AC Power Filter

ioplex · May 24, 2015

I need a simple DC supply filter for a low power JFET circuit. This is what I came up with:

Is there a circuit that provides better filtering in some way?

I don't recall seeing anything like this on the Internet so I'm wondering why this sort of circuit isn't used. Note that I don't need a regulator. I just need a DC supply that is as quiet as humanly possible for a high gain JFET circuit.

ruffrecords · May 24, 2015

KIS. Off the shelf fixed voltage regulator.

Cheers

Ian

joaquins · May 24, 2015

Maybe adding a resistor in parallel with the caps will make them sit at a lower voltage allowing higher ripple rejection, not needed if the ripple is not so big but also the entire circuit wouldn't be needed at all if that happens. Also I would add an extra cap at the output to provide faster response. In any case you still need your smoothing caps right after the diodes or all this wouldn't work at all.

JS

ioplex · May 25, 2015

ruffrecords said:
KIS. Off the shelf fixed voltage regulator.

I guess I should explain that I'm trying to get every volt out of a transformer rated for 22V at 0.175A. With the above filter and a simple JFET circuit that draws under 1 mA, the above supply actually reaches about 35V.

I did try an LM317 with the out pin connected to adj through a 270 ohm resistor but I got the same exact noise profile so I'm thinking my noise problems are maybe not from the supply.

I guess a high impedance, high gain JFET circuit on a breadboard is just asking for trouble. I think I need to put it on perf-board and maybe redesign the circuit. It's hard to get 35 dB of gain out of a single JFET and have decent current. Even with a 35V supply the current is only a couple hundred uA.

joaquins · May 25, 2015

You need to sacrifice a bit to get rid of the noise, or just add bigger smoothing caps. With your circuit you will be loosing some CE voltage, which is a bit higher than the BE voltage, so at least 1.5V. A LDO regulator will behave properly with that voltage loss, but you need in any case low ripple to start from in order to archive the desire result, so the most important to get where you want is start with big smoothing caps right on the bridge diodes.

Remember the capacitance multiplier (as you draw) multiplies the time constant or the behavior of the circuit but does not hold or multiply the energy stored in your cap, so you will need to have a big cap somewhere before those to provide the energy needed. later you could use any filter you want, LC for example works pretty good and does not loose any or little voltage for a given application.

JS