3 are generated, respectively. The pulse period Δt starts from time t, but the gate So of the irradiation start signal gate generation circuit 26 is delayed from to to Δt and opens so that S, , S2, and S3 are output. Therefore, the pulse S. is masked and not applied to the sampling/holding circuit 28a. Here, if △to < △, the pulse S2 at to2 will be the sampling
The signal is applied to the hold circuit 28b, and the current DN is read. The gates of S, S2, and S3 remain open before the start of irradiation, and are controlled by a logic circuit that closes the gates at the same time as the start of irradiation. Therefore, since DN, DN2, and DN3 are all reset to 0 before the start of irradiation, the initial Δt after the start of irradiation is
During this period, the DN is read only by the intermediate device 29M, and at to3, a positive value appears in DN3. Therefore, the control amount of the drive signal that changes x is obtained, and if the irradiation start signal gate generation circuit 26 sets the output of the flip-flop 32 in a predetermined direction before the start of irradiation, then t Control unit 3 between t and 3
Three control signals are defined to vary X. The subsequent control operations are as already described. As described above, a signal for controlling the controlled object is obtained, and the parameters are controlled to the optimum state.

Moreover, while applying the principle of differentiation, the circuit was configured with only a commonly used pulse generation circuit, an intensifier, and a sampling hold without using a differentiation circuit, so noise accompanying the differentiation circuit was generated. This can prevent malfunctions due to In the embodiments of the present invention described above, the control target is the electron beam energy, but in devices where the electron beam energy can be stabilized by other control methods, similar control can be performed using the electron beam trajectory. In order to correct the beam trajectory, a steering coil or a bending electromagnet current may be controlled.

Further, although the sampling and holding circuits 28a and 28b are used, delay circuits having different time constants may be provided instead. The radiation emitters 6 share a detector that monitors the output intensity, but a separate detector may be provided for this control circuit, for example, by detecting the radiation intensity at the center of the irradiation field and inputting it to the control circuit. A similar effect can be obtained as a signal. Furthermore, in FIG. 4, the embodiment is shown as if the timings at which the pulses S, , S2, and S3 generated from the timing pulse generation circuit 27 are fixed as t, , t2, and 3 above. Indicated. However, since the linear electron accelerator is operated in pulses, S, , S2 are synchronized with this pulse.
, S3 can also be generated.

Usually, in the intensifier 8, the pulsed output ionizing current of the radiation detector 6 is often averaged and intensified. This averaging may result in ripples determined by the circuit time constant. Therefore, it is synchronized with the operating pulse of the device (even devices that have a certain time delay with respect to the operating pulse are considered synchronized)
If the pulses S, , S2, and S3 are generated after the delay, fluctuations due to ripples can be removed from the outputs of the multipliers 29a, 29b, and 29c. FIG. 7 shows an example of the circuit.

In FIG. 7, only S, , S2 are shown. The driving pulse trigger pulse generation circuit 34 generates a pulse whose timing matches the linear electron accelerator driving pulse or a pulse with a certain time delay. Timing trigger generation circuit 35
From then on, the cycle is Δt, and a trigger pulse is generated at the timing t2. Of course, a trigger pulse may be further generated from here at timing t3 for S3. 36a,
36b is a flip-flop circuit, and 37a and 37b are gate circuits, which are provided for S, S2, respectively.

A trigger pulse generated from the timing trigger generation circuit 35 at timing tl is applied to the flip-flop 36a to operate it and generate a signal to open the gate of the gate circuit 37a. When a pulse is input from the trigger pulse generation circuit 34 while the gate of the gate circuit 37a is open, this is transmitted to the output side and becomes a pulse S.
Pulse S, is simultaneously fed back to flip-flop 36a, and at the end of pulse S, flip-flop 36a is operated to close its gate. Therefore, during the period Δt only one pulse occurs as S,. On the other hand, the trigger pulse generated from the timing trigger generation circuit 35 at timing t2 is generated by the flip-flop 36b and gate circuit 37b having the same functions as described above. In the above manner, S, S2, which are periodic to the operating pulse of the linear electron accelerator during the period Δt, are obtained.

Regarding S3, it may be similarly synchronized, or it may be a pulse delayed by an appropriate amount of Δ with respect to S2. In FIG. 8, P is the output pulse of the trigger pulse generation circuit 34, and .largecircle., , G2 are the output gate signals of the flip-flops 36a and 36b, respectively.The above operation can be understood from FIG. Furthermore, the circuit according to the present invention not only determines the maximum value of radiation intensity, but also serves as a control mechanism for controlling pulse repetition frequency, pulse duration, beam current, etc. to the minimum in order to obtain a predetermined dose. Can be applied. In this case, the flip-flop 32 is set so that the maximum control point to be sought becomes the lowest point.
The time when the output becomes an inverted output may be set as the time when the output of the multiplier 29c becomes positive. As described above, the present invention includes a control circuit that normalizes the radiation output of a radiation generation device using a linear electron accelerator and obtains the maximum value (or minimum value) of that value, and does not use a differentiating circuit. Since the configuration is configured to obtain the same functions as the normally used configurations such as the pulse generation circuit and the multiplier sampling and hold circuit, it is possible to prevent malfunctions due to noise peculiar to differentiating circuits, and to ensure stable operation, the electronic The effect of providing a highly reliable device with stable beam energy radiation output can be obtained.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an example of a conventional linear electron accelerator, Fig. 2 is an explanatory diagram showing the relationship between radiation output and parameters that define electron beam energy, and Fig. 3 is the basic device of the present invention. 4 is a block diagram showing a device according to an embodiment of the present invention, FIG. 5 is a timing chart showing the normal operation of the device in FIG. 4, and FIG. 6 is a diagram showing the control mechanism of FIG. 4. FIG. 7 is a timing chart of operations for determining initial conditions, and FIG. 7 is a diagram showing a partial modification of the apparatus of the present invention. FIG. 8 is a timing chart showing the operation of the circuit shown in FIG. In the figure, 1... accelerator, 6... radiation detector, 8... intensifier, 15...
... Radiation intensity adjustment mechanism, 16... Normalization circuit,
26...Irradiation start signal gate generation circuit, 27.
...timing pulse generation circuit, 28a, 28b
, 28c... Sampling hold circuit, 29a,
29b, 29c...Intensifier, 30...Differential intensifier, 31...Absolute value circuit, 32...Flip-flop, 33...Control Section 34...Trigger pulse generation circuit for driving pulses, 35...
. . . Timing trigger generation circuit, 36a, 36b . . . flip-flop, 37a, 37b . . . gate circuit. Note that the same reference numerals in the figures indicate the same or corresponding parts. Figure 1 Figure 2 Figure 7 Figure 3 Figure 5 Figure 4 Figure 6 Figure 8

Claims (1)

[Scope of Claims] 1. An adjustment mechanism that adjusts the output intensity of radiation in a device that includes an accelerator that accelerates an electron beam and generates radiation from the electron beam accelerated by the accelerator; a radiation detector that detects the intensity of the generated radiation; a normalizing means that normalizes the output radiation intensity detected by the radiation detector using a value corresponding to the amount set by the adjustment mechanism as a reference value; Detect the difference between the value of the output of the normalizing means at a given point in time and the value at a point before that point, and use this difference to accelerate the above normalized amount so that it reaches the maximum value or the minimum value. A radiation generating device comprising: a control device for controlling the device. 2. The control device that controls the accelerator so that the normalized quantity becomes the maximum value or the minimum value includes a first sampling and holding circuit that holds the output value at any point in time of the normalization means, and a normalization means. a second sampling/holding circuit that holds an output value at a time after the above-mentioned point in time, and a differential amplifier that produces an output according to the difference between the outputs of the first and second sampling/holding circuits; It has a third sampling and hold circuit that samples and holds the output value of this differential amplifier, and controls the accelerator so that the output value of this third sampling and hold circuit is positive. The first claim characterized in that
The radiation generating device described in . 3. The control device that controls the accelerator so that the normalized amount becomes the maximum value or the minimum value has a timing pulse that causes the first, second, and third sampling/hold circuits to perform sampling operations. 3. The radiation generating device according to claim 2, further comprising a pulse generating circuit. 4. Claim 2, characterized in that the control device is equipped with means for operating so that the first or second sampling and holding circuit does not newly sample the output of the normalization output stage at the initial stage of radiation generation. The radiation generating device according to item 1 or 3.