Grounding stuff to Vref

[GibsonGM]

Is there some sort of junction barrier there, DA? 

[ashcat_lt]

I don't love the idea of saying that ground and reference voltage are different things.  I know we're trying to keep it simple, and that you know what you mean, and that you didn't mean to start this kind of argument.  I just kind of feel like - especially when we're talking to noob - it's best to start from the understanding that what we normally call "ground" really is just a reference voltage.  That is, if we don't confuse them to begin with, then we won't have to alleviate their confusion later.  Maybe that's just me...

[antonis]

Conventionally, Earth's potentional is "zero" so we call it "ground"...
(We may name it Bob or Myriam but it doesn't change anything.. )

As long as we are familiar with the fact that we deal with different "levels" (as well said by Mike) and there is no "absolute" value, the only thing that we have to take in mind is the difference between these levels (which is the same between +9V & 0V and +18V & +9V)

[GibsonGM]

Yeah, you're both absolutely right.  I prefer the simple route in explanation to 'get' the concept, as it's not that common to find something like an opamp or transistor using an oddball ground point.  SOMETIMES you will, of course...the positive ground of a FF is one example, and some DSP stuff can be using some unexpected voltages...but getting started, what Antonis said seems logical to me...

How 'bout this?:

We're CALLING this level ground - usually it is zero volts, but that's not absolute by any means.  The thing doing the amplifying (or whatever operation you're performing) is USUALLY using zero volts as its ground, but again, not always.  This ground is typically not the same value as the Vref, which has a different purpose in the circuit (or Vref would just be called "ground").

The second (and probably more important) part of this is that since we want to work with an AC signal, and AC rides above and below zero volts, we need to somehow MOVE the AC zero point somewhere within our operating zone so that the device can equally amplify the positive AND negative excursions of that AC signal.  If we don't do this, the signal won't be able to rise and fall normally - it will be cut off on one side or the other and sound like crap if you didn't do it intentionally.   

Generally, when we're done manipulating the signal, somewhere down the line it will encounter a cap that will block this ADDED DC potential called Vref, returning the AC's reference point back to zero volts...maybe after one active device, but sometimes things are DC COUPLED and allowed to have this added voltage thru several operations.   All we've done is "mixed in" a DC level that the AC signal rides upon so an operation can be performed on it, which is totally allowed by the laws of physics.  ("superposition")

That DC level that we added is blocked at some point before the output to keep consistency with your amp or other device, which mostly use zero volts as ground AT THEIR INPUTS AND OUTPUTS (who knows what goes on inside?).  This is a convention so things can easily work with each other*...  IOW, we can shift that AC signal all over the place if we want to, to do something with it, and then bring its mid-point back to zero volts when we're done, which I'll call "conventional ground". 

* sometimes, when this convention is violated, you can get electrocuted like many players in the 60s and 70s that didn't have the additional SAFETY ground on equipment - and ended up with differences in potential between different equipment - touching both at the same time caused current to flow and them to be electrocuted!  That's a little bit of an excursion from what we're talking about here, more of a power supply problem than signal operations, but illustrates the point.

[antonis]

Millions of alive birds sitting on power cables can't be wrong...

[PRR]

Vref is 4.5V for DC but "ground" for AC/Audio.

[GibsonGM]

Yep.  Sometimes I also think of it as the 'hub' the signal rotates around, kind of.   Without adding that offset, you only can amplify part of the signal...the rest is clipped. Can't talk about that enough, so I'll show it.  Pardon my crappy art.

[antonis]

Vref is 4.5V for DC but "ground" for AC/Audio.

(allow me a more definable expression..)

ANY value of Vref is "ground" for AC/Audio, as long as the voltage source's internal resistance is much lower than the audio's one..

[greaser_au]

I don't even use the term "Ground" with it, normally, as it seems to be a bad habit. 

Consequently, using the term "Virtual Ground" should be a nasty habit..

Actually, to be correct,  splitting the power supply  with a resistor divided Vref  instead of using tracked dual supplies is the REAL perversion - unless you are using an LM324/2902!

"Virtual ground"/"virtual earth" is the EE way of describing the inverting input of an opamp. The inverting stage is an excellent place to start (the device tries to maintain the inverting input at some arbitrary 0V from the sum of the currents from Rin and Rf). This is often the first opamp circuit presented from a theory perspective, as it is the easiest to understand for a student that has just been shown the wonders of Norton.

david

[GibsonGM]

Again, yup.   But - how have we helped the OP by now putting forth so much technical info that his head may be spinning?  He wanted to know why you can't just 'ground' something to Vref.    Which necessitated explaining a bit about how WE, in our context, use a reference voltage to elevate an AC signal so we can use a single power supply.  It's ground for AC signals to 'ride around'.  Wow, that's all you need to know, isn't it?   Build enough boxes, and you'll see it over and over, it becomes part of your vocabulary after a while.

Many of us have built things that may have more than ONE Vref or use other 'fancy' power tricks!   But that doesn't help someone who's trying to get in the door.  What got me more than p*ssed off as a noob was that I seemingly needed to run a differential equation to understand something like how a cap works, etc. - when in fact, for what I wanted to do, a nice introduction was all that was needed, and the eqn's would come later when I had more interest, and need, to use them.  I still have not needed to use the dielectric constant for paper or glass (or anything else, for that matter).   

Heck, even Ohm's Law is often so over-done as to look intimidating to someone new.   Then we wonder why they're not using it when a question comes up!   I know I used to read answers like "Wow, WTMI!"  lol

>> Tracked dual supplies would be the same as using a nuke to kill a mosquite, IMO, David, LOL!  More complexity for very little gain, no?  Keeping it simple helps us to not look stupid!   

[greaser_au]

GGM,
Regrettably this will ever be the problem on fora like this. For progressing beyond the "monkey-see, monkey do" of layout replication,  some basics are required - not how to select the correct spanner to undo the bolt/nut, but  knowing that the spanner even exists - and then  which type will work on hex or square. All fine to not need deep theory in the little boxes we live in, but if you're making capacitors you 'd better *know* what the permittivity ratio if your dielectric is! if you are considering replacing coupling caps in something, 1/(2*pi*fC) will likely save you many wasted hours (yes, I know what Edison said - but there's experiment and then there is waste!).

Anyway this is all moot, because my comment was purely intended to be a humourous reponse to that one post, with just a little explanation.

david

[ashcat_lt]

I think that GibsonGM did a great job with the simple answer on the first page, and the OP expressed gratitude and it's maybe ok now to continue the discussion a bit further.  It will either enlighten or bore the OP, but he got the answer he needed.  There are obviously other folks at other levels reading this thread too...

So, I'm gonna go ahead and say that all voltage is relative, and most ground designations are arbitrary - generally being just a way of connecting a number of points on a schematic without having to draw all the wires.  Need a bunch of things to connect to the same place?  Call it ground!

Edit to avoid double post -
I believe the issue in the actual original post was that the chassis was shorting the bottom half of the Vref divider, forcing Vref to be the same as the bottom rail.  There's no reason you can't connect the chassis to Vref, but you can't also connect it to that bottom rail.  Sometimes it'll get there through paths outside of your box, so it can get tricky...

[karbomusic]

Yep.  Sometimes I also think of it as the 'hub' the signal rotates around, kind of.   Without adding that offset, you only can amplify part of the signal...the rest is clipped. Can't talk about that enough, so I'll show it.  Pardon my crappy art.

I probably still have it up on the BB and can probably grab a screenshot of the real thing. I had it occur and (thanks to lots of previous study) instantly new what it most likely was, VRef issue. On the scope the bottom half of the waveform was chopped off where I knew darn well based on the gain math I had just done that it shouldn't be that way; found the issue and fixed. Anyway, when I'm back at that particular breadboard I'll replicate it and throw it on the oscilloscope just because seeing it in action is sometimes helpful.

k

[GibsonGM]

I love looking at stuff like that on the scope, Karbo...seeing is totally believing, and learning.

You guys are 100% right, maybe it's time to go further with 'levels', and how best to view what's going on in a circuit.  I knew what you meant, greaser   OP Slasher 'gets it', too.     

Sometimes, when you look at something and realize "hey, this is all shifted, why?", and then see the schematic, and the lightbulb goes on in yer head...that's why we do this stuff!  There are no tricks being played...sadly, when I was new, it almost seemed that way, ha ha!
Carry on! 

[karbomusic]

Sometimes, when you look at something and realize "hey, this is all shifted, why?", and then see the schematic, and the lightbulb goes on in yer head...that's why we do this stuff!

Absolutely. Last week working on a compressor with a blend, it's on the BB and the scope. I couldn't (at first) understand why the volume dropped when the blend was at 50% then I saw the sine wave shifting just a little left and right as I moved the knob... Wait, that would indicate a phase issue... Turned out to be how I was shunting a gain loop cap unpolarized to ground (I think). Replaced with the cap with what I originally thought should have been there, problem solved but the biggest joy was hearing and seeing the proof of both the problem and the solution via ears/schematic/scope. Now it blends perfectly.

[GibsonGM]

Sometimes, when you look at something and realize "hey, this is all shifted, why?", and then see the schematic, and the lightbulb goes on in yer head...that's why we do this stuff!

Absolutely. Last week working on a compressor with a blend, it's on the BB and the scope. I couldn't (at first) understand why the volume dropped when the blend was at 50% then I saw the sine wave shifting just a little left and right as I moved the knob... Wait, that would indicate a phase issue... Turned out to be how I was shunting a gain loop cap unpolarized to ground (I think). Replaced with the cap with what I originally thought should have been there, problem solved but the biggest joy was hearing and seeing the proof of both the problem and the solution via ears/schematic/scope. Now it blends perfectly.

Nice!   I don't take out the scope enough...PITA to hook up my old computer that I use as a signal generator.  I really need to do that more often! 

[karbomusic]

Nice!  I don't take out the scope enough...PITA to hook up my old computer that I use as a signal generator.  I really need to do that more often!  

I wouldn't either but my scope and generator IS my breadboard so its too easy not to.

http://www.digilentinc.com/Products/Detail.cfm?NavPath=2,842,843&Prod=EEBOARD&CFID=6695924&CFTOKEN=a1183cfeafe98cbe-E46CAB17-5056-0201-02B0EDF2822E2F2B

[GibsonGM]

VERY nice!!  Do you get good results from the PC-based software scope?  At that price, I bet ya do!     {I gotta get me one of those!}

Mine, from ca. 1982, but with low miles. $2,000 then, $90 now!: 

[karbomusic]

Do you get good results from the PC-based software scope?

I do actually, it does great and loads of features that really come in handy; even has two built in VRefs for when I'm really lazy.  Best purchase I've made. I'm still jealous though. I really want one of those like the one pictured above. 

[karbomusic]

Here is the mucked up VRef on a scope screen shot.