It actually works exactly like that.
I had my doubts too but then I got myself an analogue modular (subtractive) synth and if I use a sine wave generator and run that through a filter either nothing happens or the signal gets filtered out completely since sines contain no harmonics.
However if I clip the sine I suddenly have all sorts of harmonics and the filter works pretty much exactly as if I had used a square wave in the first place.
On the other hand if you were right subtractive synths would simply not work at all.
You can create a square wave looking thing using multiple harmonics.
However this is NOT the mechanism in an amplifier. Your amplifier is NOT a synth.
Andrew, clearly a square wave is an AC waveform.
That begs the point.
For the period of time that the "flat" part of the waveform is present, it is DC.
Any finite point on an AC waveform is also DC, witness how a DAC operates?
But the point is that the "flat" part of the waveform persists, therefore is not a finite point, but a finite period.
However this is NOT the mechanism in an amplifier. Your amplifier is NOT a synth.
Andrew, clearly a square wave is an AC waveform.
That begs the point.
For the period of time that the "flat" part of the waveform is present, it is DC.
Any finite point on an AC waveform is also DC, witness how a DAC operates?
But the point is that the "flat" part of the waveform persists, therefore is not a finite point, but a finite period.
So you are saying that the physics behind clipping a low level signal are different to the clipping of higher levels?
How so and at which voltage and/or current level do the physics change?
Bear in mind that in order to clip the output of a signal generator I have to run it through one of the amplifiers which are part of my synth.
Gosh, you guys are like a dog that won't let go of a bone! 
The main reason to use a biggish capacitor on a directly coupled tweeter is to protect it from switch-on and switch-off thumps from the amplifier, isn't it?
And it might help with the low harmonics of a clipping amp a little bit.
I still persist with my feeling that an amplifier, especially affordable ones, is usually less than ideal at high frequency, so a Zobel will do something audible to roll off the top end. Not hard to connect such networks to the speaker inputs. I must try this. In fact I seem to recall I did. And it did something.
The main reason to use a biggish capacitor on a directly coupled tweeter is to protect it from switch-on and switch-off thumps from the amplifier, isn't it?
And it might help with the low harmonics of a clipping amp a little bit.
I still persist with my feeling that an amplifier, especially affordable ones, is usually less than ideal at high frequency, so a Zobel will do something audible to roll off the top end. Not hard to connect such networks to the speaker inputs. I must try this. In fact I seem to recall I did. And it did something.
Not sure what you are speaking of.
A synth uses various means to "synthesize" a close approximation of a desired waveform/wave shape.
Low level or high level, clipping an amplifier stage is the same.
Having an amplifying stage internal to your synth does not change how the amplifying stage operates. This has nothing to do with a Fourier approximation of a given waveform (square wave or other).
So when it's being DC, is it not AC anymore... or is it still AC even when it's DC?
It's quite simple really as my synth is fully modular and there is nothing internal really:
I plug a patch lead into the output of a sine wave generator and take it to an amplifier module and from there into a filter.
As long as the amplifier does not clip the filter outputs either the sine wave intact or not at all.
When I increase the gain to drive the amp module into clipping the filter starts producing an output as the number of harmonics increase and by the time the amp is clipping hard the filters output sounds practically identical to its output had I used a square wave instead of the original sine.
The synth in question looks very much like this btw:
http://www.analoguehaven.com/doepfer/a100bs2p/a100bsp2big.jpg
For the period of time when the square wave (generated either by a flip-flop or a saturated output stage in an amp) is at a constant level, until it swings back in polarity it most certainly is DC. It can not be anything else.
In theory, for an ideal squarewave, the switch between polarity + and polarity - is instantaneous, therefore IF it actually was instantaneous it would be alternating between two DC states. So, at any instant one value of DC is found, while over time (more than one half cycle) it is an alternating wave.
_-_-
In theory, for an ideal squarewave, the switch between polarity + and polarity - is instantaneous, therefore IF it actually was instantaneous it would be alternating between two DC states. So, at any instant one value of DC is found, while over time (more than one half cycle) it is an alternating wave.
_-_-
Last edited:
ur missing the point entirely, i am afraid.
you are filtering the output. that is a different matter.
If you merely clip the amplifier sufficiently hard you will eventually produce something that seen on a scope approaches an actual square wave. the discussion is about what the "flat top" of a square wave, or in this case a highly clipped amplifier actually is. Amplifier "modules" that produce all sorts of "squirrely" looking waveforms when driven hard are a different kettle of fish.
I'm reasonably aware of what goes on with analog synths... tnx.
_-_-
If I leave out the filter the clipped sine will just sound like a square wave.
The 'flat' top (it's not flat, it's rippled) is made up of the peaks of the added harmonics.
Of course you could read this published by the University of St Andrews if you refuse to believe me:
https://www.st-andrews.ac.uk/~www_pa/Scots_Guide/audio/clipping/page1.html
And here is a little gif that shows quite nicely what happens if you progressively clip a sine:
http://upload.wikimedia.org/wikipedia/commons/f/f8/SquareWave.gif
The 'flat' top (it's not flat, it's rippled) is made up of the peaks of the added harmonics.
Of course you could read this published by the University of St Andrews if you refuse to believe me:
https://www.st-andrews.ac.uk/~www_pa/Scots_Guide/audio/clipping/page1.html
And here is a little gif that shows quite nicely what happens if you progressively clip a sine:
http://upload.wikimedia.org/wikipedia/commons/f/f8/SquareWave.gif
OK, so if it's DC when it's in the flat bit, and the transition is instantaneous, then when is it AC?
A sin wave is only AC over time, that's the whole point, TIME.
You can't chop and change between one temporal point of view and another, instantaneous or continuous.
DC is DC long enough so that mathematically it can be presumed to be at the same level over all time.
A Fourier analysis is based on the idea that a waveform repeats over all time. A square wave is AC when looked at over all time, which is the only meaningful way to look at it in this context.
Everybody's out of step except oor Wullie. Get a grip.
Fourier is a method of analysis. It is NOT the actual waveform.
Someone ought to tell me what the rate of change (slope) for the top of a square wave is.
And, as I stated, moving from one state to another is AC.
Charles Darwin, what in the world does your synths method of waveform generation have to do with anything?? I know how they work. If your "square wave" has ripple on the top then it is a c**p square wave, to be blunt.
I think this discussion has reached the point where further discussion will be without much benefit.
Someone ought to tell me what the rate of change (slope) for the top of a square wave is.
And, as I stated, moving from one state to another is AC.
Charles Darwin, what in the world does your synths method of waveform generation have to do with anything?? I know how they work. If your "square wave" has ripple on the top then it is a c**p square wave, to be blunt.
I think this discussion has reached the point where further discussion will be without much benefit.
There may be an infinite number of harmonics riding on that squarewave but it still is at a DC potential. That is a fact.
Like the guys on rc groups that think the brushless motors are dc when in reality are three phase ac synchronous. They argue that because of the battery and the drive electronic only pulse dc, therefore its a dc motor. Yo reality check boyz, remove the ground reference and what do you have, ac. Besides DC cannot be coupled wirelessly (thank you Nikola)
Like the guys on rc groups that think the brushless motors are dc when in reality are three phase ac synchronous. They argue that because of the battery and the drive electronic only pulse dc, therefore its a dc motor. Yo reality check boyz, remove the ground reference and what do you have, ac. Besides DC cannot be coupled wirelessly (thank you Nikola)
Last edited:
OK, so, here's a potentially stupid question or three from a non-EE type that usually stays out of these frays:
A capacitor can be used to block at the output of an amplifier stage, right? A cap won't pass DC by definition. Let's make that cap arbitrarily big. What happens when you clip (voltage limit) the amplifier into that cap? Do the tops of the waves get flat? Does the cap momentarily pass DC? What does DC mean in that context?
If you're bugged by this conversation having started with driving a tweeter we can make the amp a capacitior-coupled OTL tube design. Let's not get bogged down with opinions on the quality of such a beast.
I'm sure I will be enlightened by someone that actually understands the math, but my (admittedly limited) understanding is that the clipped waveform can be represented as a sum of sine waves (merci, monsieur Fourier), some of which may be of very low frequency if the clip is of 'substantial' length, but that the fundamental would still have a finite period (i.e. not be actually a DC term of zero frequency).
As far as the tweeter is concerned, the LF content might as well be DC in it's ability to heat up the VC to the point of failure, but (again, in my understanding) DC current cannot actually flow through the cap.
We're pretty far from the original post. I hope the OP got a satisfactory answer.
I'll put on my Nomex now.
Bill
A capacitor can be used to block at the output of an amplifier stage, right? A cap won't pass DC by definition. Let's make that cap arbitrarily big. What happens when you clip (voltage limit) the amplifier into that cap? Do the tops of the waves get flat? Does the cap momentarily pass DC? What does DC mean in that context?
If you're bugged by this conversation having started with driving a tweeter we can make the amp a capacitior-coupled OTL tube design. Let's not get bogged down with opinions on the quality of such a beast.
I'm sure I will be enlightened by someone that actually understands the math, but my (admittedly limited) understanding is that the clipped waveform can be represented as a sum of sine waves (merci, monsieur Fourier), some of which may be of very low frequency if the clip is of 'substantial' length, but that the fundamental would still have a finite period (i.e. not be actually a DC term of zero frequency).
As far as the tweeter is concerned, the LF content might as well be DC in it's ability to heat up the VC to the point of failure, but (again, in my understanding) DC current cannot actually flow through the cap.
We're pretty far from the original post. I hope the OP got a satisfactory answer.
I'll put on my Nomex now.
Bill
If you would have had a look at the link to St Andrews University I provided you would have noticed that it actually deals with amplifier clipping a sine and the harmonics this process creates.
Your idea of a square wave appears to be based on an idealized drawing of a perfect square wave but these do not exist in reality as that would require a system with infinite bandwidth.
Since all systems existing in the real world have limited bandwidth all 'square waves' existing in the real world will have that ripple. If you can see it or not depends purely on the resolution of your method of looking at the waves.
Just read up on square waves as this is all part of Square Wave 101.
With symmetrical clipping and AC coupling, you may pass the square wave with no reduction of the lower harmonics. Typically there will be some phase shift to them and this will result in a tilt to the flat tops.
We don't typically speak of a time varying waveform as "DC for an instant". It is either DC or AC (or AC with a DC component...). If it has gone through a capacitor then any DC component (offset) will be striped off.
If you are starting with a sine wave and then you clip it (symmetrically), there is no addition of lower components. The resultant waveform is always the fundamental and multiples of it (harmonics). Symmetric clipping only gives odd harmonics.
Heating is always directly related to the current through the resistance of the coil. Frequency doesn't matter at all if you are monitoring current rather than voltage.
No Nomex required.
David
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.