Talk:Square wave

Hello, I'm wondering, Where did the animations showing the 25th harmonic of the square wave come from?

From the article:

An ideal square wave requires that the signal changes from the maximum to the minimum state instantaneously and cleanly, without any delay. Clearly, this is almost impossible to achieve. In practicle, actual square waves are only approximations, and have ringing artefacts in the transition. This is further clear from the Fourier series: note that this implies that a square wave contains infinitely many harmonic components. In practice, this cannot occur as this would require infinite energy.

Which is it? Almost impossible, or impossible? This paragraph is confused, and a clearer treatment is needed, if we are to state that square waves are unphysical. In particular, a simple square wave does not have infinite energy: it has the same energy as a DC signal of the same amplitude. What is infeasible is the infinite bandwidth. -- The Anome 08:51, 23 Sep 2003 (UTC)

Impossible, now you mention it; sorry my brain's been trying to get back into gear again :) Dysprosia

---

I added some other definitions. Someone should check that:

1. They are valid
2. They all equal the same waveform (ignoring the discontinutieis), with no scalings or time shifting or phase shifting or DC offsets, etc.

I guess these are technically "rectangular pulse trains"... - Omegatron July 2, 2005 02:53 (UTC)

what's the use of a square wave?

what's the use of square wave?

Can you be more specific? - Omegatron 15:02, July 21, 2005 (UTC)

Hmmm, it's like asking, what is the use of green, or of freezing... A square wave is just a change of state between two levels that occurs almost instantaneously, it doesn't have a use, because it's just a description of something else. However, a square wave that occurs in a physical medium, such as a change of voltage in an electronic circuit, that has millions of uses. For example, the master oscillator inside the computer you are reading this with now generates a square wave, and that is used to clock all the circuitry in the computer in lock step, so things happen in a regular progression. Square waves are ubiquitous in all forms of electronics, but especially digital circuits, because it is like a binary stream 1,0,1,0,1,0,1,0... and that is very useful for measuring time, which turns out to be universally necessary. Graham 03:59, 22 July 2005 (UTC)

---

Square wave closeup

I made an image. I was just experimenting, trying some things based off of Image:Haar wavelet.png. Considering that a square wave can have any frequency, DC offset, amplitude, phase, and discontinuity value and still be considered a square wave, I don't know how useful it actually is, since it sort of implies that those things are fixed. - Omegatron 19:58, July 22, 2005 (UTC)

I could make a more generalized one very easily. - Omegatron 20:00, July 22, 2005 (UTC)

square vs sine

"However, circuits using sine waves tend to consume more power, so square waves are used wherever possible."

Why? - Omegatron 20:24, July 22, 2005 (UTC)

My reasoning is that square wave clocks are generated by CMOS devices that are either fully on or fully off, so there is little power dissipation in the output transistors. Sine waves have to be generated and buffered by linear devices, which use more power. It's like the difference between Class A (linear) and Class D (PWM) audio amplifiers. I'm sure you know what I mean. However, I just looked at the specs of some square and sine wave clock oscillators, and to my surprise the sines were no more power-hungry than the squares. Perhaps my intuition was wrong. I'll have to look further into this. --Heron 21:44, 22 July 2005 (UTC)

Ah. Well my initial intuition was that a square wave carries more energy than a sine wave of the same amplitude. I'm sure it's more involved than either of us are thinking at first glance. I hear that computer clocks are at such high frequency they can't get anywhere near a square wave anyway, with all the parasitics and so on. - Omegatron 21:59, July 22, 2005 (UTC)

It looks as if there is no clear winner in the power stakes, so I removed the relevant sentence from the article. I was trying to explain why most circuits use square clocks and not sinusoidal ones, but I picked the wrong reason. The true reason is probably complexity: a square wave oscillator can be made of just a CMOS inverter and a crystal [1], and a clock buffer can be just a non-inverting logic gate; while sine wave clock oscillators are specialised devices, and I can't find any references to monolithic sine wave clock buffers. --Heron 12:08, 23 July 2005 (UTC)

Yeah, that's probably all it is. And the sinishness just comes about because of filtering from transmission line effects or whatever. - Omegatron 13:44, July 23, 2005 (UTC)

Oh no, I didn't mean that sine wave clocks are accidental. Some devices, like the ADCs I mentioned, explicitly support differential sine wave clocks, and you can buy sine wave clock generators to drive them. They are available from all good oscillator suppliers (like this one) and come in leaded and SMT packages that look like square-wave crystal oscillators. My point is that this is all too expensive for the average digital circuit, so designers stick to square waves where possible.

I notice we have an article on clock signals, so I'll put some information there once I get it sorted out in my mind. --Heron 14:07, 23 July 2005 (UTC)