When do opamps clip?

[BadIdeas]

Been searching around. Had some more time to fiddle with this project:
http://www.diystompboxes.com/smfforum/index.php?topic=89055.0

I've changed some values around, but my latest version looks more similar to Johan's original. The bias is offset so you can get clipping in the second stage from the opamp as well as from the LED. I think I've read that real opamps clip before hitting the actual supply voltage, but I don't know when. If I were to compare this to an imaginary diode going against the LED, what would it's bias voltage be? What supply bias would I need to use to make the clip threshold 3V? Does it depend on the opamp used? I have:
4558
JRC 4558
NE 5532
TL 072

I may not get to answer the thread soon because I may have/get to move (see personal text), but I would like to know about this.

[Taylor]

So, what you want to look for in the datasheet is output voltage swing. Some opamps are "rail-to-rail" which means that the output can swing from positive supply rail to negative supply rail, but the ones you listed are not r-to-r.

Looking at the 4558 datasheet, the output swing with a supply voltage of 15v is listed as 10v-14v with varying conditions, so 66%- 93%. Check datasheet for details of conditions. Output voltage swing also drops steeply above 10khz, but we might not care too much about that for guitar.

[PRR]

No op-amp swings all the way to the rails. Some come very close, on one or both rails, but only for light loads. Most won't come closer than 1V or 2V from the rails.

BTW, FWIW: 4558 is not specified for 9V operation. Recommended starts at +/-5V which is 10V total.

Internally, it biases with a FET-Zener string. Assuming 7V Zener (standard emitter breakdown) and a couple volts to wake-up the FET, it's fading at 9V and could pass-out for slightly lower voltage. The supply current curve shows a slow fall to +/-3V (6V total), which explains why 4558 DOES often work acceptably on 9V battery.

http://focus.ti.com/lit/ds/symlink/rc4558.pdf
Page 6

Top-left diagram shows that at +/-6V, output swing is +5V -4.5V. i.e. it acts-like it can't pull up closer than 1v to the + pin, down closer than 1.5V to the - pin. Curves are fairly straight from +/-6V to +/-17V, though drop increases a bit. We may hope the trend continues down to +/4V (typical "9V" battery). So assume 1V-1.5V drop from each supply pin.

This is typical for all IC op-amps. While the classic spec is minimum 10V swing from 15V rail, most IC opamps do better. The usual limit is a couple of emitter-followers (0.6V drop) plus some parts to drive these emitter followers. So 1V-2V drop each side.

Lower-left shows slight improvement for loads higher than 2K and steep falloff for loads below 1K.

[familyortiz]

Also look at the common mode input voltage range of the opamp. Swings beyond this range would tend to clip.
BTW, the comment about being at best 1 or 2 volts off the rails is not true. There are many flavors or rail to rail input and/or output opamps that do just that.

[BadIdeas]

So it looks like maybe I can get good results using an NE5532 and biasing between a 10k and 15k resistor (I think Johan used a 22k for the second one to get about 6V bias). Are there any clues about how hard or soft opamp clipping is?