Method and arrangement for frequency doubling

Abstract

To generate an output signal (11) the frequency of which is twice the frequency of an input signal (1, 2), a delayed signal (3, 4) which is delayed relative to the input signal (1, 2) by a quarter of the latter's cycle period is generated and the output signal (11) is then generated as the difference between the rectified input signal (1, 2) and the rectified delayed signal (3, 4). The input signal (1, 2) and the delayed signal (3, 4) are advantageously rectified by using differential signals each comprising a positive component signal (1, 3) and a negative component signal (2, 4). A respective one of two transistors connected in parallel is driven by a positive component signal (1, 3) and a negative component signal (2, 4) in such a way that a positive half-wave causes the relevant transistor (5-8) to conduct and the relevant transistor (5-8) blocks in a negative half-wave. The rectified input signal (1, 2) or delayed signal (3, 4) is obtained from the two component currents flowing through the pairs of transistors (5-8) connected in parallel when the currents are added. The rectified input signal (1, 2) and delayed signal (3, 4), which are in the form of current signals, can be passed through two resistors (9, 10) to enable corresponding voltage signals to be generated across the latter and the output signal (11) to be picked off from the two resistors (9, 10).

Description

[0001] The present invention relates to a method and an arrangement for generating an output signal the frequency of which is twice the frequency of an input signal.

[0002] In a known method of frequency doubling, a cyclic input signal is multiplied by the input signal which has been delayed by a quarter of the cycle period. A typical implementation of such a method is, for example, the Gilbert multiplier. An arrangement of this kind is known from EP 1 187 310 A2, for example.

[0003] A particular disadvantage which known arrangements of this kind suffer from is that they are made up of a relatively large number of components and have a large number of nodes to which an a.c. signal is applied. The disadvantage this has, particularly at high frequencies, is that the capacitive load becomes very high and integration becomes difficult.

[0004] This disadvantage is overcome in accordance with the invention by a method having the features given in claim 1 and by an arrangement having the features given in claim 9.

[0005] In accordance with the present invention, an input signal and a delayed signal are rectified and the difference between the rectified signals is determined. The delayed signal may be generated as a function of the input signal in this case, in which case the delayed signal is delayed relative to the input signal by a quarter of the latter's cycle period but is of the same frequency. Often however, the delayed signal will already exist, as will be the case for example when use is made of an oscillator which produces two signals which are shifted by a quarter of the cycle period and of which one can then used as the input signal and the other as the delayed signal. The present invention is based on the fact that a rectified trigonometrical function given by the following formula can be described as a series expansion of a trigonometrical function of twice the frequency, with the higher terms being severely damped in this case. 1$\begin{array}{c}\mathrm{abs}\left(\sin \left(2\pi ft\right)\right)=\frac{2}{\pi }-\frac{4}{\pi }\sum _{i=1}^{\infty }\frac{\cos \left(2\pi f·2i·t\right)}{4i^2-1}\\ \mathrm{abs}\left(\cos \left(2\pi ft\right)\right)=\frac{2}{\pi }-\frac{4}{\pi }\sum _{i=1}^{\infty }{\left(-1\right)}^i\frac{\cos \left(2\pi f·2i·t\right)}{4i^2-1}\end{array}$

[0006] Because the sine function is delayed by a quarter of the cycle period relative to the cosine function, the input signal and the delayed signal can be associated with a cosine function and a sine function respectively. Once the difference between the rectified input signal and the rectified delayed signal has been formed, what is obtained is the following formula as a series expansion of the output signal. 2$\mathrm{abs}\left(\cos \left(2\pi ft\right)\right)-\mathrm{abs}\left(\sin \left(2\pi ft\right)\right)=\frac{8}{\pi }\sum _{i=1}^{\infty }\frac{\cos \left(2\pi f·2\left(2i+1\right)·t\right)}{4{\left(2i+1\right)}^2-1}$

[0007] It can be seen from this formula that a dominant term appears on the right-hand side at twice the input frequency and that the higher-frequency terms are severely damped. Advantageously, it is only uneven multiples of the desired output frequency that appear. This fact is particularly advantageous to enable jitter to be avoided.

[0008] This method of frequency doubling is particularly suitable for implementation in a semiconductor chip. For this purpose, use is made of a differential input signal and a differential delayed signal which each comprise a positive and a negative component signal. The component signals of the input signal and the component signals of the delayed signal are, for their part, once again cyclic signals but they have different signs, which means that the positive half-waves of one component signal are in phase with the negative half-waves of the other component signal. By means of the component signals of both the input signal and the delayed signal, signals are applied to circuit-switching elements, such as transistors for example, with each current-switching element being so controlled by a component signal that the positive and negative half-waves of the component signal are converted at different gains into component currents flowing through the current-switching elements. Provision may in particular be made for the corresponding current-switching element to block at a positive or negative half-wave of one component signal. This may, for example, be achieved by controlling a transistor by means of a voltage, in which case the negative half-waves of the voltage cause a negative voltage to be applied to the control input of the transistor, which may be the base or gate, for example.

[0009] The two component currents which flow through the current-switching elements to which the component signals of either the input signal or the delayed signal are applied are added and thus form a summed signal which corresponds to the rectified input signal or the rectified delayed signal. For rectification of this kind to work, it is essential for the positive and negative half-waves of the component signals not to be converted into currents in the same way. As soon as either the positive or the negative half-waves of the component signals produce a current which is less than proportional, due in particular to blocking of the current switching element, what is produced by the summing of two component currents is rectification either of the input signal or of the delayed signal, and this in itself gives twice the frequency.

[0010] The two summed signals are current signals which generate a voltage signal across a resistor or reactor, i.e. an impedance, which means that the output signal can be generated at little cost or complication by means of two resistors or reactors to each of which one summed signal is applied. For this, it is enough for one end of each of the two resistors or reactors to be held at a constant voltage to enable the output signal to be picked off from the other ends of respective resistors or reactors. The resistors or reactors may be ohmic resistors or inductive reactors or a combination of ohmic resistors and inductive reactors. What may also be used as resistors or reactors are semiconductor structures, such as field-effect transistors for example, which are operated in particular in the linear range.

[0011] In an advantageous embodiment, the output signal may be filtered to remove unwanted signal components. What may be used in this case is in a particular a low-pass filter or a band-pass filter, in which case the filtering and the nature of the filtering will be suited to the intended field of application of the invention.

[0012] The invention will be explained in detail below by reference to a preferred embodiment and to the accompanying drawings.

[0013] The single FIGURE of the drawings shows in this case the layout of the embodiment of the circuit according to the present invention for frequency doubling.

[0014] The circuit shown in the FIGURE is used to generate an output signal 11 the frequency of which is twice the frequency of the input signal 1, 2. The input signal 1, 2 is differential and is transmitted by means of two component signals 1, 2 which comprise a positive component signal 1 and a negative component signal 2. The input signal 1, 2 is a sinusoidal signal and the two component signals 1, 2 are therefore sinusoidal likewise but are of different signs. A delayed signal 3, 4 is generated by means of an arrangement which is not shown, and it too is differential and is transmitted in the form of two component signals 3, 4, component signal 3 being the positive component signal of delayed signal 3, 4 and component signal 4 being its negative component signal. The delayed signal 3, 4 represents the input signal 1, 2 after being delayed by a quarter of its cycle period. The component signals 3, 4 of the delayed signal are likewise sinusoidal signals but they are of different signs and are delayed relative to the component signals 1, 2 of the input signal by a quarter of the cycle period.

[0015] The circuit used for frequency doubling comprises four field-effect transistors 5-8 which are each driven by one of the component signals 1-4.

[0016] The field-effect transistors 5, 6 which are driven by the component signals 1, 2 of the input signal are connected in parallel at their drain terminals and source terminals. The same is true of the field-effect transistors 7, 8 which are driven by the two component signals 3, 4 of the delayed signal.

[0017] The source terminals of all the field-effect transistors 5-8 are connected together and are connected via a constant-current source 12 to a negative pole 13 of a voltage supply source. The drain terminals of the field-effect transistors 5, 6 which are driven by the input signal 1, 2 are connected via a resistor 9 to a positive pole 14 of the voltage supply. The drain terminals of the field-effect transistors 7, 8 which are driven by the delayed signal 3, 4 are likewise connected, via a second resistor 10, to the positive pole 14 of the voltage supply.

[0018] All the component signals 1-4 are so arranged in relation to the voltage supply 13, 14 that they can cause respective ones of the field-effect transistors 5-8 to conduct in the positive half-waves and that respective ones of the field-effect transistors 5-8 block during at least part of the negative half-waves. This may, for example, be achieved by setting the mean value of the component signals 1-4 at a voltage level which is at least substantially at the same voltage as the source terminals of the field-effect transistors 5-8. For this purpose, the component signals 1-4 may have superimposed on them a direct component, in which case the setting of the component signals 1-4 will also be suited to the type and transfer characteristic of the field-effect transistor 5-8.

[0019] Because it is the two component signals 1, 2 of the input signal which drive the two field-effect transistors 5, 6, it being substantially only the positive half-waves of the two component signals 1, 2 which cause the field-effect transistors 5, 6 to conduct, the sum current which flows through the two field-effect transistors 5, 6, and thus through the first resistor 9, is one which has the waveform of a rectified sinusoidal signal. The same is true of the current which flows through the two field-effect transistors 7, 8 which are driven by the delayed signal 3, 4 and through the second resistor 10, the current flowing through the second resistor 10 being delayed relative to the current flowing through the first resistor 9 by a quarter of the cycle period which exists at the frequency of the input signal 1, 2 and the delayed signal 3, 4.

[0020] Due to the rectification, the currents flowing through the two resistors 9, 10 are already doubled in frequency, the current waveform having however a large number of harmonics. The output signal 11 is picked off from the two nodes at which the two resistors 9, 10 are respectively connected to their associated field-effect transistors 5-8. The output signal 11 which is obtained in this way corresponds to the difference between the voltage drops which are generated across the resistors 9, 10 and it is of a substantially sinusoidal waveform. By means of the present invention, it is possible in this way, at little cost or complication and in particular with only a small number of components, to obtain an output signal of doubled frequency, the waveform of the output signal 11 being low in harmonics.

[0021] A major advantages results from the reduced number of components at high input frequencies, particularly ones equal to or greater than 1 gigahertz. Because of the small number of components, it is possible for a circuit of this kind to be implemented in a semiconductor chip considerably more easily than it is with the frequency doubling circuits known in the past.

Claims

1. Method of generating a cyclic output signal (11) the frequency of which is twice the frequency of a cyclic input signal (1, 2), in which method a delayed signal (3, 4), which corresponds to the input signal (1, 2) delayed by a quarter of the latter's cycle period, and the input signal (1, 2) are rectified and the difference between the rectified input signal (1, 2) and the rectified delayed signal (3, 4) is formed and is fed out as an output signal (11).

2. Method according to claim 1, characterised in that the delayed signal (3, 4) is generated from the input signal (1, 2) by applying a delay of one quarter of the latter's cycle period.

3. Method according to claim 1, characterised in that the output signal (11) is filtered, the filtering having little or least damping at twice the frequency of the input signal (1, 2).

4. Method according to claim 1, characterised in that the input signal (1, 2) and/or the delayed signal (3, 4) are/is differential

5. Method according to claim 1, characterised in that the input signal (1, 2) and the delayed signal (3, 4) are differential and each have a negative and a positive component signal (1-4) and, to rectify the input signal (1, 2) and the delayed signal (3, 4), a current-switching element (5-8) is driven by means of each component signal (1-4) of the input signal (1, 2) and the delayed signal (3, 4) in such a way that the component currents flowing through the current-switching elements (5-8) always have the same sign and the positive and negative half-waves of the component signals (1-4) are converted into the respective component currents at different gains, and the component currents flowing through the current-switching elements (5-8) driven by the component signals of the input signal (1, 2) and the delayed signal (3, 4) respectively are added to form first and second summed signals respectively, the two summed signals respectively corresponding to the rectified input signal (1, 2) and the rectified delayed signal (3, 4).

6. Method according to claim 5, characterised in that the current-switching elements (5-8) are driven by means of the component signals (1-4) in such a way that the current-switching elements (5-8) block in the negative or positive half-waves of the component signals (1-4).

7. Method according to claim 5, characterised in that the summed signals are each passed through a resistor or reactor (9, 10) and the output signal (11) is generated as the difference between the voltage drops across the resistors or reactors (9, 10), respectively.

8. Method according to claim 5, characterised in that a constant current is applied to all the current paths through the current-switching elements (5-8).

9. Arrangement for generating a cyclic output signal (11) the frequency of which is twice the frequency of a cyclic input signal (1, 2), having two rectifying means (5-8) for rectifying both the input signal (1, 2) and a delayed signal (3, 4) which corresponds to the input signal (1, 2) delayed by a quarter of the latter's cycle period, and means (9, 10) for forming the difference between the rectified input signal (1, 2) and the rectified delayed signal (3, 4) and for feeding out the difference as an output signal (11).

10. Arrangement according to claim 9, characterised in that the arrangement has delaying means to generate the delayed signal (3, 4) from the input signal (1, 2) by applying a delay of a quarter of the latter's cycle period.

11. Arrangement according to claim 9, characterised in that the arrangement has a filter for filtering the output signal (11), the filter having little or least damping at twice the frequency of the input signal (1, 2).

12. Arrangement according to claim 9, characterised in that the input signal (1, 2) and the delayed signal (3, 4) are difference signals each having a negative and a positive component signal (1-4), and the arrangement has four current-switching elements (5-8), to each of which a component signal (1-4) is applied in such a way that the component currents flowing through the current-switching elements are always of the same polarity and the positive and negative half-waves of the component signals (1-4) are converted into current by the current-switching elements (5-8) at different gains, the arrangement being so arranged that the component currents flowing through the current-switching elements (5-8) driven by the component signals (1-4) of the input signal (1, 2) and the delayed signal (3, 4) respectively are added to form first and second summed signals respectively, and the difference between the rectified input signal (1, 2) and the rectified delayed signal (3, 4) is formed by forming the difference between the two summed signals.

13. Arrangement according to claim 12, characterised in that the arrangement is so arranged that the current-switching elements (5-8) block in the positive or the negative half-waves of the component signals (1-4).

14. Arrangement according to claim 12, characterised in that arrangement has two resistors or reactors (9, 10) to each of which a summed signal is applied, and the arrangement is so arranged that it forms the difference between the voltage drops across the two resistors or reactors (9, 10), respectively, as an output signal (11).

15. Arrangement according to claim 12, characterised in that the arrangement has a current source (12) which is connected to the current paths of all the current-switching elements (5-8).