Connecting a "virtual" to "real" ground circuit on the same supply? Is that OK?

[marg]

Ok, here is my doubt which I would rather want to solve right now than (possibly) end it in smoke

I've made some effect circuits and they work fine by themselves. They are powered by 18V single supply (all of them were made to work on the same supply - by themselves), and some use virtual grounds (Vcc/2) while some use real grounds (0V, negative pole of supply) as reference for their inputs/outputs.
Now I want to interconnect them to start combining effects but my worry is this:

If I connect, say, a circuit whose output is referenced to Vcc/2 (eg. 4.5V), to another circuit whose input is referenced to real ground (0V), will I in effect be "shorthing" a part of supply (in lack of a better term)? Pulling virtual ground down to 0V?

I know that would happen for sure if my virtual ground wasn't buffered (just split in half with resistors). I would just bypass the second resistor by connecting vgnd with ground. But my virtual ground is opamp buffered so it should maintain Vcc/2. Does it mean that I'll short an opamps output and produce current which is limited by it's output resistance? eg. current of 4.5V/2ohm ?

How to you go about connecting circuits which have different grounds (be it real or virtual) but share the same supply?

I'm sorry if my question comes out as a duplicate. I've done my efforts to find answers on other forums etc. but I'm still not quite sure. Perhaps I didn't search using the right keywords so I didn't get useful results. Anyway, if that is so, I would like to be pointed at the right direction.

[samhay]

>If I connect, say, a circuit whose output is referenced to Vcc/2 (eg. 4.5V), to another circuit whose input is referenced to real ground (0V), will I in effect be "shorthing" a part of supply (in lack of a better term)?

yes.
How big a problem this is will depend on various quirks of your circuit(s).

A solution is to isolate the input and output with capacitors so that the outside of your circuit is floating relative to your virtual/real grounds. This will work if you are sharing a common supply. If you get switch popping, then pull down resistors referenced to your 'real' ground is probably the best solution.

[Mike Burgundy]

All voltage levels are arbitrary by themselves- they only start to make sense if you reference them TO something.
The most logical reference point is ground - which is the voltage level the world around us should generally be at (hence the name), and we call that 0V. Everything is referenced to that.
Virtual ground is called this because it is NOT "real" ground, but we tell the electronics that is is, locally. To an IC, it now looks like it's getting a dual supply of + and -4.5V. "ground" has been shifted by 4.5V - and here's the trick: REFERENCED to real ground.
Hooking up real ground to virtual ground will make the opamp do it's very best to pump current into what it sees as a -4.5V sinkhole to pull that up to what it feels is ground level. This may let the magic smoke out, it may not (wise opamps use protection).
The circuit will NOT work, it's way too busy fighting voltage levels.
this is why blocking caps are used in the signal path.
Remember, GROUND in a "real ground" circuit is THE SAME as GROUND in a virtual ground circuit because virtual ground is referenced to real ground.
Does that make sense?

[antonis]

You will not "short" anything, as long as we are talking about AC signals isolated with capacitors..
(DC supply of ANY level is GND for AC signal..)


edit: Sam is faster (and wiser) as allways..

[marg]

Thanks for the replies. Straight to the point.

My inputs/outputs are already isolated with caps, just like in any other single supply circuit, and it seems that I have pulldown resistors because I'm using single supply op-amp buffers which need to have them in order to work.Hm.. But these pulldown resistors are after caps (from input perspective).
Do you mean that I also add one before input caps, should popping happen?

^^
I've seen that a lot actually (if I'm thinking correctly), and wondered what's the purpose of them, since those standard, say 1M 1M resistors were already included for the purpose of biasing an op-amp and maintaining input impedance.

It is kind of dumb that it never dawned on me that my input and outputs are already isolated and that no "shorting" can happen.  I guess it is one of those situations where you look at the apparent problem, and the solution is right there but you just don't see it. And there wasn't any problem to begin with.

To Mike Burgundy: Yes I understand what you're saying. In fact I've mentioned shorting the output of an opamp. And these things can get really hot, as I've experienced many times when I would do some kind of short. Probably not to the point of burning out, but still it would be really bad for me to screw my soldered pcb and have to do it again.

To finish it up, thanks again, you've made it all clear. I guess I just needed to be reminded of what's really going on cause I got really mixed up I haven't done electronics in a while because of other things I needed to do in life, like college etc., so now I'm recolleting it all so I can get going again.
Best wishes

[marg]

So you are suggesting that I do this?

(This is how inputs and outputs of my effects look in general)



It is obvious that my input/outputs are isolated but it would work for same grounds to prevent DC leak. Do I also need to separate the grounds by a cap? As drawn below.

I guess I misinterpreted "isolation" since it wouldn't work for different grounds without additional cap which separates ground, which I do not have. Without that cap I would still be connecting directly Vcc/2 and 0V, causing opamp short.

[Mike Burgundy]

 A cap is an open connection (so, NO connection) for DC.
in a virtual ground situation, real ground is always connected to real ground. Virtual ground is just for the opamp. Think about the connections real world vs virtual ground world:



VIRTUAL "ground" in reality +4.5V
VIRTUAL "supply +" in reality +9V
VIRTUAL "supply -" in reality 0V=GROUND

the opamp will force (or attempt to) its output to the DC level that it's + input is referenced to - this is what you're doing with the two resistors. The opamps in/outputs have a DC offset compared to real ground, which you need to isolate from the rest of the circuit with blocking caps. You have those.
Looking at your schematic, you have isolated the output of EFFECT1, and the input of EFFECT2 from this offset - the volume pot is not referenced to that offset unless you make it so - by hooking it up to Virtual ground. Hook it up to regular ground and youre golden. If that pot was inside EFF1's blocking cap (so hooked up directly to the opamp), then you should reference to Virtual ground.

[marg]

A cap is an open connection (so, NO connection) for DC.

Yeah.. Connecting two "grounds" with a cap is just like not connecting them at all. My bad.

Looking at your schematic, you have isolated the output of EFFECT1, and the input of EFFECT2 from this offset - the volume pot is not referenced to that offset unless you make it so - by hooking it up to Virtual ground. Hook it up to regular ground and youre golden. If that pot was inside EFF1's blocking cap (so hooked up directly to the opamp), then you should reference to Virtual ground.

Yes, there is no reason to put AC cap in output of effect1 and then reference the signal to VGND, because there wasn't any offset with respect to VGND to begin with, since that is the reference. But there is an offset with respect to GND. For some reason I thought that you couldn't reference signal to GND if you used VGND...whatever...

I do get it now.

My fault was that I took "whatever works" approach and was just messing up with the connections until it worked    But clearly that is not the way to go as what I did is something that a person that truly understands what is doing would not do. Not to mention how that approach falls apart when you have to combine things and realize something is very strange

Thanks for your patience. It's been useful. 

[duck_arse]

regarding pulldown resistors: in your effect2 drawn above, one side of the input cap is pulled to a voltage (your choice - Vcc/2). the other side of the cap floats free, swinging in the breeze, finding its own voltage level. that free end needs to be held to a level common between devices, like ground, so when the (not drawn) bypass switches effect2 in-circuit, the voltage is the same (ground) as the ground-referenced output of whatever preceeds.

then there is no voltage difference to cause the pop. in theory. so you'd add a high value pull-down resistor after your true-bypass switch contact, before the free end of that cap.

the vol pot in effect1 provides the pull-down function for the output cap. some form of pull-down is necessary for effect2 output cap, as well.

[marg]

so you'd add a high value pull-down resistor after your true-bypass switch contact, before the free end of that cap.

..As if you were reading my mind I just wanted to ask "what about true-bypass? Isn't it pulling it to the ground?" But clearly there is a moment of switching where it would float free.

[duck_arse]

true bypass only describes the effect being disconnected from the signal chain, it doesn't say what happens to the effect input/output. you can do what you see fit with those whilst bypassed, depending on your switch type (2 pole or 3 pole).

[GibsonGM]

Just a quick note...Isn't it easier to think of "virtual ground" as DC OFFSET, to avoid the confusion?

Simply, 'adding a level of DC to an AC signal so the opamp can deal with the zero crossing, and generally taken to be around 1/2 the supply voltage'...that's all we're actually dealing with here, and seems like the term "ground" may be a loaded term?

[Brisance]

or for that part consider virgual ground to be ground and real ground to be negative supply

[antonis]

or for that part consider virgual ground to be ground and real ground to be negative supply

Excactly what (Sir) Mike wrote about DC "offset"... 


P.S.
As long as we keep in mind that there isn't any ABSOLUTE voltage level (only vodka...) but relative voltage differences we can consider them as we like..

[marg]

I think that exactly the confusion in DC offset being called a kind of "ground" is what caused my problems in the first place. Beacuse I had the schematics and I notated that DC offset as ground (like VGND), and then got into thinking that I could connect that virtual ground with real ground because it's just "ground" And I went about searching how to do it because I had a feeling that you can't just connect it straigt because there is some difference of potentials between them.

But when you really get to think about it you realize that doing that is absurd - you're just connecting a point with some DC voltage to ground, causing a kind of short circuit if it's opamp buffered, or just bypassing it all otherwise.

I admit my knowledge is far from perfect, and that the notion of "virtual ground" kind of misleads you if you don't really know what it stands for.

Also I wasn't really sure whether that "virutal ground" thing is only used by op-amps locally or should we go acting as if we really had dual supply and then reference out input and output signals to that virtual ground. Because that was what I've done. And it works, no reason why it shouldn't, but as I've seen, the mess starts to happen when some of your circuits are referenced to real ground and some to virtual ground because they can't be interconnected. (Am I the only person here who had the practice of referencing signals to VGND? Because really I haven't seen that anywere, I just did it because it seemed logical.)

Anyway, the good thing is that I now really understand what I'm doing and that I've corrected my guitar effects and now they work perfectly, and no more mess caused by bad practices on my side. So thanks to everybody. Thanks duck_arse for pointing at that pulldown resistor thing after true bypass to avoid DC thumps.

[Mike Burgundy]

I use "Vref" (for reference), less confusing than VirtualGround.

There are circuits aplenty that reference for example a gain or volume pot to Vref - it all depends on what the "inbound" and "outbound" connections are referenced to. The trick is that you always want the SAME reference on all ends of the pot (same as with a switch).
So if you stick a volume pot "inside" the in- and output cap, you're in Vref territory and should reference the pot to Vref. Outside, reference to ground. That easy. Same goes for a resistor that is part of a filter, goes to Vref if it's in Vref referenced circuit (eg between two Vref referenced opamps), goes to ground if it's in a bit of circuit referenced to ground. There are exceptions to the rule, as always, but this is usually the best way.

[diffeq]

Hello, everyone! First time posting here.

I have a question regarding virtual ground circuit. I want to make a buffer pedal which will be first in the chain. Can I reference an input with virtual ground, like so?



Is it an ok solution or am I missing something?

P.S.: sorry for necro-bumping.

[anotherjim]

No, you can only plug something with the same "virtual ground" 4.5v into the input. If you plug in something referenced to "real" 0v ground is will short circuit the 4.5v out.
Make the input ground 0v. Keep the input capacitor. Connect the op-amp +input to 4.5v virtual ground via a resistor, usually 1Mohm if the input can be from a normal passive guitar.

[diffeq]

Alright, got it. Thanks for helping, Jim.

Here is a fixed version in case anyone needs it:


I wonder how much noise this biasing resistor will introduce, if any at all. I guess I'll find out soon.

[antonis]

For stombox purpose, TL082 is "quiet" enough...
(especially for buffer use - due to no gain at all..)

BUT you'll have to keep voltage divider resistor values (for V_bias) as low as possible (max affortable current consumption)..
(ignore the above if you use the other half of op-amp for V_bias..)

Or use something like OPA2134 or so..

P.S.
You may lower C3 value down to 10nF and C4 down to 10μF...