Split Supply Power - I am Officially Stumped

[dano12]

Ok, this is the way I've always done and drawn power supply splitters to bias an opamp. And to the right is an opamp buffer that uses this. The key point is that the inputs and outputs are grounded to the negative side of the power supply/battery.



In researching virtual ground circuits and opamp stuff, I came across this version of a split power supply:



In this version, what used to be my "vref" is now marked as ground. The resulting opamp buffer is shown.

But here's the deal that I cannot wrap my head around: this means that what I thought were "grounds" for the input and output jacks are actually vref.

This second version comes from the popular cmoy headphone amp, details here: http://www.headwize.com/projects/cmoy2_prj.htm

Argh! I've wrapped my brain around this for days and am obviously not seeing the whole picture.

[Auke Haarsma]

the site you refer to does use a dual power supply:



But it looks like it is missing a ground connection between the two batteries. For comparision, look at R.G's take on this with his humfree a/b/y:

[Auke Haarsma]

uhm, on second thought... I doubt if it makes sense what I said... those circuits are not the same..

[dano12]

uhm, on second thought... I doubt if it makes sense what I said... those circuits are not the same..

No worries.

I guess what I don't understand is that in the 2nd version, +4.5vDC is connected to the ground ring of the jacks.

That means the circuit would be putting out 4.5v on the ground?

So stumped....

[R.G.]

Ok, this is the way I've always done and drawn power supply splitters to bias an opamp. And to the right is an opamp buffer that uses this. The key point is that the inputs and outputs are grounded to the negative side of the power supply/battery.

That version of the Vref splitter is incorrect and will give you some misery in the form of debugging. It is quite important that the capacitance be from the Vref to the actual power ground. As it is, you have the equivalent of 5K between a low impedance point and all of your signal grounds. This means that all of the places your signal should be at 0V is actually alive with the mixed combination of all the signal currents across 5K.

It is possible you meant to draw C1 as connecting to the middle of R1 and R2

In researching virtual ground circuits and opamp stuff, I came across this version of a split power supply:
...
In this version, what used to be my "vref" is now marked as ground. The resulting opamp buffer is shown.
But here's the deal that I cannot wrap my head around: this means that what I thought were "grounds" for the input and output jacks are actually vref.

You're not thinking about impedances, only DC voltages.

Think about a capacitor. Its impedance declines steadily with frequency (to a first order at least) so that at some frequency we can consider it to be a short circuit. You can tie any point to "ground" with a big enough capacitor, and to AC signals it is a short circuit. However, it will support a DC voltage difference. We are free to call any point "ground" that we like. But we had better make that point be one with a low impedance to the other points we want to call "ground" if we are to connect it to the outside world "ground" and pass signals back and forth.

[ayayay!]

It is possible you meant to draw C1 as connecting to the middle of R1 and R2

Or also an additional cap...?

Dano, did you also catch Vcc on the Virtual Ground Version is +4.5v, not 9v?  (I think you did, just checking...)

[R.G.]

I guess what I don't understand is that in the 2nd version, +4.5vDC is connected to the ground ring of the jacks.
That means the circuit would be putting out 4.5v on the ground?
So stumped....

That's because you're not getting the idea of "ground". For an isolated circuit, we can call any point at all "ground"; in isolated circuits, no interconnection to anything else, "ground" is only the place you call zero volts and measure everything else from.

When you hook something else up to the previously-isolated circuit, it now matters what point you call "ground", because both "grounds" will become the same voltage by connection of the two by wires.

For the circuit with the midpoint of the 9V battery connected to the ground ring of input and output jacks, you have defined the zero volt point as the middle of the two resistors because you're going to connect the ground rings of the two jacks to other stuff. So no, there is not 4.5V on the "ground". There is zero volts on the ground, and the opamp's negative pin is sitting at NEGATIVE 4.5V, not zero volts.

Another way to look at this is that the negative terminal of a 9V battery is not zero volts. Zero volts is whatever you connect the meter black lead to.

[ayayay!]

What happens if you remove the resistors on the Virtual Ground version?   

[dano12]

It is possible you meant to draw C1 as connecting to the middle of R1 and R2

Yep missed that part on my redraw. C1 is as is, there is another cap from vref to ground.

I'll now attempt to digest all that you posted.

Thanks so much.

[Auke Haarsma]

Here's some more to digest. I have bookmarked threads related to several topics of interest to me. Ground / PSU is one of them. Hope this helps!

im confused about dc- and grounds
http://www.diystompboxes.com/smfforum/index.php?topic=71221.0

How do I establish 2 grounds...Sewer ground and signal ground?
http://www.diystompboxes.com/smfforum/index.php?topic=58244.0

Noob question: Battery- = Gnd?
http://www.diystompboxes.com/smfforum/index.php?topic=71542.0

Re: Neovibe - a question about the GGG board (reply #3 by RG about Ground)
http://www.diystompboxes.com/smfforum/index.php?topic=71555.msg578920#msg578920

[valdiorn]

Ground is just some point in your circuit that you decide to call ground, it's just a reference point to which you assign the value of zero volts.

And real Earth-ground is something different, something that comes from your power socket. It's a pathway to the actual soil outside your house. When using batteries this isn't really something you need to think about, since there isn't actually a "ground" on your battery, just a positive and negative end of a voltage cell (if you wanted, you can call the positive terminal ground, and use 0V ground, -4.5V vRef and -9V... it doesn't matter, it's just another way of thinking of the same problem).

[Sir H C]

I don't see circuit two working.  The whole circuit is floating, so what is the actual voltage for those nodes.  Now if you get rid of the input capacitor, then all is good.

[Ardric]

On circuit one, the input bias connection from Vref to pin 3 should be a resistor, not a wire.  This resistor in parallel with the opamp's impedence makes the input impedence of the circuit.  Tying it directly to Vref won't work.

On circuit two, the input bias resistor is missing.  There should be a resistor from U1a pin 3 to ground.  It's the exact same circuit whether we call the mid-point reference voltage Vref or GND.

+1 on Auke's links.

[tempus]

Something else that may help: Voltage is also known as "potential difference". So when you measure a voltage with your meter and it reads, say, 9v, that means that it's 9v higher than where you referenced from, i.e., there is a 9v difference from the black lead to the red lead. Now, if the black lead happened to be 3v higher than whatever our reference is, you would only measure a difference (or voltage) of 6v. That's how a virtual ground works. Earth ground (as in the ground you stand on) is at 0v (I think), so if something is actually connected to earth ground, the difference between the ground you're standing on and the voltage coming out of a wall receptacle (which is referenced to earth ground) is 120v, the voltage that comes out of it is indeed (dangerously) 120v because to your body, earth ground is the reference point. If you touch the hot lead of an AC receptacle (please don't) there will be a voltage on your finger 120v higher than the one at your feet, and you will get a shock (or much worse). If we go back to our virtual ground opamp setup, you've set the reference at 4.5 volts. IOW, the signal is now sitting 4.5v higher than actual earth ground. Since there is a total of 9v available at V+ of the opamp, the AC signal can swing down 4.5v (to 0v) or up 4.5v (to 9v), minus the leeway needed by the supply rails.

Hope that made some sense.