# OpAmp Floating Dual Supply

1. Mar 11, 2013

### Rudinhoob

How to power an OpAmp using two batteries, so there's no connection to mains ground or neutral? This circuit requires ground to force the node connecting the batteries to be at zero potential. How can I achieve the same result without grounding?

thanks.

#### Attached Files:

• ###### circ.jpg
File size:
12.3 KB
Views:
222
2. Mar 11, 2013

### Averagesupernova

One end of the load connects to the output of the op-amp and the other end of the load has to connect to the node between the two batteries. The inputs may or may not be referenced back to this node depending upon configuration. What else you connect that node to is up to you. Did you think it is impossible to use op-amps in cases where there is no connection available to the earth such as in airplanes, cars/trucks, and portable devices?

3. Mar 11, 2013

### Rudinhoob

I'm pretty sure they can be used without earth connection. What confused me really that many texts show the dual power supply using batteries with a connection to earth, which makes no sense.

But how using two power sources is different than using one? Cannot just combine both into one source?

4. Mar 11, 2013

### f95toli

"earth" is often just another word for "ground".

There are sometimes sublte differences in how the words are used (especially when dealing with equipment connected to mains, or for interference/shielding); but in from a theoretical point of view earth is just a convenient place to ground to your circuits.

It is important to understand that the ground point in a circuit is just an arbitrary point in a circuit, where we DECIDE that the potential (voltage) is zero.
Now, if you connect to batteries is series (- + - +), and choose the ground point to be the connection between them you will find that you've created a bipolar source.

5. Mar 11, 2013

### Rudinhoob

Are you referring to "reference node"? If so then this has no physical impact on circuit it's merely used for calculations.

Not sure what do you mean by "bipolar source" I thought every source is bipolar with + and - poles.

6. Mar 11, 2013

### jim hardy

Terminology, terminology....

I see nothing on your sketch requiring a connection to "Ground".

"Circuit Common" is a better term than "Ground" because it does not imply "Earth".

Simply take your eraser and remove that wire between the three-line "Ground" symbol and the junction of your natteries. In fact, erase the "Ground" symbol too.
Then hook your voltmeter's black wire to that battery junction.That'll do it.
Put a big round dot where the batteries join if it makes you comfortable.

Get over the notion that "ground" is magic. It's just another wire.

7. Mar 11, 2013

### Rudinhoob

Is the inverting configuration on the left valid? What makes it different than the one on the right?

#### Attached Files:

• ###### circ.jpg
File size:
15 KB
Views:
238
8. Mar 11, 2013

### jim hardy

Yes.

On right: The + input of your opamp is not tied externally to any point of reference, so it'll find its own according to whatever is just inside the triangle representing the opamp.
Some opamps will self bias satifactorily, some won't.
When running a dual supply opamp from a single supply as you show, it is usual practice to create a reference of ~ Vsupply/2, perhaps by placing two equal resistors in series across the supply and using their midpoint as reference.
EDIT late addition... in yuor right hand circuit you'd use the reference thus formed, Vsupply/2, as signal reference point and connect the opamp's + input there.

That is an interesting application for self-biasing single supply audio amplifiers like the LM383(TDA2002) - its output is biased halfway between V+ and V- . So it makes a good reference for dual supply opamp circuits because it'll absorb or deliver substantial current.

Have fun

old jim

Last edited: Mar 11, 2013
9. Mar 11, 2013

### f95toli

I am not sure what you mean by "physical impact". The point is that voltage is relative: it is always the difference in potential between two points. Hence, it is perfecly valid to assign "V=0" to whatever point you want in the circuit and call that point ground as long as you are consistent. An obvious example would be a circuit with voltage source (lets say a12V source ) where the negative terminal is often assigned to be ground. However, from a physical point you could also just say put ground at the positive terminal, in which case the negative terminal would be at -12V.

Note that this also has practical implications. It is e.g. why you can amplify an AC voltage using a circuit with only one battery; as in your circuit above.

No, a bipolar source means a source where one terminal has a potential lower than ground (i.e. it has three terminals: plus, ground and minus).

10. Mar 12, 2013

### sophiecentaur

Yes, that's the way to do it, basically.
It is not good practice to let inputs 'float', the way they are doing on the right hand diagram. A good OP Amp will have a certain amount of 'common mode rejection' and will not mind too much where its signal inputs lie with respect to the supply volts; it will do its best to amplify the voltage difference. However, there is always a possibility that, somewhere in the signal paths, there will be a 'stray connection' to pull the inputs one way or another, via some very high resistance path to one or other of the 'rails', which will limit the output current available and, thus, the feedback will stop working.
It is much better to provide a common potential, somewhere near the mid point of the power supply rails, to give the OP Amp the best chance of working properly.
If you use the left hand arrangement and find the circuit doing what you want and then suddenly disconnect that mid-point (bias) line, it can work fine for hours until things drift and take you totally by surprise. This can also happen at switch on.

There is a good principle in electronics and that is to 'define' the operating conditions of all devices and don't let anything 'float' if you can help it. (That goes for logic gate inputs, too.)

11. Mar 12, 2013

### Staff: Mentor

You could view it as a single power supply with a centre tap. http://physicsforums.bernhardtmediall.netdna-cdn.com/images/icons/icon6.gif [Broken]

Last edited by a moderator: May 6, 2017
12. Mar 12, 2013

### sophiecentaur

The point about having two power supplies and a 'ground' is that the 'mid point' is at the same potential as it would be in another different piece of equipment, with different voltage supplies, would have the same ground reference. A distinct advantage.

13. Mar 12, 2013

### Rudinhoob

And the ground of this "bipolar" battery has to be connected to "ground" or "earth"? or can be connected to what point in the circuit?

14. Mar 12, 2013

### f95toli

In most circuits this midpoint (or "centre tap" as NascentOxygen called it) IS the ground, i.e. everything else that should be at zero potential is connected to this point.

15. Mar 12, 2013

### sophiecentaur

It can be connected to anywhere you like but choosing the same point for all your equipment makes sense and a common connection between all circuits on one board, all circuit boards in a rack / unit, the equipment case and (but not always) the mains Earth is pretty common. From your questions, I gather that you don't have a lot of actual hands-on experience of this stuff. Once you do, you will find things make much more sense.

The point about using a bipolar supply is that your signals can go positive and negative and the amplifier can handle them much easier. It is not so important for RF circuitry where the signals are usually fed through series capacitors and the actual DC voltage is not relevant.

16. Mar 13, 2013

### Averagesupernova

The diagram on the right cannot work. There really is no feedback path. It appears that there is, but when the inverting input is pulled in a direction by the resistor network coming from the output, there is nothing that is keeping the non-inverting input from following it by the same amount. The non-inverting input needs to be 'held' to something as it is in the schematic on the left.

17. Mar 13, 2013

### sophiecentaur

By George, you're right. No explicit DC feedback - as I was hinting in post 10.

18. Mar 13, 2013

### Rudinhoob

Do I need to consider the internal circuitry of the OpAmp?

19. Mar 13, 2013

### sophiecentaur

No. That's the beauty of an OP Amp. You can assume that it produces a massively magnified version of the difference between the two inputs (thousands of times). Different manufacturers often make their version of the generic Op Amps differently - they just do what the spec says they should. You just drop it into a circuit with the right operating conditions and it will behave ' perfectly' if you stick to the frequency and load limits.
Open circuit gain is treated as infinite for many applications.

20. Mar 13, 2013

### Averagesupernova

You should not need to consider internal circuitry for the examples that you gave from the standpoint of whether the circuit will work at all or not. Of course you need to know enough about the device to know that it is a bad idea to use feedback resistors that are in the 100 Megohm range or in the less than 10 ohm range. Do you understand more than when you first posted in this thread? We are here to help!

21. Mar 13, 2013

### Rudinhoob

Then how could you tell the circuit on the right does not have "real" negative feedback?

22. Mar 13, 2013

### jim hardy

Good circuit design means you don't leave things up to chance, so in a respectable design you'd provide DC feedback as the fellows are saying.

I was probably remiss in not pointing out that self biasing opamps are the exception rather than the rule..
The LM386 for example is a small audio amp that can work withpout DC feedback..
But it's not a general purpose opamp instead it's made for AC signals. So its designers removed the need for DC feedback.

old jim

23. Mar 13, 2013

### sophiecentaur

Can you see any DC feedback which would keep the output at mid rail? If you can't see it then you have to assume it's not there and you have to provide your own. If it had some vestigial bias, it could be either way.

24. Mar 13, 2013

### Rudinhoob

Then this DC feedback is OpAmp type dependent, which means it has to do with the internal circuitry. Although not necessarily exposed in the analysis, it seems that there are some decisions that are dependent on this and to justify them, we need to include the details of the OpAmp in analysis.

25. Mar 13, 2013

### sophiecentaur

I wouldn't put it that way. You asked a question about Op Amps and not about specialist circuits that are aimed at a particular application. Read up something about the characteristics of an OP Amp and you are unlikely to find any mention of internal DC feedback. The LM386 is not sold as an Op Amp. Why are you looking for loopholes here?