How Often DO Amplifiers Clip?

[Ralf11]

Are there any data on how often amps hit clipping?

[jabbr]

Do you mean if you present a sufficiently large input signal?

 

Absent a limiting circuit, this is a function of the input, gain and power supply voltage...

[Ralf11]

12 minutes ago, jabbr said:

Do you mean if you present a sufficiently large input signal?

 

Absent a limiting circuit, this is a function of the input, gain and power supply voltage...

 

No, I mean how often "real world" conditions - in a home audio system.

[jabbr]

37 minutes ago, Ralf11 said:

 

No, I mean how often "real world" conditions - in a home audio system.

Look up the output specs of your DAC. Do you have a preamp? Look up the input specs of your amp. That will generally tell you. In the absence of a preamp perhaps at the highest volume level for most equipment.

[pkane2001]

1 hour ago, Ralf11 said:

No, I mean how often "real world" conditions - in a home audio system.

 

Mine never clip. I have the old Pass Aleph 1.2s. According to Nelson Pass these will happily drive a maximum current into a complete short for as long as you want without clipping or overheating and continue to operate in class A region while doing so. He jokes that you can arc weld with these ... I've never tried 

[jabbr]

32 minutes ago, pkane2001 said:

 

Mine never clip. I have the old Pass Aleph 1.2s. According to Nelson Pass these will happily drive a maximum current into a complete short for as long as you want without clipping or overheating and continue to operate in class A region while doing so. He jokes that you can arc weld with these ... I've never tried 

Nice

So with 60V supply and gain of 23dB = 14x -> 4.2 x 14 = 60 (ish) so your amp can tolerate a 4.2(ish) input V without clipping, or so... but assuming you were to supply an 8V sine wave, it can't hit 100v (at the 23 dB gain) and so would clip... you shouldn't see that under normal circumstances...

[esldude]

Mine never clip.  Not because of some particular property of their awesomeness.  They never clip because I properly gain stage the system.  I find out how much input can push them into clipping and make sure they do not see that.  I also have put together pieces so they couldn't clip terribly hard without that attention to detail. 

 

Common booby trapped systems are tube amps paired with tube pre-amps.  The tube amps often are rather sensitive requiring little input to reach clipping and having higher gain than solid state amps.  Many old school tube pre-amps can put out large voltage levels.  More than 20 volts being not uncommon.  They also have somewhat larger than average gain than most pre-amps.  The reason is it was needed to get really low distortion at the nominal 1.5 volts output.  It also gave the pre-amp lots of headroom before actual clipping happens.  Pairing tubed pre and power amps however meant you would need to run the volume turned way down and that clipping the amp was very easy to do. 

 

So properly gain staging a system is worthwhile to do. Makes sure you don't push it into clipping and that at nominals listening levels there is plenty of headroom not to suffer when the volume goes up. 

[jabbr]

I'll just add that many times you don't need voltage amplification rather current amplification and that's typically called a buffer. No gain, but reduced output impedance. You could use a passive attenuator with that for volume control.

[esldude]

9 minutes ago, jabbr said:

I'll just add that many times you don't need voltage amplification rather current amplification and that's typically called a buffer. No gain, but reduced output impedance. You could use a passive attenuator with that for volume control.

Yes, that is what I did for many years.  Purchased DACs like a Wadia that had a beefy 4 or 5 volt output stage with low output impedance, and used passive volume controls situated right at the amp input.

[STC]

When I was using the high gain Supratek Preamp, I suspect it had clipped often but there was no way i could tell except when I hear distortion. 

 

Now, I don't use separate preamp.  Mytek output of 1.228V is matched with Crown XLS 1.4V sensitivity. 

 

Although, crown comes with its clip protection, I have disabled them. The unit clipped only once during the Jazz variant of Manger track but I was playing very loud where the peaks were touching more than 104dB. It was partly my fault for not compensating the volume when adding DSP. 

 

I think technically, my output and input is well matched and the Amp should not clip. At least not seen the red LED yet. 

 

 

 

[GregWormald]

Amplifier output:

Consider a speaker with sensitivity of 84 dB (1 watt at 1 metre). This is not unusually low.

Every 3 dB increase in sound level requires twice the amplifier power.

So 87 dB--2 watts, 90 dB--4, 93--8, 96--16, 99--32, 102--64, 105 dB (orchestral crescendo)--132 watts. And that's at 1 metre.

 

Stereo gives a 3 dB increase, each doubling of the distance loses 6 dB.

 

So at 3 metres, a stereo of 132 watts/channel will produce a 'clean' sound of (105+3-9.5)  98.5 dB. To reproduce the orchestral crescendo will require (105-98.5=6.5 dB---let's say 6 dB for simplicity, then double twice132--264--528) 528 watts/channel.

 

Given the above, many, many amps will clip when the music is loud.

 

Greg

[Tony Lauck]

My Snell A's were 84 dB efficient and were powered by 200 wpc.  They would clip when played loud, but in practice this only happened when excessive alcohol had been consumed.  I had a large room and sat more than 12 feet from the speakers.

 

Now I have Focal Twin 6's and a Sub 6.  This has a total of 1150 watts into 90 dB efficient drivers, and can easily produce a peak SPL of 118 dB.  There is a question of hearing damage, but no question of clipping.  It is possible to turn the gain up 10 dB beyond the loudest setting I ever use and, if safely out of the room, confirm by ear that there is no clipping.  These are near field monitors and I sit about 5 feet from the drivers.

 

[gmgraves]

On 5/9/2017 at 5:45 PM, Ralf11 said:

 

No, I mean how often "real world" conditions - in a home audio system.

 

The answer should be NEVER. If your amp is clipping under real world conditions then, your amp is too small for the speakers you're running. I have a friend who bought a pair of Maggie MMGs to play from his Pioneer receiver. The Maggies kept blowing fuses. I tried to tell him that it was his amplifier clipping that was blowing the fuses, but he was convinced that Maggies were terrible speakers and the company should improve their quality control. The idea that his amp was too small simply did not register with him and neither did my testimonial that I had owned 6 sets of Maggies over the years, including MG-2s, Tympany-3Cs, MG-3s, MG-3.1s and MMGs and MG-0.7s and I have never popped a fuse in any of them.

[Ralf11]

I'm pretty sure my amp is not clipping, but I asked because of this:

 

http://tom-morrow-land.com/tests/ampchall/

 

"Amplifier Challenge"  - no info on what speakers have been used or what source material

[GregWormald]

30 minutes ago, Ralf11 said:

I'm pretty sure my amp is not clipping, but I asked because of this:

 

http://tom-morrow-land.com/tests/ampchall/

 

"Amplifier Challenge"  - no info on what speakers have been used or what source material

 

I didn't bother reading past the first paragraph or so because:

 

1. Tests are not good indicators, eg. there are lots of medical drugs that passed their double blind tests with flying colours and then after a couple (or more) years on the market were withdrawn because they caused fatalities.

 

2. I know from my own experience that amps do sound different--sometimes immediately, sometimes only after days and weeks of listening.

 

Greg

[esldude]

Close to half of the audiophiles I know will blissfully listen to their system with amps occasionally clipping.  Only two I know are the type to listen without clipping.  I knew one who somehow thought a fair amount of clipping was good, fast, energetic sound.  No matter the amp he wasn't happy until he turned it up to get that zippy sound he loved. So how often your amps are clipping is all down to their capabilities, the speaker load and how loud you listen to it.

 

Even sighted many thought to be differences disappear with careful level matching.  In the case of amps, the fact you have to put in and take out the amp causes a time delay which makes it hard to do good comparisons.  Too many other things can influence your opinion, your feelings etc. no matter how much you think it is not so.

 

As for Richard Clark's challenge I don't think I have ever seen what speakers he uses.  I have seen one long discussion where people were saying he should use a challenging load like Maggies.  That is funny.  Maggies require a fair bit of power and some current, but they are very nearly 100% purely resistive.  They have the least reactance of probably any speaker made.  This means if the amps don't clip then you are least likely to hear a difference with Maggies than any other speaker.  They won't alter response from output impedance, they won't require huge amounts of instantaneous current like some designs, all in all they are an easy load.

 

The Swedish AES has published amplifier tests of a different sort.  They developed an artificial loudspeaker load much like that of the speakers they were using.  They loaded an amp with that, tapped the voltage, attenuating it and fed it into another amp powering speakers.  The carefully level matched test was to see if inserting the amplifier in the chain sounded the same or different.  They would first do this sighted and then blind ABX style.  They had exactly two amplifiers sound the same to them in the sighted test. One was a large Audio Research SS amp from a few years back.  It sounded the same to them sighted, but in the blind test they could spot it. The other was the large Bryston amp.  It passed both the sighted and blind tests.  No other amp they tested managed this.

 

Now before I knew of their testing I did the same thing.  I haven't done it in years, but back when I owned Quad ESL63s I did that for several different amps.  I did it sighted only (with very precisely matched levels and listening to the same recordings to know the level was okay each time).  I found exactly one amp I judged to be fully transparent and undetectable.  That was a Spectral DMA50.  Some of the good SS amps came pretty close with only minor signatures.  Tube amps were all very clearly colored.  That was one big learning experience.  The big, dynamic, 3D, airy, smooth musical sound of quality tube amps was all a coloration.  People had the idea those were all things a SS amp was incapable of doing.  I could put the tube amp in front of the Spectral and you heard all of that undiminished. 

 

So I don't know how Richard Clark's test would fare, but if he had relatively easy or consistent to drive speakers it is certainly possible just as he describes.  If you don't clip something, you don't distort something, and you don't allow frequency response variations to go uncorrected, there isn't much left to hear different. 

[plissken]

1. Speakers that and room that are appropriately sized for each other

2. Amp that is properly sized to produce the needed output with out having to jam on the pre-amp signal level to get the output desired.

 

I never clip my amp. 

[Ralf11]

esl - were the good SS amps that came pretty close MOSFET designs by any chance?

 

 

I have to say that amps should come pretty far down the line in terms of changing the ultimate SQ (as long as they can drive the speakers and control them).  I am thinking of holding off an amp purchase/test until after I replace my DAC (DacMagic original) and disc player (a Cal Audio with PowerBoss which does HDCD, but won't do SACDs).

[esldude]

22 minutes ago, Ralf11 said:

esl - were the good SS amps that came pretty close MOSFET designs by any chance?

 

 

I have to say that amps should come pretty far down the line in terms of changing the ultimate SQ (as long as they can drive the speakers and control them).  I am thinking of holding off an amp purchase/test until after I replace my DAC (DacMagic original) and disc player (a Cal Audio with PowerBoss which does HDCD, but won't do SACDs).

Generally no.  This was a decade or more back mind you.  MOSFETs from then really did have a bit of grain or mist in the mids or upper mids.  A very strange combination was a smooth triode feeding a MOSFET.  It was a grainy smooth sound.    The best of the few I tried in Mosfets were the big CJ amps. 

 

One of the best were the BEL 1001 mkII amps from Richard Brown.  I seem to recall they had a MOSFET driver stage and bipolar outputs. 

[jabbr]

9 hours ago, esldude said:

 

The Swedish AES has published amplifier tests of a different sort.  They developed an artificial loudspeaker load much like that of the speakers they were using.  They loaded an amp with that, tapped the voltage, attenuating it and fed it into another amp powering speakers.  The carefully level matched test was to see if inserting the amplifier in the chain sounded the same or different.  They would first do this sighted and then blind ABX style.  They had exactly two amplifiers sound the same to them in the sighted test. One was a large Audio Research SS amp from a few years back.  It sounded the same to them sighted, but in the blind test they could spot it. The other was the large Bryston amp.  It passed both the sighted and blind tests.  No other amp they tested managed this.

 

Now before I knew of their testing I did the same thing.  I haven't done it in years, but back when I owned Quad ESL63s I did that for several different amps.  I did it sighted only (with very precisely matched levels and listening to the same recordings to know the level was okay each time).  I found exactly one amp I judged to be fully transparent and undetectable.  That was a Spectral DMA50.  Some of the good SS amps came pretty close with only minor signatures.  Tube amps were all very clearly colored.  That was one big learning experience.  The big, dynamic, 3D, airy, smooth musical sound of quality tube amps was all a coloration.  People had the idea those were all things a SS amp was incapable of doing.  I could put the tube amp in front of the Spectral and you heard all of that undiminished. 

 

So I don't know how Richard Clark's test would fare, but if he had relatively easy or consistent to drive speakers it is certainly possible just as he describes.  If you don't clip something, you don't distort something, and you don't allow frequency response variations to go uncorrected, there isn't much left to hear different. 

I mostly agree with what you've written.

 

The "perfect" amp would be perfectly linear with flat eq. That amp doesn't exist and so there are both linear differences i.e. Imperfectly flat frequency response but also *nonlinear* differences that essentially form a fingerprint that identifies electronic devices including amps. Given a good FFT each amp can be fingerprinted and identified so Clark's unfounded assertion is that these fingerprints are inaudible but common experience tells us they are.

 

Amp designers can tell you how they design for a specific signature.