What is sinewave distortion? is it top and bottom flattening or is it unequal mark space ratio? i.e the rise time being different to the fall time?
Neil.
Neil.
sine wave distortion
- Thread starter Neil Groves
- Start date
It is any deviation from an ideal sinewave.
One way to measure this is to look at the wave with a spectrum analyzer. A perfect sinewave shows only one frequency. Any harmonics displayed are distortion products.
Another way to measure is to suppress the fundamental of the sinewave with a sharp notch filter. Anything that's left is the distortion part.
Good audio amplifiers have harmonic distortion less than 0.1% of the fundamental.
One way to measure this is to look at the wave with a spectrum analyzer. A perfect sinewave shows only one frequency. Any harmonics displayed are distortion products.
Another way to measure is to suppress the fundamental of the sinewave with a sharp notch filter. Anything that's left is the distortion part.
Good audio amplifiers have harmonic distortion less than 0.1% of the fundamental.
There is more to distortion that so far offered.
When a perfect signal is presented to a real circuit (say an amp) many things happen during its passage through the amp.
The result is that the signal at the output is not a perfect copy (maybe magnified) of the input. There is either something missing, changed or added to the original.
We normally reserve the term distortion for one contributor to this difference.
The first contributor is called noise. Noise is a random signal added to the original. It is never the same twice. There are however different forms of noise with different characteristics.
The second contributor is phase shift. The phase of the signal out is usually different from the phase of the signal in. Worse, the phase shift often changes with the frequency of the signal. So a signal with components at several frequencies will suffer variable phase shifts between the components. This is sometimes called phase distortion.
The human ear is pretty insensitive to such changes at audio frequencies.
The third contributor is distortion. Distortion occurs when the instantaneous output amplitude is an incorrect match with the input.
As with noise there are several forms and sources of distortion.
The form of distortion called harmonic distortion is what I think you mean by sine wave distortion. Here the Fourier series of the output is different from the Fourier series of the input. There are other forms of distortion, for instance cross over distortion, that cannot be represented in this manner.
When a perfect signal is presented to a real circuit (say an amp) many things happen during its passage through the amp.
The result is that the signal at the output is not a perfect copy (maybe magnified) of the input. There is either something missing, changed or added to the original.
We normally reserve the term distortion for one contributor to this difference.
The first contributor is called noise. Noise is a random signal added to the original. It is never the same twice. There are however different forms of noise with different characteristics.
The second contributor is phase shift. The phase of the signal out is usually different from the phase of the signal in. Worse, the phase shift often changes with the frequency of the signal. So a signal with components at several frequencies will suffer variable phase shifts between the components. This is sometimes called phase distortion.
The human ear is pretty insensitive to such changes at audio frequencies.
The third contributor is distortion. Distortion occurs when the instantaneous output amplitude is an incorrect match with the input.
As with noise there are several forms and sources of distortion.
The form of distortion called harmonic distortion is what I think you mean by sine wave distortion. Here the Fourier series of the output is different from the Fourier series of the input. There are other forms of distortion, for instance cross over distortion, that cannot be represented in this manner.
ok thanks people, all I need it for is to stuff into filters and buffers just to experiment with transistor biasing, so what I have is perfectly adequate.
Neil.
Neil.
It can be any and/or all of the above. The bottom line is harmonic distortion. A pure sine wave has one and only one frequency and no harmonics. Causes of distortion can be legion.
ERic
The harmonic distortion are create because of a nonlinearity in amplifier.
For example the voltage gain of a Common emitter is a signal level dependent.
Av = gm*Rc ≈ 40 * Ic *Rc
For example if we have CE amplifier with Vcc = 10V; Rc = 10K and Ic = 500μA and Vce = 5V.
So the average gain is equal to Av ≈ 40 * 500μA * 10KΩ = 200V/V
If the input signal swings positive so that the collector current increases to 800μA and the collector voltage falls to Vce = 2V, the incremental gain
will be 40*800μA*10K = 320V/V
If the input signal swings negative so that the collector current
falls to 200μA and the collector voltage rises to 8V, incremental gain will fall to 40*200μA*10K = 80V/V. The incremental gain of this stage has thus changed by over a factor of 4 when the output signal has swing 6V peak-to-peak. This create a high level of distortion in the output signal.
Last edited:
Cross over distortion will indeed show up as harmonics in the Fourier series of the waveform. The harmonics will be different, of course, from those caused by other forms of non-linear distortion.
I think this is a language mismatch.
By mismatch I mean the output voltage waveform is not a faithful or exact copy of the input, except for scaling. This is not like an impedance mismatch which leads to power loss and or voltage loss, although the waveform may still be the same shape.
Consider an amplifier output
When there is zero input the output may still not be zero, but some random (usually low) output. This output is noise not distortion.
There may be a systematic (=non random) output due to poor earthing. This will show itself as hum.
Or there may be an output due to the amplifier itself oscillating. This is called motorboating when the oscillation is at low frequency.
Where is an input the output waveform may be a perfect copy but shifted in phase.
Finally there may be an output waveform which is an imperfect copy of the input in a systematic way ie the same thing happens every time. This is distortion.
Thanks Jony
Actually, I intended to ask about this.
I don't know if this is what Studiot meant.
Av= -gm.Rc
And transconductance, gm, of transistor depends on DC operating point. If the Q point moves towards the direction so that Ic increases, then gm will increase, respectively.
Relating to input level, because the Ic-Vbe characteristic only be considered as linear in input signal are small. As input signal are large the characteristic is no longer linear anymore.
Are my understanding correct?
Actually, I intended to ask about this.
This makes me imagine about transmission line. If wires are very long compared to wavelength (longer than 1/10 wavelength) then the instantaneous signal voltages at the source and the load is not equal and this maybe causes reflection.
I don't know if this is what Studiot meant.
Av changes according to operating point.
Av= -gm.Rc
And transconductance, gm, of transistor depends on DC operating point. If the Q point moves towards the direction so that Ic increases, then gm will increase, respectively.
Relating to input level, because the Ic-Vbe characteristic only be considered as linear in input signal are small. As input signal are large the characteristic is no longer linear anymore.
Are my understanding correct?
I did say it was complicated. There are several forms of distortion, not all of which show up in a harmonic analysis.
No harmonic series can either introduce frequencies that are not harmonically related or add up to any waveform with an infinite number of discontinuities.
If there are two or more harmonically unrelated frequencies in the input wave then the output wave may contain additional frequencies, not present in the input.
This type of distortion is known as intermodulation distortion and the additional frequencies introduced will be found from the usual cross products in the trigonometric modulation formulae.
Now a pure sine wave has no Fourier series or you can consider the series to have only one term. Either way the maximum slope of the waveform is 1.
Waveforms with a steeper slope than this may be subject to a form of distortion, known as slew rate distortion, which will not affect a pure sine wave.
A repetitive waveform that is continuous or contains only a finite number of discontinuities may be represented by a Fourier series. All the terms in such a series have harmonically related frequencies, that is they are all an integer multiple of the fundamental. Other frequencies are not present. That is why I referred to harmonic distortion as affecting amplitude and Jony is correct it generally arises form nonlinearity in the transfer function.
In crossover distortion there are an infinite number of discontinuities. Further the distortion is concentrated in a small portion of the waveform near zero. so serious crossover distortion can register as negligible harmonic distortion on a harmonic distortion meter, but sound horrendous to the human ear.
But even for small signal Vin = 10mV peak the harmonic distortion will be quite large (THD<10%). Because the voltage gain is not constant, gain varies with the input signal.
Again Jony is correct about non linearity.
The usual method of dealing with this, is of course, negative feedback. However no amount of negative feedback can cure crossover distortion. Again this is because the effect occurs near zero amplitude. Since there is zero or near zero output there is nothing to feed back!
Incidentally, anhnha, did you understand my point about intermodulation distortion?
That is why I mentioned instantaneous amplitude.
Harmonic distortion will introduce exactly the same amount of amplitude error to each cycle of a waveform.
Intermodulation distortion will introduce a periodically varying amount of error to each cycle, so each cycle will be differently 'wrong'. This effect cannot be modelled by Fourier analysis.
The usual method of dealing with this, is of course, negative feedback. However no amount of negative feedback can cure crossover distortion. Again this is because the effect occurs near zero amplitude. Since there is zero or near zero output there is nothing to feed back!
Incidentally, anhnha, did you understand my point about intermodulation distortion?
That is why I mentioned instantaneous amplitude.
Harmonic distortion will introduce exactly the same amount of amplitude error to each cycle of a waveform.
Intermodulation distortion will introduce a periodically varying amount of error to each cycle, so each cycle will be differently 'wrong'. This effect cannot be modelled by Fourier analysis.
I disagree with you, negative feedback can reduce this "deadzone".
For example if our push pull amp has Vbe_on = 0.7V we have a crossover region from +/-0.7V, 1.4V of a "deadzone".
But now we use opamp and we include push pull amp in the feedback loop
As you know op amp has a very large open loop voltage gain Aol.
And that means that, even very small changes in input voltage cause a relatively large change in the output voltage.
For example if op amp has a open loop gain equal to 1000V/V, 1mV change in the input voltage will result 1V change at the output voltage.
And our amplifier circuit, when both transistors are off, op amp will work in open loop. So small change in input will result much large change in Vx node voltage. And this large change in Vx node voltage will turn-on the transistor. So from the outside world it looks as if there was no crossover region.
Our example amplifier will reduce crossover region from +/-0.7V to +/-0.7mV. And it's all thanks large open loop voltage gain.
In this simulation I use op amp with open loop gain Aol = 10V/V So deadzone is reduce +/-0.5V to +/-0.05V
Last edited:
Jony, I didn't say reduce, I said cure (= eliminate).
Yes negative feedback helps and the greater the open loop gain the more you have to play with at low signal levels.
Your graphs, still show crossover distortion, just smaller.
It is however theoretically posible to eliminate non linear distortion with NFB, provided certain limits are observed.
For instance say the open loop gain is 100. If NFB is used to control the gain to reduce it to down to 30 for all input amplitudes, then the gain margin can be used to force the gain to be independent of amplitude as required.
Yes negative feedback helps and the greater the open loop gain the more you have to play with at low signal levels.
Your graphs, still show crossover distortion, just smaller.
It is however theoretically posible to eliminate non linear distortion with NFB, provided certain limits are observed.
For instance say the open loop gain is 100. If NFB is used to control the gain to reduce it to down to 30 for all input amplitudes, then the gain margin can be used to force the gain to be independent of amplitude as required.
Last edited:
How is AM or FM modulation, distortion? Distortion is usually consider to be the unwanted portion of a waveform. I would not call the intentional periodic variation of the amplitude or frequency of a carrier wave in response to a modulation signal "distortion".
The Electrician
- Joined Oct 9, 2007
- 2,970
Where are the infinite number of discontinuities? I don't see any discontinuities; at most I see discontinuities in the first derivative:
You May Also Like
-
Keysight Grows Its Line of Signal Source Analyzers to Cover Up to 54 GHz
by Jake Hertz
-
ST Drops Two Time-of-Flight Sensors for 3D Depth Sensing
by Jake Hertz
-
Remembering Caroline Haslett, a Staunch Advocate for Women in Engineering
-
Qualcomm Updates Audio Portfolio With Two New Sound Platforms
by Duane Benson
