Transistorized power supplies for induction heating
Transistorized power supplies for induction heating: In the last five years power MOSFETS have become available which have good radio frequency capability, low drive power requirements, and ease of paralleling for higher power output. An induction heating power supply using power MOSFETS as switches are more efficient is more easily controllable, and has a lower capital cost than a valve supply, and the MOSFET power supply is not limited in frequency as thyristor supplies are. This paper gives details of three different types of transistorized power supplies developed for induction heating applications and compares their performances.
I . Introduction
Transistorized power supplies for induction heating
In the simplest form, the output of an induction heater consists of a water-cooled copper tube carrying an alternating current and surrounding the metallic workpiece which is to be heated. The pulsating magnetic field set up by the induction heating work coil induces eddy currents in the metallic workpiece and, as these eddy currents meet resistance to their flow, heat is generated by the Joule heating effect.
most important parameters
The two most important parameters of induction heating power supplies are their frequency and power output. Figure 1 shows typical operating frequencies and power supply ratings associated with industrial applications of induction heating.
Operating frequencies and power ratings associated with induction heating techniques vary so widely that many different types of power sources have been used (Davies et a/. 1979). Mains frequency systems are used for metal melting, heating metal billets before forging, rolling, or extrusion at powers up to 100MW. Magnetic frequency multipliers extend the frequency range to 150Hz by extracting the third harmonics generated when the outputs of three single-phase mains frequency transformers are connected in an open delta configuration. Later units used the ninth harmonic to extend the operating frequency to 450Hz.
induction motor driving
For frequencies between 1 kHz and 10 kHz motor-alternators have been used. These units consist of a standard squirrel-cage induction motor driving a Guy or Lorentz type inductor alternator via a single rotor shaft. These units are extremely robust and reliable, but the high maintenance costs and low efficiency, especially at low power outputs, have meant that over the last decade they have largely been superseded by solid-state thyristor inverters. In the last five years advances in device technology and inverter circuitry have raised the operating frequency capabilities of thyristor inverters and units are available at 1 kHz up to 1 MW, at 10 kHz up to 400kW, and 50kHz up to 100kW (Hobson 1984).
For applications requiring operating frequencies between SO kHz and MHz, triode valve oscillator power supplies are almost universally used. A high direct voltage is applied between the anode and the cathode of the valve, and the high-frequency output is usually fed to a parallel resonant circuit possibly incorporating a matching output transformer.
Even operating in a class C mode, as is used, operating efficiencies of valve oscillators rarely exceed 60%. The capital cost of valves has risen sharply over the last decade, whereas the costs of solid-state devices have fallen and their availability increased. Hence for some years, attempts have been made to replace valve oscillators with transistorized power supplies. Units using parallel combinations of bipolar transistors are available within this frequency range, but they have limited power output capabilities and poor reliability. They have made little impact industrially, usually being associated with applications within laboratory-type environments.
high-power field-effect transistors
The development of high-power field-effect transistors (MOSFETs) has made high-power solid-state units for induction heating feasible. The MOSFET is a majority-carrier device and hence has far shorter turn-off times and hence lower switching losses than bipolar transistors and thyristors. Turn-off times of less than 20011s for a device capable of carrying 15 A and sustaining 400V are typical.
A MOSFET is switched on by charging the input capacitance, and hence it is a voltage-driven device. The drive power requirements for a MOSFET are therefore much less than that of a bipolar transistor and inputs directly from CMOS logic are possible.
MOSFET power supplies
Because induction heating applications require relatively large power outputs, a major advantage of MOSFETs is that their positive temperature coefficient of resistance makes it easy to use devices in parallel for higher power outputs. MOSFET power supplies should, therefore, be far more reliable than bipolar transistor systems which require special circuit configurations to avoid thermal runaway. An induction heating power supply using power MOSFETs should therefore be more efficient, more easily controllable, and have lower capital costs than existing valve oscillator power supplies.
2. Inverter configurations
There are many types of solid-state inverter circuits associated with induction heating (Hobson 1983). but the three most common are (a) the current-fed (or load resonant) inverter, (b) the voltage-fed (or swept-frequency) inverter, and (c) the cycloconverter or AC-AC inverter. The cycloconverter produces a single-phase high-frequency output from a three-phase 50Hz input without the use of a DC link. It has been used with thyristors at frequencies up to 3 kHz and power levels over 1 MW largely for induction melting furnaces. At first sight, the elimination of the DC link including the separate rectification and energy storage elements should reduce the number of components, and therefore should increase the operating efficiency of the unit and possibly decrease the capital cost. On the other hand, the increased complexity of the control logic and the higher level of supply system harmonics usually associated with the cycloconverter must be taken into account. To assess the cycloconverter form of transistorized power source a small prototype unit was built (Hobson et al. 1985).
2.1. The cycloconuerter
The basic design of the power circuit is derived from existing thyristor cycloconverter circuits (Havas et al. 1970) but using power MOSFETs as the switching elements. The reverse voltage which can be sustained across a MOSFET is very small because its construction includes a parasitic diode in parallel with the transistor, which provides a short-circuit path for any reverse voltage between drain and source. When using an AC supply, therefore, each transistor requires a blocking diode in series with it to prevent a short-circuit.
One possible circuit arrangement would be that of a totem pole. However, the reverse recovery time of the parasitic diode is rather slow (290x1s). compared with MOSFET switching times which can be as fast as 15-2511s. Therefore it was decided to use discrete fast-recovery diodes to provide the reverse blocking, with a diode in series with each MOSFET. The operation of the circuit can be explained by considering the variation of the three-phase supply voltages over one cycle of the red phase, and a simplified schematic of the cycloconverter is shown in Fig. 2.
At O” in the red voltage cycle, the most positive phase is the blue and the most negative is yellow. Thus switching on T5 will allow a current to flow in direction I, through the load and return to the supply through the three capacitors; whichever are at a more negative potential, i.e. in this case the ones connected to the yellow and red phases. Then if T5 is switched off and T4 switched on, a current will flow in me, Three-phase the direction I, again through the load, and the capacitors at the most positive potential. Therefore by switching between T4 and T5 in antiphase an alternating current through the load can be obtained.
At 30″ the red phase becomes the most positive mains cycle, so TI is switched on for the positive half-cycle instead of T5. At 90″ blue becomes the most negative mains cycle, and so T6 is switched on for the negative half-cycle instead of T4. Hence by monitoring which mains cycles are most positive or most negative and switching the transistors in those phases on and off in antiphase at a high frequency, a continuous high-frequency single-phase current will flow through the load. Figure 3 shows a block diagram of the cycloconverter.
A cyloconverter was built using International Rectifier IRF 610 devices which have a maximum drain-to-source voltage rating of 200V and a maximum drain current of about I A. A conventional series LCR circuit was used to represent the resonant induction heating load. The introduction of power MOSFETs into a high-frequency cycloconverter configuration suitable for use with an induction heating load proved successful. Operating frequencies in excess of 100 kHz were achieved.
However, the cycloconverter configuration did require additional control circuitry and more complicated protection circuitry than the more common current or voltage-fed inverter circuits. Mains harmonics were a significant problem even at the low powers used in this unit. It is difficult to extrapolate the observations made with the small unit, but the greater rates of change of current and greater stray inductances usually associated with larger power units would suggest that the problems of harmonic suppression would outweigh any advantages from the lower inherent component count of the cycloconverter.
2.2. Voltage-fed inverter
A voltage-fed inverter was constructed and the circuit diagram is shown in Fig. 4. At first sight, the replacement of thyristors by MOSFETS as the switching elements could mean the use of the internal parasitic diode of the MOSFET rather than an external diode to carry reactive currents.
However, the relatively slow reverse recovery time of the internal diode can cause problems if one transistor takes over conduction from the diode of an adjacent transistor in the same leg. The transistor may switch on faster than the diode switches off, and a ‘shoot-through can result, reducing the lifetime of the devices. Therefore the load needs to be driven above its resonant frequency to provide a lagging current, as shown in Fig. 5.
prototype voltage-fed inverter
A prototype voltage-fed inverter has been built and used to develop 3kW of power into a resistive load. The voltage-fed approach realized the advantages of easy shut-down by switching off the MOSFETs in the bridge, and power control using swept-frequency was straightforward. However, problems were encountered in protecting the transistors against the short-circuit fault conditions commonly associated with induction heating applications. The large capacitors inherent in the voltage-fed inverter have sufficient charge to maintain the short-circuit fault current for a long time. Also, an excessive rate of rising voltage problems was encountered when the load was driven off-resonance.
The use of power transistors at frequencies in excess of lOOkHz places great importance on minimizing the switching losses within each device. It is inherent in swept-frequency power control that the load is driven off its resonant frequency, and hence the switches have to interrupt large currents and reactive currents are carried by parallel diode circuits. Figure 6 shows that as soon as S1 starts to turn off and
the current through it is reduced, the load current will start to flow through D3. This will bring the voltage across S1 to V, minus the conduction drop of diode D3, making the switching losses of S1 appreciable.
2.3. Current-fed inverters
Current-fed circuits using thyristors are the most common inverters used for medium-frequency induction heating applications. They have a variable rather than the fixed direct voltage which is then inverted to give the desired output frequency. The inverter always operates at approximately the resonant frequency of the load, and power control is obtained by varying the direct voltage supplied to the inverter circuit.
high-frequency current-fed inverter
A high-frequency current-fed inverter circuit using power MOSFETs, as shown schematically in Fig. 7, has a number of advantages, including a low component count, low switching losses and inherent short-circuit protection provided by the large smoothing choke. The choke is necessary to feed a constant current into the inverter stage, so causing a square wave of current, as shown in Fig. 8, to be fed into the load. A 3 kW prototype current-fed inverter has been developed to a stage of industrial application. Details are now given of the design of the unit and its performance.
2.3.1. The rectification stage
The direct voltage is derived from the output of a fully controlled six-pulse bridge. Phase-back of thyristor firing angle is performed by a digital timing circuit which receives an eight-bit delay word from the front panel for open-loop control or from a microprocessor in closed-loop control. For shut-down, the timing logic takes a hard-wired delay word and the direct voltage output will be inverted to a maximum of 60″. After the energy has been drawn out of the choke a contactor is taken out. The phasing-back of firing angle can be used for power control, or the temperature of the workpiece can be monitored and a temperature control loop implemented.
A direct-current current transformer (DCCT) has been used to monitor current in the DC link (Tebb era/. 1985). The DCCT has advantages over a current shunt in that it provides isolation and does not reduce efficiency. The direct current needs to be monitored for estimation of power input, and for overcurrent detection.
A microprocessor can be incorporated in the control system for closed-loop control of power output or workpiece temperature, as well as for housekeeping tasks, e.g. supervision of interlocks and monitoring of cooling-water temperature. The timing circuitry for the thyristors in the rectification stage has been designed to take eight-bit words as its input and so facilitate interfacing with a microprocessor bus.
2.3.2. The load circuit
Since the loaded work coil has a very low power factor (<0.1), the coil is resonated and the load for the inverter is, therefore, a resonant tank circuit. The square current wave fed to the load has high-frequency components which see a low impedance with a parallel resonant circuit and a high impedance with a series resonant circuit. A current-fed inverter feeding a series resonant load would cause voltage spikes, and so it feeds a parallel resonant tank circuit.
To select the semiconductors for the rectification and inversion stages, the Q and inductance of a loaded and unloaded induction heating coil and the required power input to the workpiece need to be known. From these values, the necessary tank circuit capacitors to resonate the coil at a given frequency can be calculated, as can the impedance of the tank circuit at resonance and so the voltage required across the tank circuit. This specifies a minimum drain-to-source breakdown voltage of the MOSFETs, but to find the values of V, and I, and so the voltage and current ratings of thyristors and MOSFETs, a relationship between the tank circuit voltage (V,) and V, must be derived. The current waveform switched through the tank circuit is shown in Fig. 9.
2.3.3 The inverting stage
Transistorized power supplies for induction heating
A current-fed inverter has been constructed in which each switch in the inverter stage is implemented by two transistors in parallel (type IR450). The switching sequence of the inverter stage is as follows:
S1 and S2 on
SI, S2, S3, and S4 on
S3 and S4 on.
The overlap period when all switches are on is necessary to prevent the choke from being open-circuited. This unit is fully protected and has fed 3 kW into a resistive load at 150 kHz.
Feeding the induction heating
Feeding the induction heating load caused problems with parasitic lead inductance ringing with the drain-to-source capacitance of the MOSFETs. More careful attention to layout to reduce the inductance of connections has greatly reduced this problem. A filter has also been used to reduce ringing. This is placed in series with the path of the ringing currents and the results of this work will be published shortly. Free of voltage spikes on the MOSFETs, the inverter has been successfully applied to industrial work coils, and waveforms from these tests are given in Figs. 10 and 11.
Transistorized power supplies for induction heating
The prototype cycloconverter worked with limited success at operating frequencies in excess of IOOkHz and power levels approaching 3 kW. However, this cycloconverter required additional control circuitry and more complicated protection circuitry. Mains harmonics were also significant, and although it is difficult to predict the level of harmonics expected at larger power levels from results taken from the prototype unit, the greater current levels and large stray inductance values associated with larger units suggest that the problems of harmonic suppression would limit the use of this circuit configuration.
A prototype voltage-fed inverter has been built which again developed approximately 3 kW at 100kHz. In induction heating applications the method of power control is by swept frequency, i.e. requiring the driving of the load away from its natural resonant frequency; and therefore diodes are required in parallel with transistors to carry the reactive currents. It was not possible to use the internal diode of the MOSFET at frequencies around 100 kHz. Problems were encountered with the protection of the transistors under short-circuit faults across the load circuit (which is common in induction heating applications) and with an excessive rate of rising of voltage when the load was driven off-resonance. Switching losses within the transistors could also be high when working significantly off-resonance.
The current-fed inverter design was found to have a number of advantages, including lower switching losses, better short-circuit load protection (inherent in the use of a large choke to restrict the rise of current), fewer components, and simpler control and protection circuitry. A current-fed inverter has now been developed to the stage of industrial application, and further work is at present underway to enhance the performance of the original design with a view of making larger-scale units suitable for the whole range of induction heating applications.
Induction Heating Handbook