Author Topic: Yet another Audiophile Question.  (Read 7714 times)

0 Members and 1 Guest are viewing this topic.

Offline helius

  • Super Contributor
  • ***
  • Posts: 3688
  • Country: us
Re: Yet another Audiophile Question.
« Reply #25 on: August 07, 2020, 12:38:46 am »
Is recovery from clipping that slow? Or am I missing something else?

In your second (clipped) waveform, the shape approximates a square wave, which as you know has odd harmonics up to very high frequencies.
In a typical two- or three-way loudspeaker, the crossovers provide a high-pass filter for the signal going to the tweeter; but there is no high-cut filter, so ultrasonic frequencies go to it as well. They do not go to the other drivers.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17429
  • Country: us
  • DavidH
Re: Yet another Audiophile Question.
« Reply #26 on: August 07, 2020, 01:27:25 am »
Back in the 70's hifi days 40 watt amps were good at blowing tweeters. Klipsch is saying a low powered amp could blow theirs while clipping. Likely they put a ballpark figure to not blast/play loud them with less than a 60w amp.

And worded not well with some tech support offshore rep.

I think that is exactly the situation.  The third harmonic distortion from a *lower* power amplifier driven into clipping can damage the high frequency speaker because it is producing high frequency content that did not exist and twitters specifically have low power handling capability.

The representative missed the implications of limiting distortion to 1%.

As usual, Rod Elliott explains things better than I ever could- https://sound-au.com/tweeters.htm
Relevant part here:

A persistent myth in the audio industry is that clipping damages tweeters, so you should use a bigger amp to ensure more headroom so the amp won't clip.  This claim is simply bollocks!  Take the 100W amp described above, and replace with an amp big enough to prevent clipping ... even with the additional 12dB input signal as shown in Figure 7.  Since a 100W amp was just below clipping with an average output of 16W, if we add 12dB that takes the peak amp power to 1.6kW (near enough) and the average power will be 254W.

Do you imagine for an instant that this amp won't blow the tweeters (and everything else) if the input level is increased by 12dB (until it's just below clipping)?  Everything will fail, and usually fairly quickly if the speaker was designed for a 'nominal' 100W input.  It is simply nonsense to imagine that the loudspeaker drivers in a 100W speaker can survive an average power of over 250W and peak power of up to 1.6kW.

But the situation he describes is not realistic.  One might turn a lower power amplifier up causing clipping but a higher power amplifier would not have been turned up to full power in the first place because it would already have been loud enough.  If I need 20 watts of low and midrange for comfortable listening, and use a 10 watt amplifier, then I turn it up past the clipping level generating more high frequency distortion while a 40 (or 20) watt amplifier turned to 20 watts produces practically none.

 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Yet another Audiophile Question.
« Reply #27 on: August 07, 2020, 01:58:24 am »
Another interesting fact about the 3116D class D amp is that it's distortion is 10% max.

If you look at my graphic above and the sound level, it's so loud at 1% THD that it would never need to be louder, and that's at 50% power at 19V.  If I pushed it to max power and THD 10%, at 19V, I could get 95dB out of it at 10'.

Isn't 95dB at 10% THD considered pretty decent?

The point is that even if I push it to max at 19V, the dBs don't change. (24 watts to 28 watts)

At 10'.
More interesting is that:
 if I push the watts up from 19V at 1% THD: 23 watts = 94Db
 if I push the watts up from 24V at 1% THD: 33 watts = 96Db
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 9003
  • Country: us
  • Retired, now restoring antique test equipment
Re: Yet another Audiophile Question.
« Reply #28 on: August 07, 2020, 02:41:50 am »
What is the basis for your statement that 10% distortion is “pretty decent”?  When I measured distortion on my amplifiers, 10% THD was found when the amplifier was noticeably clipping on a CRO.
« Last Edit: August 07, 2020, 02:44:26 am by TimFox »
 
The following users thanked this post: tooki

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Yet another Audiophile Question.
« Reply #29 on: August 07, 2020, 08:13:16 am »
What is the basis for your statement that 10% distortion is “pretty decent”?  When I measured distortion on my amplifiers, 10% THD was found when the amplifier was noticeably clipping on a CRO.

Hey Tim,

10% measured at full amp power. If think the 3116D is 10% at full power, whatever that is for voltage up to 26V. It's less than .1% at 50% power, and I don't know what the increase is for each step above 50%. So, it might not even clip at 10%. Spec Sheet:
https://www.ti.com/lit/ds/symlink/tpa3116d2.pdf?ts=1596787799909&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FTPA3116D2
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
Re: Yet another Audiophile Question.
« Reply #30 on: August 07, 2020, 12:45:14 pm »
Another interesting fact about the 3116D class D amp is that it's distortion is 10% max.

If you look at my graphic above and the sound level, it's so loud at 1% THD that it would never need to be louder, and that's at 50% power at 19V.  If I pushed it to max power and THD 10%, at 19V, I could get 95dB out of it at 10'.

Isn't 95dB at 10% THD considered pretty decent?

The point is that even if I push it to max at 19V, the dBs don't change. (24 watts to 28 watts)

At 10'.
More interesting is that:
 if I push the watts up from 19V at 1% THD: 23 watts = 94Db
 if I push the watts up from 24V at 1% THD: 33 watts = 96Db
The maximum distortion can go almost unlimited on any amp — it just depends on how much you overload it.

An amp power rating is so many watts at a chosen THD. Companies will advertise using a power level at what I consider absurd THD (like 10%) because it’s a bigger number. But some advertise at less THD, and ultimately the data sheet tells you more. Bear in mind that for a given THD, the maximum output power is dependent on the amp’s design, on the supply voltage, and on the load characteristics.

As for what’s decent or not: is 10% THD acceptable to you? (It’s not for me.) And is 95dB enough? Well are you playing in a bedroom or a stadium? Cause a TPA3116 isn’t going to fill a stadium.
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 9003
  • Country: us
  • Retired, now restoring antique test equipment
Re: Yet another Audiophile Question.
« Reply #31 on: August 07, 2020, 01:13:19 pm »
I suspect that the manufacturer was specifying power at 10% THD as a measure of the clipping level.  This is "eyballed" on a CRO be noting the flattening on the crests of the sine wave, and the THD increases rapidly above that power level, as no more voltage can be supplied by the amplifier.  This is well below a square wave output, which has about 46% THD, based on the fundamental waveform included in the Fourier series of the square wave.
Have you considered my suggestion on how to quantitatively estimate the power you actually need for your purposes?  I assume the point of this thread is that you are interested in a particular chipset to make an amplifier for your existing speakers, and you want to determine how much power is required for your application.
« Last Edit: August 07, 2020, 01:57:19 pm by TimFox »
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Yet another Audiophile Question.
« Reply #32 on: August 07, 2020, 05:20:56 pm »
Another interesting fact about the 3116D class D amp is that it's distortion is 10% max.

If you look at my graphic above and the sound level, it's so loud at 1% THD that it would never need to be louder, and that's at 50% power at 19V.  If I pushed it to max power and THD 10%, at 19V, I could get 95dB out of it at 10'.

Isn't 95dB at 10% THD considered pretty decent?

The point is that even if I push it to max at 19V, the dBs don't change. (24 watts to 28 watts)

At 10'.
More interesting is that:
 if I push the watts up from 19V at 1% THD: 23 watts = 94Db
 if I push the watts up from 24V at 1% THD: 33 watts = 96Db
The maximum distortion can go almost unlimited on any amp — it just depends on how much you overload it.

An amp power rating is so many watts at a chosen THD. Companies will advertise using a power level at what I consider absurd THD (like 10%) because it’s a bigger number. But some advertise at less THD, and ultimately the data sheet tells you more. Bear in mind that for a given THD, the maximum output power is dependent on the amp’s design, on the supply voltage, and on the load characteristics.

As for what’s decent or not: is 10% THD acceptable to you? (It’s not for me.) And is 95dB enough? Well are you playing in a bedroom or a stadium? Cause a TPA3116 isn’t going to fill a stadium.

The amp isn't meant to fill stadiums. I live in a small house, 1200 sq ft, and I could never listen to it at 50 power with the Klipsch R-51M bookshelf speakers. Way, way too loud. I can hear it fine all over the house at a few watts power. The living room is 20x20 and PLENTY of volume at a few watts.

Now, they do offer an amp with TWO 3116D chips in it. I had one of those. I sent it back. No need to blow the windows in the house. At  Way too much power for my needs.
 

Offline Bassman59

  • Super Contributor
  • ***
  • Posts: 2501
  • Country: us
  • Yes, I do this for a living
Re: Yet another Audiophile Question.
« Reply #33 on: August 07, 2020, 05:35:18 pm »
But the most common mechanism for driver failure is thermal overload. Too much power. Area under the curve.

It's common to think about audio power amplifiers as being "power" devices, because their outputs are given in watts and all of it. But they are really voltage amplifiers that can source or sink a lot of current.

This means that there's an upper limit to an amplifier's output, which is the supply rail. Hit the rail and you clip, we all know that, right?

What happens when you clip a signal? The simple answer is that the output stays at the rail for the duration of the time that the input voltage times the gain wants to exceed the rail (plus, perhaps, an overload recovery time).

And all of the time sitting at the rail means you're just dumping current, in a DC sense, into the voice coil and you're baking it.

The amplifier with the higher voltage rails ("more power") would allow a higher-level output to go through unclipped, and because real music waveforms don't sit at a peak for very long, a peak doesn't damage the driver because it's not generating enough heat to do so. The amplifier with the lower voltage rails ("less power") clips and stays at the rail and thus a constant voltage level, generating more heat in the voice coil and eventually the driver dies.

Area under the curve.

As for the Klipsch minimum-power recommendation, it's silly. As long as the amplifier output is clean, not clipped, a 5 W amplifier won't damage the speakers.
I'm confused, sorry. I do understand that a long duration DC signal will fry a tweeter. I'm just confused on why the exact same amplifier with, say, a +-12V rail can damage a tweeter more than the twin with a +-24V rail, assuming the output signals are identical amplitude (hence the 12V one is clipped)

I created two horrible Paint JPGs, showing what I mean. Same signal overall amplitude, in one case clipped at an arbitrary height we can call 12V, another full swing. In both cases, the area under the curve is in red. And in the clipped version, the area under the curve is smaller. More complex waveforms will look similar, unless the recovery time from clipping exceeds the highest audio frequency. But even at 44KHz, that would mean a recovery slower than 22 micro seconds


Again: area under the curve.

Your picture does show clipping and its effects, but it doesn't show units. Driver destruction is a function of heating. You can clip all you want, but as long as the total heat dissipated in the voice coil remains below "danger" levels, the driver is fine. For example, look at guitar amps -- their outputs are highly distorted but drivers don't die because generally the speakers installed are rated properly.

Remember that coil heating is integration. The driver might survive a minute of clipped inputs at/above its ratings, but will it survive the DJ night at your local club?

I think that for the low amplifier power levels relative to the rating of the drivers being discussed here, driver destruction is not a concern.
 

Offline Bassman59

  • Super Contributor
  • ***
  • Posts: 2501
  • Country: us
  • Yes, I do this for a living
Re: Yet another Audiophile Question.
« Reply #34 on: August 07, 2020, 05:40:37 pm »
Which is exactly why I was confused by the statement that a higher voltage rail will not cause problems. The contrary is true: assuming two identical ideal amplifiers with the same input and amplification factor, one with 12V rail one with 24V rail, the one with 24V rail will blow the tweeter faster.

If the gain of your two amplifiers is the same, then the output for any given input is the same, if you are not clipping. The "larger" amplifier will not blow the tweeter faster. The output level (volume) will be identical.

When you get into clipping then all bets are off, of course. The point is that the 12 V amplifier will clip before the 24 V amp, and you have to look at the specific conditions when clipping is occurring. How much clipping? What is the waveform being clipped? Lopping off the peak of a snare drum hit is very different from clipping a sine wave.
 
The following users thanked this post: tooki

Offline robca

  • Frequent Contributor
  • **
  • Posts: 258
Re: Yet another Audiophile Question.
« Reply #35 on: August 07, 2020, 09:30:45 pm »
Again: area under the curve.

Your picture does show clipping and its effects, but it doesn't show units. Driver destruction is a function of heating. You can clip all you want, but as long as the total heat dissipated in the voice coil remains below "danger" levels, the driver is fine. For example, look at guitar amps -- their outputs are highly distorted but drivers don't die because generally the speakers installed are rated properly.

Remember that coil heating is integration. The driver might survive a minute of clipped inputs at/above its ratings, but will it survive the DJ night at your local club?

I think that for the low amplifier power levels relative to the rating of the drivers being discussed here, driver destruction is not a concern.
I didn't add units, but assume that the two signals are identical (same input, same amplification factor, same amplifier, just one with 12V power rail, another 24V. So the curves are identical, in the 12v case they are simply "lopped off" at 12V, in the other case can swing much higher). Just assume both graphs have the same curve, same units

The area under the curve of the clipped signal is much smaller than the area under the curve of the non-clipped one. So, much less power dumped into the tweeter (there might be a bit extra due to the crossover handling of the higher frequencies generated by the clipping, but still smaller than the more powerful aplifier)

I think we are saying the same thing: a low power amplifier even if with some clipping and distortion (and higher frequency harmonics dumping more energy into the tweeter) is unlikely to fry a tweeter that can handle 4x the continuous power. In the original question, the max power of the amplifier was 22W and used at half volume, and he speaker was rated 85/340W.

Can anyone here see a case where that low power amplifier would cause problems, for normal music (i.e. not signals designed to create problems)
 

Offline JohnG

  • Frequent Contributor
  • **
  • Posts: 586
  • Country: us
Re: Yet another Audiophile Question.
« Reply #36 on: August 07, 2020, 09:50:57 pm »
A speaker rated for 85/340 W has some assumptions about the overall spectra of the input signal resembling that of music. If its all above the crossover point, and you put it all in the tweeter, you will blow the tweeter, your ears, or both, real fast, for the vast majority of tweeters. If the crossover has some attentuation of high frequency in the response curve, you may blow the crossover as well, although the tweeter will probably protect it. Clipping is a non-linear operation that will put a lot of power into the high frequency band.

John
"Reality is that which, when you quit believing in it, doesn't go away." Philip K. Dick (RIP).
 

Offline robca

  • Frequent Contributor
  • **
  • Posts: 258
Re: Yet another Audiophile Question.
« Reply #37 on: August 08, 2020, 12:22:35 am »
A speaker rated for 85/340 W has some assumptions about the overall spectra of the input signal resembling that of music. If its all above the crossover point, and you put it all in the tweeter, you will blow the tweeter, your ears, or both, real fast, for the vast majority of tweeters. If the crossover has some attentuation of high frequency in the response curve, you may blow the crossover as well, although the tweeter will probably protect it. Clipping is a non-linear operation that will put a lot of power into the high frequency band.

John
I understand that, but...

I built a couple of low power speakers (below 60W max) and looked at many designs before settling on one. Out of curiosity I looked at the specs of the tweeters used in those designs. Pretty much all tweeters I looked at could handle more than 40W continuous (only one specified the frequency as 6KHz, so definitely frequency can influence). The amplifier in the original question was 22W max. Even assuming that only high frequencies exist and the crossover is a perfect crossover sending all the power to the tweeter at the maximum amplifier power, there is sill a significant margin of safety

Now, granted, there are no specs for the LTS 1" tweeter used by the Klipsch speakers, and it's a horn loaded tweeter, so it's entirely possible it can't handle as much. But an 80W continuous speaker must be able to handle 80W of a 22KHz signal without too many problems. People routinely play speaker calibration CDs at high volumes, and those frequencies are common. And the max rating is still 340W, so there is a margin

I have a hard time understanding why, in this specific situation, using a 20W amplifier, even with clipping, can be a problem. "Area under the curve" doesn't explain it, nor higher harmonics. And, yes, clipping is non-linear and generates a ton of harmonics, but it won't generate a continuous 22W 40KHz signal dumping 100% of the power available into the tweeter.

In general, I understand perfectly how a poorly designed lower power amplifier might damage a tweeter. But I simply cannot see how in this specific scenario (22W amplifier of decent quality, hence low-ish clipping, 85/340W speaker) is a problem
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Yet another Audiophile Question.
« Reply #38 on: August 08, 2020, 09:21:51 pm »
A speaker rated for 85/340 W has some assumptions about the overall spectra of the input signal resembling that of music. If its all above the crossover point, and you put it all in the tweeter, you will blow the tweeter, your ears, or both, real fast, for the vast majority of tweeters. If the crossover has some attentuation of high frequency in the response curve, you may blow the crossover as well, although the tweeter will probably protect it. Clipping is a non-linear operation that will put a lot of power into the high frequency band.

John
I understand that, but...

I built a couple of low power speakers (below 60W max) and looked at many designs before settling on one. Out of curiosity I looked at the specs of the tweeters used in those designs. Pretty much all tweeters I looked at could handle more than 40W continuous (only one specified the frequency as 6KHz, so definitely frequency can influence). The amplifier in the original question was 22W max. Even assuming that only high frequencies exist and the crossover is a perfect crossover sending all the power to the tweeter at the maximum amplifier power, there is sill a significant margin of safety

Now, granted, there are no specs for the LTS 1" tweeter used by the Klipsch speakers, and it's a horn loaded tweeter, so it's entirely possible it can't handle as much. But an 80W continuous speaker must be able to handle 80W of a 22KHz signal without too many problems. People routinely play speaker calibration CDs at high volumes, and those frequencies are common. And the max rating is still 340W, so there is a margin

I have a hard time understanding why, in this specific situation, using a 20W amplifier, even with clipping, can be a problem. "Area under the curve" doesn't explain it, nor higher harmonics. And, yes, clipping is non-linear and generates a ton of harmonics, but it won't generate a continuous 22W 40KHz signal dumping 100% of the power available into the tweeter.

In general, I understand perfectly how a poorly designed lower power amplifier might damage a tweeter. But I simply cannot see how in this specific scenario (22W amplifier of decent quality, hence low-ish clipping, 85/340W speaker) is a problem

The watts for the TA 3116DD amp chip were about 2 x 24 watts at <.1% THD @ 19V.

Spec sheet:
https://www.ti.com/lit/ds/symlink/tpa3116d2.pdf?ts=1596921462063&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FTPA3116D2
1043608-0
« Last Edit: August 08, 2020, 10:09:09 pm by DW1961 »
 

Offline Bassman59

  • Super Contributor
  • ***
  • Posts: 2501
  • Country: us
  • Yes, I do this for a living
Re: Yet another Audiophile Question.
« Reply #39 on: August 09, 2020, 07:06:10 pm »

In general, I understand perfectly how a poorly designed lower power amplifier might damage a tweeter. But I simply cannot see how in this specific scenario (22W amplifier of decent quality, hence low-ish clipping, 85/340W speaker) is a problem

In this specific scenario, the underpowered amplifier is not a problem.
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 9003
  • Country: us
  • Retired, now restoring antique test equipment
Re: Yet another Audiophile Question.
« Reply #40 on: August 09, 2020, 07:08:22 pm »
Those graphs show power at 1% THD+N, not 0.1%.  Do you have the graph for THD vs power at a fixed supply voltage?
« Last Edit: August 09, 2020, 07:10:39 pm by TimFox »
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
Re: Yet another Audiophile Question.
« Reply #41 on: August 09, 2020, 08:32:44 pm »
Those graphs show power at 1% THD+N, not 0.1%.  Do you have the graph for THD vs power at a fixed supply voltage?
Figures 6-10, 21, and 25 of the datasheet, for various combinations of VDD, bridging mode, and speaker impedance.
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Yet another Audiophile Question.
« Reply #42 on: August 09, 2020, 09:01:30 pm »
Those graphs show power at 1% THD+N, not 0.1%.  Do you have the graph for THD vs power at a fixed supply voltage?

Page 7 in spec sheet:
https://www.ti.com/lit/ds/symlink/tpa3116d2.pdf?ts=1596921462063&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FTPA3116D2

Image:
1044092-0
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 9003
  • Country: us
  • Retired, now restoring antique test equipment
Re: Yet another Audiophile Question.
« Reply #43 on: August 09, 2020, 09:08:04 pm »
By the way, should you measure the THD of these devices, you have two practical problems:
1.  In bridged mode, you cannot connect one side of the output to ground, but the THD meter is probably grounded.
2.  Even with the low-pass filters in the THD meter, you will probably find that the high-frequency ripple exaggerates the THD, although it is inaudible through the speakers.
 

Offline Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 2078
  • Country: us
    • The Messy Basement
Re: Yet another Audiophile Question.
« Reply #44 on: August 09, 2020, 09:42:55 pm »
Maybe OT, but if you want to save tweeters, don't use simple low order crossovers. Again, area under the curve.
 

Offline bson

  • Supporter
  • ****
  • Posts: 2497
  • Country: us
Re: Yet another Audiophile Question.
« Reply #45 on: August 10, 2020, 05:55:54 pm »
Is recovery from clipping that slow? Or am I missing something else?
The tweeter sits on the high-passed output from the crossover filter(s).  When clipped the output doesn't stay at the level to the tweeter but drops off very quickly, in effect producing a transient.  So you replace a higher peak with a lower one with less area under the curve (= less energy).
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf