| ??? 11/29/08 11:47 Modified: 11/29/08 11:54 Read: times Msg Score: +1 +1 Informative |
#160404 - Big caps at output were used in the past Responding to: ???'s previous message |
Richard said:
If you're worried about the Vcc-to-Gnd capacitance damaging your regulator, you should consider putting a diode in reverse bias (from output to input) on your regulator, so that if the power goes down the capacitance on Vcc won't reverse bias the regulator. Problems can occur when repeatedly switching power on and off, particularly if there's a bleed resistor on the unregulated power supply's capacitor, as the resistor will discharge that cap once power is turned off, while the Vcc-to-Gnd caps may discharge more slowly, though that can occur any time power is turned off. The reason for these big caps that can sometimes be seen at output of 78XX-regulators is the proper design practise in the past with discretely built well known "3-leg-regulators": http://de.wikipedia.org/wiki/Bild:Ser_stab.svg Here a big cap at output indeed helped to filter out noise and hum (ripple) and to improve the load regulation. Just because of the fact, that these old discrete 3-leg-regulators showed rather high source impedance, where the big cap at output worked like a powerful low pass filter. When the first 78XX-like integrated 3-pin-regulators appeared in the scene, engineers also used big caps at output of these regulators, but which isn't actually needed, because 3-pin-regulators provide very low source impedances by itself, due to the internal multi-stage design of regulating feedback loop. So, you won't see an improve of regulation charactertistics that is worth mentioning, when changing the output cap from 47µF to 1000µF or even more. This reverse diode from output to input is mainly needed, when the input can be short-circuited to ground, while the output caps still holds a high output voltage. But normally there is a big storage cap at input of regulator, which cannot be short-circuited due to design, because it's permanently connected to the rectifier. And normally this storage cap is much larger than the output cap, so that the input potential is always higher than the output potential. Also, there's already an internal pn-junction from output to input on the die of 78XX/79XX-series, which can serve as protection for little currents. Higher currents need the external reverse diode, though. Richard said:
As for switchers ... I doubt you'll ever find one that has sufficiently low output noise/ripple to warrant their use without a downstream linear regulator and considerable filtering on both sides of that in order to provide sufficient line and load regulation. Hf-noise isn't a problem usually, because nearly always a LRC-lowpass-filter can be used to clean the output voltage. But I have seen huge 120Hz ripple at output many times! A good alternative is to use a switcher as pre-regulator (LM2674 e.g.) and a 78XX-regulator for the fine regulation. The pre-regulator to minimize the heat dissipation in the 78XX-regulator and the 78XX to provide a stable, noise- and ripple-free supply voltage. To take all tolerances of involved parts into consideration I design the input voltage of 78XX-regulator (which is the output voltage of switcher) to be 3.5V higher than the output voltage. This gives a very powerful supply modul of brillant performance. Kai |
