OK RG, a similar question if you have time...

[csmatt45]

What about power filtering. Some say, 100 ohm+100ufcap, red llama is 1K, 330 uf cap, some say 220 ohm 220uf cap etc....

thanks in advance everyone!

matt

[R.G.]

Once again, this is a matter of how much is enough. 

A power filter is designed to do two things. First, it is designed to present a low AC impedance to the circuit which uses the power. Second, it is designed to isolate that circuit from other possibly noise-producing parts of the circuit.

The classic low AC impedance is a capacitor across the power supplies. As I mentioned, this provides a low power supply impedance because the capacitor can supply required power pulses locally so they don't have to come from the remote power supply.The value of the cap depends on the frequency of the pulses of power needed (that is, the signal frequency) and the size of the current pulses. It can also depend on how susceptible the circuit being decoupled is to oscillation if the power supply isn't right. Some circuits are almost immune, some oscillate at the drop of a hat. Low frequency circuits need big capacitors. High current circuits need big capcitors. High frequency circuits need good capacitors with low impedance at high frequencies, something that electros don't do all that well, so ceramic is a better choice. Touchy opamps and other high gain circuits may need both good low frequency bypassing/decoupling and good high frequency decoupling, so you may need to put in both an electro and a ceramic.

Once again, the exact value tends to be somewhat arbitrary unless you know the circuit's application very well indeed, so putting in more than you can possibly need is good practice since caps are cheap.

The series resistor (or inductor, as in tube amps) is to give you some electrical distance from other parts of the circuit. This can be a rectifier/filter or just a noisy other section, like a switching section or a ticking LFO.  In this case, any noise voltage riding on top of the desired DC voltage is reduced by the voltage divider effect of the resistor/cap.

You can compute noise reduction if you know the noise frequency. For instance, in the full wave rectified supply you have made, the principal noise source is the 120Hz rectification ripple. There are other, higher frequency components in the ripple, but the biggest and lowest frequency - and therfore the hardest to reduce - is the 120Hz component. So if you use 1K/330uF, the impedance of 330uF at 120Hz is Xc = 1/(2*pi*120Hz*330E-6F) = 4.02 ohms. So the reduction is Vripple out/Vripple in = 4.02/(1000+4.02) = 4/1004 = 0.00398. The ripple is reduced to about 0.4% of what it was, or a 48 db reduction. Not bad for a resistor and cap, eh?

That's great. But there are also DC losses. Each milliampere of current through the 1K resistor loses you 1V of DC voltage as well. So you can't stand much of that and still have enough voltage to run your circuit if it was designed to run on 9Vdc.

If you use a 100 ohm resistor instead of a 1K resistor and that same 330uF cap, you get only 1/10 of the reduction, or the noise being 4% of what it was, -28db. That may be enough if your ripple wasn't all that big to start with.

On the other end of the frequency scale, if you have for instance a noisy rectifier diode that's making recovery spike transients once per AC line cycle that are 100nS wide, you have a different problem. That 330uF cap may only get down to 1 ohm of impedance before its internal inductance makes its impedance go up again with increasing frequency. Now as frequency increases, the filtering gets worse, and the noise spikes ride right through the series resistor, unattenuated.

For this instance you need a cap that has a relatively high capacitance, but also keeps that capacitance well up into high frequencies. Ceramic caps do this well. Film caps are OK-ish, but wound films have their own self-inductance issues, so cheap ceramic caps are the common choice. Ceramic caps usually have impedances that decrease right up until the self inductance of their straight lead wires becomes the limiting factor, usually at several hundred MHz. This makes a good choice of decoupling in general be a series impedance - resistors are cheap and easily available - a big, low frequency cap and a paralleled ceramic high frequency cap.

The variation of values of the resistor and caps reflects what the designer hopes will work. The "designer" in the case of effects generally has no clue what the actual conditions the effect will see, so they put in whatever they feel like, based on experience or wishful thinking. The designer (note the lack of quotes) of a high speed logic or radio frequency circuit in general knows, simulates, and calculates how much is enough, then doubles, triples or more that value.

[csmatt45]

Cool thanks. Sometimes the theory makes me feel like I'm trying to learn a new language, during a Metallica concert.

To be more specific with both of my questions, I've built a tyco clone. I was getting weird sounds and such when I would roll my volume back on my guitar. I was trying to figure out if the weird sounds were just part of the tweaked tyco sounds, or not. Finally I borrowed a 1998 Fulltone octa-fuzz to a/b it with. His did not have the same problems. I went over mine, everything checked out fine. I popped open his (fulltone), and everything matched component for component. His circuit board is glued down, so I couldn't get a look at the underside, but on top there were no differences. EXCEPT, I could find no Power Filtering, but his was more quiet than mine, and I could find no pulldowns, but his didn't pop. His also has a much thicker fuzz. I socketed my trannys and had a handful of MPS6519 and A18s and couldn't get that sound, and still had the weirdnes. I was useing Jeorge Tripps schem, so even tried flipping Q1 arourd. That quieted things a bit, but also drasticly reduced my fuzz.

I don't mean to saymine is unusable. It is functioning great in all other fashions, it's just the squeely thing, and the power supply issues.

To be a bit focused:

1) I get a squeely noisy thing if the pedals fuzz is maxed and I roll the guitar volume back.
2) could this be my 1M pull down? (FYI I was going, guitar-tyco-amp, no other pedals, and my strat has VERY low output pickups CS '54's) must be impedance right?
3) these characteristics worsen when I use a power supply, filtered with a 150 ohm in series+a 100uf cap+a 4001 type diode reverse biased, paralled to ground. It's not the power supply, works fine on the fultone. I know the next step would probably be to bypass my filtering and see what happens, I just didn't want to go cutting and unsoldering unless I had to.
4) How the @#$% is his so quiet, with NO filtering or pull downs

help

matt