Re: Mosfetový kapacitní násobič
Si tu nieco odlozim:
Capacitance Multiplier Using A MOSFET
A capacitance multiplier doesn't have to use bipolar transistors, but they will usually be the easiest option. The ready availability of good complements (NPN and PNP), low prices and ease of use mean that most people will use this option. However, a MOSFET version may be attractive in some cases, but initial testing (by simulation) indicates that transient performance is very poor - much worse than a circuit using otherwise identical values for passive parts but with bipolar transistors.
If the load current is steady and you can tolerate the higher voltage drop across a MOSFET, then by all means try it for yourself. There are MOSFETs with a low gate threshold voltage (VGS) and this reduces the power dissipation. However, these will almost certainly not be available in complementary versions (N-Channel and P-Channel).
The greatest advantage of using a MOSFET is that the filter section can be higher impedance, meaning that less capacitance is needed for a given hum attenuation. The trade-off will usually be higher dissipation, so a larger heatsink is needed. Heatsinks are bigger and more expensive than capacitors, so there's no economic benefit. Because the gate insulation layer of MOSFETs is sensitive to over-voltage, you also need to use a zener diode between gate and source to prevent damage under fault conditions.
A sample circuit is shown above. The time constants of the filter network are identical to those shown in the other examples, and hum attenuation is improved by around 10dB - but only if the load current is constant. As noted, the hum breakthrough when transient loads are applied is far worse than a bipolar version (by at least 20dB), and it lasts longer as well (around 500ms for the MOSFET, less than 200ms for the bipolar transistor version). These figures depend on the load current, amount of current change, rate of change (etc.), and should be considered as representative only.
Overall, the MOSFET version is interesting, and it may well be useful for providing a very low noise unregulated supply. However, it performs badly if the load current isn't constant, and it also dissipates more power for a given output current than an equivalent circuit using bipolar transistors.