AMAZON multi-meters discounts AMAZON oscilloscope discounts
|
|
1 INTRODUCTION
Very often, an auxiliary power supply will be required to provide power for control and drive circuits within the main switchmode unit.
Depending on the chosen design approach, the auxiliary supply will be common to either input or output lines, or in some cases will be completely isolated. A number of ways of meeting these auxiliary requirements are outlined in the following sections.
The method chosen to provide the auxiliary needs should be considered very carefully, as this choice will often define the overall design strategy. For example, in "off-line" sup plies, if the internal auxiliary supply to the control and drive circuits is common to the input line, then some method is required to isolate the control signal developed at the output from the high-voltage input. Often optical couplers or transformers will be used for this purpose.
Alternatively, if the internal auxiliary supply is common to the output circuit, then drive transformers to the power transistors may be required to provide the isolation. For this application, they must meet the creepage distance and isolation requirements for the various safety specifications. This makes the design of the drive transformers more difficult.
When the specification requires power good and power failure signals or remote control functions, it may be necessary to have auxiliary power even when the main converter is not operating. For these applications, impulse start techniques and auxiliary supply methods that require the main power converter to be operating would not be suitable. Hence, all the ancillary requirements must be considered before choosing the auxiliary supply method.
2 60-HZ LINE TRANSFORMERS
Very often small 60-Hz transformers will be used to develop the required auxiliary power.
Although this may be convenient, as it allows the auxiliary circuits to be energized before the main converter, the 60-Hz transformer tends to be rather large, as it must be designed to meet the insulation and creepage requirements of the various safety specifications. Hence, the size, cost, and weight of a 60-Hz auxiliary supply transformer tend to make it less attractive for smaller switchmode applications.
In larger power systems, where the auxiliary transformer size would not have a very dramatic effect on the overall size and cost of the supply, the 60-Hz transformer can be an expedient choice.
An advantage of the transformer approach is that fully isolated auxiliaries can easily be provided. Hence, the control circuitry may be connected to input or output lines, and the need for further isolation may be eliminated. Further, the auxiliary supplies are available even when the main switching converter is not operating.
3 AUXILIARY CONVERTERS
A very small, lightweight auxiliary power supply can be made using a self-oscillating high-frequency flyback converter. The output windings on the converter can be completely isolated and provide both input and output auxiliary needs, in the same way as the previous 60-Hz transformers.
Because auxiliary power requirements are usually very small (5 W or less), extremely small and simple converters can be used. A typical example of a nonregulated auxiliary converter is shown in FIG. 1. In this circuit, a self-oscillating flyback converter operates from the 150-V center tap of the voltage doubler in the high-voltage DC supply to the main converter.
FIG. 1 Auxiliary power supply converter of the single-transformer, self-oscillating
flyback type, with energy recovery catch diode D3.
4 OPERATING PRINCIPLES
Initially, Q1 starts to turn on as a result of the base drive current in resistors R1 and R2. As soon as Q1 starts to turn on, regenerative feedback via winding P2 will assist the turn-on action of the transistor, which will now latch to an "on" state.
With Q1 on, current will build up linearly in the primary winding P1 at a rate defined by the primary inductance and applied voltage (dI/dtsVcc /Lp). As the current builds up in the collector and emitter of Q1, the voltage across R3 will increase. The voltage on the base of Q1 will track the emitter voltage (plus Vbe), and when the base voltage approaches the voltage developed across the feedback winding P2 (about 5 volts), the current in R2 will fall toward zero, and Q1 will start to turn off.
Regenerative feedback from P2 will now reverse the base drive voltage, turning Q1 off more rapidly. By flyback action, the collector of Q1 will fly positive until the clamp diode D3 is brought into conduction. Flyback action will continue until most of the energy stored in the transformer core is returned to the 300-V supply line.
However, at the same time, a small amount of energy will be transferred to the outputs via D5 and D6. Because D3 conducts throughout the flyback period, and the same primary winding is used for both forward and flyback actions, the flyback voltage will be equal to the forward voltage, and the output voltage will be defined by the supply voltage. Also, the flyback period will be the same as the forward period, giving a 50% mark space ratio, that is, a square-wave output.
The inductance of the transformer primary should be chosen by gapping the core such that the stored energy at the end of an "on" period LPp is at least three or four times greater than that required for the auxiliary outputs, so that clamping diode D3 will always be brought into conduction during the flyback period. This way, the secondary flyback voltage will be defined by the turns ratio and the primary voltage. This can be an advantage when the auxiliary voltage is to be used for power failure/power good indications and to provide low-input-voltage inhibit actions.
As the primary turns are very large (300 to 500 turns typically), there is a considerable distributed interwinding capacitance in the primary. The relatively small primary inductance, given by the large air gap in the transformer core, improves the switching action in these small converters.
Although the current in the primary may appear quite large for the transmitted power, the overall efficiency remains high, as the majority of the energy is returned to the supply line during the flyback period. The mean off-load current of these converters is often only 2 or 3 mA, although the peak primary current may be as high as 50 mA.
Because a DC current is taken from the center tap of the input capacitors C1 and C2, this simple converter is suitable only for voltage doubler applications. If full-wave input rectification is used, a DC restoration resistor is required across C1.
5 STABILIZED AUXILIARY CONVERTERS
Many variations of this basic self-oscillating converter are possible. By using a high voltage zener on the input side, it is possible to provide stabilized auxiliary outputs and also maintain a constant operating frequency.
FIG. 2 shows one such modification to the basic circuit. This circuit is also more suitable for dual-voltage operation, as the flyback energy is returned to the same input line as the primary load.
The input voltage is stabilized by ZD1 and gives constant-frequency operation. This auxiliary converter may be used as the basic clock for the control circuit, providing drive directly to the power switching transistors. Very simple and effective switchmode supplies may be designed using this principle.
An energy recovery winding P3 and diode D6 have been added to the transformer so that the spare flyback energy is returned to the same supply capacitor as the primary winding P1 (C3). This makes the mean loading current on ZD1 very low, allowing simple and efficient zener diode pre-regulation. Both forward and flyback voltages are regulated by ZD1, providing a regulated flyback voltage and hence regulated outputs. It is important to use bifilar windings for P1 and P3 and to fit the energy recovery diode D6 in the top end of the flyback winding P3. In this position it isolates the collector of the switching transistor from the interwinding capacitance in T1 during the turn-on edge of Q1.
The DC supply to the auxiliary converter is taken from the main 300-V line via R1 and R2, which have been selected so that input link changes for dual input voltage operation will not affect the operating conditions of ZD1. Further, an extra transistor Q2 has been added to the base of the converter transistor Q1 to permit external synchronization of the converter frequency.
It should be noted that the frequency may only be synchronized to a higher value, as Q2 can terminate an "on" period early, but cannot extend an "on" period. Turning on Q2 results in immediate flyback action for each sync pulse, giving higher-frequency operation.
FIG. 2 Stabilized auxiliary power converter of the self-oscillating
flyback type, with energy recovery winding P3 and synchronization input
Q2.
6 HIGH-EFFICIENCY AUXILIARY SUPPLIES
FIG. 3 shows a more efficient version of the previous circuit, in which the loss incurred in the feed resistors R1 and R2 has been eliminated by using a separate bridge rectifier D5-D8 to supply the converter. This arrangement is particularly useful for dual 110-220-V applications, as the rectifiers D5-D8 are effectively fed with 110 V for both positions of the voltage selector link; that is, for both 110- and 220-V operation. This effective 110-V ac line is also used in this example to supply the 110-V cooling fan. Hence, the same fan may be used for both input voltages. (To meet the safety requirements, the insulation rating for the fan must be suitable for the higher-voltage conditions.) Note: When operating from 220-V line inputs, the link to the center of C1-C2 is removed.
Under this condition, the load on the 300-V DC line must exceed the fan and auxiliary loading, to ensure DC restoration of the center point of C1-C2. Hence this circuit is suitable only for applications in which a minimum load is maintained on the output. Capacitors C1 and C2 must be selected to accommodate the additional ripple current loading provided by the fan and the auxiliary converter, although this will probably be a small percentage of the total loading in most applications.
FIG. 3 Auxiliary power converter of the self-oscillating flyback type
with a 110-V ac cooling fan supply, suitable for use with 110-V ac fans
in dual input voltage applications.
7 AUXILIARY SUPPLIES DERIVED FROM MAIN CONVERTER TRANSFORMER
When the main converter is operating, it is clear that a winding on the main converter transformer can provide the auxiliary supply needs. However, some means are required to provide the auxiliary power to the control circuits during the start-up phase. The following SECTION describes a number of starting methods.
8 PROBLEMS
- 1. Explain why the characteristics of the auxiliary power supply systems are sometimes fundamental to the operation of the main power section.
- 2. What is the major disadvantage of using small 60-Hz transformers for auxiliary power systems?
9 LOW-NOISE DISTRIBUTED AUXILIARY CONVERTERS
9.1 Using a High-Frequency Sine Wave Power Distribution System
The previously described auxiliary power systems use hard switching methods (square waves). A disadvantage of such systems is that they have fast switching edges, and there is a tendency for high-frequency noise to be radiated and coupled into the distribution lines and transformers that feed the various parts of the host control system.
Modern large power systems now tend to use embedded microcontrollers and industrial interface systems, linking them to other power supplies and computer control systems.
Such systems often use multiple PCBs and may require several different and isolated local ancillary power supplies. Clearly, it would be very undesirable to have any high-frequency noise injected into internal or external ancillary power and control lines.
Although good layout and filtering may reduce the unwanted noise in hard switching systems to acceptable limits, a more effective approach is to use a high-frequency sine wave system that does not have fast switching edges and hence does not generate excessive high frequency noise. In this SECTION we examine a suitable 20-watt sine wave system.
10 BLOCK DIAGRAM OF A DISTRIBUTED AUXILIARY POWER SYSTEM
FIG. 4 shows a block diagram of the proposed 20-watt semi-stabilized
50-kHz sine wave distributed power system.
FIG. 4 shows the block diagram of a regulated distributed ancillary power system intended to provide multiple, distributed and isolated auxiliary supplies to multiple control PCBs in a large power system. Sine wave power is distributed via two pin connectors and twisted pairs at 50 kHz, 20 volts. This provides for low-noise, semi-regulated local supplies on each PCB, using local, small, high-frequency transformers and secondary rectifiers.
In the proposed system, we assume a single phase line input on the left, derived from two phases of a 208V 60Hz three phase system, or a single phase system. This is rectified and voltage stabilized in Block 1 to provide a regulated 215 volts DC. This DC is applied to the input of a 50-kHz sine wave inverter in Block 2. The inverter outputs a semi-regulated 20-volt RMS 50-kHz sine wave to drive a range of output modules providing the needs of the various sub-sections of the overall system. Although only four sub-modules are shown, clearly more or fewer modules can easily be provided, limited only by the system needs and the total power limitation of 20 watts.
The main advantage of providing a 20-volt 50-kHz sine wave output is that the transformers in the sub-modules can be quite small and efficient. The 50-kHz sub-module transformers provide isolation and reduce common mode noise. Further, the secondaries can have any voltage required by the sub-module, and an unlimited number of isolated outputs can easily be provided. Further, with the output windings isolated from the input, and by using sine wave drive with low harmonic content, the electrical noise on the outputs and distribution lines is very low.
A typical application for a system of this type is shown, in which the 10-kW power system described has eleven separate PCBs that are provided with a wide range of isolated ancillary supplies, ranging from 3 volts through 5 volts, plus and minus 12 volts, plus and minus 15 volts, 18 volts, and 24 volts.
The 50-kHz, 20-volt RMS output from the inverter (Block 2) is best connected to the various PCBs using two pin connectors and twisted pairs to minimize radiation. However, good results have been obtained using two adjacent lines of a ribbon cable with a grounded line each side.
We will now consider the details of the various building blocks starting at the left with Block 1 the "DC Regulator Section."
FIG. 5 shows the schematic for the 60Hz single phase line rectifier,
open loop linear regulator, and current limit section of the 20-watt
pre-regulator for the 50-kHz sine wave inverter shown in FIG. 8. The
inverter is used to provide the high-frequency distribution lines to
local on-board transformers, producing isolated, multiple DC outputs
for various control functions in a distributed ancillary network.
11 BLOCK 1, RECTIFIER AND LINEAR REGULATOR
FIG. 5 shows the 60 Hz line input, the rectifiers, and the series linear regulating circuit. Diodes D1 through D4 and C1 rectify the nominal 208 volt 60 Hz input to provide a nominal 300 volts DC at node "A," the input to the linear regulator IGBT Q1. Resistors R1, R2, and R3 provide current to the zener diode chain Z1, Z2, and Z3 to develop about 220 volts at node "B" with respect to the common input at node "E". This voltage is applied to the gate of the IGBT Q1 to produce an output via R7 at node "D" of 215 volts. This voltage is open-loop stabilized by the zener diode chain and the IGBT buffer to maintain the voltage near constant for load current and line voltage changes.
11.1 Output Voltage Regulation
FIG. 6 shows the start-up characteristics and output voltage at node "D," for input voltages from zero to 250 VDC. Good output voltage regulation is seen for the normal working voltage range, from 250 VDC through 400 VDC.
FIG. 6 shows the DC voltage regulation at the output of the linear regulator
at node "D," for input voltages measured at node "A" in
the range zero to 400 volts. Notice the linear transfer ratio from zero
to 250 volts, followed by good output regulation at 215 volts for inputs
from 250 to 400 volts DC. The characteristic provides a good low-voltage
start up action for the 50 kHz sine wave inverter section shown in FIG.
8, and good voltage regulation over the working range.
11.2 Fold-back Current Limiting
In FIG. 5, we see that the load current flows through the current-sensing resistor R7 to the output at node "D," so that the voltage across R7 is proportional to the load current. Overload current limiting is provided by R7 and Q2. The output voltage at node "D" decreases with increasing load current, with respect to the voltage at node "C." This decrease in voltage is applied directly to the emitter of Q2 while the base of Q2 (node "F"), is maintained constant at minus 2.2 volts with respect to node "C" via R4. When the voltage on Q2 base exceeds the emitter by 0.6 volts, Q2 turns "on" progressively, diverting the gate drive away from Q1 to prevent any further increase in current as the load resistance decreases. As a result, the output voltage decreases. Resistors R5, R6, and R7 produce the 2.2-volt negative bias at node "F" such that the voltage across R7 must exceed 2.8 volts (130 ma in 22 ohms), before Q2 can begin to turn "on." As the output voltage drops under current limiting, the negative bias at "F" decreases such that Q2 turns "on" harder, and the current limit is reduced to give a "fold-back" current limiting characteristic. This fold-back action reduces the stress in Q1 under short circuit conditions.
FIG. 7 shows good output voltage regulation for the normal working load range from zero to 100 mA, and current limiting above 130 mA (a load of 1,600 ohms). For overload conditions above 130 mA, a fold-back current limit characteristic is shown for load resistances from 1,600 ohms down to zero (a short circuit). The current limit point at 130 mA is there to provide a working margin; the current under normal conditions should not exceed 100 mA.
FIG. 7 shows the load regulation and foldback current limiting action
for the open loop linear regulator shown in FIG. 5. Notice the good load
regulation in the working range from zero to 100 milliamps (21.5 watts).
Overloads above 130 milliamps drive the unit into foldback current limiting.
The nominal dissipation in Q1 is 8.5 watts at full load, compared with
a short circuit dissipation of 27 watts. For extended protection under
short circuit conditions thermal shutdown may be required, depending
on the size and efficiency of the heat sink.
Although the open loop regulator design is relatively simple, it can be seen from FIG. 6 and 7 that it has a good start up characteristic, and provides adequate performance for this application.
Since the input voltage to the sine wave inverter is stabilized at 215 volts, the 50-kHz output from the converter will be semi-stabilized to within acceptable limits, such that it can be used for many ancillary applications without needing any additional regulation.
12 BLOCK 2, SINE WAVE INVERTER
In its simplest form, a high-frequency self-oscillating sine wave converter is not difficult to design and implement. Such systems are in common use for electronic ballasts. This type of inverter is described more fully in the Supplementary section Part 4 under the heading "Resonant and Quasi Resonant Power Supplies." FIG. 8 shows the basic schematic for a non-regulated, current fed, self-oscillating, sine wave inverter, designed to run at a nominal 50 kHz from the stabilized 215 VDC supply.
Since the input voltage is stabilized, it will be shown that with this type of inverter, the 50-kHz 20-volt output will be also be semi-stabilized.
FIG. 8 shows the schematic for the current fed, self-oscillating, sine
wave inverter. This simple inverter provides an output voltage that is
a direct linear function of the input voltage. With a voltage stabilized
input, the output will also be semi-stabilized. The transformer secondary
S1 provides 20 volts RMS at 50 k Hz to drive several remotely located
high-frequency transformers and rectifier circuits, thus providing isolated
and semi-regulated DC voltages to various control circuits in the main
system.
12.1 Input Choke L1
We describe the type of inverter shown in FIG. 8 as being "current fed" because the input current flows through the input choke L1, which is sized such that the ripple current at the working frequency is typically less than 10% of the full load input current. For simplicity in this example, the input current will be considered essentially constant at the mean working current defined by the 20 watt load.
The term "choke" will be used for L1, rather than inductor, because a relatively large DC current flows in the winding. There is also a large ac voltage stress across the choke.
The input choke has two optional positions, L1 or the alternative position LA. We will first consider the choke in position L1, since the circuit function is best understood this way. We will see later that the optional position LA does not change the function and will reduce capacitive coupling from primary to secondary windings, reducing output noise.
12.2 Choke Waveform and Peak Output Voltage
With L1 chosen to be large (typically 5 mH or more), the current in L1 can be considered constant throughout a cycle. The 215 volt DC input is applied to left of L1, and nearly constant current flows through it into the center tap of the transformer/inductor T1 at node "A." The actual current depends on the applied load and losses. (In this example, the current in L1 is near 100 mA at 20 watts load).
The choke input voltage is 215 volts DC at node "D" and the voltage waveform at the output of the choke at node "G" is shown in FIG. 9(B). It will be shown that this haversine waveform is well defined by the resonant action of the tank circuit formed by the inductance of P1 and P2, and C2. Further, the peak voltage at node "G" is also well defined, because the steady state forward and reverse volt seconds impressed across L1 are constrained to equate in the longer term to maintain constant input current.
12.3 Start Up
On start up, 215 volts DC appears at node "G" and current flows via R1 to the base of transistors Q1 and Q2. Diodes D3 and D4 block current flow to common node "E." The transistor with the highest gain will start to turn "on" (say Q2). As Q2 turns "on" the voltage at the finish of winding P2 (node "J") will go low, and the starts of all windings will go high. Hence, winding S1 start (node "K") takes R3 and the base of Q2 to an even higher voltage, giving regenerative turn "on" action for Q2. This starts a resonant half cycle of operation as shown in FIG. 9(A).
Notice that the voltage on the base of Q2 with respect to common cannot exceed the base emitter voltage of Q2, plus the diode drop of D2. This clamps the base voltage to about 1.2 volts at node "K." Since the voltage developed across S1 at mid cycle is about 10 volts, the clamping action at node "K" forces the finish of S1 to go negative by about 8.8 volts, turning Q1 hard "off." This waveform is shown in FIG. 9(C). At the same time D3 conducts, taking the top end of L2 negative while D4 blocks.
Current now flows upwards through L2 and R2, forming an inductive current loop from S1 start via R3, Q2 base-emitter, D2, L2, R2, D3 and through the S1 winding, back to the start of S1. At this stage of the cycle, we now have Q2 fully saturated "on," and Q1 fully "off."
The current established in the 1 mH inductor L2 will continue to flow by inductive action in the upwards direction throughout a cycle, limited only by R2 and the mean negative voltage at the top end of L2. During steady state operation, this current is the main drive to Q1 or Q2 during the turn-"on" action.
12.4 Resonant Action
With Q2 "on" and its collector near zero volts, the 215 volt supply and L1 continue to force current into the center tap of T1 (node "G"), and by resonant action the start of all windings will follow a half sine wave action, as shown in FIG. 9(A). At the end of the half cycle the voltage on all windings will be zero, and will start to reverse such that the finish of S1 will now go positive and the start negative. The previously established current up through L2 will now flow via D4, Q1 base-emitter, and D1 back to the lower end of L2. This cur rent, together with the current from R1 and S1 finish, now turns Q1 "on." The negative voltage on S1 start now turns Q2 "off " as D4 conducts. As before, this regenerative action is reinforced by the voltage generated across S1. A second half cycle is now established with Q1 fully "on" and Q2 fully "off," and a half sine wave voltage will appear on the finish of primary P2 at node "J." The voltage across the two primary windings (and hence across C2) is as shown in FIG. 10(A).
FIG. 9 shows the waveforms to be expected from the inverter circuit
with, respect to the common line node "E" with the input choke
in the L1 position and LA not fitted. The first trace (A) is the half
sine wave seen at node "H" (Q1
drain). A similar waveform will be found at node "J" (Q2 drain).
The second trace (B) is found at node "G" (the output of the
choke L1, and the center tap of the transformer). The third trace (C)
is found at node "K" (or the base of either Q1 or Q2). The
forth trace (D) is the current in the transformer primary P1 or P2, and
the final trace (E) is the collector current of Q1 or Q2.
FIG. 10 shows the waveforms expected across the tank circuit (the voltage
across C2) node "H" to node "J." (A similar voltage
will be found across the output winding S2, but the amplitude will be
much smaller).
Tip: To prevent damage to the oscilloscope, make sure it is isolated from the inverter line input when making these measurements. Use two X100 probes and two channels, with the oscilloscope ground on the common line node "E" when making the node "H" to node "J" measurement.
As a result of the above action the circuit self-oscillates at the resonant frequency, the base drive being provided by feedback winding S1, R1, and the forcing action of L2, with Q1 and Q2 switching "on" and "off " at zero voltage with low switching loss.
Tip: You will require a line isolation transformer on the input to the unit (better), or on the oscilloscope. Do not connect the oscilloscope ground to node "H" or "J," even with the isolation transformer, as these are high-voltage nodes and the oscilloscope will distort the waveforms and may even be damaged. Use two x100 probes and two channels, and connect the oscilloscope ground (common) to the lower common line of the ballast circuit at node "E" (provided the input choke is in position L1).
Special NPN transistors are designed for this application, such as the BUL216 or similar. For best results use these special devices, as they are designed for the reverse collector base current that flows in this application during Q1 and Q2 turn-"off " action.
12.5 Overlapping Conduction
In any current fed system using an input choke, it is essential that a current path is always provided through L1 via Q1 or Q2. It is essential that there is no part of a cycle when both turn "off " at the same time. If this happens, L1 will force excessive voltage to appear across the transistors, leading to breakdown. To prevent this happing under any conditions of load and line, a short conduction overlap is introduced.
In this design, the transistor storage time provides a short overlap period when Q1 and Q2 are both "on" at the same time. This shows up as a step on the waveform of Fig 10(A). This overlap is achieved by overdriving the base emitter junctions. Notice that although there is an "off " bias voltage available across S1, emitter-base reverse current flow is blocked by D1 or D2 so that the transistors turn "off " by internal recombination (a relatively slow process). Hence by adjusting R2, the overdrive can be adjusted to get the required overlap period.
Full wave rectification is recommended to balance the loading on the sine wave inverter.
13 OUTPUT MODULES
FIG. 11 shows a typical output module. Clearly the actual design of the modules and the output voltages will depend on the application.
FIG. 11 shows a typical output module. In this example semi-regulated positive
and negative 12-volt outputs are provided, and a fully regulated 5-volt output.
Many other arrangements are possible, depending on the application.
13.1 Module Transformers Designs
Normally the module transformers can be quite small, because the 50-kHz 20-V RMS primary input requires a relatively small number of primary turns. For example, we will see that a small EE19 core with a center pole area of 20 mm2 requires only 30 turns on the primary as shown below.
Because the waveform is a sine wave, the primary turns may be calculated directly using a dimensionally modified version of Faraday's Law as follows:
Ae _ Effective core area (mm2)
13.2 Design Example Using an EE19 Ferrite Core
The EE19 core has a center pole area Ae _ 20 mm2
. We can calculate the turns required at 20 volts RMS and 50 kHz as follows.
Minimum primary turns
When calculating secondary turns, remember the rectifiers will respond to the peak value of the sine wave where VV peak RMS 2 _
Also, be sure to allow for diode forward voltage drops.
Tip: If the correct ratio cannot be found with 30 turns on the primary, then increase the primary turns to obtain the nearest integer secondary turns as required.
13.3 Construction
Use a two section bobbin, and wind the primary in one section and the secondaries in the other. This has the advantage of reducing capacitive coupling and provides good primary to secondary isolation. The leakage inductance primary to secondary will be quite high, but this is a rare example where leakage inductance is helpful. It reduces the peak current into the output rectifiers and increases the conduction angle, but has little effect on the output voltage and regulation.
14 SINE WAVE INVERTER TRANSFORMER DESIGN
14.1 Operating Quality Factor Q
We start the design by defining the quality factor Q. This is an important parameter as it sets the magnitude of the primary current and allows the core size and wire sizes to be defined.
Q defines the ratio of reactive current, which does not contribute to output power, to real (resistive) current. Another way of looking at "Q" is that it is proportional to the ratio of energy circulating in the tank circuit, to the energy taken away by the load each cycle.
High Q circuits have nice clean sine waves, which are maintained by the larger cur rents circling in the tank circuit, as the tank circuit is loaded (rather like the flywheel in a mechanical system). This clean sine wave is obtained at the cost of high circulating current in the tank circuit, leading to high loss in the resistance of the primary windings P1 and P2. Conversely, low Q circuits have more distortion of the sine wave, but less loss. In this type of self-resonating inverter circuit, a compromise value will be found with a Q between 2 and 5.
Since the ac voltage across the capacitor C2 tends to remain constant, the value of C2 and the frequency tend to set the reactive current and hence Q. (The larger C2, the higher the Q). This is explained more fully in the supplementary section.
In this example C2 is 1000 pF, the frequency is 50 kHz, and the RMS voltage of the tank circuit is 477 volts. We can calculate the reactive current as follows:
At 477 volts RMS the current will be IVXC RMS RMS // mA. __ _ 477 3180 150 The effective input current reflected to the primary is I_load power/tank voltage / mA. ___ 20 477 42. The working Q will be Q/ / load ___ II C 150 42 3 6.
T1 provides two functions. First, it is a transformer, stepping down the 477-V RMS primary voltage to the required secondary voltage of 20 volts RMS. The primary windings P1 and P2 and the effective core permeability form the resonating inductor, and C2 the resonating capacitor in the parallel tank circuit. These are chosen to set the natural resonant frequency to 50 kHz.
The preferred design approach is to select C2 for the required Q (see above). The core gap is then adjusted, to get the effective permeability of the core for the correct inductance, so that P1 in series with P2 will resonate at 50 kHz with the selected capacitance C2.
Tip: In general with ferrite cores, providing the core size is not too small, it will be found that when the minimum primary turns (as defined by the frequency and primary voltage) have been calculated and wound, the inductance of the primary will be too high. Hence, a core gap will normally be required to reduce the effective core permeability and inductance. This is convenient, because it allows the inductance to be adjusted for optimum performance by simply adjusting the core gap. (If the inductance is too low, increase the primary turns).
14.2 Primary Voltage
It is shown in that the voltage across the tank circuit (the primary voltage across P1 and P2) is well defined and for our purpose here, we can assume that the peak voltage will be near P Vin.
In this example:
P(215 VDC) _ 675 volts peak. (FIG. 9 (B)).
The RMS value will be VV peak peak /2 or (. ). 0 707
In this example:
(675)0.707 _ 477 volts RMS.
Note: The "on" state overlap on Q1 and Q2 will increase this voltage by about 10%.
14.3 Core Size
The working "Q" (quality factor of the tuned circuit, Section 14.2) defines the ratio of reactive current to real (primary) load current. To find the primary current, we can use the reactive current IC flowing in the resonant capacitor C2, and the real current IL reflected into the primary from the load.
A Q of 3.6 was used to provide a good sine wave and optimum waveforms. However, the primary current in P1 and P2 will be almost four times larger than it would be in a non resonant system, (the reactive component does not contribute to the output power but does cause heating in the primary winding). Hence for our 20-watt sine wave system, we would choose a larger core. In this example we use a core recommended for a 60 watt system.
Hence in this example an ETD 35 was chosen.
14.4 Calculating Primary Turns
The ETD 35 has a core area of 92 mm^2.
As shown above in a sine wave system, the primary turns may be calculated as follows:
Where Nmin _ Minimum primary turns VRMS _ Winding voltage (Volts RMS)
f _ Frequency (Hz)
B_ Flux density (Tesla, typ. 150 mT) Ae _ Effective core area (mm^2)
It was found convenient to wind 6 layers of 28 AWG at 35 turns per layer (with a center tap at three layers), giving a total of 210 turns with a tap at 105 turns.
(This is acceptable, because the increased turns reduce the flux density to 111 mT, reducing core loss).
14.5 Turns Ratio (Primary to Secondary)
The primary voltage is 477 volts RMS and the secondary is required to be 20 volts RMS so the ratio is:
14.6 Core Gap
With the frequency ( f0 _ 50 kHz) and resonant capacitor (C2 _ 1000 pF) defined, the required resonating inductance (Lp) may be calculated.
With the primary turns defined and the core size and permeability known, the core gap may be calculated. However, I find it much faster to simply adjust the gap to get the required frequency in the working prototype as follows:
Start with a small core gap (say 0.010 inches) and hold the cores in place with an elastic band. Then apply sufficient input voltage for oscillation and check the frequency. Adjust the gap for the required frequency, and note the gap size.
Note: A "butt gap" is preferred. (That is, a gap the goes right across all three legs of the EE core). In this example a butt gap of 0.018 inches was required for 50 kHz operation. A butt gap reduces the local heating caused in the windings due to gap flux fringing. Cores that are gapped in the center pole only will have considerable local fringing that will cause eddy current heating in nearby windings. (See construction details in Part 4).
14.7 Calculating the Turns for the Drive Winding S1
Good regenerative starting will be found with peak drive voltages between 6 and 10 volts.
The primary volts/turn _ 477/210 _ 2.3 volts per turn, so 3 turns will give 6.8 volts RMS, (9.6 volts peak) and this was used in the prototype shown here.
15 REDUCING COMMON MODE NOISE
With the choke located in the L1 position, in series with the center tap of the 50-kHz transformer, the 100-kHz 337-V haversine peak voltage (FIG. 9 trace (B)) will capacitively couple to the 20-volt sine wave output winding, introducing common mode noise.
Moving L1 to the alternative position LA in the common return of the DC supply (this does not change the function of the circuit) has the advantage that the center tap now has a DC voltage on it, and the common mode injection point is removed.
Although the function is unchanged, the waveforms will look quite different.
Also see: Our other Switching Power Supply Guide