A basic smoothing circuit is shown in figure P1, and it is in all sorts of trouble: The output voltage is at the mercy of the load impedance, which sets the discharge rate of the capacitor; a 10% variation in the mains voltage is also within specification. The next circuit will use a zener diode and series regulator to contain such effects.
A short-circuit protection feature is a must for all commercial equipment, but there is one aspect of it the site does not like. The fast acting fuse shown has prevented the circuit from sustaining permanent damage whenever I was careless.
If the zener voltage looks high for the target output voltage, mind the twin voltage drop across the base-emitter junctions of the Darlington. The Led is more than an indicator; also, it is a small load permanently attached to the output. It is intended to prevent damage to high-impedance, voltage sensitive circuits resulting from the leakage current of a transistor 'below spec'. The power transistor will be placed on a heatsink. It is often bolted on to the bottom of the metal case, if an insulting washer and cap are employed.
While this circuit minimises the effects of mains voltage and load variations, the base-emitter voltage of a transistor is an ill-specified parameter- and worse, it increases as the logarithm of the current does. You will sustain some 0.2 Volts of drop per decade of current range because of the Darlington. Therefore, this circuit is basically useful for loads which present a more or less constant impedance. Now the next circuit will reduce that figure.
A voltage comparator has been placed at the output; you will notice the different zener voltage. By the way, a supply of 5.5 volts is the acceptable maximum for TTL devices, which will then run slightly faster, if at a rather higher current, too.
In the prototype, a drop of 160 millivolts was registered for more than one thousand times increase in current, from 0.5 mA to 670 mA. The input capacitance was increased from 2,200 microFarads to about 5,500 µF (4,700 nominal) for this test. This figure assumes you do not increase the value of the 1,000 Ω resistor or reduce the value of the 4.7k resistor.
If you change the input or output voltage, you may have to adjust the 2.7 k resistor, so that the currents through both transistors of the differential pair are approximately equal. If you are pedantic, you may want to do that for your preferred output load. It is not a good idea to reduce the headroom between the input and output voltage.
All other items in the forthcoming table being equal, the ripple on the output is mainly dependent on the value of the input capacitor: The latter should be large enough to sustain your projected highest current during the low parts of the rectified mains cycle. If it is not, compensation will not work. Too large a value of the capacitor, on the other hand, appears to be a waste of money. The 100 nF capacitors are effective at high frequencies.
Incidentally, some people claim that an electrolytic capacitor which is not used at least once a month will lose 10% of its value. However, the site could not verify that: An axial capacitor of 2,200 µF nominal value has been found, which has been mislaid in a drawer for about 15 years. It was measured at about 3,300μF.
The (chiefly triangular) ripple voltage on the output was measured at less than 10 millivolts peak to peak. More details of the relevant conditions can be found in the following table:
| Secondary Rms voltage (Volts) | 8 |
| Secondary (DC) resistance (Ω) | 1 |
| Input capacitor (µF) | 5,500 |
| Output voltage (V) | 5.53 |
| Output current (mA) | 250 |
Both circuits have been powering a portable CD player for several years without ill effect. The discman consumes about 600 mA at power on, but only about half the current thereafter (excluding headphones). As they have not been tested extensively under such conditions, the circuits are not necessarily recommended for loads over one Amp.
Always measure the output voltage of a new power supply before using it. As multimeters only display average values, it is a good idea to monitor the output on the oscilloscope for the ripple voltage, too.
The input voltage will be at least two Volts higher than the output voltage for stabilisation to be effective. At the low voltages usually required of a regulator, this implies that an undue part of total power is lost on the stabilising transistor. For high currents, this is not acceptable. A switched mode power supply (SMPS) will be more environmentally conscious under such conditions. Or it may be the case of battery operated equipment in a remote location, or an implant.
Higher efficiency is achieved by using an input voltage which is much higher than the intended output and switching the Darlington fully on (or fully off) for part of the time only. It is plain the output voltage will have to be well filtered.
While the site has designed such a circuit and it has powered an old microcontroller board for a short period without obvious trouble, this certainly is not long enough testing time. The output filtering capacitor tends to run warmer than capacitors usually do, though this seems to be the case for other SMPS's, too. The site is hoping to present the circuit in about a year.
- Site map
- Jpeg Converter
For my own code up to this point: Valid XHTML 1.0!