Perhaps the most crucial node in the transceiver is GPA, which determines the frequency stability and noise performance. This article is an attempt in a popular form to state what is well described in the textbook [1].
The entire mathematical apparatus is omitted, so as not to frighten the unprepared readers formulas and vector diagrams.
Frequency instability of oscillators has many causes. We could divide all the causes of instability in two directions:
- the causes that affect the tuning frequency of the driving circuit;
- the causes that affect the frequency due to the change mode of the active element.
The simplest cause first direction is the mechanical fragility of the structure. The next obvious reason the same direction - temperature instability. The heating of the parts of the oscillator causes changes of inductance and capacitance. For example, heating coils, copper wire wound on a ceramic frame, causes dilation of copper, increasing the length of the wire, the increase in the winding diameter. This entails increasing the inductance and decreasing frequency. The same heating coil, wound on a Teflon frame, causes an increase in diameter of the turns, but because of the excessively large linear expansion of PTFE coil is stretched in length so much that more than offset the increase in diameter, and as a result, the inductance does not increase, but decreases, and the frequency increased. For this reason, PTFE is completely unsuitable for high stability circuits.
The magnetic permeability of most ferromagnetic materials when heated increases. Increase during heating and the capacitance of the varactors. The capacitance of the capacitors during heating may increase or decrease depending on the materials of the plates and dielectric. Sometimes (unfortunately not always) on the capacitors to write the value of temperature coefficient of capacitance (TKE), which shows how much ppm changes the capacitance of the capacitor while it is being heated at 1°C. the sign of the change (plus or minus) is denoted by the letters "M" or "P". Designation M means that during heating for every degree the capacity is reduced by g-6. Designation P33 means an increase in the heat for each degree on h-6. If a capacitor with TKE M had at rated temperature capacitance 1500 pF, when additional heating 20°C capacity becomes equal to 1500-1500x750xl0-6x20 =1500-22,5=1477,5 pF. If the oscillator was working for example at a frequency of 500 kgci the frequency is determined by this capacitor, the frequency deviation will be 3,79 kHz, which is clearly a lot.
Radical method in this case is temperature control. But simpler and cheaper - choice parts with the smallest temperature deviations. The so-called temperature compensation can reduce to certain limits of temperature instability, but does not resolve it completely. There are two reasons. First, the GPA is a circuit tunable, and the percentage of fixed and variable capacitors in the restructuring changes. Therefore, the compensation achieved at one frequency to another frequency is disturbed. Secondly, changes in capacitance and inductance occur during heating according to different laws. Therefore, the compensation achieved during heating at 10°C will be violated if we heat up the generator by another 10°C.
As details for the programme can be recommended coil wound heated during winding silver plated wire on ribbed ceramic frame. Capacitors can be used KM5 (five-layer, small-sized) with TKE M47 or M75. If you have set the GPA applied to the varactors, TKE capacitors should be even more, since TKE varactors positive and depending on the offset (i.e., frequency settings) vary from 70...h'6 at high voltages up to h"6 for small. Therefore, it is unacceptable to use the varactors at a bias voltage less than 8...9 V. If the capacitance of the varactors is insufficient for a given circuit, you should either use varactors with large capacities (e.g. LW 105), or put two or three varactors in parallel. The author does not recommend the use of coils with wojenny silver. Yes, they have good temperature stability, but... low quality factor. But the more important factor.
Another reason affecting the frequency of the circuit, is the instability of the parasitic capacitance of the active elements, which are connected to the circuit and serve as integral parts of its capacity. During operation of these parasitic capacitance change, and directly lead frequency of the circuit. Previously addressed temperature drift of frequency occur slowly, they can be podgorecki to adjust on a numerical scale or compensate. The impact of instability parasitic capacitance occurs quickly, often in time with the modulation, and is accompanied by a characteristic distortion of the signal. Parasitic interelectrode capacitance in the transistors are conventional barrier capacitance of p-n junctions, reconstructing, when changing the applied voltage to them. The influence of parasitic capacitance can to some extent reduce, but not eliminate it.
To reduce their impact need to ensure that the percentage of parasitic capacities in the total capacity of the circuit was less so in the presence of large total capacitance of the circuit a few picofarads of parasitic capacities had less impact. Here, however, there are two limitations. First, too large capacity with low inductance leads to the decrease of the quality factor of the circuit. Secondly, too much constant capacity requires a proportional increase in variable capacity, otherwise you will not be provided within the adjustment circuit. In any case, you cannot do GPA on almost only the parasitic capacitances, as is done in [2], where in the circuit...7 1.8 MHz applied VA-recap SWС111 with small capacity. And to get the restructuring, the author applies a large inductance and a small permanent capacity. In this case the parasitic input capacitance of the transistor 20%(!!) from the total capacitance of the circuit. Parasitic capacitance had little effect on the frequency, if the voltage and mode of operation of the generator were perfectly stable, which is really unattainable.
One of the methods that solve to some extent the problem is the decoupling of the cascades between circuit GPA and active element. In Fig.1 shows a simple scheme of inductive treatacne, and in Fig.2 - treatacne with the addition of an isolation source follower.
Fig.1
The voltage difference "between the gate and the source is 10 times less than the input voltage. But if the voltage difference is small, the input capacitance of the repeater runs an alternating current is 10 times less, which is equivalent to 10 times reduction in input capacitance.
But that's not all. A repeater (Fig.2) has deep EP DC. When the supply voltage varies, the current in the transistor is changed many times less than it could be modified without source resistor, i.e., parasitic capacitance is more stable.
In the first case (Fig.1) generating transistor takes the current to create automatic offset from the circuit and degrades its quality. In the second case (Fig.2) this current is taken from the repeater and on the quality factor has no effect. Due to the large gain of the power generating source of the transistor is connected to a lower portion of the turns of the loop (1/10 1/20...) and less effect on the circuit. The best results are obtained if the repeater field-effect transistor with left-hand characteristics, without submitting to the gate bias. We can recommend KPI. The circuit parameters should be selected so that the repeater transmitted amplitude of vibration or noise, or even limit the top and bottom. There is another mechanism of destabilization of frequency, not so obvious. The oscillator runs continuously due to the fact that his high-q circuit "rings" and maintains oscillations. The energy in the circuit is replenished jerks only when the peaks of the positive half waves of the gate. For stable operation of the generator is necessary to maintain the amplitude balance and balance phases. The first requires that for each period of the oscillations in the circuit replenished energy as much as her out of the loop is spent (on gate currents, the losses in the capacitors and resistors radiation into the surrounding space). This balance is maintained by the automatic offset. As soon as the oscillation amplitude is slightly reduced, reduced and offset, the transistor is opened a little more, and the portions of the supply of energy increase. And Vice versa.
The second requires that the supply current pulses come in the contour strictly in tact with existing vibrations - not earlier or later. The balance of the phases is also supported automatically, but to understand this process more difficult. For simplicity, we describe it in the case of the oscillator on the vacuum triode. When opening the lamp bundle of electrons starts moving from the cathode to the anode. The current in the anode circuit at this time there. The current pulse will follow in the anode circuit only after a bunch of electrons will reach the anode. For this, in General, negligible time, the phase of the oscillations in the circuit will change, and pushing a pulse of current will lag the voltage pulse on the grid. This gap is expressed as a phase angle of a few degrees. This so-called angle of the slope (not to be confused with the steepness of the current-voltage characteristics!). The angle of the slope, shows the magnitude of the delay of the signal depends on the distance between the electrodes and the velocity of the electrons, which, in turn, depends on the magnitude of the anode voltage.
Thus, the pulses enter the loop with a delay. The generator adapts to this? It turns out that it generates is not exactly on the frequency of the circuit, and just below this frequency.
If using an oscillating circuit, an alternating current, the voltage across the circuit is exactly in phase with the current in one case: when the current is exactly in resonance with the frequency of the circuit. In all other cases the voltage at the circuit ahead or current, or behind him. Now, the oscillator automatically chooses the frequency at which the voltage at the circuit ahead of the supply current pulses precisely at such a value, which then holds the lamp. It is known that high-q circuit very sharply reacts to the deviation of frequency. A very small frequency deviation causes a large phase deviation. Accordingly, to compensate for phase lag in the lamp, the generator need only slightly away from the resonant frequency of the circuit. If the anode voltage is changed, is changed and the delay in the lamp. The generator switches to another frequency, which again would ensure the balance of the phases. The frequency shift will be negligible if the factor of the circuit is high. When discoloration generator circuit to compensate for such a delay is necessary to modify the frequency is much stronger.
Delays are not just lamps, but also in the transistors and chips. Only there their physics are not so obvious. Thus, changing the mode of operation of the lamp or of the transistor, we can change the oscillation frequency, it is even used for frequency modulation. But what if not only can, but also do not want - and frequency "floats"! First, it should be possible to stabilize the power, and secondly, to use the oscillating circuit is the maximum possible q-factor, which to wind the coil sufficiently thick silver plated wire on the ribbed frame of radioparty or polystyrene. If the frame has not forced notches, it is necessary to wind the heating wire from the step-down transformer. After cooling, the wire shrinks and tightly hugs the frame, fixing the coils.
Coating the coil with this purpose, varnishes, paints, etc. is absolutely unacceptable. If the oscillator operates at frequencies above 10 MHz, the circuit elements should not be soldering into the PCB. Used in the circuit capacitors and varactors should be soldered directly to the ends of the coil, without additional wires. If the oscillation frequency is high and the parasitic capacitance of the transistor will inevitably form a significant part of the capacity of the circuit, and the transistor need to solder on the coil mounted Assembly. Thirdly, it is necessary to use GPA for transistors with minimum parasitic capacitances. Often to prevent self-excitation of the oscillator at УSW used anti-parasitic resistors in the circuit of the gate or base. Along with damping spurious oscillations, they reduce the quality factor of the primary circuit. Therefore, the resistors, even if the scheme provides, first set is not necessary. If parasitic oscillations do arise, it is necessary to look for other ways to eliminate them, but if it will not give effect, only then to put anti-parasitic resistor minimum values starting with a few Ohm. Parasitic excitation on УSW not only creates additional reception channels and spurious emissions, but also disturbs the stability of the core generation. Parasitic circuit can have a low quality factor, parasitic oscillations are unstable amplitude. The mode of the oscillator is continuously changing, causing changes in the fundamental frequency and resulting in bewilderment of their creators.
Frequency instability can be caused by so-called "tightening". If the oscillator is poorly shielded, when transferred to the circuit is subjected to high interference, which combine with the fundamental vibrations, result in a complete mess the phase at the input of the transistor. Accordingly starts to "walk" the generation frequency. Prevention - screening. power supply decoupling and compliance with the chart of the levels at which the amplitude of oscillations is many times higher than the amplitude of the noise.
It may be objected that much of what was said here na so important. Are the same transceivers whose GPA is made contrary to many thoughts expressed here. Yes, they are. But how? Take a particular GPA, change the voltage by 10% and see the frequency shift by the frequency counter. Of course, in actual use it does not change by 10%, and much smaller, but it's more convenient for greater clarity. Then you can see all their mistakes - what frequency instability gives the coil coating lacquer, gives much unsoldering capacitors and varactors on the printed circuit Board, etc.
Generator with high electronic frequency stability respectively, and has a small phase noise. This does not apply, however, to the case when the stability is achieved by using a digital scale and CAPC, not a good scheme of the GPA.
Literature
Author: G. Gonchar (EW3LB), Baranovichi; Publication: N. Bolshakov, rf.atnn.ru