Fullscreen HTML5Med Res (??MB)HTML4

Circuit Surgery Regular clinic by Ian Bell Royer oscillators Safety warning This article will discuss the principles of the Royer converter, including some issues relating to simulation. We are not presenting a complete design (project), but of course designs can be found elsewhere. Anyone interested in building these circuits must be aware that (depending on the design) they may be capable of producing dangerous output voltages and high levels of heat dissipation leading to hot components. There is a risk of electrocution and burns, so appropriate safety precautions must be taken when working with such circuits. Simulation files Most, but not every month, LTSpice is used to support descriptions and analysis in Circuit Surgery. The examples and files are available for download from the PE website. 58 Origins of the Royer oscillator Magnetic saturation The Royer converter was first described by GH Royer in 1954. A basic Royer oscillator circuit is shown in Fig.1. It is based around a transformer with multiple windings driven by two transistors. The oscillator is formed by driving the centre-tapped primary winding L1-L2 and receiving feedback via a relatively small centre-tapped secondary winding L3-L4. The oscillation will occur with just these two windings, but for obtaining a step-up output voltage a third winding is required (L5) – this is a secondary of the transformer that can deliver a stepped-up output voltage. The voltage induced in the secondary depends on the supply voltage (VCC) and the transformer turns ratio from winding L1-L2 to L5. In order to use the transformer correctly in the Royer oscillator it is necessary to know which way round the terminals are connected. This is indicated by using dots on the device and on the schematic symbols – these are known as ‘phasing dots’. The dotted terminals have the same instantaneous voltage polarity during transformer action. For example, when a dotted primary is being driven by the positive half of the AC cycle (or positive voltage pulse) a dotted secondary will have a positive polarity with respect to its non-dotted terminal. During this positive cycle the current will flow into the dotted primary terminal and out of the dotted secondary. Operation of the Royer oscillator involves magnetic saturation, so it worth defining what we mean by this. Applying an external magnetic field to a magnetic material magnetises it. Magnetic saturation occurs when an increase in an applied magnetic field cannot further increase the magnetisation of the material. This happens in some types of material, particularly ferromagnetic materials such as iron. In these materials there are microscopic domains, which act like very small magnets that can change their direction of magnetisation. In the presence of an external magnetic field the domains have a tendency to align with the applied field. The stronger the field the more aligned the domains become; however, there is a limit to this process after which any further increases in the applied magnetic field will not cause more domain alignment (magnetisation). At this point, magnetic saturation has occurred. If the core material of an inductor exhibits saturation, then the inductor will become nonlinear at currents which approach or exceed the saturation point. This means that the inductance will vary with the current flowing through the inductor (ideally, the inductance of an inductor does not change with the current through it). The inductance reduces as the core enters saturation; and this will also reduce coupling between Vcc Q1 L1 Q2 L2 Vout L5 R1 L3 R2 L4 Fig.1. Basic Royer power converter circuit. Inductance (arbitrary units) Recently, user Frostyjams posted a question about the Royer oscillator on the EE Web forum. We will not address the specific question asked – there is a reference to schematics which are not available at the time of writing; however, this is an interesting circuit so we will take a look it in more general terms. The Royer oscillator is also referred to as the ‘Royer converter’ because it is commonly used in inverter power converters which generate relatively high AC voltages from a DC supply (of course, the AC output can also be rectified to give DC-to-DC power conversion). There are also several similar circuits which can be used for the same purpose. A common use of Royer (or similar) converters has been for generating the high voltages needed for display backlighting using Cold Cathode Florescent Lamps (CCFL). However, use of CCFL backlighting has been decreasing for the last decade as LEDbased approaches take over. Nevertheless, there are a variety of reasons why stepup voltage converters might be needed where a Royer converter, or similar circuit might be considered. 1.0 0.8 0.6 0.4 0.2 0 0 0.25 0.5 0.75 Current (arbitrary units) 1.0 Fig.2. Example of typical inductance vs current curve for an inductor exhibiting hard saturation. The specific shape, inductance and current values will depend on the size of the inductor and core material. Practical Electronics | May | 2022 windings in a transformer. The shape of an inductance vs current graph varies depending on the properties of the core material. An example is shown in Fig.2 – this shows a characteristic referred to as ‘hard saturation’, where the inductance reduces rapidly above a certain current level. In most applications, saturation is avoided as the reduction in inductance may lead to damaging increases in current, waveform distortion, or poor efficiency. However magnetic saturation is exploited in some applications, and it plays a part in the operation of the Royer oscillator. Waveforms and frequency Royer circuit oscillation is ‘square wave’ in nature rather than sinusoidal because the transformer is driven into saturation (an appropriate transformer, with suitable core material, must be used to achieve efficient operation). In mains power supplies the input to the transformer primary is a 50Hz or 60Hz sinewave. For DC-to-DC converters, neither a sinewave nor a frequency as low as this has to be used. Higher frequencies enable smaller transformers to be used, and if the frequencies are above the audio range (>20kHz) then they will give silent operation (you may hear a hum or whine from power circuits operating in the audio range). Pulsed (square-wave) inputs to a transformer (or other inductor) are commonly used in modern switching power supplies because they are relatively easy to generate using control logic. This logic often uses pulse modulation (switching pulses on or off, or modifying their length) for feedback control of the output voltage as the load varies. Of course, such converter circuits deliver better performance than the basic Royer oscillator converter (which does not regulate the output), but Royer oscillators (or similar circuits – see later) can be used in conjunction with modern switching regulators to provide a more stable output voltage. This was a common approach in the heyday of CCFL backlighting. Circuit operation The resistor network (see Fig.1) provides bias and ensures that one of the transistors switches on and the circuit starts oscillating when power is applied. When the circuit powers up, real-world asymmetry will result in one of the transistors switching on faster than the other (the two halves of a real circuit will never be exactly the same, despite the symmetrically designed structure). The circuit has a ‘bistabletype’ behaviour, with the transistors switching on and off out of phase with one-another, with a duty cycle of 50%. An advantage of the Royer circuit is that the transformer is directly part of the oscillator, so it uses relatively few components to achieve a step-up converter (compared with using separate oscillator and driver stages). However, the circuit has quite a few issues, which we will discuss later. Transistor operation Appropriate transistors should be used, which must have high gain (hFE), low saturation voltage V CE(sat) , low on-resistance (RCE(sat)) and high collectorbase breakdown voltage. Transistors specifically designed for high-current switching applications should be used. As noted above, the transistors alternate being on and off and one will switch on first at power up. So, we can use the assumption that one transistor has just switched on as a starting point for describing circuit operation. If we assume Q1 switches on first then L1 will have voltage across it with the dot end positive, and by transformer action there will be voltages produced on the L3-L4 secondary with polarity corresponding with the dotted ends of the windings. Thus, Q1’s base will be driven positive – turning the transistor on more – this is positive feedback, which will ensure definite switching. After Q1 switches on, the magnetic flux in the core will be increasing (the rate of change can be found from Faraday’s law, which relates the rate of change of magnetic flux to electromotive force). Vcc However, after some time in this state the Q1 C1 R1 R2 transformer will L1 saturate, that is the DC Vout magnetic flux with Q2 + stop increasing, at L4 which point the C2 – majority of the transformer action L3 will stop, and with it the base current to Q1. With the base current removed, Q1 Fig.3. A variation on the Royer circuit with an example will turn off, causing secondary side circuit. Practical Electronics | May | 2022 www.poscope.com/epe - USB - Ethernet - Web server - Modbus - CNC (Mach3/4) - IO - PWM - Encoders - LCD - Analog inputs - Compact PLC - up to 256 - up to 32 microsteps microsteps - 50 V / 6 A - 30 V / 2.5 A - USB configuration - Isolated PoScope Mega1+ PoScope Mega50 - up to 50MS/s - resolution up to 12bit - Lowest power consumption - Smallest and lightest - 7 in 1: Oscilloscope, FFT, X/Y, Recorder, Logic Analyzer, Protocol decoder, Signal generator 59 Royer oscillator is challenging to simulate accurately. However, it is worth looking at a couple of the issues involved. Fig.4 is an LTspice schematic of a Royer oscillator based on Fig.3. The component choices are somewhat arbitrary – simulating this is not going to be realistic, so it really isn’t worth worrying about it too much. To create a transformer in LTspice we need to define a mutual inductor (add a mutual inductor component to the netlist), which is done by adding the netlist line for a mutual inductor as text to the schematic. The syntax of the mutual inductance statement is: Kxxx L1 L2 [L3 ...] <coefficient> This is a name starting with K, followed by the list of inductors which are coupled together (the windings of Fig.4. LTspice Royer oscillator schematic (which is not an accurate model). the transformer), followed by the coupling coefficient. For ideal transformers, the coupling coefficient is 1, but it can be set in the range 0 to 1 to model transformers where its collector current to drop. This reversed voltage change not all of the magnetic flux perfectly links the coils. Here we will, via the remaining transformer action, produce a positive use 1 to keep things simple. voltage at Q2’s base, turning it on, and again the positive The results of simulating the circuit in Fig.4 are shown in feedback action will push this switching action, further Fig.5. Note that there are oscillations, but they take a long time turning Q2 on. to get started. When simulating oscillators like this, start-up As Q2 turns on the resulting change of voltage polarity is often a problem – some circuits do not oscillate at all in on the primary (check the dots on the schematic) will cause simulation. The reason in this case is the perfect symmetry the transformer to fi rst come out of saturation and then of the simulated circuit, something which does not occur saturate with the fl ux in the opposite direction to when in real circuits – earlier we assumed one of the transistors Q1 was on. When Q2 saturates it will switch off, in the switched on first, which started the oscillation, but this may same way Q1 did, and consequently Q1 will start to switch on and the cycle will repeat. Thus, we have oscillation. The frequency of oscillation depends on the transformer core’s crosssectional area, number of turns in the primary, its magnetic properties (saturation flux), and the supply voltage. This makes it relatively difficult to set a precise frequency. There are a number of possible variations of, or additions to, the basic Royer circuit in Fig.1. These include the addition of diodes to protect transistors from voltage spikes and variations of the bias and secondary feedback circuit. An example of this is shown in Fig.3 – this circuit Fig.5. Simulation results for the circuit in Fig.4. does not need a centre-tapped feedback winding. The circuit in Fig.3 includes a supply decoupling capacitor (C1) – supply decoupling is generally good practice, particularly with switching circuits such as this which may experience large transient current demands. Fig.3 also shows a full-wave rectifier circuit and smoothing connected to the secondary – this is just an example – different secondary circuits can be used depending on the application. Royer simulation We often simulate circuits for Circuit Surgery articles, but the Fig.6. Setting initial conditions can help start oscillator simulations. 60 Practical Electronics | May | 2022 .IC V(bq1)=0.6 v(bq2)=0 This forces the base voltages at the start of the simuation to a situation where Q1 is on and Q2 is off, which will start the oscillation immediately, as shown in Fig.6. The statement is added to the schematic using the ‘.op’ SPICE directive button. Simulating transformer saturation Fig.7. LTspice schematic to demonstrate inductor saturation. not happen in the simulation with two identical transistor models. For the simulation, numerical differences between node calculations may be sufficient to start the simulation, but probably not in a realistic way. Forcing simulator oscillation Oscillation start-up in symmetrical oscillator circuits can be achieved by introducing asymmetry in the circuit, or by forcing the circuit voltages to initialise in a state that will start the oscillation. For the first approach, we could change component values – for example, in this circuit, setting slightly different resistor values may work. For the second approach we can try the .IC and .NODESET SPICE directives. These control the DC initial conditions which LTspice calculates at the start of every simulation, and are used to determine the starting point for transient simulations. The .IC directive forces the initial voltage at the start of the simulation, whereas .NODESET is more of a hint to the simulator during its initial calculations. Although use of .IC requires caution because it can give misleading results, it can be useful for starting oscillators if .NODESET does not solve the problem. For the circuit shown in Fig.4, consider: It would be easy to assume that Fig.6 is a useful result – it shows a ‘spikey’ square wave oscillation, which is what we might expect from a Royer oscillator. However, we need to be very cautious as there is a very significant factor lacking from the simulation model, which was a key element of our description of the circuit’s operation – that is saturation of the transformer. The inductors in the simulation are close to ideal (apart from the series resistance specified) and so do not exhibit saturation, so it is not safe to assume the results are realistic. It is reasonably straightforward to include saturation of individual inductors in LTspice. This is done by replacing the normal inductance value on the schematic with a mathematical expression for flux to create a behavioural inductor (known as the Arbitrary Inductor model). The most basic form of the expression is: flux=L*Is*tanh(x/Is) Here, L is the inductance in henries (when not saturated), Is is the saturation current in amps and x is the inductor’s current (referred to as a ‘special keyword’ in LTspice documentation). tanh is the hyperbolic tangent function. Thus, for an inductor of 100µH and a 500mA saturation current, the expression above becomes: flux=0.0001*0.5*tanh(x/0.5) This example is used in the LTspice schematic in Fig.7, along with a standard inductor of 100µH. This example is to illustrate the behaviour of LTspice inductor models and does not represent specifi c real components. The flux expression can also be more complex to model more details of the inductor saturation; for example, sharpness of the current-inductance curve. The piecewise linear (PWL) voltage sources used in the circuit in Fig.7 create a current into the inductors which is either constant or changing at a constant rate of ±200A/s (1.0A in 5ms) – see the top trace on Fig.8. The characteristic equation of an inductor – the relationship between current through (i) and voltage across it (v) is: v = L di/dt Fig.8. Simulation results from the circuit in Fig.7. Practical Electronics | May | 2022 Here, L is the inductance. The term di/dt is a differential, which represents the rate of change of the current (i) with time (t). Passing a current through an inductor results in a voltage 61 across the inductor that is proportional to the rate of change of the current. The voltage for the standard inductor model in the simulation is shown in the middle trace in Fig.8. When the current is constant (either at 0 or at 1.0A) the inductor voltage is zero. When the current is changing at ±200A/s the voltage across the inductor is 100µH × 200A/s = 20mV, as given by the above equation, with the polarity of the voltage dependent on the direction of change of the current. The behaviour of the inductor with saturation is shown in the bottom trace in Fig.8. It is obviously very different from the middle trace. The current-voltage-inductance equation still applies, but the inductance decreases with increasing current – as shown in Fig.2, but not necessarily following exactly the same curve. The decrease in inductance with increasing current results in lower voltages at higher currents – note, for example, the peak in negative voltage as the decreasing current passes through zero at 20ms in Fig.8. Simulation limitations Unfortunately, the Arbitrary Inductor model does not cover all aspects of inductor saturation – specifically saturation hysteresis. Another approach can be used, which is based on the Chan model (named after John Chan et al. who published it in the IEEE Transactions on Computer-Aided Design, Vol. 10. No. 4, April 1991). This model is also built into LTspice but requires several magnetic parameters to be set for the core (in a similar way to the Arbitrary Inductor model, the list of parameter values is entered in place of the normal inductance value), so you need to know these values to use it. Furthermore, the LTspice Chan model does not support mutual inductors, so cannot be used in a straightforward way for transformers. More comprehensive transformer models exist but typically require complex equivalent circuits to represent the transformer’s behaviour. Thus, we conclude our simulation discussion without a detailed model of the transformer in Fig.3, but with some insights into the difficulties. Drawbacks of the Royer oscillator The Royer oscillator has a few undesirable characteristics. First, it tends to produce large simultaneous voltage and current spikes which may damage the transistors. Second, the frequency of oscillation is dependent on the supply voltage, which may be problematic in some designs. Third, the large spikes, and possible high frequency oscillations (not the main oscillations) after each switch event, can generate a lot of radio frequency interference (RFI). Adding an inductor between the centre tab of the primary and the supply can reduce the spikes and improve efficiency. Baxandall oscillator Fig.9. Baxandall oscillator LTspice circuit used to produce illustrative waveforms. Fig.10. Simulation results from the circuit in Fig.9. 62 Before finishing our discussion it is worth taking a quick look at the circuit in Fig.9 (also an LTspice schematic). This circuit is frequently referred to as a Royer oscillator, but is also called the ‘Baxandall oscillator’ after Peter Baxandall, who published a paper on it in 1959. Many of the ‘Royer’ circuits in the CCFL step up converters mentioned at the start of the article are Baxandall oscillators. A different name is probably preferable as the operating principle is very different. Of particular relevance to the preceding discussion is that the transform does not saturate in the Baxandall oscillator. A key difference from the Royer circuit is the capacitor (C2) across the primary winding (L1-L2), which creates a resonant LC circuit and leads to sinewave oscillation, unlike the square wave for the Royer circuit. The LC resonant oscillator drives the alternate switching of the transistors via the feedback winding in a similar way to the Royer circuit. The transistor switching in turn sustains the resonant oscillation. Simulation results for the Baxandall circuit are shown in Fig.10. The voltages at the two collectors (top two traces) are half sinewaves, but the voltage across the whole primary (L1-L2) is a full sinewave (bottom trace). The circuit in Fig.9 is not a suggested design, it is just a configuration that produced illustrative waveforms for a typical circuit of this type. Practical Electronics | May | 2022