I just designed and built an audio amp that puts 1 Watt (8 volts peak to peak) into an 8 ohm load. Now that it's built I'm testing it and documenting it. I noticed that at its maximum power, my amp shows a slightly bent triangular or sawtooth wave. It seems to me a perfect amplifier would not have this bend and and that there would be no bend if the ideal linear waveform coming from my function generator was perfectly reproduced by my amp at full power. I was wondering if the midpoint's actual position on the oscilloscope vs the mathematically correct position are used to create a percentage distortion figure that could be put in a table or a graph showing power vs percent distortion. Does anyone know about this?

 

An amplifier with no distortion produces an output that looks like a sine-wave if its input is a sine-wave, produces an output that looks like a triangle-wave if its input is a triangle-wave and produces a sawtooth when its input is guess what.

Straight parts that look bent indicate a frequency response that cuts low frequencies or high frequencies.

 

I noticed that at its maximum power, my amp shows a slightly bent triangular or sawtooth wave. It seems to me a perfect amplifier would not have this bend and and that there would be no bend if the ideal linear waveform coming from my function generator was perfectly reproduced by my amp at full power.

The tell tale signs of slew rate induced distortion? This assumes that your input is what you expect it to be.

I was wondering if the midpoint's actual position on the oscilloscope vs the mathematically correct position are used to create a percentage distortion figure that could be put in a table or a graph showing power vs percent distortion. Does anyone know about this?

If this is possible, I have never heard of this. I have always used a distortion analyzer. You can do it the old fashion way - by hand - when measuring the harmonics if you have a spectrum analyzer handy.

 

I use my ears to hear if the distortion is bad enough to see on a 'scope. About 1% to 3%is obvious. 10% or more is horrible. Less than 0.05% is not heard.

An error in frequency response is not distortion.

 

Thanks to both of you. I actually fixed the problem by putting a 10 pf capacitor in a place I never would have expected to put one. I found it experimentally, by touching the circuit board here and there while watching the output. Don't ask me to explain it.

It's funny how a design works well in the lab with breadboards and long leads then you make a tight unit and use solder and short leads and then pow! spurs! I've seen many a commercial board with capacitors added on. These were such fixes, too, I think.