What are the differences between single supply and dual supply op amps?

[0xDEADBEEF]

I did a bit of searching and I don't get it. What is really the difference between single supply and dual supply op amps?

Voltages are just a potential difference. Op amps usually do not have a ground pin so they should have no idea if I am connecting +-15 V or +30V,0V it should still drive the output such that the difference between the inputs is minimal.

I understand that you want to keep the input and output voltages close to the mid point of the rails, so the transistors are in their sweet spot. But how can the op amp help with that?

And maybe there are issues with power supply/common mode rejection, but if I buy a single supply or a dual supply device what is different in the device that I get (except for the labelling of the pins)?

Edit:
I had an idea. Is a single supply op amp just a rail-to-rail amplifier that can go to the lower rail, but not necessarily to the upper rail?

 

[I_am_learning]

Dual supply op-amps can't go near any of the rails but single supply op-amps can go very near to the lower rail. And in dual supply types you NEED dual supply because it needs +, - and GND supply.
You can look up the datasheet of each type of op-amps to confirm.

 

[0xDEADBEEF]

When I select op amps at linear technology all rail to rail op amps are also single supply op amps.

LT1006 is single supply but not rail to rail, and can only go to ground I suppose

LT1001 is dual supply but doesn't have a ground pin same for LT1007, and LT1008

So far my guess seems right

 

[Averagesupernova]

I have used both in both configs. I typically prefer rail to rail op-amps and don't really care if they are single or dual supply or whatever. The circuit can be configured to fit.

 

[AlephZero]

Voltages are just a potential difference. Op amps usually do not have a ground pin so they should have no idea if I am connecting +-15 V or +30V,0V it should still drive the output such that the difference between the inputs is minimal.

That's correct so far as it goes. But if you are working with analog signals, you probably need a low impedance "signal ground" connection at the signal's "zero" level, and for symmetrical signals that is half way between the + and - supply voltages.

This won't show up in a computer simulation unless you are simulating noise immunity etc, but it will show up in real life, especailly with high impedance and/or high frequency signals.

"Rail to rail" is a different issue - basically, it's a way of squeezing a more signal (either analog or digital) out of a system with a low supply voltage. If your op amp supply is 5V and the output can only swing between 1V and 4V, that takes a big chunk out of the noise immunity of CMOS logic for example, where the specificaiton for logic 0 and logic 1 voltages is "less than 1.67V" and "greater than 3.33V".