Does a virtual ground use headroom of opamp?

[Max999]

Ok, my basic question is stated as the title. I know that the audio ground is always relative, and doesn't has to be at 0V.
The thing that got me confused though is thinking about the difference between a split supply ( plus and minus) and a virtual ground, and a "stacked singe supply". The example is about 18v circuits:
Split: -9/+9, ground at 0
virtual ground: 0/18, Vref at 9V
"stacked single supply" ( like stacking two 9V batteries, and connecting the +9 and 0V in the middle for ground): 0V/18v, ground at 9V

Does having the ground or Vref at 9V makes the opamp have half the headroom at input and output .. please enlighten me!

[GibsonGM]

Well, you can think of it this way: your signal can ride from whatever point you set as ground to the upper and lower power rails...minus a bit for most opamps, of course (the standard ones can't go all the way out to the rails, altho you can buy some models that come d@mn close).

So, if you have -9V and +9V below and above the reference point of 0, your signal can swing -9 to +9 (again, not quite all the way due to opamp limits).   The 0 to 18V, ref. at 9, does the same thing as far as your signal is concerned...9V above and below the reference.

The thing with 2 batteries is the same as a bipolar supply...and the 0 to 18V with ref. at 9 accomplishes the same thing...

Oh, welcome to the forum, good question   

[PRR]

For fun, I hop up and down 16 feet. (??)

I need a room 18 feet high.

Does it matter if it is built 9 feet up and dug 9 feet down from ground level, or is all above ground 18 feet?

The choice is just about what's more convenient to build, dig&stack or all-stack. Once I am INside the room, all I know is I got my 18 feet to hop in.

[Max999]

Glad on be on board of this great forum GibsonGM.

The follow up question is: why don't I need an output capacitor aften an opamp operated with the stacked batteries supply ( or bipolar) when the ground is at 9v?

[GibsonGM]

I thought you did need one, Max...can you clarify, like posting a schematic you see this in or something?   

When you bias a device with DC, you must isolate that from whatever comes before or after it (there are times you don't, you can USE that DC to bias the next stage, but let's keep it simple and general).    The cap blocks DC and lets AC (your signal...) thru.    If it's not there, the DC will wreak havoc down-stream, like making output pots scratchy and is generally very undesirable...

[Max999]

At https://tangentsoft.net/elec/vgrounds.html you can find a page related to referincing opamps. The first image shows a good example: Vground is at 9V measured from V- .. but if if you take Vground as the center of your measurement, V- is at -9, and V+ is at +9. Image is shown below too.

I am under the assumption that dual power supply fed opamps ( +/-) don't need an output cap because only a small DC offset, caused by the op amp, will appear at the output.

So why is a virtual ground circuit different as we might as well call +9 volt for example "ground", because "ground" is relative.

I am clearly missing something because all virtual ground circuits I see use coupling caps, but I am trying to understand if changing my stompbox designs to dual supply, or the easier to do center tapped stacked isolated supplies will add any sound quality.

[GibsonGM]

I see what you are saying now.  Oh, you may not "need" the output cap, but it is good practice to include it, as only a few mV are required to make things scratchy down the line...you'd just 'pretend' it's the same as all the other circuits you're looking at.   If ALL your stuff was run on that dual power supply, in theory you'd always have references etc. at zero volts...so yup, ground.   But there can be variance....battery internal resistance and so on.  I think you'd find that zero volts may be shifting around just a little.

In practice, many tube amps don't have an input cap....I add one anyway (it does zero harm), in case an FX pedal is leaking DC...they can do this if one of THEIR output caps gets damaged etc.   I don't want ANY DC getting in there and messing with my bias....

A real 'virtual ground' circuit (let's say 4.5V Vref as that is so common) WILL have an offset to deal with, and that is how it's different. 

Question: why are you running on 18V, + and -  , when you can use 9V and gain all of the simplicity of it?   I'm not aware of any real 'sound quality' benefits.  What you'll get that way is HEADROOM - a whole lot more space to do what you want to do without clipping.  But most of us don't need it - it would be related to very clean sounds, usually.   Studio equipment...things that need ultra clean processing.....

[Max999]

Well, I am trying to get all my gain from tubes and keep the effects as clean as possible. I have done quite a few experiments with different supply voltages, and also by making the bipolar supply by stacking two isolated power outputs .. I have even been running at 36V.
There is something that I cannot explain: every voltage sounds different. 9 volt is thinner, higher voltages produce more bass but also make the mids more tinny ( or more real, depending on whats going on).
Every circuit, in this case buffers, is listened to by headphones at output, guitar straight in.

I would like to understand why the supply voltage applied changes the sound, so I am trying to see which parameter of the opamp is violated .. when sufficient headroom is reached all supply voltages should sound the same!

So with this thread I am also trying to get information if the biasing method can be the reason for this, and if I have to resort to building an actual dual power supply instead of a "stacked one" to see if this gives a consistent sound using different supply voltages.

The design idea for this buffer is: keep it totally transparant .. totally clean. I never thought a simple buffer circuit could be so hard to get right.

[GibsonGM]

That's probably above my pay grade, ha ha!  You should talk to PRR, who has done a lot of work in similar areas.  Yes, I think you're on to something about opamp parameters....ESP (Elliot Sound Products, has great web site) goes into this quite a bit, you might want to check them out! 

I've rarely gotten more advanced than a +/- 9v supply, myself...I've never felt our equipment was hi-fi enough to warrant it, and I don't like 'special' requirements and so on.  But your requirements may vary! 

[Max999]

Well thank you for taking the time to explain me a few things you know. The esp site is good, so far what I have seen of it it is still understandable for a guy like me who wasn't really interested in math at high school 
I am surely not afraid to give my equipment "special" requirements if it gives me the tone I have in my head at that time ( moving target), and it is fun to learn about electronics doing it. It's so cool that we as electric guitarists can influence our tone so much by external factors.

[antonis]

Components values DO matter.. 
(e.g. IN & OUT caps, voltage divider & bias resistors, IN & OUT impendances, etc..)

Every passive component forms (by itself and/or in combination with something before or after it) a signal attenuator which is frequency dependable in aspect of attenuation degree together with possible phase shift..
(even a humble DC voltage divider does it with attenuation ratio of R2/(R1+R2) and zero phase lug..)

[Max999]

Ok, but when using the exact same components, and only changing the supply voltage, with all supply voltages enough to not clip the opamp, how can that make the sound change?

[amptramp]

Ok, but when using the exact same components, and only changing the supply voltage, with all supply voltages enough to not clip the opamp, how can that make the sound change?

The input stages are run from a current source so the input stage will not know anything about the change in voltage.  The output stage of an op amp is run directly from the rails so as voltage goes up, collector to base capacitance declines and the highs come in better.  Higher voltages will mean more power dissipation meaning higher internal temperatures and more gain from individual transistors within the op amp.  All of these are undocumented second-order effects (meaning a change of voltage does not make anywhere near the same or even a predictable change) so they cannot be relied on as a design feature, but they will be there.

[Max999]

Wow that is an awesome reply amptramp. I thought I understood that opamps were designed to have the lowest change in sound. For instance, the opamp I use now has a THD+N of 0.00005%, and a IMD of about 0.0007% ( depending on test situation the IMD varies).
They spec this behaviour for supply voltages between 4.5 V to 36 (±2.25 V to ±18 V).

It puzzles me how they can spec such low error numbers for the opamp when the change in sound is easy to hear for a somewhat trained ear.

[Transmogrifox]

My first gut feeling is to ask to see a schematic of your actual test setup (what is the headphone amp, how are amplifier stages coupled together etc).

The specs on the datasheet tell me there shouldn't be an audible error once you get headroom beyond your signal's limits.

At the same time, these specs don't cover all the corner cases or specific application...but...many external factors can interact with the way the op amp works (like what load resistance and capacitance you are driving may or may not be on the curve for spec'd operation) (are you driving a cable directly from the op amp output?).  So many external factors can play a role in the sound and even though the op amp might maintain its specified operation under one condition it may be things like PCB layout and external components have an effect on things.  For all I know your "buffer" might be unstable at 50 MHz and you have a completely different beast than what the datasheet specs.

[Max999]

I hope I found the answer: I decided to make recordings of the circuit at different voltages to check if the 32 ohm headphone load was messing with the opamp.
I will do this experiment again because I don't want to jump to conclusions too quick but the sound of the circuit at different voltages were, when above 9v, the same! 9v was a little different, but, ok.
The 2K minimum load mentioned in the datasheets seems to be there for a reason 

I hope you guys can check the schematic anyway, to see if I made any mistakes.
Do I really need the 150R after the Vbias buffer or will this buffer never oscillate anyway under capacative loads?
The small 0.01uF ceramic for bypassing the power rails as close as possible to the opamps, does this go like in the schematic from V+ to ground or to Vbias? I only need one 0.01uF capacitor for a virtal ground circuit right?

the buffer/ splitter

[PRR]

> if the 32 ohm headphone load was messing with the opamp.

32 Ohms is a SEVERE load on most/all "opamps", and sure would explain different sounds.

Use opamps for 2K and up (includes all guitar-cord work).

Use a proper headphone amp to drive headphones.

[antonis]

Do I really need the 150R after the Vbias buffer or will this buffer never oscillate anyway under capacative loads?

IHMO, you can't predict it so let it be there..

The small 0.01uF ceramic for bypassing the power rails as close as possible to the opamps, does this go like in the schematic from V+ to ground or to Vbias?

All single supply bypass caps should be tied between supply rails (i.e. Vcc & GND or Vee & GND in case of negative supply)
For bi-polar supply, you can find them tied between Vcc & GND and Vee & GND or just between Vcc & Vee..

I only need one 0.01uF capacitor for a virtal ground circuit right?

Actually, you don't need any 10nF - 100nF ceramic cap for virtual ground..
Some people like to place a 10μF - 47μF electro cap on Vref output (U4 OUT) and some other don't..

[amptramp]

I don't see the need for R10 when your only load is R2.  As PRR said, op amps were never intended to drive a 32 ohm load.

[ashcat_lt]

Does it matter if it is built 9 feet up and dug 9 feet down from ground level, or is all above ground 18 feet?

Well, I mean, atmospheric pressure and gravitational pull will be ever so slightly different. 

Also, are you only 2 foot tall?

Sorry.  I'll stop.