clipping vs. limiting

[Rockin Mike]

This may sound like a strange question, but why is hard limiting better than clipping? Both are just truncating the loudest part of the signal, right?

We are taught that clipping is bad and limiting is good, but isn't the limiter just shaving the tops off the waveform even more severely than the clipping it is there to prevent?

I'm not trying to make a case for clipping, just trying to ensmarten myself.

 

[dave64o]

I'm going to guess that you mean compressing rather than clipping? Clipping is what happens when you overdrive the amp too much and it completely clips the signal it sends to the speakers. Compression and limiting are variations of the same basic sound processing technique, but varying the parameters of the effect differently to get different results.

Ovnilab has GREAT info on compression techniques, compressor types, reviews of lots of different compressors,m etc. It can explain things far better than I could.

 

[alembicguy]

Clipping tends to eat my speakers if left to its own devices.

 

[seamonkey]

Clipping is a form of limiting. Diodes are used all the time as limiters.

A good limiter is a good safety net for HF drivers.

Waveforms are complex. The high frequency is carried on the signal of lower frequencies. Limiting prevents the HF from rising as the LF compresses. i.e. it has the effect of turning down the HF when the LF would clip. Limiters have recovery time built in so they keep the HF down for a while longer than the actual clipping would occur.

Here's an interesting read with pictures of what clipping looks like:
http://www.st-andrews.ac.uk/~www_pa/Scots_Guide/audio/clipping/page3.html

For example:

Given this image of a couple of frequencies playing, both even and odd harmonics are produced. If you don't want extra harmonics introduced, then limiting helps to prevent this.

Mild clipping happens all the time and nobody cares. In cars, On TV's, on your bass, etc. It's only at the extremes that everybody seems to focus on in MI forums.

 

[Rockin Mike]

I'm going to guess that you mean compressing rather than clipping?

No, I mean clipping versus hard compression (limiting).

What does clipping do to the waveform? Takes the top off of it.
What does limiting do to the waverform? Takes the top off of it.

Why is one bad and the other good?

 

[Einherjar]

I was under the impression that limiting actually downsizes the decibel value of the signal in order retain as much volume as possible without losing any part of the wave form due to clipping which can eventually damage speakers (not to mention clipped waveforms usually sounds like ****, unless it is very carefully done on purpose for a specific effect)

 

[Rockin Mike]

Compression keeps some of the signal shape after it kicks in.
The point where the compression kicks in is called the "knee",
and above the knee a higher compression ration results in a flatter waveform.

Limiting is compression with a very high ratio, resulting in a "flat on top" waveform similar to (or just like?) clipping.

 

[Rockin Mike]

Clipping is a form of limiting. Diodes are used all the time as limiters.

A good limiter is a good safety net for HF drivers.

Waveforms are complex. The high frequency is carried on the signal of lower frequencies. Limiting prevents the HF from rising as the LF compresses. i.e. it has the effect of turning down the HF when the LF would clip. Limiters have recovery time built in so they keep the HF down for a while longer than the actual clipping would occur.

Here's an interesting read with pictures of what clipping looks like:
http://www.st-andrews.ac.uk/~www_pa/Scots_Guide/audio/clipping/page3.html

For example:

Given this image of a couple of frequencies playing, both even and odd harmonics are produced. If you don't want extra harmonics introduced, then limiting helps to prevent this.

Mild clipping happens all the time and nobody cares. In cars, On TV's, on your bass, etc. It's only at the extremes that everybody seems to focus on in MI forums.

That image would appear to be two different amplitudes.
The first one peaks at +Vc, and the second one would have peaked much higher than +Vc if it hadn't clipped.

I wish I could compare three images, one of an unmodified wave, one of the same amplitude but clipped, and another of the same amplitude but limited. How would the clipped wave look different from the limited wave?

 

[Hactar]

One way to think of clipping is that it is limiting with a compression ratio of infinity:1 and a threshold of 0 dB (or +12dB, +18dB, depending). Thus, signals that rise over that threshold are "clipped", completely prevented from rising over that threshold.

When limiting, typically a very high compression ratio is used, but infinity:1 is uncommon. Also, the threshold is typically set a little lower the maximum level the hardware can handle. This allows a little bit of headroom, so that when a signal rises above the threshold, it can still rise, but it is made to rise slower.

However, the primary difference between clipping and limiting is that limiting is done with by looking at the average of the signal. That is, a limiter does not consider each peak (electrical peak) of a signal, but rather takes a small average. Additionally, a limiter has response times on both its attack and release. These give the signal more freedom, resulting in a still musical sound.

Clipping, on the other hand, is something that occurs at the immediate amplitude of the signal. In other words, instead of looking at an average and dealing with attack and decay, a piece of equipment that is clipping simply cannot deal with such a high signal level and therefore at each electrical peak of the signal, it maxes out. It simply cannot produce a voltage high enough. In a sense, clipping is at "less refined" level, if you will. Each individual peak is clipped without consideration of other parts of the signal, whereas a limiter considers a larger part of the time domain.

Hope this helps, I just realized how difficult this is too explain.

 

[willbassyeah]

Than how does distortion pedals work? My assumption from reading the post is they produced signal that is hot enough causing the speaker to not be able to handle it thus produce distortion. And because the signal is hot, the square wave is formed? Therefore a hard clipping is produced?

 

[megafiddle]

Distortion pedals produce the distortion (clipping) right in the pedal itself. The distorted
signal is then passed on to the amp.

 

[Rockin Mike]

One way to think of clipping is that it is limiting with a compression ratio of infinity:1 and a threshold of 0 dB (or +12dB, +18dB, depending). Thus, signals that rise over that threshold are "clipped", completely prevented from rising over that threshold.

When limiting, typically a very high compression ratio is used, but infinity:1 is uncommon. Also, the threshold is typically set a little lower the maximum level the hardware can handle. This allows a little bit of headroom, so that when a signal rises above the threshold, it can still rise, but it is made to rise slower.

However, the primary difference between clipping and limiting is that limiting is done with by looking at the average of the signal. That is, a limiter does not consider each peak (electrical peak) of a signal, but rather takes a small average. Additionally, a limiter has response times on both its attack and release. These give the signal more freedom, resulting in a still musical sound.

Clipping, on the other hand, is something that occurs at the immediate amplitude of the signal. In other words, instead of looking at an average and dealing with attack and decay, a piece of equipment that is clipping simply cannot deal with such a high signal level and therefore at each electrical peak of the signal, it maxes out. It simply cannot produce a voltage high enough. In a sense, clipping is at "less refined" level, if you will. Each individual peak is clipped without consideration of other parts of the signal, whereas a limiter considers a larger part of the time domain.

Hope this helps, I just realized how difficult this is too explain.

I don't see that. If the threshold were zero and the ration was infinity, the signal would never rise to any level. It would be a flat top starting at 0. It would be silence.

also, I don't see how a limiter could reasonably look at an average of wave heights. It would have to allow some history of potentially damaging waves before acting on subsequent waves. In other words, it would have to allow some non-limited high peaks through so it could measure them and limit the next waves. By then it's too late, the spike damaged your gear.

 

[Rockin Mike]

It may be that clipping produced as a result of maxing out the hardware's capacity can damage the hardware, and limiting is a way of "early clipping" to prevent that damage.

In that case, how come they sound different? preamp clipping sounds like overdrive/distortion/fuzz, but heavy limiting just sounds squashed and un-dynamic.

Again, just trying to get edumacated.

 

[Bassamatic]

Yes - Hactar is correct. Limiting has some attack and release time to keep some sound quality. Clipping is instantaneous attack and release with a hard threshold. FYI - clipping generates tons of upper harmonics and can blow out your tweeters that are not protected.

Diodes are often used as a final limiter to catch spikes, after the limiter circuit has done it's best. However, the amount of signal that is clipped should be only a tiny amount relative to the main signal. \

Distortion pedals basically overdrive some component until it reaches it's threshold and clips. Then this clipped signal is mixed back into the main signal. what frequencies are clipped and how they are EQ'd, etc. makes the differences in various devices.

 

[Rockin Mike]

clipping generates tons of upper harmonics and can blow out your tweeters that are not protected.

That brings up another question...
When we look at one of these wave graphs, what exactly are we looking at? I get that it's volume on the Y axis and time on the X axis, but volume of what frequency? I can change the volume of different frequencies with my EQ, how can that be graphed? In fact I imagine it may be possible to clip some frequencies and not others. I know there are multi-band compressors out there (like the TC amps have).

 

[4Mal]

Time and frequency are 1 in the same. Narrower wave form, less time, higher frequency...

 

[megafiddle]

Actually, it's not volume on the Y axis, it's amplitude as it varies over very short periods
of time. The Y axis of those types of wave graphs usually represents voltage amplitude.
Volume is not totally wrong, it's just that volume is usually veiwed over longer time periods,
as a waveform "envelope".

Clipping occurs when the waveform tries to exceed the limits of the circuit; it reaches a
certain level and simply cannot go any higher (or lower in the negative direction). This
usually occurs in the power amp output stage, but can occur in any stage that is driven
hard enough.
Clipping chops off each and every peak of the waveform that is large enough to reach
the clipping level. As long as your signal is below this level it will remain clean. Any time
the level tries to exceed this level it will clip and distort.

Limiting acts on the overall level to prevent it from exceeding a certain level, but it does
this by varying the gain. If the signal tries to exceed the limit, the gain is reduced. So the
signal is not allowed to rise above a certain point.

Imagine that someone extremely fast was controlling your volume control as you were
playing. Every time you tried to play louder than a certain level, they turned down the
volume. That's what the limiter does electronically.

 

[Rockin Mike]

This is starting to make sense, thank you.

So far the most enlightening thing (to me) is learning that the difference between limiting and clipping does not have to do with the flatness of the top of the wave, but is actually about the reason for the flatness. In clipping the electronics are freaking out and acting abnormally, in limiting the amplitude is intentionally kept low enough to prevent such freaking out.

I generally think of amplitude and gain as the same thing, both are measures of voltage. Is that right?

I'm pretty sure volume and level are the same thing, measures of sound pressure level in the air, but loudness is a measure of perceived volume and factors in things like the ear's sensitivity to different frequencies and its temporary desensitization after prolonged exposure to high volume. Is that right?

 

[seamonkey]

I wish I could compare three images, one of an unmodified wave, one of the same amplitude but clipped, and another of the same amplitude but limited. How would the clipped wave look different from the limited wave?

If the input signals goes up enough to cause amplifier clipping as on the right waveform, then the limiter turns the volume down so it looks more like the left waveform.

It's not instantaneous to kick in, and not instantaneous to release.

It ends up somewhere in-between - depends on the limiter.

 

[megafiddle]

This is starting to make sense, thank you.

So far the most enlightening thing (to me) is learning that the difference between limiting and clipping does not have to do with the

flatness of the top of the wave, but is actually about the reason for the flatness. In clipping the electronics are freaking out and

acting abnormally, in limiting the amplitude is intentionally kept low enough to prevent such freaking out.

Yes, that would be a common use for limiting.

I generally think of amplitude and gain as the same thing, both are measures of voltage. Is that

right?

Amplitude is usually measured in volts, but gain is a measure of amplification. An amplifier stage will increase the amplitude
of a signal by some certain amount. For example it may increase the signal 10 times. Therefore it would have a gain of 10.
2 volts in would produce 20 volts out, 0.3 volts in would produce 3 volts out, etc.
Notice that gain is just a value, in this case it is simply "10", not 10 volt or "10 something".

I'm pretty sure volume and level are the same thing, measures of sound pressure level in the air,

but loudness is a measure of perceived volume and factors in things like the ear's sensitivity to different frequencies and its

temporary desensitization after prolonged exposure to high volume. Is that right?

You could use those terms like that. "Level" is also a general term, so you could speak of sound pressure level, input level,
clipping level, etc.