and
Now, assuming that
then,
Equation (2b) can also be written
SHORT-TERM DRIFT
The short-term drift of a supply is defined by the National Electrical Manufacturers Association (NEMA) as \change in output over a period of time, which change is unrelated to input, environment, or load [5].\of the material described in the section on temperature coefficient is applicable here as well. It has been determined experimentally, however, that thermal air drafts in and near the vicinity of the power supply contributes enormously to the short-term characteristics. The cooling effects of moving air are quite well known, but it is not often recognized that even extremely slow air movements over such devices as zeners and transistors cause the junction temperature of these devices to change rapidly. If the TC of the supply is large compared to the regulation, then large variations in the output will be observed. Units having low TC's achieved by compensation-that is, by canceling out the effects of some omponents by equal and opposite effects of others may still be plagued by these drafts due to the difference in thermal time constants of the elements.
Oftentimes, a matched transistor differential amplifier in a common envelope is used for the first amplifier just to equalize and eliminate the difference in cooling
effects between the junctions. Approximations to this
method include cementing
or holding the transistors together, imbedding the transistors in a common metal block, etc. Excellent results were achieved by the author by placing the input stage and zener reference in a separate enclosure. This construction is shown in Fig. 4. The improvement in drift obtained by means of the addition of the metal cover is demonstrated dramatically in Fig. 5.
Fig. 5. Short-term drift of a power supply similar to the one shown in Fig. 4 with and without protective covers. The unit was operated without the cover until time tl, when the cover was attached. The initial voltage change following t, is due to a temperature
rise inside the box.
Fig. 5. Short-term drift of a power supply similar to the one shown n Fig. 4 with and without protective covers. The unit was operated without the cover until time tl, when the cover was attached. The initial voltage change following t, is due to a
temperature rise inside the box.
If potentiometers are used in the supply for output adjustment (e.g., RI), care
should be used in choosing the value and design. Variations of the contact resistance can cause drift. It is not always necessary, however, to resort to the expense of high-resolution multiturn precision units to obtain low drift. A reduction in range of adjustment, use of low-resistance alloys and low-resolution units which permit the contact arm to rest firmly between turns, may be just as satisfactory. Of course, other considerations should include the ability of both the arms and the wire to resist corrosion. Silicone greases are helpful here. Periodic movement of contact arms has been found helpful in \
THERMAL DRIFT
NEMA defines thermal drift as \changes in internal ambient temperatures not normally related to environmental changes. Thermal drift is usually associated with changes in line voltage and/or load changes [5].\
Thermal drift, therefore, is strongly related to the TC of the supply as well as its overall thermal design. By proper placement of critical components it is possible to greatly reduce or even eliminate the effect entirely. It is not uncommon for supplies of the 0.01 percent(regulation) variety to have drifts of between 0.05 to 0.15 percent for full line or full load variations. In fact, one manufacturer has suggested that anything better than 0.15 percent is good. Solutions to reducing thermal drift other than the obvious approach of improving the TC and reducing internal losses include a mechanical design that sets up a physical and thermal barrier between the critical amplifier components and heat dissipating elements. Exposure to outside surfaces with good ventilation is recommended. With care, 0.01 to 0.05 percent is obtainable.
TRANSIENT RESPONSE
Most power supplies of the type being discussed have a capacitor across the load terminals. This is used for stabilization purposes and usually determines the dominant time constant of the supply. The presence of this capacitor unfortunately leads to undesirable transient phenomena when the supply is used in the remote sensing mode①. Normally, transistorized power supplies respond in microseconds, but as the author has pointed out [6], the response can degenerate severely in remote sensing .The equivalent circuit is shown in Fig. 6. The leads from the power supply to the load introduce resistance r. Is is the sensing current of the supply and is relatively constant.
Under equilibrium conditions,
A sudden load change will produce the transient of Fig. 7. The initial \is caused by an inductive surge Ldi/dt; the longer linear discharge following is the result of the capacitor trying to discharge (or charge). The discharge time is
where
and
The limitations of I,, are usually not due to available drive of the final amplifier stages but to other limitations, current limiting being the most common. Units using pre regulators of the switching type (transistor or SCR types) should be looked at carefully if the characteristics mentioned represent a problem.
①Remote sensing is the process by which the power supply senses voltage directly at the load.
Fig. 6. Output equivalent circuit at remote sensing.
Fig. 7. Transient response, remote sensing.
Fig. 8. Block diagram.
Preregulated supplies are used to reduce size and losses by monitoring and controlling the voltage across the class-A-type series passing stage (Fig. 8). Since the main regulator invariably responds much quicker than the preregulator, sufficient reserve should always be built into the drop across the passing stage. Failure to provide this may result in saturation of the passing stage when load is applied, resulting in a response time which is that of the preregulator itself.
SWITCHING PREREGULATOR-TYPE UNITS
The conventional class-A-type transistorized power supply becomes rather bulky, expensive, and crowded with passing stages, as the current and power level of the supply increases. The requirement of wide output adjustment range, coupled with the ability of the supply to be remotely programmable, aggravates the condition enormously. For these reasons the high-efficiency switching regulator has been employed as a preregulator in commercial as well as military supplies for many years. The overwhelming majority of the supplies used silicon controlled rectifiers as the control element. For systems operating from 60-cycle sources, this preregulator responds in 20 to 50 ms.
Recent improvements in high-voltage, high-power switching transistors has made the switching transistor pproach more attractive. This system offers a somewhat lower-cost, lower-volume approach coupled with a submillisecond response time. This is brought about by a high switching rate that is normally independent of line frequency. The switching frequency may be fixed, a controlled variable or an independent self-generated (by the LC filter circuit) parameter [7], [8]. Faster response time is highly desirable since it reduces the amount of reserve voltage required across the passing stage or the amount of (storage) capacity required in the preregulator filter.
A transistor suitable for operating as a power switch has a high-current,
high-voltage rating coupled with low leakage current. Unfortunately, these characteristics are achieved by a sacrifice in thermal capacity, so that simultaneous conditions of voltage and current leading to high peak power could be disastrous. It