Tying filter to ground or Vb

[strungout]

Oy.

I'm piecing up together a distortion design and I have a questions:

What's the difference, if any, between tying a filter to ground or Vb (read it was called "virtual ground")? I tried both, but didn't hear any change. I was also wondering the same about clipping diodes.

Ciao.

[Mark Hammer]

Vb exists so that there can be a positive and negative half cycle with some headroom in a single-ended power circuit.  

In some parts of the circuit, you don't need to "float" the signal midway.  For instance, once the signal comes out of the last active stage and goes to, say, a passive tone stack and output volume control, everything is references to actual ground rather than Vb because you don't need to "add 4.5v" to whatever is going on to make it work right.

Another instance where the reference point would be different is when the bias voltage from an earlier stage is carried through to a subsequent stage, rather than eliminated with a DC-blocking cap.  Perfect case in point will be chorus or other BBD-based effects where a bias voltage is added to an op-amp input and carried through the lowpass filter after it to the BBD input.  Because a lowpass filter has caps to *ground*, the signal path from the op-amp to the BBD is essentially a couple of resistors and a transistor, none of which are expected to eliminate the DC voltage that the signal is riding on.

The need to rebias, I suspect, is tempered by the tendency of the op-amp being used to introduce a DC offset.  If the DC offset coming out of the op-amp is something other than V+/2, then the smart thing is to use a DC-blocking cap at the output of stage N and rebias for stage N+1, rather than assume the Vb can be carried through to serve another stage.

[strungout]

Thnx, I think I understood most of that Thnx for the explanation.

Maybe it was because I was too tired yesterday, but I tried switching my clipping stage refferences from Vb to ground again, and if the first stage is tied to ground and the second to Vb, I get some added hum. Both tied to ground is still a bit noisier, so I'm leaving them on Vb.

Ciao.

[Jay Doyle]

A case where this does sort of matter is the case of the "arm" of a non-inverting opamp amplifier configuration.

The resistor and cap that string from the "-" input are seen going to both ground and Vb. In simulations and in my own listening tests putting it to Vb allows for greater gain.

I guess what is going on here is if the cap is tied to ground instead of Vb the cap has to constantly be recharged because the "-" in put is always at Vb.

....Yikes, there is some kernel of truth somewhere in the above, but I am wracked out on painkillers for a bum wisdom tooth so you might want to ignore what I said until someone with a clearer head can clean it up...

[Mark Hammer]

What confuses me about that, Jay, is that just about all the time there is a cap on that "arm" of the circuit, which would necessarily block DC.  So how could Vb play any role in that part of the circuit?  Indeed, the primary function of that "arm" is to serve as one half of an attenuator to reduce the amount of feedback from the output to the input of the op-amp.  In which case, it's primary function is to simply "bleed" signal to ground.  Or at least that's my understanding of it.

As with most things, it's usually more complicated than "what I understand of it".

[R.G.]

What confuses me about that, Jay, is that just about all the time there is a cap on that "arm" of the circuit, which would necessarily block DC. So how could Vb play any role in that part of the circuit?

The reason it might matter is that Vb is not perfect.

The common resistor divider in effects is a pair of 10K to 100K's from +9 to ground, filtered by an electro cap of 1uF up to as high as 100uF. At DC the impedance is the two resistors in parallel, but as soon as you get over the RC constant for the resistors in parallel and the cap, only the cap matters.

But the cap is not perfect. If you have a 1uF cap and a 0.1uF cap in the "arm", then the 1uF cap on Vb adds 10% to the arm cap value and you probably can hear it. With 10uF to 100uF, this effect goes away... except for the fact that the 100uF is not acting like a cap anymore above about 5kHz or so.

An active VB is much better: either an opamp buffered Vb or the LM386 Vb I posted a ways back have output impedances in the units-of-ohms or less all the way up to high audio frequencies, so there is much less difference between ground and Vb.

In-the-rut effects design has some almost unspoken rules, and minimalism is one of them: never put in a fancy part if you can do it cheaper, or dirtier or with fewer parts somehow. That leads to lower costs, simpler builds - and dirtier power supplies and reference voltages. Sometimes it works fine, others not.

[puretube]

in spite of minimalism where neccessary, there are manufacturers
nowadays, that are aware of the benefits of that quarter quad-opamp
low-Z Vb supply...  :wink:

[R.G.]

....shhhhhh... don't let the others hear...