Sine wave distortion in simple op-amp circuit- bypass capacitor?

Question

I've built a simple unity-gain buffer op-amp circuit from the UCSB lab here, with a 12V power supply and a 1V, 1 kHz input sine wave.

^{simulate this circuit – Schematic created using CircuitLab}

Everything seems to be going well- except for that the output sine wave has a drop-off on the lower part of the signal, which will start if you leave the circuit alone for a couple seconds. Here's a gif I made of the problem (YouTube here):

The signal stays in the strange raised state indefinitely, but if I short the leads of the +V bypass capacitor, the signal returns to a normal sine wave. If I short the leads of the -V bypass capacitor, the signal turns into the raised state.

This seems to me like there is a problem with the values of the bypass capacitors (currently 104 or 0.01 uF). I know from playing around with the circuit earlier that two 0.0001 uF bypass capacitors just don't "fill" the sine waves out all of the way.

Is something wired wrong? (I connected all of the earth grounds together... is that good?) Is 0.1 uF just not large enough for the bypass caps? Or should I add a resistor somewhere? If so, why?

EDIT 11/16/15:

Here is my shot at a working circuit. Is this what is recommended?

^{simulate this circuit}

Answer 1

Andy's comment is correct. It answers the question and explains what you are seeing, assuming you have swapped the + and - opamp inputs in your schematic : as drawn it has positive feedback instead of negative feedback!.

You are feeding the opamp with 12V, but there is no connection between that 12V supply and the 0V reference for your audio input, output, and scope probe.

Therefore you do not know whether it is +11V/-1V, +/-6V, or +1V/-11V.

And in fact it's probably all of these at different times. When it is +11V/-1V there isn't enough voltage on the -V supply rail for the opamp to function, and so the -ve peaks of the output are clipped.

If you had measured either of the supply rails as I suggested, you would see quite a large variation on them, probably a 50Hz noisy sinewave, whose peaks correspond to the distortion you see.

So you have to define the relationship between that 12V supply and your 0V. As a quick and crude test, take a pair of identical resistors, 1kilohm will probably do, and connect them between ground and +V, and between ground and -V. And observe with the scope probe that V+ is now about 6V (but probably still with a little noise on it).

As long as you aren't loading the opamp output with more than a couple of miliamps this will work, but in future, it's better to use +6V, 0V, and -6V supplies instead.

This arrangement with two resistors is often used on single supplies, but usually with a voltage regulator or even another opamp to provide a better "ground" than 2 resistors alone. It requires some care to remember that "0V" is not the negative supply rail!