Power Supply Sag Questions

[Electron Tornado]

I'm trying to get a better understanding of power supply sag. From what I've read, it's a drop in power supply voltage due to a large transient signal.

Here are a few questions:

1. How does a large signal cause sag?

2. How large, in relative terms, does that signal need to be to cause sag?

3. Is it possible to look at a schematic for an amp or a pedal and determine that it might be susceptible to sag?

4. How can sag (or at least the effects of it) be eliminated at the power supply? (filter cap, or make the supply regulated)

5. Can sag also be eliminated by modifying the circuit being powered?

[petemoore]

1. How does a large signal cause sag?
In an amp, the output lets out a 'batch' of power over a small time frame, and the 'saggy' power supply voltage at the plates drops for 'a time' while the rectifier refills the partially emptied ]of voltage] capacitors.

2. How large, in relative terms, does that signal need to be to cause sag?
Large enough to cause large currents which tap the capacitors faster than they can be refilled.

3. Is it possible to look at a schematic for an amp or a pedal and determine that it might be susceptible to sag?
Other than tube amps, or 'doctored to sag' power supply [somehow the power supply voltage is manipulated to drop when signal peaks occur], it would be mostly textual reference to the battery, ie 1/2 dead battery. A saggy power supply in a tube amp is cool, especially when the conditions actually cause sag to even happen or be heard.

4. How can sag (or at least the effects of it) be eliminated at the power supply? (filter cap, or make the supply regulated)
. Stiffen the supply capacitors ability to deliver larger currents than it expects, or use a voltage regulator. Simply increasing the supply voltage could make sag disappear or be less apparent.

IIRC The Tone God was it, won a contest with a 'sag circuit' which is worth checking out, and something about an envelope filter [from Nurse Quacky or Dr. Quack] wired to cause an adjustable voltage regulator response [voltage drop] instead of frequency sweep. Worked great with everythiing dialed in, might build a couple more [takes some dialing in of input, voltage drop, bias of the circuit it's applied to etc.,  but something like a Jfet or Ge HF compressor, cool sounding fuzzes etc. Peak signals being 'de-peaked' to positive/interesting results...such as accentuated drop in ability to channel high frequencies

5. Can sag also be eliminated by modifying the circuit being powered?
  Make it run on less current, or a fraction of the voltage...
  There is no perfect DC power supply, draw sufficient current and the voltage part of the ohms law equation exponent becomes proportionately larger.
  The trick to making sag un noticeable is to make sag small enough that it isn't detectable.