SMPS Design--Theory + Practice: SELF-OSCILLATING DIRECT-OFF-LINE FLYBACK CONVERTERS

Home | Articles | Forum | Glossary | Books

AMAZON multi-meters discounts AMAZON oscilloscope discounts

1. INTRODUCTION

The class of converters considered in this section relies on positive feedback from the power transformer to provide the oscillatory behavior.

Because of their simplicity and low cost, these converters can provide some of the most cost-effective solutions for multiple-output low-power requirements. With good design, extremely efficient switching action and reliable performance are obtained. A number of problems which beset the driven converter-in particular, cross conduction and transformer saturation-are overcome as a natural consequence of the self-oscillating topology. As the mode of operation is always complete energy transfer, current-mode control can very easily be applied, giving fast, stable single-pole loop response.

Self-oscillating converters are extremely low cost because few drive and control components are required. Attention to good filter design and screening of the transformer make this converter suitable for computers, video display terminals, and similar demanding applications.

There is a tendency to assume that these simple units are not real contenders in professional power supply applications. This misconception is probably due to the rather poor performance of some of the early self-oscillating designs. Also, the variation of operating frequency with load and input voltage has been considered undesirable in some applications. However, since the output is DC, there should not be a problem with the operating frequency in most applications, so long as efficient input and output filtering and magnetic screening have been provided.

2. CLASSES OF OPERATION

There are three classes of operation:

Type A, fixed "on" time, variable "off " time Type B, fixed "off " time, variable "on" time Type C, variable "on" time, "off " time, and repetition rate (frequency)

The major differences in the performances of these classes are as follows:

Type A will operate at an extremely low frequency when the load is light.

Type B will have a low frequency when the load is maximum.

Type C has a more desirable characteristic, as the frequency remains reasonably constant from full load down to approximately 20% load. Below 20% load the frequency will usually become progressively higher. (See FIG. 1.)


FIG. 1 Typical frequency variation of Type C self-oscillating converter as a function of load.

3. GENERAL OPERATING PRINCIPLES

In the self-oscillating converters considered here, the switching action is maintained by positive feedback from a winding on the main transformer. The frequency is controlled by a drive clamping action which responds to the increase in magnetization current during the "on" period. The amplitude at which the primary current is cut off, and hence the input energy, is controlled to maintain the output voltage constant. The frequency is subject to variations caused by changes in the magnetic properties of the core, the loading, or the applied voltage.

Figure 2 shows the major power components of a unit of type C. The converter is self-oscillating as a result of the regenerative feedback to the power transistor base from the feedback winding P2. The circuit functions as follows.

After switch-on, a voltage is developed across C1, and a current will flow in R1 to initiate turn-on of transistor Q1. As Q1 starts to turn on, regenerative feedback is applied via the feedback winding P2 to increase the positive drive to the base of Q1. The base current flows initially in C2 and then in D1 as the drive voltage is established. Consequently, Q1 will turn on very rapidly, and its maximum drive current will be defined by resistors R2 and R1 and the voltage across the feedback winding P2.


FIG. 2 Nonisolated, single-transformer, self-oscillating flyback converter, with primary current-mode control.

As these units are operating in the complete energy transfer mode, when Q1 turns on, current will build up from zero in P1 (the primary winding of the main transformer) at a rate defined by the primary inductance Lp. Hence

d Ip/ dt = Vcc / Lp

where Ip= primary current

Vcc= primary voltage

Lp = primary inductance

As the collector current, and therefore the emitter current, of Q1 increases, a voltage will be developed across R4 which will also increase at the same rate toward the turn-on voltage of Q2 (approximately 0.6 V). When Q2 has turned on sufficiently to divert the majority of the base drive current away from the base of Q1, Q1 will begin to turn off. At this point, the collector voltage will start to go positive, and regenerative turn-off action is provided by the snubber current flowing in D2, C5, R5. The voltage developed across R5 will assist the turn-on action of Q2 and turn-off of Q1. Further, by flyback action, all voltages on the transformer T1 will reverse, providing additional regenerative turn-off of Q1 by P2 going negative; the reverse current flow in C2 assists the turn-off action of Q1.

Although this drive system is extremely simple, it operates in a very well defined way.

Examination of the base current of Q1 will show almost ideal drive waveforms (see FIG. 3). The reason for the turn-off slope shown in FIG. 6.3 when Q2 conducts is that toward the end of the "on" period Q2 responds to a base drive voltage that is ramping upward; therefore Q2 turns on progressively, giving the very desirable downward ramp to the base drive current of Q1. This is the ideal drive waveform for most high-voltage transistors, and it is present here because regenerative turnoff action does not occur until all carriers have been removed from the base of Q1 and the collector current has started to fall. This turn off waveform prevents hotspot generation in the transistor Q1 and secondary breakdown problems. (See Part 1, Sect. 15.)


FIG. 3 Base drive current waveform of self-oscillating converter.

The system also has automatic primary power limiting qualities. The maximum current that can flow in R4 before transistor Q2 turns on is limited to Vbe/R4, even without drive from the control circuit. Consequently, automatic overpower limitation is provided without the need for further current-limiting circuitry.

In normal operation, the control circuit will respond to the output voltage and apply a drive signal to the base of Q2, taking the Q2 base more positive. This will reduce the current required in R4 to initiate turn-off action. Consequently, the output power can be continuously controlled so as to maintain the output voltage constant in response to load and input variations.

In foldback current-limiting applications, additional information on output current and voltage is processed by the control circuit to reduce the power limit under short-circuit conditions. Note that a constant primary power limit (on its own) would provide very little protection for the output circuitry, as there would be a large current flowing in the output when the output voltage is low or a short circuit is applied.

4. ISOLATED SELF-OSCILLATING FLYBACK CONVERTERS

A more practical implementation of the self-oscillating technique is shown in FIG. 6.4.

In this example, the input and output circuits are isolated, and feedback is provided by an optical coupler OC1.

Components D3, C4, and R8 form a self-tracking voltage clamp (see Sec. 3.2). This clamp circuit prevents excessive collector voltage overshoot (which would have been generated by the primary leakage inductance) during the turn-off action of Q1.

Components D1 and C3 are the rectifier and storage capacitor for the auxiliary supply line that provides the supply to the control optocoupler OC1.


FIG. 4 Isolated-output, single-transformer, self-oscillating, current-mode-controlled flyback converter. The control loop is closed to the output using an optical coupler.

5. CONTROL CIRCUIT (BRIEF DESCRIPTION)

A very simple control circuit is used. The diode of the optical coupler OC1 is in series with a limiting resistor R9 and a shunt regulator U1 (Texas Instruments TL430).

When the reference terminal of the shunt regulator V1 is taken to 2.5 V, current will start to flow into the cathode of V1 via the optocoupler diode, and control action is initiated. The ratio of R12 and R11 is selected for the required output, in this case 12 V.

The optocoupler transistor responds to the output control circuit so as to apply a bias current to R3. A voltage divider network is formed by OC1, R3, and the base of Q2 as the optocoupler current increases, and so the ramp voltage required across R4, and hence the collector current required to turn Q2 on and Q1 off, will be reduced. (A more complete description of this control circuit is given in Sec. 7.4.) As Q1 starts to turn off, its collector voltage will go positive, and the collector current will be diverted into the snubber components D2, C5, and R5. The voltage across R5 results in an increase in base drive voltage to R3 and Q2, as R5 has a higher resistance than R4, more than compensating for the drop in voltage on R4. This gives a further regenerative turn-off action to Q1. (The action of the snubber components is more fully described in Part 1, Sect. 18.) This simple circuit has a number of major advantages.

First, the unit always operates in a complete energy transfer mode. Consider the switching action: When Q1 turns off, flyback current will flow in the output circuitry. The transformer voltages will be reversed, and the drive winding P2 will be negative. Consequently, Q1 will remain turned off until all the energy stored in the magnetic field has been transferred to the output capacitors and load.

At that time, the voltage across all windings will decay toward zero. Now C2, which would have been charged during the flyback period, will track the positive-going change of voltage on P2 and take the base of Q1 positive. Once again, by regenerative action, augmented by the current drive through R1, Q1 will turn on. Consequently, a new "on" period will be initiated immediately after the stored energy has been transferred to the output capacitors and load. Complete energy transfer will take place irrespective of the loading or input voltage.

The transformer design is simplified, as there will be no DC component to consider in the design process, and the full flux capability of the core can be exploited with confidence.

There is a further protection mechanism should the core begin to saturate for any reason.

This saturation effect will be recognized by the increase in current in R4, and the "on" pulse will be terminated earlier. As a result of this action, there will be an increase in the frequency of operation such that saturation no longer occurs. This allows the designer to confidently utilize the full flux excursion ability of the core without the need for excessive flux margin to prevent saturation.

A typical plot of frequency against load for this type of converter is shown in FIG. 1.

Note that at very light loads, very high operating frequencies are possible. To prevent excessive dissipation in the switching transistor and snubber components, this high-frequency mode should be avoided by using the power unit for applications where the minimum load is not less than 10%. Alternatively, dummy load resistors may be applied.

The normal snubber arrangements and voltage clamps described in Part 1, Sect. 18 and 3.2 would be used. A transformer designed for the fixed-frequency flyback converter (Sec. 2.2) will be found to operate quite satisfactorily in this variable-frequency unit. However, some improvement in efficiency may be realized by making use of the extra flux capability and reducing the primary turns accordingly. To give good regenerative action, the drive voltage generated by P2 should be at least 4 V.

In the final design, extra circuitry will often be used to improve the overall performance; for example, a positive bias may be applied in series with the drive winding P2 to speed up the turn-on action, which would tend to be rather slow in the example shown. A square-wave voltage bias may be applied to the base of Q2 by a capacitor (shown dashed in FIG. 6.4) or a resistor to improve the switching action under light load conditions. This decreases the minimum loading requirements by reducing the switching frequency under light loads and thus reducing the current at which squegging occurs.

6. SQUEGGING

In this application, "squegging" refers to a condition in which a number of pulses are generated, followed by a quiescent period, on a repetitive basis. The cause of "squegging" is that for correct switching action at light loads, a very narrow minimum "on" period is necessary.

However, as a result of the transistor storage time, under light load conditions, the actual minimum "on" period will have more energy content than is required to maintain the output voltage constant. Hence there will be a progressive increase in output voltage as a number of pulses are generated. Because progressive control is lost at some point, the control circuit has no option but to turn off the switching transistor completely, and a quiet period now follows until the output voltage recovers to its correct value. With good drive design, giving minimum storage time, this "squegging" action will not occur except at loads below 2% or 3%. In any event, it is a non-damaging condition.

7. SUMMARY OF THE MAJOR PARAMETERS FOR SELF-OSCILLATING FLYBACK CONVERTERS

The component count is clearly very low, giving good reliability at economic cost.

The converter transformer may be designed to operate very near the maximum flux density limit, as the power transistor switches off at a well-defined current level. Any tendency to saturate is recognized by the control circuit, and the "on" pulse is terminated. (The frequency automatically adjusts to a higher value at which the core will not saturate.) This self-protecting ability leaves the designer free to use the maximum flux range, if desired, giving a more efficient power transformer with fewer primary turns.

The unit always operates in the complete energy transfer mode, so that by using current mode control, automatic protection for overloads is provided and performance improved.

(The current control mode is explained more fully in Part 2, Sect. 7 and Part 3, Sect. 10.) The complete energy transfer mode (discontinuous mode) avoids the "right-half-plane zero" stability problems. Additional windings on the main transformer will provide isolated auxiliary supplies for the control circuit, or additional outputs. The winding used for regenerative feedback can also be used to generate an auxiliary supply.

Input-to-output isolation may be provided by optocouplers or control transformers in the control feedback path.

A possible disadvantage of the technique is that the frequency will change with variations of load or input voltage. This should not be a problem so long as adequate input and output filtering is provided and the supply is located or screened so that magnetic radiation from the wound components will not interfere with the performance of adjacent equipment.

This type of supply has been used very successfully in video display units and has replaced fixed-frequency or synchronized power units in many applications.

8. QUIZ

1. What are the major advantages of the self-oscillating off-line flyback converter?

2. What is the major disadvantage of the self-oscillating technique?

3. The flyback self-oscillating converter operates in a complete energy transfer mode.

Why is this an advantage?

Top of Page

PREV. | Next | Guide Index | HOME