Single supply vs Dual supply op-amp?

[hermitengineer]

I've seen datasheets for various op-amps.  Some advertise prominently that they are single-supply, while others seem to act as if they only work as dual supply.

But in the end, what, if any, is the difference?  Sure, I could have a +5 and a -5 supply voltage for my dual-supply op-amp.  But if I shift my reference point to where -5V is ground instead, then I have a +10V single-supply op-amp with a rail splitter to provide a bias voltage.  And the op-amp is none the wiser.

So is there any real reason to care if an op-amp says it's "single-supply"?

[SeanB]

Those advertised as "single supply" tend to include one of the power rails in the input voltage range, and some even also include excursions slightly out of the power rail, typically having -0.5v to Vcc-1V as input voltage range. Dual supply ones tend to have large areas which the inputs must not go within, typically 2V or so from each power rail, making a dual supply more convenient to use. Yes mostly a semantics thing, but in single supply applications you tend to have inputs referenced to a supply rail in most cases, and it can be both hard and inconvenient having to scale and apply offsets to all the inputs to keep within the CMRR range, and many opamps can do odd things when you go out of the range, from the output inverting, to the opamp latching up either destructively or non destructively.

[hermitengineer]

So, it sounds like the intent of advertising as "single supply" maybe implies that it is more useful than dual-supply in single-supply systems then, not that it cares either way.  In your example of one that requires a 2V buffer from each rail, it would be almost useless in a +5V and gnd situation since there's only 1V of useful range.  I've noticed that "single supply" and "rail to rail" are often used together.  Perhaps R2R is what typically qualifies it as single-supply?

[thm_w]

So, it sounds like the intent of advertising as "single supply" maybe implies that it is more useful than dual-supply in single-supply systems then, not that it cares either way.  In your example of one that requires a 2V buffer from each rail, it would be almost useless in a +5V and gnd situation since there's only 1V of useful range.  I've noticed that "single supply" and "rail to rail" are often used together. 

Correct. If you wanted a opamp for your 5V microcontroller system, best to look at "single supply" types, they will often be optimized for 3.3V to 5V supply for example.
https://www.ti.com/lit/an/sboa059/sboa059.pdf

Note that you won't find it in parametric search: https://www.ti.com/amplifier-circuit/op-amps/products.html so that tells you how useful it is as wording.

Perhaps R2R is what typically qualifies it as single-supply?

Probably one of the important parameters yeah.

[David Hess]

That brings up the question of why all operational amplifiers are not made to be single supply.

Single supply operation entails some design compromises.  Output stages must be common emitter/source instead of common collector/drain resulting in higher output impedance and greater stability problems.  With an input range which extends to either or both supplies, input bias current cancellation cannot be used.

[macboy]

I'll just add that "single supply" usually implies use of ground plus a positive supply, therefore "single supply" implies that the inputs and outputs can go to (or close to) ground, but it does not imply anything about the positive rail. You might still have some limitation of how close the input and/or output can get to the positive rail. You can see then that "Rail to Rail" ("R2R", "RR") denotes something different, think of it as a superset of single supply. Also note that some are Rail to Rail for input only or for output only, but not both. If both, then sometimes the term "RRIO" (Rail to Rail Input Output) is used. You might also encounter OTT (Over The Top) inputs, which allow the input common mode range to exceed the positive rail, sometimes significantly (maybe tens of volts). These unusual amps can be used for things like high side current sensing.

[rstofer]

Depending on how you look at it, all op amps are single supply.  There is no reason you can't take a +-15V op amp and run it on 30V relative to the bottom rail which we might as well call ground.  All we have to do is bias the inputs up to Vcc/2 and this is easily done with a couple of resistors.

The real issue is the low voltage op amps.  How much room is there for the input and output to swing given, say, 3.3V power supply.

The older op amps, like the uA741 had dual supplies of +-15V but signal swing was +-10V which left 5V on either end for the internals to work.

Here is a great video on the topic:

[exe]

All we have to do is bias the inputs up to Vcc/2 and this is easily done with a couple of resistors.

This will negatively affect input impedance.

[Zero999]

All we have to do is bias the inputs up to Vcc/2 and this is easily done with a couple of resistors.

This will negatively affect input impedance.

Then use bootstrapping to increase the input impedance.

At signal frequencies, the input impedance of this circuit will be well above the 1M1 suggested by R6 + R5. The output is coupled back to the input via R7 and C5 and is attenuated by R5, to avoid oscillation. R5 = R1 and R7 = R2. The left hand side of R6 follows the input voltage, so virtually no current flows though it.

[David Hess]

I'll just add that "single supply" usually implies use of ground plus a positive supply, therefore "single supply" implies that the inputs and outputs can go to (or close to) ground, but it does not imply anything about the positive rail. You might still have some limitation of how close the input and/or output can get to the positive rail.

The LM301A and several of the older JFET input operational amplifiers have an input common mode range which includes the positive supply but their output does not go to either supply so a level shift may be required.  This still makes them very useful in some applications.

Depending on how you look at it, all op amps are single supply.  There is no reason you can't take a +-15V op amp and run it on 30V relative to the bottom rail which we might as well call ground.  All we have to do is bias the inputs up to Vcc/2 and this is easily done with a couple of resistors.

Asymmetrical power also has implications for input precision because the specified input offset voltage assumes specific supply voltages and common mode and power supply rejection are finite.

[schmitt trigger]

Back in the good old days, systems required both “digital” and “analog” Bipolar supplies.

These added cost and complexity, and constrained its use on battery powered units.

Thus, there was a big push to develop opamps which operated from popular 5 volt digital supplies.
The very first requirement, besides maintaining reasonable performance at lower voltages, was that the common mode input range included ground, like the ubiquitous LM324/ LM358. That the output is capable of swinging very close to ground was an additional requirement.

As others have mentioned, the single supply families have spawned additional subcategories, like R2R and micro power, designed specifically for battery operated products.