An elusive transformer problem m PWM systems

Pulse width modulation is a popular way of controlling power when working from a dc source. This method is widely used in regulated power supplies, in electric vehicles, and in other motor-control systems. Load power control is accomplished by varying the on or off time of a switching device, then averaging or filtering the chopped waveform. High operating efficiency can be realized for the simple reason that the switch is cither in its on or in its off state of conduction, with minimal time allotted to the transition between the two conduction states. A peculiar trouble often arises because of a common oversight in the driver transformer for such a PWM switch.

These days, relatively few engineers have had extensive exposure to the theory and design of transformers because training focus has been on computer themes. Moreover, even those who learned about transformers used texts that generally confined discussion to sine waves, and sometimes square waves when electronic circuits were studied. The insights presented by such constraints are fine but lead to invalid assumptions when the duty cycle of the square wave is other than 50 percent. Thus, in a PWM waveform, where the duty cycle is constantly varying, transformer operation departs from both sine-wave and symmetrical square-wave behavior. This is where the mysterious poor performance of PWM systems often asserts itself, and the reason can be quite elusive if you do not know what to look for.

The first problem deals with sine- and symmetrical square-wave operation of a transformer. Assuming a one-to-one ratio between the windings for simplicity, the voltage waveform across the secondary is, essentially, a replica of the waveform impressed across the primary winding. For a square wave, a certain amount of distortion can be visible at low frequencies, but the symmetry of the secondary waveform remains intact. The dc component of the primary waveform, if any, is lost in the secondary because a transformer cannot transfer dc. This is usually understood and need cause no trouble.

The second problem when the duty cycle at the primary is no longer 50 percent, the amplitudes of the positive and negative excursions of the secondary voltage are no longer equal. If, as is often the case, a power MOSFET is being driven from the secondary winding, the enhance­ment voltage for the gate will vary with the duty cycle. Thus, the power MOSFET is likelyto operate from hard-driven to under-driven conditions. In the latter case, its dram-source voltage drop will increase, giving rise to faulty operation and excessive power dissipation. Unfortunately, if diagnosis is made under the wrong duty-cycle drive conditions, the reason for the poor performance might remain concealed.

If a wide range of control is required from a PWM format, something must be done to keep the amplitude shifts of the secondary voltage in the drive transformer within bounds. Otherwise a driven power MOSFET can either be destroyed by breakdown of its gate insulation, or can be subjected to overheating because of insufficient gate drive. A brute-force remedy is occurred. Here, the positive and negative excursions of the secondary voltage in the driver transformer are both clamped at approximately 16 V by zener diodes. (15 V of zener breakdown voltage plus 1 V due to forward voltage drop in the alternate zener diode results in an effective clamp of about 16 V.) Most power MOSFETs saturate in the vicinity of 10 V applied to the gate but tend to develop a lower R& at a somewhat higher gate voltage. In any event, it is usually necessary to keep the gate-source voltage well below twenty volts in either polarity.

An objection to the technique of the brute-force remedy is that it dissipates drive power, making greater demand on the transformer size and the driver circuit. After all, one important reason for using the power MOSFET as a switch is that relatively little drive power is needed to put it through its paces. With this in mind, a more sophisticated approach can be achieved. Inspection reveals that something is done to the primary, rather than the secondary, circuit of the transformer in order to make the positive and negative induced voltage independent of duty cycle. This is realized through the operation of a logic circuit that presents the primary of the transformer with a symmetrical bipolar waveform while being actuated from a unipolar PWM source.

Note that when one gate is on, the other is off, and there is no possibility of overlap of their conduction states. Suitable logic gates can be used from the D469 1C, which actually comprises four independent gating circuits. An additional secondary winding on the transformer can be used to drive a second power MOSFET of a totem-pole output stage. A negative power supply would then also be needed. This scheme can provide trouble-free operation over a wide duty-cycle range because the transformer always sees a true ac wave. (4900)

Перескажите преподавателю данный текст по-английски, пользуясь своим планом.

Дата добавления: 2015-07-20