In general, a modulator combines a carrier input signal and a modulating input signal to generate a modulated carrier output signal. In a radio frequency (RF) transmitter, the modulating input signal will typically be a baseband message signal which is required to be transmitted over one of a range of radio frequency (RF) channels. In one type of RF transmitter, the modulator modulates the message signal onto a sinusoidal carrier signal having a frequency equal to the frequency of one of the RF channels. This process is known as direct-at-carrier modulation and is characterised in that the desired RF channel is determined by the frequency of the carrier signal. In another type of RF transmitter, the modulator modulates the message signal onto a carrier signal having a fixed intermediate frequency and the modulated carrier output signal is then frequency up-converted to an appropriate RF channel frequency. This process is known as intermediate frequency (IF) modulation and is characterised in that the desired RF channel is determined by the frequency up-conversion process.

The modulation techniques used for RF transmission may also be used in radio test equipment for the generation of RF signals which may not necessarily be broadcasted.

There exist various methods by which modulators combine a carrier signal and a modulating signal. In amplitude modulation (AM), the modulating or message signal is impressed on the amplitude of the carrier signal. In quadrature amplitude modulation (QAM), the message signal consists of a in-phase (I) modulating signal and a quadrature phase (Q) modulating signal which amplitude modulate an in-phase carrier and a quadrature phase carrier respectively. A quadrature amplitude modulator typically comprises a separate amplitude modulator for the I modulating signal and the Q modulating signal.

Figure 1 shows a typical double-balanced mixer that is capable of being used as a modulator.

The mixer has three ports, LO, RF and IF. In a typical application the local oscillator is applied to the LO port and a signal is applied to the RF port. The output is taken from the IF port. In many respects the ports are interchangeable, but the IF port is fundamentally different in that it is DC- coupled whereas the other two ports are not due to their transformer coupling. In a modulator application where the modulation contains low-frequency and possibly DC-components the modulation would be applied to the IF port and the modulated output would be taken from the RF port.

The applicant has performed investigations into the possibility of using such a mixer as a direct-at- carrier modulator. During these investigations the following problems and limitations were revealed : Firstly, balance is only good at low carrier frequencies and this degrades with increasing local oscillator frequency. This is due to the fact that the LO and RF ports are connected to an unbalanced source and load respectively, and capacitive coupling between the transformer windings causes signal breakthrough even in the absence of any stimulus being applied to the IF port.

Secondly, intermodulation performance too is only good at low carrier frequencies and also degrades with increasing local oscillator frequency. There is also considerable variation of intermodulation performance with frequency in a way that is complicated and difficult to predict, and at high frequencies it is possible for the performance to change from good to bad or vice-versa over a relatively small change of frequency.

It is believed that this performance change is due to an interaction of the switching action of the diodes and the imperfect nature of the transformers. The transformers only work correctly at frequencies where the electrical length of the windings are short in comparison with the wavelength of the frequencies involved, in this case the local oscillator frequency. However, the switching action of the diodes gives rise to a series of spectral components at harmonics of the local oscillator frequency, and at high local oscillator frequencies these will have wavelengths that are not short in relation to the winding-lengths of the transformer.

Ideally, these harmonics are reflected back through the transformer at the LO port and are ultimately absorbed by the source resistance of the local oscillator. It is often in fact customary to place some resistive attenuation between the LO source and the mixer to assist with this when using local oscillators with poorly controlled source impedance. If this is not possible due to limitations in the transformer, then the harmonics will be re-reflected back to the diodes in an amplitude and phase that may vary as a function of frequency, and in a way that may be detrimental to their switching action.

This effect is illustrated in Figures 2a and 2b. Figure 2a is a plot of a third order harmonic signal A added in-phase to a fundamental frequency signal B and illustrates a steep transition C between the positive half cycle and the negative half cycle. In contrast, Figure 2b is a plot of a third order harmonic signal A- added in anti-phase with a fundamental frequency signal B and illustrates a shallow, ill-defined transition C"between the positive and negative half cycles.

Because the phase difference between the fundamental signal and the harmonic components varies with frequency so the switching characteristics of the diodes will also vary with frequency.

The performance of the modulator is therefore increasingly frequency dependent as the frequency of the local oscillator signal increases.

In practice the mixer in Figure 1 works very well up to approximately 200MHz. Above this there is a gradual deterioration of balance and intermodulation performance. Above about lGHz balance is very poor and intermodulation performance is heavily frequency-dependant.

There exist more complicated mixer topologies designed so as to overcome one or more of the difficulties experienced by the mixer of Figure 1, involving the use of additional transformers, diodes, or in some cases resistors. Unfortunately not all of these have a DC-coupled IF port, and are therefore not suitable as a modulator. Of those that remain, there do not appear to be any that fully satisfy the requirements of balance and good intermodulation performance. Furthermore, the relative complexity of such mixers make them costly to construct and thus unsuitable for volume production.

According to the present invention there is provided a modulator for modulating a carrier signal with a modulating signal, the modulator having an input stage for receiving a directly coupled balanced modulating signal and for providing a grounding path for the carrier signal, a switching network for switchably supplying, in accordance with a switching cycle controlled by the carrier signal, the modulating signal in either forward or reverse sense to an output stage, a further input stage for providing an input for the carrier signal, and an output stage for converting the balanced modulated signal into an unbalanced output signal.

A modulator in accordance with the invention has the advantage that it provides input for both the modulating and carrier signals without the need for transformers, which fact helps to maintain good intermodulation performance without dependency on frequency.

A further advantage is that the output stage can be designed for optimum balance and bandwidth without the need for it to incorporate either an input or a ground for the carrier signal. This helps to maintain optimum balance.

Advantageously, the balanced modulating signal may be directly coupled by means of a conductive or resistive coupling in the input stage.

Preferably, the input stage comprises first and second input terminals for the modulating signal.

Preferably, the input stage provides first and second conductive or resistive couplings between the first and second input terminals and the switching network, and first and second resistive couplings from the first and second input terminals to ground.

Preferably, the modulating signal is supplied by the switching network to first and second output terminals of the output stage. Ideally, the switching network comprises two pairs of signal paths, a first pair of signal paths for supplying the modulating signal from the first and second input terminals respectively to the first and second output terminals, and a second pair of signal paths for supplying the modulating signal from the first and second input terminals respectively to the second and first output terminals. Suitably, the first and second pairs of signal paths are switched 'on' and 'off' (conducting and non-conducting) in anti-phase with one another in accordance with the switching cycle. The first and second pairs of signal paths may constitute a diode ring whereby the switching action is performed by diodes in the signal paths which are either forward or reverse biased.

Preferably, the further input stage couples the carrier signal to the switching network such that the carrier signal operates in a common mode on the two pairs of signal paths. Preferably, the carrier signal is an unbalanced signal which is coupled to the first and second output terminals via parallel resistive couplings. This provides an advantage in that a transmission line supplying the carrier signal can be optimally terminated with a resistive load which matches the characteristic impedance of the transmission line.

Suitably, the output stage comprises means for transforming the balanced floating modulated signal at the first and second output terminals into an unbalanced, grounded signal for outputting on a transmission line. Preferably, the transforming means is a balun which may be formed by having a coaxial line output having a central conductor connected to the first output terminal and a outer conductor connected to the second output terminal, and a cylindrical sleeve surrounding the coaxial line made from a high permeability lossy magnetic material such as a ferrite. This transforming means has the advantage that it maintains a constant impedance with respect to frequency for the output signal, but it exhibits a high and largely resistive impedance to the carrier signal.

The output stage may further comprise a compensating impedance coupling between the first output terminal and a reference potential such as ground so as to maintain an overall balance in the modulator. Preferably, the compensating impedance coupling comprises a conductor of equivalent diameter to that of the outer conductor of the balun coaxial line with an equivalent surrounding cylindrical sleeve of magnetic material.

Further features and advantages of the invention will be apparent from the following description.

Embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, in which : Figure 1 is a circuit diagram of a known double-balanced mixer ; Figure 2a is a plot of a third order harmonic signal added in-phase to a fundamental frequency signal ; Figure 2b is a plot of a third order harmonic signal added in anti-phase with a fundamental frequency signal ; Figure 3 is a circuit diagram of a modulator in accordance with the present invention ; Figure 4a is a diagram of a modified output stage for the modulator of Figure 3 ; Figure 4b is a diagram of another modified output stage for the modulator of Figure 3 ; Figure 5 is a diagram showing the layout and configuration of the modulator of Figure 3 including the modification of Figure 4a ; and Figure 6 is a circuit diagram of a quadrature amplitude modulator which employs two modulators in accordance with the invention.

Referring to Figure 3, there is shown a circuit diagram of a direct-at-carrier modulator 50 in accordance with the invention suitable for modulating a carrier signal CS with a modulating signal MS. The modulator 50 is particularly suitable for use as a balanced amplitude modulator in a quadrature amplitude modulator. The modulator 50 has three stages, an input stage 51 which receives the baseband modulating signal MS, a switching stage 52 which is switched by the relatively high frequency carrier signal CS, and an output stage 53 which outputs a high frequency modulated carrier signal RF.

Input Stage The modulating signal MS is a baseband signal typically, though not necessarily, having a bandwidth in the order of 30 MHz and frequency components down to zero frequency ie. DC.

The modulating signal MS is supplied via a pair of input terminals 8,9 to a pair of symmetrical signal paths 21,22 of the input stage 51 by means of a differential voltage drive 54. The modulating signal MS on each signal path 21,22 is resistively coupled via a respective resistor R, to a pair of terminals 10,11.

Each terminal 10,11 of the input stage is resistively coupled along respective signal paths 23,24 to a common ground potential G. A series resistance R2 is provided in each of the signal paths 23,24 to ground. These resistors provide a path to ground for the carrier signal and have as low a value as possible consistent with providing adequate input for the modulating signal. The combination of the signal paths 21,23, 24,22 in the input stage 51 forms a closed loop having a resistive impedance equal to 2RI + 2R2 around the differential voltage drive 54. There is also an additional load from the switching stage that appears in parallel with 2R2 but the effect of this is small if R2 has a low value. The value of R, can be conveniently chosen so that the combination of 2R,, and 2R2 matches the source impedance of MS, typically 50ohms.

The differential voltage drive 54 is provided by, for example, a complementary pair of op-amps to give a differential signal which is suitably buffered.

In an alternative embodiment, the differential voltage drive 54 is replaced by a differential current drive. In this embodiment, the resistance matching is not required and the resistors R, may be replaced by direct conductive connections to the terminals 10,11.

Switching Stage The switching stage 52 comprises a network of signal paths 25,26, 27,28 which electrically couple the input terminals 10,11 of the input stage 51 to output terminals 12,13 of the output stage 53. The signal paths 25,26, 27,28 are each provided with fast switching Schottky diodes 35,36, 37,38. Each output terminal 12,13 of the output stage is resistively coupled along respective signal paths 29,30 to a common signal input terminal 14 which is supplied with the carrier signal CS. A series resistance R3 is provided in each of the signal paths 29,30. The carrier signal CS is supplied to the signal terminal 14 as an unbalanced voltage signal relative to the ground potential G. The unbalanced signal CS is a sinusoidal signal at a frequency F and is supplied at a higher power than the modulating signal MS. The frequency F of the carrier signal can be shifted to other desired frequencies from time to time in order to vary the frequency of the modulated carrier signal RF at the modulator output.

The switching network operates as follows.

During positive cycles of the carrier signal CS, the voltage difference between the signal terminal 14 and ground potential G is positive, which results in a voltage drop across the resistors R3, the diodes 35,36, 37,38, and the resistors R2. The voltage drop across the diodes causes the outer diodes 35,36 to conduct and the crossover diodes 37,38 to be reversed biased so as not to conduct.

Conversely, during negative cycles of the carrier signal CS, the voltage difference between the signal terminal 14 and ground potential G is negative, which results in an opposite voltage drop across the resistors R3, the diodes 35,36, 37,38, and the resistors R2. Thus the crossover diodes 37,38 are conducting and the outer diodes 35,36 are switched off.

In summary, the carrier signal CS causes the outer pair of signal paths 25,26 to conduct during positive halves of the sinusoidal signal and the crossover pair of signal paths 27,28 to conduct during negative halves of the sinusoidal signal. In other words, the pairs of signal paths are switched'on'and'off in anti-phase with each other at the frequency F of the carrier signal.

The effect of switching the signal paths in the switching stage 52 is an electrical coupling of the input terminal 10 to the output terminal 12 and the input terminal 11 to the output terminal 13 during positive cycles of the carrier signal CS, and an electrical coupling of the input terminal 10 to the output terminal 13 and the input terminal 11 to the output terminal 12 during negative cycles of the carrier signal CS. Effectively, the modulating signal MS at the output terminals 12, 13 is reversed in polarity every half cycle of the carrier signal CS.

Because the carrier signal CS is at a higher voltage than the modulating signal MS, the relatively lower voltage of the modulating signal MS has a negligible effect on the switching of the diodes 35,36, 37,38. Across the switching network 52, the modulating signal MS appears as a small voltage fluctuation on the main voltage difference produced by the carrier signal CS. However, across the output terminals 12,13 the voltage differences produced by the carrier signal in the signal paths 29,30 cancel to zero whilst the modulating signal produces a voltage difference proportional to the voltage difference across the input terminals 10,11. Because the carrier signal is at a higher level than the modulating signal the signal paths 29,35, 37, and 23 are designed to be symmetrical with respective signal paths 30,36, 38 and 24. This avoids carrier signal CS energy appearing across the output terminals 12,13. In other words, the modulator 50 is designed to be correctly balanced. The resistive impedance between the signal terminal 14 and the ground potential G is defined approximately by the following equation : Rcs=(R3+Rd+R2)/2 Where Rd is the resistance of each of the diodes 35,36, 37,38.

The diode switching process is made faster and more accurate if a large current flows through the diodes. This is achieved in the modulator 50 by minimising the resistance Rcs between the signal terminal 14 and the ground potential G and by increasing the level of the carrier signal CS.

The carrier signal CS enters the modulator via a transmission line having a characteristic impedance of, for example, 50 ohms. In order to prevent reflection of the carrier signal back along the transmission line, the transmission line is preferably terminated by a matching resistive load, in this case 50 ohms. This resistive load is effectively Rcs.

The signal paths 29,30 produce a series resistance of 2R3 across the output terminals 12,13 which shunts the switched modulating signal MS present at the output terminals 12,13 away from the output stage 53. This effect is reduced by choosing a relatively large value for the resistance R3. Some of the enhancements to the modulator 50 referred to above require conflicting variations of the resistor values Ri, R2 and R3. However, a choice of values for the resistors Ri, R2 and R3 exists whereby the modulator performance is optimised. In the modulator 50 of Figure 3, suitable values of the resistors are R, = 20 ohms, R2 = 5 ohms, and R3 = 56 ohms.

Output Stage The output stage receives a balanced modulated signal from the output terminals 12,13 which is periodically reversed at the frequency F of the carrier signal CS. A Fourier Series analysis of this signal shows that it is effectively a modulated carrier signal RF with additional higher frequency modulated harmonic terms.

The output stage includes a balun 55 which comprises a section of coaxial cable 56 surrounded by a close fitting ferrite sleeve 57 with an annular cross-section. The balun 55 converts the balanced signal across the terminal outputs 12,13 to an unbalanced signal RF relative to a ground potential at the balun output. The unbalanced output signal can then be transmitted over a transmission line such as a coaxial cable.

The output terminal 12 is conductively coupled to the inner conductor 58 of the coaxial cable 56 and the output terminal 13 is conductively coupled to the outer conductor 59 of the co-axial cable 56. The output end of the balun 55 provides the unbalanced modulated carrier signal RF on the inner conductor 58 of the coaxial cable relative to a ground potential coupled to the outer conductor 59 of the coaxial cable.

The performance of the balun is impaired by a parasitic impedance to ground over the outer conductor 58 of the coaxial cable. This parasitic impedance is increased in the modulator 50 by decreasing the diameter of the coaxial cable which increases the current density along the axis of the ferrite sleeve 57 and hence the magnetic flux generated inside the ferrite sleeve.

The parasitic impedance generates an imbalance in the switching stage 52 which causes a general imbalance in the modulator 50. In a preferred embodiment, the output stage 53 is modified as shown in Figure 4a to include an impedance coupling 65 which couples the output terminal 12 to a ground potential. This impedance coupling 65 comprises conductor 69 surrounded by a ferrite sleeve 67 and is designed to complement and compensate for the parasitic impedance coupling in the balun 55. The modified output stage 53 has the effect of restoring balance in the modulator 50.

Figure 4b shows another modified output stage in which the ferrite sleeves 57,67 have been removed. This solution relies on the specific wavelength characteristic of the RF signal to unbalance the signal at the output and is thus band limited. In contrast, the output stage shown in Figure 4a uses broadband components and is thus a wide band solution.

A suitable configuration of the balun 55 includes a 0. 5 mm diameter coaxial cable having a characteristic impedance of 50 ohms.

Because the output signal RF is properly terminated, there is no problem with the signal reflecting back into the modulator and affecting performance.

The chosen physical layout of the modulator 50 is an in-line configuration as shown Figure 5.

This configuration achieves good balance in the modulator 50 by ensuring that interactions between electrical components are symmetrical and will generally cancel out. The signal paths 23 and 24 have been modified to a parallel arrangement of resistors coupled to a common ground potential. This reduces the parasitic impedance (inductance) presented to the carrier signal.

The modulator 50 is particularly suited for use in a quadrature amplitude modulator. This is in part due to the accuracy and repeatability of the electrical lengths present in the baluns, which assists in the maintenance of accurate phase-quadrature when using a pair of circuits. Figure 6 illustrates such a quadrature amplitude modulator which includes a first modulator 50 for modulating an I modulating signal MS (I) onto an in phase carrier signal CS, and a second modulator 50 for modulating a Q modulating signal MS (Q) onto a quadrature phase carrier signal CS.

It will be evident in view of the foregoing description that various modifications may be made within the scope of the present invention. For example, the carrier signal input and the radio frequency output may be interchanged such that the carrier signals CS is supplied via the balun and the radio frequency signal RF is output via the terminal 14.