Nrthugu

Let's say we have an AC signal that can swing from -15v to +15v max peak to peak. If I use an opamp with the power rails to -5v and +5v, can I get a non clipped output signal if I use a gain of 0.333? Or will op amp failed to recognize anything above/below +/-5 V at it's input even if the required output is within range?

Do I need to have my opamp's power rails and specs within the input range or the output range provided that the output have less peak-to-peak swing than the input with my gain of 0.333?

The signal I'm talking about is a low, signaling level (about 1mA max) with varying frequencies, like an audio signal.

The op-amp will have a common mode input range specification. This means that there is a range of voltage relative to the supply rails in which you need to keep the inputs for the op-amp to function correctly. For example, with LM324 you can take the input down to the negative rail but not within 1.5V of the positive rail. If you use an inverting amplifier configuration then the inverting input of the op-amp is a virtual earth so you might have 0V on both inputs, even with a 15 V signal. A non-inverting configuration where you apply the signal directly to the non-inverting input would not work.

Figure 1. Inverting and non-inverting amplifiers. All About Circuits.

The inverting amplifier gain is given by $ A = -frac {R_2}{R_1} $ and the inverting input is at virtual ground so it will be fine.

You can see that the non-inverting amplifier has it's V_in connected directly to the input so this can't be directly fed. Instead you would use a potential divider on the input to bring the voltage down to within the common-mode voltage for the op-amp. The gain is given by $ A = left(1 + frac {R2}{R1} right)$ and if you set R1 to infinity (open circuit) the gain will be 1.

If the input pins of the op-amp don’t exceed the common mode specifications or the maximum specifications (e.g., Vdd-1V, 0V or whatever these are) then there is no danger. Adding protection diodes (if you don’t fully trust the op-amp’s pad protection) and using high enough resistors can make sure of this.

However, I don’t really understand why people use OP-Amps with fractional gains. It is actually more precise and less noisy to simply use a resistive voltage divider.

The input stage of a simple bipolar OP-Amp looks like this:

^{simulate this circuit – Schematic created using CircuitLab}

If your V+ is +5V and your input signal is >5.7V, the collector diode of the input transistor becomes conducting in its forward (the transistor's reverse) direction. The input stage cannot operate the intended way.

If your V- is -5V and your input signal is <-4.3V, the input transistor will never conduct. The input is at the rails. If your input signal is about 5V lower than the V- rail, the emittor diode will break through and the transistor is toast.

I would suggest a voltage divider to get the voltage swing down to within your power supply rails, and put the resulting V-div output through a unity gain opamp buffer. That gives you a low source impedance signal for whatever further conditioning you want to do.