kevmurray

Not wrong, just over simplified (as I see it, I make no claims of being an authority). A clipped waveform does not contain a DC component unless you sample a very small window in time which contains only one peak of the waveform. In reality a clipped signal is the original waveform plus added higher frequency harmonics.

If you took a 100Hz sine wave for example, and summed it with a 300Hz sine wave at 1/3 voltage, summed with a 500Hz sine wave at 1/5 voltage, (ad infinium)... You would have a square wave. All from sine waves, no DC. This becomes difficult to explain without getting into mathematics. I wish I had a photo as it becomes easy when you can see it.

Edit: Some one who can explain it better than I can:
http://www.allaboutcircuits.com/vol_2/chpt_7/2.html

macguyver

A DC square wave is functionally identical to a clipped signal. As an electrician with 15+ years in both car audio and and Interprovinicial Electrical certification, I'm here to tell you that a clipped sine wave contains significant DC, in both polarities as a function of time (as you've said). As the interval between voltage at peak vs. no voltage slips towards the voltage side, damage occurs.

Try pulsing a tweeter with a DC source, like a cordless drill battery, and see how long it lasts.

The reason that the tweeters failed is not because the ratings for power handling were bogus, it was because the amount of DC present heated the voice coils. Most tweeters even in those days were liquid cooled, and some would remain warm for hours after they smoked.

The amount of energy present to a tweeter when running music power from an amplifier playing full-range signals is much less than would be present in the lower frequencies. When an amplifier runs grossly into clip, it distorts all waveforms at all frequencies, and the DC (yes, clipped sine waves are functionally DC) goes everywhere and passive filtering networks are often powerless to stop it.

Want a simple demonstration of this? KEF manufactured passive networks for their component sets that use a light bulb as a drain for DC to protect the tweeter. Other companies have used this as well. Hook up a set with various amplifiers in different sizes, and compare how often that it lights up. Lower powered amps will glow that sucker like crazy, and higher powered amps will make significantly more output before they light it up.

Even though you may think square wave are not DC in theory because their is a reversing-polarity component, it's effects on equipment are the same as DC. Simple phase rectification will turn a square wave into pulsed DC, and is done this way all the time in DC electronics for battery chargers, entry solenoids, and 1,000's of other things.

TragicMagic

So when a speaker is sent a clipped signal, it still tries to reproduce that signal, right. So the speaker will play normally, until it reaches the squared off part, then it just stands still for the duration of that square portion of the signal, then picks up on the tail end of the signal.
Now, what happens when a portion of the music I'm listening to has 100Hz sound, a 400Hz sound, and a 800Hz sound? Speakers reproduce these simultaneously (sort of, I think).
If certain information contained on a CD leads my head-unit to send a strong signal to reproduce a given sound, and my amp's sensitivity is too high, that signal can't be entirely reproduced by the amplifier, and it gets squared off. Is that about right?

Has any manufacturer/brand invented a "smart" amplifier that adjusts its sensitivity to match the input from the head-unit, so clipping will never happen? Is that something that circuitry could be designed to do?

kevmurray

In theory it would, but in reality it can't sit still. It will continue to resonate.
Yep.
Yep.
I know of no manufacturer that has done this. If gain is properly adjusted it would never kick in anyway. If you are interested there are clipping detector circuits on hobby electronics sites that can be used to retro fit an amp.

TragicMagic

Thanks, I think I get it a little more now.

What I meant by a "smart amp" is that the gain self-adjusts, perhaps based off a calibrating CD that comes with the amp...? It would use the frequencies on the CD, in conjunction with the voltage output of the given head-unit to setup its own sensitivity.

Okay, so lets take those 100, 400, and 800Hz tones that are playing simultaneously... If the 400Hz tone is sent as a strong enough signal that causes the amp to clip it, what happens to the 100 and 800Hz tones at that time. Does the amp still send them properly to the speaker, and the speaker attempts to play them while having a clipped 400Hz tone?

kevmurray

A waveform with a flat section does not automatically qualify as having a DC component. I respect your opinion and certification journeyman, but if one performs a fourier analysis on a square wave one would see there are many harmonics and no DC. It is an over simplified approach to say that switching polarity of a DC source still gives you DC. Once a square wave is rectified then yes it is DC, but a rectified pure sine would have a DC component also.

Consider a square wave achieved by switching a 12v power supply on and off (like in your cordless drill example). The output frequency is 5Hz (at 12vpp) and the polarity is never reversed. My position is that the output would have a 5Hz 12vpp fundamental frequency and many harmonics (diminishing in voltage with rising frequency). It would also have a 6volt DC component. If the polarity were reversed instead of just switching the supply off, then there would be no DC.

PS: I never said the tweeter's rating was bogus, only that it was for amp selection.

kevmurray

Not a bad idea.

They all will be sent to the speakers plus some nasty harmonics as a result of the clipped 400Hz tone.

godzilla1978

I'm oversimplifing here but the above statement is close but not quite. A clipped signal will blow a speaker because it will present to much power to the speaker. A clipped signal at high volume will(as far as the voicecoil is concerened)present around 2x the power, so if the signal was unclipped the speaker would see 50watts the same signal clipped would see 100watts. If the sub cannot handle that power damage will occur. Too much power heats up the voice coil beyond what it is built to handle. It's not the waveform it's how the clipping of the signal changes the amount of power over time. I found a link somewhere that explained it perfectly but it's on my ps3 I'll find it and post it here.

macguyver

What interesting to see, is that at higher frequencies, a clipped signal functionally becomes more like DC than one would think. At lower frequencies, the "off-time" of waveform is proportionally larger, and the speaker has a chance to cool. When the frequency increases, this off time becomes less and less. Now, you have to include how the amplifier itself reacts. It does not always shut off the output devices in a nice, smooth fashion, and this adds even more harmonics onto the system.

The effective "on time" seen at higher frequencies with a clipped signal is what causes the tweeter especially to build up heat, even if it's driven with an amplifier far below it's rated power handling.

I do seem to recall that some companies used to use anti-clipping circuitry in their amplifiers to help solve this. I'm digging deep in the mind here, but did the not the Blade SE series have this?

When you move into the pro-audio arena, many amplifiers have clip indicators, and mixers/processors have them as well, as it's just as destructive no matter what the venue.

In the AC world, we deal with 60hz as our fundamental frequency, but we have many orders of harmonics that are very devastating to all manners of things, from wiring itself to connected equipment. We have logging power quality meters to find them and diagnose problems. And many items we use manufacture modified waveforms to power equipment, household light dimmers and VFDs are good examples. But I digress a bit.

This is what I have come to understand as is happening when for many years seeing problems that should not exist (blowing 100w tweeters with 30w amplifiers).

kevmurray

There is no off time. The current simply reverses. This is where my DC comparison comes in. Power has only a magnitude, no polarity so with a square wave it might as well be DC. I think here we are in agreement, it's just that we see the cause to be different (and of course you are wrong and I'm right )

Class A and B amps do not shut the devices off for this reason and class D amps have output filters because of their switching nature.

Tweeters heat up because the harmonics pass through their high pass filters and add to their power dissipation. Remember that tweeters with passive filters are AC coupled.

Not sure about Blade. I know of "soft clipping" circuits used in some amps which dramatically reduce the harmonics but they induce clipping earlier in order to achieve it. The lesser of two evils I suppose.

We have the same trouble at work with the larger VVVF drives. You can appreciate that I'm sure.

Regarding the blowing of 100w tweeters by a 30w amp, consider the heating power of a 60w light bulb. It will burn your hand no problem (and only ~80% of that power is converted to heat). Imagine a square wave from the same amp at peak output (~60w) being dissipated in an enclosed voice coil with nothing but radiation cooling. It's an extreme example of course since no tweeter sees a square wave (for very long hopefully), but it makes my point.