Virtual Ground Confusion

[Tunemeister]

Hi,
Forgive the ignorance but I am getting confused by this one.
I am putting together a reverb unit which requires a -9 0 +9v supply & have an 18v dc supply which I was planning to use in conjunction with a TLE2426 splitter rail to give 0 9 18v using 9v as virtual ground. Can I do this or will there be a problem when I plug the guitar in via a chain of boxes all with ground at 9v. Similarly I will be coming out of the reverb & into an amp again with ground at 9v (assumed). I will be running the stomp boxes from separate power supplies but these will be -ive terminal to 0v ground.

Help!!!

[Mark Hammer]

The difference between circuits powered with bipolar supplies and those powered with single-ended supplies with a floating ground is what each "sees" on their input.

Since a virtually-grounded circuit has to "invent" a midpoint of signal variation, that midpoint is a steady DC voltage on top of which the signal is superimposed.  So, you will see the input/s of a chip connected to both the signal source AND the reference/bias voltage.  It combines them both.

Since that invented midpoint is a DC voltage, however, it is not the sort of thing you want to have snowballing error.  Consequently, what you will also see often are DC-blocking caps in the circuit all over the place, in tandem with rebiasing.  This is just to assure that the DC bias that was fed to the input doesn't magically turn into something a little weird on the output.  Best practices are to eliminate the DC virtual ground from the preceding stage with a blocking cap, and then re-bias.  It's a bit like a needle-exchange program; if you always inject with a fresh needle, you don't risk "catching" anything.  As a result of this need to bias, DC block, and re-bias, single-ended circuits tend to have higher parts counts than bipolar circuits.

This is the long way of saying that yuo can certainly run the circuit off your intended supply, as long as you do the requisite biasing and DC blocking.

The one caveat I would add is that not ALL semiconductors are happy with the same rail-to-rail difference any way they can get it.  Some ICs are comfortable with 18V on their supply lines but others aren't.  If there are any CMOS chips in the circuit, you will need to assure that they get no more than 15vdc fed to them or else they will be ex-CMOS chips.

[Tunemeister]

a chain of boxes all with ground at 9v. Similarly I will be coming out of the reverb & into an amp again with ground at 9v

Of course these should read 0v!

I was planning on using the Virtual Ground Driver (noise reduction version) from http://tangentsoft.net/elec/vgrounds.html

& the Reverb Driver circuit is the Stage Center Reverb from
http://www.generalguitargadgets.com/diagrams/stage_center_reverb_lo1.gif with a TL074 chip.

I am very much a novice at this with just a TS808 clone under my belt so far. That was just a case of following instructions whereas I am having to think about this a bit more & would be grateful for any guidance as to where the blocking caps should go, values & how to go about biasing.

Thanks

[toneman]

Since a virtually-grounded circuit has to "invent" a midpoint of signal variation

Maybe U should stop thinking of this as a "virtual ground".
Ground is ground and there can be no "virtual" ground.
Not, in reality.
In reality, U have a *midpoint*.   A voltage midpoint.
(most)Opamps need to have a midpoint for all of their biasing.
With a bipolar opamp pwrsupply(PS), the midpoint is Zero, and *this* is also (signal)gnd.
This is the easiest to understand.
With unipolar supply, U have to invent a PS "midpoint" for the opamp.
The opamp will not bias correctly nor work without it.
U could(and should)  call it Vref1.
This would be more a more accurate description/nomenclature than "virtual gnd".
After all, it *is* a reference voltage; not a gnd of any sort.\
afn
StayGrounded.... in your reference to reality.