Power supply filtering

[Roberto]

Well, Marshall still uses this configuration:
http://www.drtube.com/schematics/mar...03)%20iss7.pdf

[R.G.]

The idea to use little caps to bypass big ones is a consequence of the physical construction differences in caps. An ideal capacitor has an impedance that decreases with rising frequency all the way to infinity. The problem is that the inductance of the capacitor leads and plate structure increases with frequency all the way to infinity too. It was well recognized by RF practitioners back in the 1940s and 1950s that all capacitors eventually reached a frequency where the inductance of the leads and construction increased until it equaled, then ovewhelmed the decreasing impedance of the cap itself.

So all capacitors have a 'V" curve for impedance. They start out at open circuits for DC and from there their impedance is inversely proportional to frequency. So double the frequency, half the impedance. But their inductive impedance starts at very nearly zero, and doubles with frequency. At some point the inductive and capacitive impedances are equal, and the cap has an impedance of just its ESR. From there on up, it's an inductor and impedance rises. This was particularly bad for electrolytics because they were just strips of aluminum foil wound (!!) into a spiral. Modern electros use modified winding techniques and connector tabs to make this much better than it used to be.

The question of whether paralleling big caps with small ones is another place where the devil is in the details. Like with race cars - does a hand polished airbrushed finish showing flying flames make the car run faster? Yes, maybe. If the fancy paint job is super-polished and reduces air friction, it could. But in the driver's mind, it's HUGE and the car feels like it's flying. If little caps bypassing big caps reduce the power supply impedance at a frequency that matters to the circuit, it really does work better. If they just make the use think it sounds better - well, music is a feather of the brain, not the ears. I can make a case for both sides of this.

If the amp in question has a response up to frequencies where the baby cap's capacitance can help it, AND that frequency is in the audio signal range, OR that frequency is important for not causing feedback in the audio or near-ultrasonic range, it can help. Miss any of these, and the hand-painted flames just look pretty and count for bragging rights.

[catalin gramada]

Better to decouple HF as close by its stage as you can. Fit those small cap directly from cathode ground point to B+ point plate resistor.keep the wire short