what does a cap across the power lines do ?

[Hal]

i see it in lots of circuits...whats it for ?

[phillip]

You mean something like a 22uF - 100uF capacitor from +V to ground?  Those are there to help further filter the incoming power to help reduce hissing of humming, especially in high gain circuits (like in the ToneBender MKII).

Phillip

[Hal]

cool.  Thanks :-D

[Ge_Whiz]

So why is it that when op-amp circuits use a resistive divider to give a Vb supply mid-way between the rails, folks put a big capacitor between Vb and earth, but never between Vb and +ve? I've never understood this.

[mikeb]

The idea is that the cap acts as a (small) local energy store that takes up the slack caused by transient current demands the 'real' DC supply can't respond to. Circuits have different current draws from their power supply over time. In the case of the resistive divider you want to (try and!) make sure that Vb is moderately constant relative to ground. In some circuits a simple local current store isn't enough, and you need to move to the next stage where actual active regulation (not 'passive' regulation as implemented by the capacitor) is needed - this can be done easily by connecting the middle of the resistive divider to the non-inverting input of an opamp, connecting the inverting input to the output (to form a unity buffer) and using this output as the Vb. It's also possible to add a decoupling cap (or two) to this output to further improve the supply characteristics vs current draw.

Mike

[Ge_Whiz]

So, in a resistive divider circuit, it's more important that Vb stays the same relative to earth, rather than maintaining mid-way between earth and supply voltage, as two equal resistors alone will ensure? I'd always assumed that the mid-way criterion was the intention... :?

[mikeb]

In theory, if the resistors have the same value, then the voltage differential between V and Vb will be the *same* as that between Vb and ground.

Mike

[R.G.]

i see it in lots of circuits...whats it for ?

To lower the effective impedance of the power supply.
Your next question is undoubtedly - uh... what is lowering impedance doing?

Power supplies should be perfect voltage sources - that is, you can pull out any amount of current, no matter how little or much, and the voltage stays the same. No load will change the voltage. If we use Ohm's law, we can compute that the resistance inside the power supply is the change in voltage divided by the change in current. But we just said that the voltage doesn't change at all, so the resistance we calculate is always 0, because the voltage never changes no matter what current is supplied. So another way to say "perfect voltage source" is to say "zero (or low) impedance voltage source.

This matters because if any of the loads change the power supply, the changes (usually signal currents) get coupled back to earlier stages, where Murphy's law says that it will cause uglinesses to happen.

Unfortunately, no power supply is perfect. There is some internal impedance. You can think of this as a resistor just inside the battery terminals that you can't get at. Capacitors let us fake out a perfect battery by storing electric energy at an **almost** constant voltage. They get better as frequency goes up. A capacitor is a good way to make the power supply look like a low impedance at signal frequencies even if it has some DC internal resistance. The capacitor soaks up energy from the power supply and then lets it out to even out the voltage. The higher the frequency and bigger the cap, the better it works.

Since a power supply capacitor is there to short AC signals, intentional or otherwise, to both side of the power supply, it prevents any power supplied carried coupling from one stage to the other, leading to the term "de-coupling capacitor". It literally prevents coupling from being carried through the power supply.

So why is it that when op-amp circuits use a resistive divider to give a Vb supply mid-way between the rails, folks put a big capacitor between Vb and earth, but never between Vb and +ve? I've never understood this.

So, in a resistive divider circuit, it's more important that Vb stays the same relative to earth, rather than maintaining mid-way between earth and supply voltage, as two equal resistors alone will ensure? I'd always assumed that the mid-way criterion was the intention...

Both of these rely on a subtlety - if the power supply is low impedance as it normally should be, then you can fix any intermediate point to either side, and it's the same as fixing it to both the high and low side.

In single supply systems, all the voltage measurements are made to the nominal "ground" lead, so it makes sense to decouple the Vb to the ground side. You can do exactly the same thing with two capacitors for Vb, one to +V and one to ground, but that takes another cap. Also, since ground is assumed to be perfectly 0V (which may or may not be true, depending on how you do your measuring), then using a decoupling cap to fix the reference voltage to the side assumed to be perfect makes sense.

[Ge_Whiz]

Okay R.G., I get that - thanks. But in the case of a resistor divider splitting the incoming voltage, doesn't it make more sense to decouple the incoming supply between +V and earth? Surely, then, no further decoupling is especially necessary for Vb?

[Tim Escobedo]

Those resistors in the resisive divider make a high impedance reference voltage (Vb). Even if the main supply is stiff. Thus the need for a cap for the reference, to stabilize things.

You could experiment to find out the effect of the cap. Wire up a high gain op amp circuit. Note the behavior with and without the cap there. With low gain circuits, the cap may not be necessary. And different op amps are worse than others.

[csj]

Um...take this with a very large grain of salt. I am still unfamiliar with much of the world of current flow in SS circuits.

I think the culprit here is the common impedance of the single power supply. The ac currents of each stage will flow through the power supply so the internal resistance of the PS will act as a common impedance for each. No problem with the currents which are out of phase but the ones which are in phase will add up and cause signal oscillations in other stages.
Several names for it.
Modulating the power supply...
feedback...
motorboating...
really bad farting sounds

We want to "mismatch" the impedances in a way allowing for a low reactance route around the PS and a way to reduce the ac current through it. This is important in tube amplification where the ac plate current will absolutely play havoc on successive stages if it is not properly isolated from the power supply. I imagine that this is close to what is happening here.

I think what RG is saying is (amoung other things) ...how you alter the impedance for the Vb supply... Vb to + or Vb to - doesn't really matter. The circuit doesn't care so long as it's done. In this case why not connect it to the "0" side just to make it easier for voltage measurement.

doesn't it make more sense to decouple the incoming supply between +V and earth

This just simply wouldn't provide for the necessary impedance variation. It will alter the the total impedance path to the power supply but it won't create the different path necessary for isolation.

[Ge_Whiz]

Now I get it - thanks to all who explained this to dull li'l ol' me.  :oops: