A wider band is provided by BUT on connected lines. In the decimeter and long-wave part of the centimeter range, tandem couplers and Lange couplers are used. BS with such NOs (Fig. 17.10, e) provide an isolation of more than 15 dB with an SWR of at least 1.5 in a band of several octaves. A high level of isolation in a wide frequency band in the GIS BS is provided by hybrid connections based on the connection of transmission lines of different types. In the decimeter range, to reduce the size of the BS, microminiature passive elements with lumped parameters are used. Balanced mixers, unlike unbalanced ones, tend to operate at zero diode bias.

For the practical use of mixers, a higher isolation of the signal and heterodyne inputs is often required. In BS with quadrature bridges, the isolation is quite small and does not exceed 10 dB. This is due not only to the imbalance of the circuit, but also to the fact that if the diodes are not fully matched to the waveguide, the local oscillator oscillations reflected from them are sent to the signal input. To avoid this shortcoming, mixing diodes are connected to the inputs of the quadrature bridge with a shift of Λ/4. Figure 17.10, c shows the topological diagram of such a BS.

Fig. 17.10, e shows the BS circuit on the Lange bridge with additional suppression of the mirror channel using selective circuits that implement the idle mode, in Fig. 17.10, f - a circuit with the implementation of a short circuit on the AF. The noise figure of such mixers can be reduced to 3.5–2.5 dB. The use of mixers with selective circuits is limited due to their narrow bandwidth.

Summarizing the above, we can distinguish the following advantages of the BS over the NBS: 1) due to the phase suppression of the local oscillator noise, the noise figure ksh is reduced by 2 - 5 dB; 2) all the power of the local oscillator signal goes to the diode, so you can use a lower power local oscillator; 3) due to the suppression of even harmonics of the local oscillator in the balanced circuit, the level of side signals is much lower, as a result, noise immunity and dynamic range increase; 4) the electrical strength of the mixer increases, since the power is supplied to 2 diodes; 5) when one diode fails, the circuit remains operational, however, the output signal level drops by ~3dB, and kw increases by ~5–6dB; 6) losses of the received signal due to leakage of energy into the local oscillator circuit are insignificant due to the high decoupling of bridge circuits.

17.6. Double balanced mixers

Double balanced mixers (DBS) make it possible to provide phase suppression at the mirror channel frequency ω3K and restore the energy of AF oscillations in the IF without using an input filter, which reduces losses and provides a wider bandwidth of operating frequencies.

17.7. Ring balanced mixers

The best electrical parameters are provided in ring tanks

lance mixers(KBS), thanks to the use of a diode bridge (DM) of four diodes and broadband differential transformers. KBS

a - diode bridge; b - designation on the diagrams; c - electrical circuit of the COP;

G – CS with matching transformers; e - equivalent circuit of the COP

With matching transformers; e - electric circuit of the BCS

more broadband than DBS, since there are no connecting lines between pairs of diodes. Oscillations of the signal u C (t) and local oscillator u G (t) lead to

orthogonal diagonals of a balanced diode bridge, which has the form of a ring of four diodes made on the same chip with almost the same parameters (Fig. 17.12, a), therefore, the decoupling of the signal and local oscillator circuits reaches 25–30 dB. Due to the symmetry of the circuit, the even harmonics of the local oscillator and the signal are compensated, as a result of which additional suppression of undesirable combination conversion products is carried out and the dynamic range of the mixer increases. Figure 17.12, b shows the symbol DM on the electrical circuits.

Figure 17.12, c shows the electrical circuit of the KBS. The received signal is fed to one of the DM diagonals through a matching balancing transformer TV 1, the local oscillator voltage is supplied to the other diagonal of the black

cut TV 2 . The IF output loaded with resistance R 0 is shunted to the microwave capacitor C 1 and connected to the midpoints 1 and 2 using identical chokes L 1 -L 4, the resistance of which is large at high frequencies and low at the IF. Decoupling capacitors C 2 must pass microwave signals and prevent the FC currents from closing through transformers in case of circuit asymmetry. The local oscillator voltage from the secondary winding TV 2 opens diodes VD 1 and VD 2 in positive half-cycles, and VD 3 and VD 4 in negative half-cycles, connecting in turn output 4 or 3 of the secondary winding of the signal transformer TV 1 to case 2 through open pairs of diodes and chokes

L 1 and L 2.

The difference between the oscillation frequencies of the signal and the local oscillator is equal to the IF, and ω IF<< ω С ≈ ω Г , таким образом, мгновенные фазовые сдвиги между

voltages u C and u G change slowly in comparison with the period of their oscillations. If the voltages u C and u G are in-phase, then in the positive half-cycle u G under the action of voltage u C /2 s L 4 in the FC circuits, current flows from point 1 through the load R 0, point 2, chokes L 1 and L 2 and open diodes VD 1 and VD 2 to point 4, and in the negative half-cycle - from point 1 in the same direction through R 0, point 2 to the inductors L 1, L 2 and further through open diodes VD 3, VD 4 to point 3. The low-frequency component of such a pulsating current is the IF current, the low-frequency components are shunted by the capacitor C 1. The IF current is maximum at common-mode u C and u G, then with increasing phase difference between them it decreases, in the case of orthogonal u C and u G, the IF current is zero, since now the current passing through R 0 and C 1 changes direction every quarter of the period signal. Further, the IF current changes sign, reaches a maximum at the opposite

in-phase u C and u G, etc.

Effective use of CBS in microwave technology is possible only with a high degree of symmetry of differential transformers and diodes. When designing integrated circuits for mixers of the decimeter and lower frequency ranges, the so-called "long line" type transformers (LTL) are used, in which one or more transmission lines are used, made in the form of twisted conductors, or pieces of coaxial cables. Such transformers have a wide operating band in the high frequency ranges compared to conventional multi-turn conductor transformers.

To reduce the unevenness of the frequency response in the high frequency region, the line length is selected from the relation l = Λv /8, de Λv is the wavelength in the transmission line at the upper frequency in a given range. The lower limiting frequency of the TDL, which is determined by the inductance of the primary winding of the transformer, can be significantly reduced by using a core with high magnetic permeability at low frequencies. Difficulties in the implementation of TDL on ferrite cores with twisted conductor transmission lines increase with increasing operating frequencies due to an increase in active losses in the cores and an increase in the influence of the irregularity of transmission lines. Therefore, when constructing

Due to their simplicity, high sensitivity and selectivity, good reliability, direct conversion receivers and transceivers are popular with radio amateurs. But not always in the apparatus, even made according to a well-established scheme, the capabilities and parameters inherent in it are realized from the very beginning.

As a result of many years of operation by the author of the article of this group of communication equipment, it turned out that low-frequency nodes (mainly bass amplifiers) remain operational when the supply voltage drops to 2 ... 6 V (at a nominal value of 9 ... 12 V). At the same time, their gain, as a rule, decreases.

The main reason for the unsatisfactory operation of direct conversion receivers and transceivers is the non-optimal operation of the mixer. High parameters are achieved only with careful selection of the heterodyne high-frequency voltage across the mixer diodes. It should be within 0.6 ... 0.75 V on silicon diodes and 0.15 ... 0.25 - on germanium. At lower local oscillator voltages, the mixer gain decreases. It also decreases at high voltages, since the diodes are open almost all the time. This increases the noise of the mixer.

The stability of the frequency and amplitude of the voltage supplied to the mixer from the local oscillator (especially on the HF amateur bands) largely depends on the stability of the supply voltage.

In almost all the circuits given in the literature, there is no circuit for adjusting the heterodyne voltage on the mixer diodes. It is recommended to select a local oscillator coupling capacitor with a mixer or change the number of turns of the coupling coil. But this process is very time-consuming and, moreover, does not give confidence that the device has been set up properly.

The disadvantage of this method is also that in the process of establishing it is necessary to turn off the receiver (transceiver) and solder the capacitor or rewind the coil. But during this time, the amateur station, the reception volume of which is being tuned, often stops working, and therefore it is impossible to know whether the sensitivity of the device being adjusted is increasing or decreasing. It is more expedient to carry out tuning according to the signals of a "weak" station during a stable passage of radio waves, i.e. when there are no noticeable fluctuations in the level of the received signal.

Due to the lack of necessary measuring instruments, direct conversion receivers and transceivers are often tuned "by ear", which is not the best way to reflect on their parameters.

Fig.1

On fig. 1 shows a diagram of a voltmeter probe, modified in accordance with the recommendations given in. It allows you to quite accurately measure the local oscillator voltage directly on the mixer diodes.

Consider simple ways to tune and refine direct conversion receivers and transceivers, which allow you to eliminate the above design flaws.

Figure 2

First of all, when finalizing, it is necessary to introduce a circuit for stabilizing the supply voltage of the local oscillator. The stabilizer circuit is shown in fig. 2. Zener diode VD1 is selected with a stabilization voltage 1.5 ... 2 times less than the nominal supply voltage of the receiver (transceiver). Resistor R 1 sets the optimal current through the zener diode. The resistance of the resistor R1 must be such that the stabilization current of the zener diode VD1 does not exceed the maximum allowable value. Capacitor C1 reduces the "leakage" of zener diode noise, resulting in reduced noise modulation of the local oscillator voltage, and reduced overall receiver noise.

It is convenient to change the RF voltage on the mixer diodes with a tuning non-inductive resistor connected in parallel or in series with the coupling coil (R1, respectively, in Fig. 3 and 4).

In the latter case, you can use both transformer (Fig. 4, a) connection of the local oscillator with the mixer, and autotransformer (Fig. 4.6). With a more precise adjustment of the local oscillator voltage (for example, when receiving signals from hard-to-hear stations "by ear"), the RF voltmeter is turned off.

It should be noted that if the above improvements are applied, the number of turns of the coupling coils should be slightly increased, since the introduction of a tuning resistor reduces the output voltage of the local oscillator. This is especially true for the variant, the scheme of which is shown in Fig. 3. Together, the number of turns of the coupling coil, the resistance of the resistor R1 and the capacitance of the capacitor C2 must be such that the voltage on the silicon diodes of the mixer can be adjusted from 0 to 1.2 ... 2 V, on germanium - from 0 to 0.5 ... 1 V. In this case, the optimal voltage is achieved approximately at the middle position of the resistor R1 slider.

You can regulate the output voltage of the local oscillator by changing the supply voltage, as, for example, done in [3]. However, this is only suitable at frequencies up to 3...4 MHz. At higher frequencies (above 7 MHz), such an adjustment can lead to a significant shift in the local oscillator frequency.

On fig. 5 shows a diagram of a local oscillator with a buffer node, in which an output voltage regulation circuit is introduced. When repeating, it should be taken into account that the emitter follower does not provide voltage gain, and therefore the high-frequency voltage on the coupling coil must be twice as high. than required for normal operation of the mixer.

In amateur radio practice, diode balanced mixers are most widely used. Their main advantages are the simplicity of design and configuration, the absence of high-frequency switching when switching from reception to transmission. Balanced mixers on field-effect and bipolar transistors are used much less frequently.

In simple balanced diode mixers, the local oscillator voltage and some output conversion by-products can be suppressed by 35 dB or more. But such results are achieved only in one direction: in that in which the mixer is balanced. In the author's design of the transceiver, the mixer is balanced only towards the power amplifier. If a double balanced mixer is used, noise will decrease, sensitivity will increase, and noise immunity will improve.

Dual balanced mixers are balanced on both inputs (outputs). They suppress not only local oscillator oscillations, but also the converted signal, leaving only the products of their mixing and thus ensuring the purity of the spectrum. The use of such mixers makes it possible to reduce the requirements for the cleaning filter included at the mixer output, and even to abandon it altogether by connecting the mixer output directly to the IF amplifier, at the output of which there should be a main selection filter (for example, an EMF or a quartz filter). A significantly higher signal level can be applied to a double mixer during reception, since it sharply weakens the effect of direct signal or interference detection, i.e. there is no detection without the participation of local oscillator oscillations, as is the case in a conventional amplitude detector.

Most often in amateur radio designs, a double balanced mixer is used, the diagram of which is shown in Fig. 6. It is also called ring, since the diodes in it are included but in the ring.

When working on low-frequency ranges, high-frequency transformers are wound, as a rule, on ferrite rings of size K7x4x2 with a magnetic permeability of 600 ... 1000 with three twisted (3-4 twists per 1 cm of length) PELSHO 0.2 wires between themselves. Approximately about 25 turns are made (until the ring is completely filled). When installing a transformer, its windings are phased according to Fig. 6 and 7.

There are two main options for incorporating a dual balanced mixer into a transceiver. In the first one, the signal passes both during reception and transmission in one direction from the input to the output of the mixers. So, for example, it was done in the well-known transceivers "Radio-76" and "Radio-76M2". Numerous experiments carried out by the author revealed that when the heterodyne voltage is less than the optimum, the sensitivity in the receive mode deteriorates significantly, and at a higher voltage, the carrier suppression in the transmit mode decreases significantly (the sensitivity also drops, but this is less noticeable to the ear than in previous case). The qualitative dependence of the main parameters of the transceivers on the voltage level of the local oscillator supplied to the mixer is shown in fig. 8 (curve 1 - sensitivity during reception, determined by ear, 2 - sensitivity measured by devices, 3 - carrier suppression during transmission).

In the second variant, the signal in the receive mode is fed to the input of the balanced mixer, and in the transmission mode - to the output. With this inclusion, the principle of reversibility of the mixer is used. This is how the RF path of the transceiver described in. Setting up the mixer in this case also comes down to setting the optimal heterodyne voltage and carefully balancing it. It should be especially noted that the adjustment operation does not depend on the principle of constructing the RF path of the transceiver.

First of all, you need to set up the mixers. Previously, the engines of the balancing resistors in them are set to the middle position. Next, the GSS is connected to the antenna jack of the transceiver and the heterodyne voltage on the mixers is gradually increased. The signal from the GSS is supplied with a level exceeding the sensitivity of the receiving path by several times. It is necessary to achieve signal reception. If there is no generator, the operation is performed by ear, receiving a signal from an amateur radio SSB radio station or a noise generator on a low-power zener diode.

Then alternately adjust each of the mixers. First, the optimal heterodyne voltage is selected. To do this, it is gradually increased and evaluated by ear: whether the volume of the reception of the GSS signal, the radio station or the noise generator is increasing. As noted by the author, as the heterodyne voltage applied to the mixer increases, the aural reception volume first increases, reaches a maximum, and then practically does not change (Fig. 8, curve 1). The heterodyne voltage should be set in such a way that when it is slightly reduced, the reception volume drops, and when it is slightly increased, it does not increase. In practice, this is realized by moving within a small range of the resistor engine that controls the level of the output voltage of the local oscillator. If there is no such possibility in the transceiver, then the device should be modified.

As a rule, an emitter follower is switched on at the output of one or another local oscillator. In this case, the refinement turns out to be very simple: the constant resistor in the emitter circuit of the transistor is replaced with a non-inductive trimming resistor of the same value as the constant one.

After optimizing the heterodyne voltage, the mixers need to be more carefully balanced again. An RF millivoltmeter or an oscilloscope is connected to the input or output (depending on the construction of the transceiver), and by moving the slider of the resistor R1, and then adjusting the capacitors C1 and C2 (see Fig. 7), a minimum of readings is achieved. If devices with high input impedance are used, then resistors close in resistance (within 50 ... 100 Ohms) should be connected to the input and output of the mixer.

Preference should be given to balancing towards the output of the transmitting path. The difference in balance between the input and output of the mixer should be small (a few decibels). If it reaches 10 dB or more, then this, as a rule, is a consequence of the fact that the heterodyne voltage applied to the mixer is much higher than the optimal one.

To check and balance mixers, the author created simple devices. On fig. 9, a shows a circuit of an RF amplifier, to the input of which a mixer is connected, and a high-frequency voltmeter is connected to the output for coarse tuning (Fig. 9, b), for fine tuning - an RF probe (Fig. 9, c). At the same time, it is not necessary to install additional resistors with a resistance of 50 ... 100 Ohm in the mixer.

Finally, the mixers are configured after they are installed in the transceiver (it is put into transmission mode). The device must first be set to receive mode. To prevent microphone noise from interfering with balancing, the input of the microphone amplifier is short-circuited. The lowest-frequency mixer is balanced first, and then the rest in the order of the signal passing through them in the transmission mode, achieving a minimum of RF readings on the dummy load (Fig. 10) connected to the transceiver power amplifier. After that, adjust the settings of the remaining nodes. It is advisable to repeat this procedure two or three times.

Vladislav Artemenko (UT5UDJ), Kyiv. Ukraine

LITERATURE

1. Polyakov V.T. Radio amateurs about the direct conversion technique. - M.: Patriot, 1990, p. 264.
2. Stepanov B. Measurement of small RF voltages. - Radio, 1980, N 7, p. 55-56.
3. Artemenko V. A simple SSB mini-transceiver for 160 m. - Radio amateur, 1994, N 1.c. 45, 46.
4. Artemenko V.A. A simple transceiver with EMF. - RadioAmator, 1995, N 2, p. 7-10.
5. Bunin S.G., Yaylenko L.P. Handbook of amateur shortwave. - K .: Technique, 1984, p. 264.
6. Stepanov B., Shulgin G. Transceiver "Radio-76". - Radio, 1976, N 6, p. 17-19, No. 7, p. 19-22.
7. Stepanov B., Shulgin G. Transceiver "Radio-76M2". - Radio, 1983, N 11, p. 21-23, N 12, p. 16-18.
8. Vasiliev V. Reversible path in the transceiver. - Radio, N 10, p.20,21.

The described method makes it possible to improve the characteristics of a two-balanced active mixer in terms of intermodulation components by introducing negative feedback, thus reducing the nonlinearity of the active elements. As a result, in terms of its characteristics, a two-balanced active mixer becomes comparable with such previously known 1,2 mixer circuits as an annular diode mixer and a mixer based on powerful key field-effect transistors with an insulated gate ( MOSFET).

Introduction

Mixers and modulators are an important component in the construction of radio frequency communication systems. To implement such necessary functions in communication systems as frequency conversion, modulation and demodulation, many different mixer circuits are used, built using diodes, powerful key field-effect transistors with an insulated gate ( MOSFET), double-gate field-effect transistors, as well as the very popular so-called “transistor tree” or “Gilbert Cell” developed at the time by Barrie Gilbert. But in all these circuits, the non-linearity of the semiconductor devices used, directly or indirectly, causes distortion when two or more different signals interact in the mixer - a phenomenon known to professionals as the occurrence of intermodulation distortion (IMD - intermodulation distortion).

The sources of intermodulation distortion are the subject of a separate discussion, which has received much attention in the specialized literature, and the continuation of which is not the subject of this article. More precisely, the reader will be invited to a brief discussion of two of the most well-known mixer designs, such as the ring diode mixer and the transistor tree, to identify their main characteristics and then compare with the new negative feedback mixer circuit mentioned earlier, in which the undistorted useful signal can be achieved by applying simple negative feedback circuitry, known from the transistor amplifier circuit with parallel negative voltage feedback, which significantly improves the performance of the mixer in terms of 3rd order intermodulation products (IIP 3) and compression point (P 1dB).

Ring Diode Mixer

Ring diode mixers began to be used with the widespread use of semiconductor diodes in the late 1940s, and the non-linearity of their characteristics immediately became apparent 3,4 . This phenomenon still continues to be the object of close study in the specialized literature 5,6,7.

The construction of a class I ring diode mixer is illustrated by a diagram on fig.1. Here, four diodes are connected in a ring and alternately switch to the state "ON" And "OFF" supplied from the local oscillator (LO) signal.

Fig.1. A typical class I ring diode mixer.

The local oscillator signal power required for normal operation of such a mixer is usually +7 dBm, for circuits of ring diode mixers of the following classes, the required power of the local oscillator signal reaches +17 dBm and more, which is due to the desire for higher quality indicators for intermodulation components.

For the purpose of subsequent comparative analysis, let us consider the qualitative characteristics of the intermodulation components and the compression point of a common class I type ring diode mixer SBL-1 produced by the company Mini Circuits. This mixer is very popular among amateur radio developers, and its commercial counterpart SBA-1 distributed even more widely, and therefore was chosen for this study.

According to the test conditions, the level of the local oscillator signal with a frequency 10 MHz made the required +7 dBm, while the other input of the mixer received two signals with frequencies 500 kHz And 510 kHz. These frequencies were chosen based on the operating frequency range of the mixer SBL-1 and will also be used for subsequent comparative testing of other mixer circuits.

Mixer quality parameters SBL-1 illustrates fig.2, and their numerical values ​​are summarized in table 1.

Fig.2. Intermodulation distortion of ring diode mixer SBL-1, 10 dBm/div.

These are objectively typical characteristics of a class I ring diode mixer, but as will be shown below, higher levels of IIP 3 - and P 1dB -parameters can be achieved with significantly lower local oscillator signal power in an active mixer built on the basis of two negative feedback amplifiers. .

Table 1.

Mixer on powerful key field-effect transistors with insulated gate (MOSFET)

Fig.3.

High-quality ring mixers use key insulated-gate field-effect transistors instead of diodes ( MOSFET). A typical diagram of such a mixer is shown in fig.3.

For mixers of this type, the intercept point for intermodulation products of the 3rd order (input intercept points - IIP 3) above +40 dBm, but at the cost of a very high local oscillator power level, usually +17 dBm and higher, which in practice often prevents their use in portable radio equipment. However, it is superior in performance to a class III ring diode mixer.

In professional and amateur radio literature 8,9,10,11,12,13,14 the topic of building ring mixers on powerful key field-effect transistors is very widely discussed, and it is rather difficult to pay enough attention to this topic without digressing from the actual purpose of this article.

Mixer according to the "transistor tree" scheme

On fig.4 the functional diagram of the "transistor tree" type mixer is given. Originally patented in 1966 by Howard Jones as a synchronous detector 15 , this very popular active mixer is better known as the "Gilbert Cell" in accordance with a later patent and using this circuit as the basis for building analog multipliers 16 . This mixer is, in its design, a derivative of the family of vacuum tube synchronous demodulators 17 .

Fig.4. A transistor tree mixer, also known as a Gilbert Cell.

Here the intermediate frequency (IF) input signal through the transformer T2 out of phase controls a differential current source on transistors VT2 And VT5. To stabilize the mixer conversion ratio over a wide range of input signal levels, as well as to reduce the effect of transistor non-linearity VT2 And VT5 series negative current feedback resistors are connected to the emitters and between them R4..R6.

The output currents of the differential current source, that is, the collector currents of the transistors VT2 And VT5, switched out of phase by transistors of differential pairs VT1:VT3 And VT4:VT6, alternately switched to the "ON" state. and "OFF" signal supplied from the local oscillator LO through a transformer T1. The collectors of transistor pairs are mutually cross-connected, therefore, due to the summation of the currents on the load resistors R3 And R7, the local oscillator and intermediate frequency signals are suppressed, and their mixing products, including the useful radio signal RF, are separated on the primary winding of the transformer T3.

In order to check the characteristics shown in fig.4 the mixer was assembled on the manufactured by the company Harris microchip CA3054(now manufactured by Intersil- approx. translator) containing two identical differential amplifiers. With a supply voltage equal to +12 V and resistor resistance R4..R6 equal to 100 ohm(a resistor assembly of three resistors was used) voltage at the bases of transistors VT2 And VT5 was set to +2.1 V, while the collector bias current of these transistors was 15 mA. Voltage at the bases of transistors VT1, VT3, VT4 And VT6 was set to +4.7V. Thus, the operating point of transistors VT2 And VT5 remained in the linear section of their characteristics over the entire range of input signal levels 18 . All transformers T1, T2 And T3 Fair Rite 2843-002-402(binocular transfluctor). With a ratio of windings 1:1:1 the input and output impedances of the mixer are 50 ohm.

The test conditions for the mixer were the same as for the ring diode mixer, except for the local oscillator signal level, which was 0 dBm (1 mW). This level was set for all active mixers considered in this article, which work quite satisfactorily even at such low levels of the local oscillator signal as -6 dBm (0.25 mW).

Fig.5 And table 2 illustrate the qualitative characteristics of the mixer according to the "transistor tree" scheme. compression point P 1dB characteristics of such a mixer is higher than that of a ring diode mixer, and the intersection point for intermodulation components of the 3rd order ( IIP 3) - below. However, despite the fact that the local oscillator signal level required for the operation of a “transistor tree” type mixer is significantly lower than for a ring diode mixer, its qualitative characteristics in terms of intermodulation distortion are slightly inferior to a ring diode mixer.

Fig.5. Intermodulation distortion of the mixer according to the "transistor tree" scheme, 10 dBm/div.

Table 2.

For a long time, it was believed that the main obstacle to obtaining higher characteristics in terms of the level of introduced intermodulation distortion in the mixer according to the "transistor tree" scheme are control transistors. VT2 And VT5 operating as voltage-controlled current sources. 19,20 A number of methods described in the literature have been successfully used to correct this deficiency. 19,21,22 But all these methods ignore other sources of intermodulation distortion such as current gain non-linearity hfe control transistors, as well as the nonlinearity of the characteristics of four transistors switching their current VT1:VT3 And VT4:VT6. These shortcomings can be overcome by applying a combined series-parallel negative feedback circuit ( series/shunt feedback), covering all transistor nodes of the mixer, by analogy with transistor amplifier stages.

Amplifier with combined series-parallel negative feedback ( series/shunt feedback)

On fig.6 a diagram of a transistor amplifier with a combined series-parallel negative feedback (OOS) is shown.

Fig.6.

Sequential OOS ( series feedback) is formed by a resistor R2 included in the emitter circuit of the transistor VT1. Parallel OOS ( shunt feedback) is formed by a resistor R1 connected between the collector and the base of the transistor VT1.

The input and output impedance of such an amplifier is determined by the ratio 23.24:

and the power gain:

Such a negative feedback topology makes it possible to increase the linearity of the transistor amplifier by simple means and, moreover, is easily implemented in a "transistor tree" type mixer circuit.

(option 1)

A diagram of a linearized active mixer according to the "transistor tree" scheme, covered by a deep FOS, is shown in fig.7. The first linearized "amplifier" with a combined series-parallel FOS is formed by including separate parallel FOS resistors ( shunt feedback) R2:R3 between collectors of transistors of a key transistor pair VT1:VT3 and the base of the control transistor VT2 through a decoupling capacitor C1. Sequential OOS ( series feedback) is formed by a chain of three resistors R5:R9:R13. As a result, the "amplified" intermediate frequency signal IF, which is suppressed in the basic "transistor tree" circuit, is here distinguished as common mode at the load resistors and through the parallel feedback circuit. R2:R3:C1 fed into the base of the control transistor VT2. At the same time, the signals of the local oscillator LO and the resulting radio frequency RF based on the transistor VT2 are suppressed. Thus, the circuit works as an amplifier only for the intermediate frequency signal IF, and since the combined series-parallel OOS circuit covers all three transistors, the distortions introduced by them, due to their non-linearity, are compensated.

Fig.7.

Similarly, the second transistor pair VT4:VT6 with second control transistor VT5 and the corresponding parallel and series OOS circuits form a second linearized "amplifier". Note that three resistors R5:R9:R13 play the same role as the resistor R2 in the diagram for fig.6 and expressions and .

Output transformer T3 connected to collectors of transistors of transistor pairs VT1:VT3 And VT4:VT6 through four 100 ohm resistors R7:R8:R10:R11 in such a way that signals with a local oscillator frequency LO and an intermediate frequency IF on its primary winding are suppressed and only their mixing products are present at the mixer output.

To test the active mixer linearized in this way, a circuit was assembled from the same elements as the previous mixer circuit, with the same DC modes. With the resistance of parallel OOS resistors R2, R3, R15 And R16 equal to 330 ohm the input and output impedance of both "amplifiers" was approximately 100 ohm, and the amplification of each "amplifier" of the intermediate frequency signal IF was about +6.7 dB.

Fig.8. Intermodulation distortion linearized active mixer (option 1), 10 dBm/div.

Table 3.

Given on fig.8 and in table 3 test results show that, in comparison with the previously considered "transistor tree" type mixer, the circuit of which is shown in fig.4, collected according to the fig.7 circuit, a linearized active mixer with combined feedback has higher characteristics in terms of the level of introduced intermodulation distortion and outperforms a ring diode mixer SBL-1 firms Mini Circuits at a significantly lower level of the local oscillator signal LO. The compression point suffers somewhat P 1dB, - this is caused by incomplete suppression of the local oscillator signal LO on the collectors of transistors VT1:VT3 And VT4:VT6, which leads to their saturation too early. This is due to four 100 -ohm resistors R7:R8:R10:R11 in the crosshairs between the collectors of these transistors, while in the mixer "transistor tree" on fig.4 the corresponding collectors of the transistors are directly connected to each other and the local oscillator signal is almost completely suppressed on them. In addition, this circuit of resistors introduces excessive attenuation of the output signal - about 6 dBm. This shortcoming was avoided by combining the output signals of the mixer not with resistors, but with the help of a so-called "hybrid" transformer.

Combining Signals with a "Hybrid" Transformer

Hybrid transformers 25,26,27 (also known as bridge transformers or symmetrical transformers) were previously widely used in telephone amplifiers, but with the use of appropriate ferromagnetic materials, they easily found their way into high-frequency circuits.

In the diagram for fig.9 A hybrid transformer is used to extract the difference signal from two common-mode signals. Having a common-mode component, the signals are fed to the opposite terminals of the primary winding of the transformer, which has a tap from the middle and is isolated from the secondary. With this inclusion, the common-mode component appears at the midpoint of the primary winding of the transformer, and the difference signal is isolated on its secondary winding. This happens because the current in the primary winding flows only at different potentials at the opposite terminals of the winding.

Fig.9 Isolation of the difference signal using a "hybrid" transformer.

Let the primary and secondary windings of such a transformer have 2N And M turns, respectively. Then, in order to match the load, the resistance values ​​in the circuit on fig.9 should be related by the following relationships:

Use to combine output signals in a mixer circuit on fig.7 chain of four resistors R7:R8:R10:R11 led to a decrease in the mixer transfer coefficient by 6 dBm. The use of a hybrid transformer for the same purpose negates these losses, therefore, when talking about such a circuit topology, the term “lossless” is often used (i.e., “lossless” or “without attenuation”).

Linearized active mixer without useful signal loss (option 2)

On fig.10 a diagram of a linearized active two-balanced mixer is shown, in which lossless-topology using hybrid high-frequency transformers. The circuit contains two identical balanced active mixers, so it is enough to consider the operation of one of them.

Fig.10.

To begin with, imagine that the mixer as a whole is loaded by the RF output on the load resistance R L(not shown in the diagram). Then the reduced value of the load resistance for each of its constituent balanced mixers will be equal to 2R L. At the same time, if the windings of hybrid transformers T3 And T4 made with the ratio of the number of turns 1:1:1 , then the resistance at the midpoint of their primary winding will also be 2R L, and the resistance at the ends of this winding will be equal to 4R L.

Periodic anti-phase switching of transistors VT1 And VT3 the local oscillator signal LO modulates the collector current of the transistor VT2, thereby creating a differential signal in the primary winding of the transformer T3. Load resistance in the collector circuit of the transistor VT2- constant value, equivalent to parallel-connected resistances in the collector circuits of transistors VT1 And VT3 and equal to the resistance at the midpoint of the hybrid transformer, i.e. 2R L. Thus, in this circuit, it is possible to implement an "amplifier" with a combined series-parallel OOS ( series/shunt feedback).

Let us assume that the secondary windings of both output hybrid transformers are disconnected from each other and each is loaded with its own load resistance. In this case, the voltages on the collectors of four transistors VT1, VT3, VT4 And VT6 are defined respectively by the expressions , , and :

AIF is the amplitude of the intermediate frequency signal;
G- defined by the expression gain "amplifier";
- the value of the local oscillator frequency;
- the value of the intermediate frequency;
I bias- collector bias current of the transistor VT2.

The rightmost term in the equalities is the differential carrier signal of the local oscillator in the primary winding of the transformer T3. It is equivalent to the signal in the primary winding of the transformer T4, but opposite in phase (equalities and ). The balance of these two signals, with an appropriate connection of the secondary windings of these two transformers (see. fig.10), provides effective suppression of the local oscillator signal and separation of mixing products, including the useful radio signal RF, at the output of the mixer. In the ideal case (that is, in the absence of losses), the expressions describing the voltages on the collectors of the same four transistors take the following form:

Reconstructed intermediate frequency signals at the midpoints of the primary winding of output hybrid transformers T3 And T4 are described by expressions:

and the signal at the output of the mixer is described by the expression:

which, under the condition of equality M=N, takes the form:

The circuit for testing was assembled, again, from the same elements as the previous mixer circuit, with the same DC modes. Two hybrid transformers T3 And T4 had the same design as the input transformers T1 And T2, and with the winding ratio 1:1:1 contained four turns of a trifilar winding on a core of the type Fair Rite 2843-002-402. Therefore, the input and output resistance of each of the balanced mixers was 100 ohm. Accordingly, taking into account the parallel connection of the secondary windings of transformers T3 And T4, the input and output impedance of the mixer is 50 ohm.

The circuit was tested for fig.10 at the same frequencies and local oscillator signal level as the previous one. Fig.11 And table 4 illustrate the quality indicators of the mixer. As a result of the fact that the level of third-order intermodulation products was -53 dBc, intersection point IIP 3 reaches a quite satisfactory level +29.5 dBm. Also the compression point P 1dB rose to +10.5 dBm. Thus, the use of a hybrid transformer in the circuit made it possible to design an active mixer that competes in its low level of intermodulation distortion with a class III ring diode mixer, but at the same time requires much less local oscillator signal power.

Fig.11. Intermodulation distortion linearized active mixer (option 2), 10 dBm/div.

Table 4.

Reactive Load Sensitivity

In view of the foregoing, a lumped selection band pass filter with a central frequency was assembled 10.7 MHz and bandwidth 500 kHz, the diagram of which is shown in fig.12. The measured intrinsic attenuation of the filter was 5.5 dB and was taken into account in the results of subsequent measurements.

Fig.12.

Of those given in table 5 measurement results show that the ring diode mixer SBL-1 indeed very sensitive to the connection at its output instead of a purely active terminating load of a narrow-band IF filter: third-order intercept point IIP 3 while falling to 11.5 dB, and the compression point P 1db on 3 dB. Active mixers, without exception, showed substantially less sensitivity to frequency dependent loading, compression point P 1db at the same time, it remained in the same place, and the intersection point for the third-order intermodulation products IIP 3 fell no more than 1 dB in all three cases.

Table 5.

There is nothing surprising in the results obtained. In the case of a ring diode mixer, the signal energy from the unloaded output is reflected back into the diode circuit, where it can then interact with the non-linearity of the diode junctions. Conversely, the signal energy reflected back into the active mixer is quenched in the load resistances of the switching transistors, and the non-linear base-emitter junctions are isolated due to the small reverse current transfer coefficients of the transistors.

Conclusion

So, an active mixer with a combined series-parallel OOS circuit showed such qualitative characteristics that are also desirable in the development of high-quality radio-frequency transceiver systems. Further improvements, including the use of alternative negative feedback topologies to improve mixer noise performance, will result in a very wide dynamic range mixer that does not require excessive power levels from the local oscillator.

©Christopher Trask, 1998.

Translation © Zadorozhny Sergey Mikhailovich, 2006

Literature:

1. Trask, Chris, "Feedback Technique Improves Active Mixer Performance"; RF Design, September 1997.
2. Patent pending.
3. Belevitch, V., "Non-Linear Effects in Ring Modulators"; Wireless Engineer, Vol.26, May 1949, p.177.
4. Tucker, D. G., "Intermodulation Distortion in Rectifier Modulators"; Wireless Engineer, June 1954, pp.145-152.
5. Gardiner, J.G., "An Intermodulation Phenomenon in the Ring Modulator"; The Radio and Electronics Engineer, Vol.39, No.4, April 1970, pp.193-197.
6. Walker, H.P., "Sources of Intermodulation in Diode-Ring Mixers"; The Radio and Electronics Engineer, Vol.46, No.5, May 1976, pp.247-253.
7. Maas, Stephen A., "Two-Tone Intermodulation in Diode Mixers"; IEEE Transactions on Microwave Theory and Techniques, Vol.MTT-35, No.3, March 1987, pp.307-314.
8. Evans, Arthur D.(ed), "Designing with Field-Effect Transistors"; McGraw-Hill/Siliconix, 1981.
9. Rohde, Ulrich L., "Recent Developments in Circuits and Techniques for High-Frequency Communications Receivers"; Ham Radio, April 1980, pp. 20-25.
10. Rohde, Ulrich L., "Key Components of Modern Receiver Design"; QST, May 1994, pp.29-31 (pt.1), June 1994, pp.27-31 (pt.2), July 1994, pp.42-45 (pt.3).
11. Rohde, Ulrich L., "Recent Advances in Shortwave Receiver Design"; QST, November 1992, pp. 45-55.
12. Rohde, Ulrich L., "Performance Capability of Active Mixers"; Ham Radio, March 1982, pp.30-35 (pt.1), April 1982, pp.38-44 (pt.2).
13. Rohde, Ulrich L., "Performance Capability of Active Mixers"; Proceeding WESCON 81, pp. 24/1-17.
14. Rohde, Ulrich L. and T.T.N. Bucher, "Communications Receivers: Principles and Design, 1st ed."; McGraw-Hill, 1988.
15. Jones, Howard E., "Dual Output Synchronous Detector Utilizing Transistorized Differential Amplifiers"; U.S. Patent 3.241.078, March 15, 1966.
16. Gilbert, Barrie, "Four-Quadrant Multiplier Circuit"; U.S. Patent 3.689.752, 5 September 1972.
17. Schuster, N.A., "A Phase-Sensitive Detector Circuit Having High Balance Stability"; The Review of Scientific Instruments, Vol.22, No.4, April 1951, pp.254-255.
18. Sullivan, Patrick J. and Walter H. Ku, "Active Doubly Balanced Mixers for CMOS RFICs"; Microwave Journal, October 1997, pp. 22-38.
19. Chadwick, Peter, "The SL6440 High Performance Integrated Circuit Mixer"; WESCON 1981 Conference Record, Session 24, pp.2/1-9.
20. Chadwick, Peter, "More on Gilbert Cell Mixers"; Radio Communications, June 1998, p.59.
21. Heck, Joseph P., "Balanced Mixer With Improved Linearity"; U.S. Patent 5.548.840, 20 August 1996.
22. Gilbert, Barrie, "The MICROMIXER: A Highly Linear Variant of the Gilbert Mixer Using a Bisymmetric Class-AB Input Stage"; IEEE Journal of Solid-State Circuits, Vol.32, No.9, September 1997, pp.1412-1423.
23. Meyer, Robert G., Ralph Eschenbach, and Robert Chin, "Wide-Band Ultralinear Amplifier from 3 to 300 MHz"; IEEE Journal of Solid-State Circuits, Vol. SC-9, no. 4, Aug 1974, pp. 167-175.
24. Ulrich, Eric, "Use Negative Feedback to Slash Wideband VSWR", Microwaves, October 1978, pp. 66-70.
25. Gross, Tom, "Hybrid Transformers Prove Versatile in High-Frequency Applications", Electronics, March 3, 1977, pp. 113-115.
26. Sartori, Eugene F., "Hybrid Transformers", IEEE Transactions on Parts, Materials, and Packaging (PMP), Vol. PMP-4, no. 3, September 1968, pp.59-66.
27. Bode, Hendrik W., Coupling Networks, U.S. Patent 2,337,965, December 28, 1943.
28. Yousif, A.M. and J.G. Gardiner, "Distortion Effects in a Switching-Diode Modulator with Tuned Terminations", Proceedings of the IEE, Vol. 119, no. 2, February 1972, pp. 143-148.

original text:

Trask, Chris, “A Linearized Active Mixer”, Proceedings RF Design 98, San Jose, California, October 1998, pp. 13-23.

In the case of practical implementation of unbalanced diode mixer circuits, it is necessary to decouple the local oscillator signal paths and the input RF signal, usually performed using RF transformers, directional couplers or diplexers (Fig. 5).

Rice. 5. Scheme of an unbalanced diode mixer

Most diode mixers use unbiased diodes, however, when applying a forward bias voltage to the diode to produce a small current ID , can be reduced mixer conversion loss. This is especially desirable when using a low signal LO. The diode is biased to establish a static operating point close to the region of maximum non-linearity in the operating characteristic, to be in the square of the diode characteristic when the LO signal is low.

Advantages of unbalanced mixers:

• can operate over a very wide frequency range
• circuit simplicity

Disadvantages of unbalanced mixers:

• they do not provide acceptable isolation between ports
• the power of the useful output signal depends on the levels of both the input and the reference heterodyne signals

Rice. 6. Balanced mixer with hybrid transformer

In a balanced diode mixer ( Single-balanced Diode Mixer, SBM ) two diodes are used. The signals from the local oscillator and the RF source are added in antiphase, thus reducing the level of unwanted signal components at the output of the IF mixer and their suppression. The level of suppression depends on the amplitude and phase symmetry of the transformer, which ensures the symmetry of the signals and the matching between the two diodes. In high-quality mixers made on discrete elements, suppression by 20-30 dB is possible. One of the other benefits of balanced mixers is the rejection of even spurious components and the rejection of LO amplitude (AM) noise. In early microwave AM receivers, noise was a serious problem, as the local oscillator signals were very noisy. However, in modern RF units of SSPO devices, frequency synthesizers are used as local oscillators, and the phase noise of the reference signals is a more serious problem than AM noise.

Rice. 7. Double balanced mixer

IN double balanced diode mixers (Double-balanced Diode Mixers, DBM ), often called ring, usually 4 diodes are used, connected in a ring or a star with balanced inputs of the local oscillator and the RF signal. All mixer outputs are actually isolated from each other. When making diode rings inside the IC, it is possible to achieve very good matching and symmetry, since the diodes are made of the same material, on the same substrate, and have the same parameters. Such structures are balanced for both heterodyne and RF inputs.

Advantages of dual balanced diode mixers:

• increased linearity, greater dynamic range of the device;
• RF and local oscillator signals are suppressed at the output;
• at the output of the mixer, the combination products of the local oscillator and RF signals of even orders are suppressed;
• good mutual isolation of mixer ports.

Disadvantages of dual balanced diode mixers:

• the use of two balancing RF transformers, which are technologically complex elements, and because of this, it is difficult to implement such mixer structures in integrated structures;
• the real range of operating frequencies is limited by the technological symmetry of RF transformers achieved;
• the need to use a powerful local oscillator signal;
• it is necessary to use semiconductor components with identical characteristics.

Double double balanced mixer ( Double Doubly Balanced Mixer, DDBM ) or inline mixer ( Triple Balanced Mixer, TBM ) is a set of two ring balanced mixers. On fig. 9 shows a schematic diagram of such a mixer. The main advantage of the circuit is increased linearity, because The use of two diode rings and additional RF transformers allows the device to expand the dynamic range by approximately 3 dB and increase, by at least 6 dB, the isolation between the input ports of the local oscillator and the RF signal.

Rice. 9. Schematic diagram of a double balanced mixer

The main disadvantage of this mixer is the increased complexity, because 3 balancing transformers and 8 diodes are used. In addition, it is necessary to increase the power of the local oscillator signal by 3 dB, compared with a ring balanced diode mixer. An alternative implementation of highly linear mixers is FET implementation, described below. This can provide even greater linearity than in diode mixers, using a simpler device circuit.

In practice, such a complex transformer system is not used, since a more practical solution is to combine ring balanced mixers using hybrid combiners and splitters.