Integrated Amp discussion...

Status
Not open for further replies.
M Code

M Code

Audioholic General
Incredible all of the attention toward THD...
Its been my experience that even more crucial than THD is IM distortion...
Often when evaluating competitive Class A-B amplifiers we found many times the audible, sonic differences were from having a quasi-complementary output devices rather than a full-complimentary output stage design. The full-complimentary design tended to cancel out some of harmonics.

Another crucial audible parameter was how were the output devices biased...
If the output devices were set at a lower level bias, then slight distortion could be audible as that device turned on. Certain amplifier design engineers for Marantz, Harman/Kardon and others would instead design their amplifiers to have the bias level higher so the output device was never totally turned off... The down side was that the amplifier would run hotter especially at idle.. So the amplifier required having significant over-design of the heat sink area & power supply.

Additionally certain amplifier designs had separate power supplies for the driver and output stages. As if the amplifier is being pushed hard and the power supply started to crap out by lowering the available VA to the driver stage this increased the probability of having audible distortion. Key point of having an independent power supply for just the driver stage, it that isolates the driver stage of contributing to the increased distortion...

Just my $0.02... ;)
 
E

<eargiant

Senior Audioholic
You don't see sales brochures like this anymore. It covers a lot of what's been discussed in this thread. TIM is right on the cover page and discussed extensively throughout.

https://www.vintageshifi.com/repertoire-pdf/pdf/telecharge.php?pdf=Sansui-AU-X1-Brochure.pdf

I believe it was Matti Otala that got the basis for the IM idea from someone else before him and ran with it.

Incidentally, Ayre Acoustics seems to have revived the Diamond Output stage developed by Sansui. It is currently used in their top amplifiers. Charles Hansen once said that in his estimation "the diamond circuit, used as an output section, simply sounds better."

All I know is that my fully restored AU-X1 is not going anywhere.
 
Last edited:
GrimSurfer

GrimSurfer

Senior Audioholic
That's why I'm a dinosaur; it concerns me. Sticking a second-order analog L-C filter in the amplifier output path, with whatever associated side-effects they might entail, and then specifying that the published measurements will be strictly limited to 20KHz feels like don't-pay-attention-to-the-man-behind-the-curtain. Class AB audio amplifiers are just too high performing and well understood for me to evolve into a bird and fly with the Class D products.
I'm of exactly the same view, Irv. It is about manufacturers making good choices based on what technology offers, not choosing tech and then having to apply extraordinary measures to make it work. Kind of...

Like you, I feel that AB represents a sweet spot of sorts. This is likely the result of forty years of continuous development.

Class D may get there some day, but it still has a way to go before it's ready for general amplification duty.
 
Matthew J Poes

Matthew J Poes

Audioholic Chief
Staff member
It really has arrived. The LC filter on the output is far from an extraordinary measure. Many of the worlds best Class A and AB also have LC filters on the output stage. They are used to avoid oscillation from external feedback into the feedback loop itself.

Modern Class d amplifiers offer extraordinary advantages over Class AB amplifiers. The ultrasonic noise is nothing more than a measurement artifact. It has no impact on the sound quality of performance of the amplifier anywhere near its operating bandwidth.
 
highfigh

highfigh

Seriously, I have no life.
An amplifier designed with high feedback will often sound terrible when driven hard by a high dynamic range source.... Now U are stepping in John Curl's territory for TIM.. :cool:
Haven't heard about this- care to explain, with specifics?
 
Matthew J Poes

Matthew J Poes

Audioholic Chief
Staff member
Haven't heard about this- care to explain, with specifics?
It has been shown that the addition of high feedback (all else held constant) reduces THD and some forms of IM dramatically but increase a far more audible TIM or transient internodulation distortion. Such designs can measure well but sound bad.

I would say that there is no debate about this issue other than to what degree it matters. This phenomena is well known amongst designers but some feel that modern methods of optimizing the overall design can keep TIM at reasonable levels. Research into all of this is severely lacking and whatever difference it does make is bleeding edge. Audible but likely swamped by bigger problems.

You can process wav files with added TIM to get a sense of what it sounds like. If I get a chance to do this I can try to create some.
 
M Code

M Code

Audioholic General
It has been shown that the addition of high feedback (all else held constant) reduces THD and some forms of IM dramatically but increase a far more audible TIM or transient internodulation distortion. Such designs can measure well but sound bad.

I would say that there is no debate about this issue other than to what degree it matters. This phenomena is well known amongst designers but some feel that modern methods of optimizing the overall design can keep TIM at reasonable levels. Research into all of this is severely lacking and whatever difference it does make is bleeding edge. Audible but likely swamped by bigger problems.

You can process wav files with added TIM to get a sense of what it sounds like. If I get a chance to do this I can try to create some.
Certain experienced analog Class A-B amplifier design engineers understand very well about negative feedback.. Thats why they design the amplifier 1st with NO feedback and then add slightly some just to get a better, marketable THD spec. The poor sonic performance of an amplifier with high negative feedback is very apparent when driving loudspeaker with a high dynamic range, lossless source.. Google TIM, John Curl and U can find tons of info about this subject..

Just my $0.02... ;)
 
Last edited:
Matthew J Poes

Matthew J Poes

Audioholic Chief
Staff member
Certain experienced analog Class A-B amplifier design engineers understand very well about negative feedback.. Thats why they design the amplifier 1st with NO feedback and then add slightly some just to get a better, marketable THD spec. The poor sonic performance of an amplifier with high negative feedback is very apparent when driving loudspeaker with a high dynamic range, lossless source.. Google TIM, John Curl and U can find tons of info about this subject..

Just my $0.02... ;)
Of course some do. Bob cordell and even Marshall Leach extensively studied and discussed this topic. John Curl is just one of many engineers that took this seriously. I’m not suggesting class AB amplifiers all have high TIM and sound bad. Far from it. However there are a lot of low cost well-measuring amplifiers that achieve this feat through more basic high open loop gain designs. While they may look good on paper, I think they may sound bad.

Remember my originally point came from pointing out that Denon had some amplifiers that had exceptionally low distortion and they were being argued as really good amplifiers. I was making the point that while they might be good, you can achieve exceptional measured performance (since few reviewers ever bother to measure TIM) while still having a poor sounding amplifier.

There are many amplifiers in existence, even today, which high measured TIM. Some studies examining the audibility of TIM established a threshold as low as .3%. I don’t currently have the tools to measure this, but when I first learned of the topic I copied the proposed approach to measuring TIM using a voltmeter, oscilloscope, and a digital tone generator to create the modulated square wave. Assuming what I did was right I was measuring TIM on most of the amps I had in my possession at rates all higher than that.

I plan to purchase some measurement equipment to use in reviews and I’d love to add more valuable distortion metrics to my reviews. TIM could be interesting and possibly enlightening.
 
M Code

M Code

Audioholic General
Of course some do. Bob cordell and even Marshall Leach extensively studied and discussed this topic. John Curl is just one of many engineers that took this seriously. I’m not suggesting class AB amplifiers all have high TIM and sound bad. Far from it. However there are a lot of low cost well-measuring amplifiers that achieve this feat through more basic high open loop gain designs. While they may look good on paper, I think they may sound bad.

Remember my originally point came from pointing out that Denon had some amplifiers that had exceptionally low distortion and they were being argued as really good amplifiers. I was making the point that while they might be good, you can achieve exceptional measured performance (since few reviewers ever bother to measure TIM) while still having a poor sounding amplifier.

There are many amplifiers in existence, even today, which high measured TIM. Some studies examining the audibility of TIM established a threshold as low as .3%. I don’t currently have the tools to measure this, but when I first learned of the topic I copied the proposed approach to measuring TIM using a voltmeter, oscilloscope, and a digital tone generator to create the modulated square wave. Assuming what I did was right I was measuring TIM on most of the amps I had in my possession at rates all higher than that.

I plan to purchase some measurement equipment to use in reviews and I’d love to add more valuable distortion metrics to my reviews. TIM could be interesting and possibly enlightening.

Mattew... The guy I remember meeting many years back in Europe for making me aware of TIM was Matt Otala from Finland. Besides TIM there are other crucial specs affecting how an amplifier sounds including rise time, slew rate..
IMHO.. The sonic differences often mentioned about different amplifiers is also dependent upon the actual load the loudspeaker puts on the amplifier. Depending upon the loudspeaker's X-over design/specs, it can throttle back an amplifier's output and even trigger the protection circuitry. And the audible differences can be the action of the protection circuitry. Here is a link for his Otala's AES presentation.

http://hifisonix.com/wordpress/wp-content/uploads/2010/10/Measuring-TIM.pdf.pdf

Just my $0.02.... ;)
 
Matthew J Poes

Matthew J Poes

Audioholic Chief
Staff member
Mattew... The guy I remember meeting many years back in Europe for making me aware of TIM was Matt Otala from Finland. Besides TIM there are other crucial specs affecting how an amplifier sounds including rise time, slew rate..
IMHO.. The sonic differences often mentioned about different amplifiers is also dependent upon the actual load the loudspeaker puts on the amplifier. Depending upon the loudspeaker's X-over design/specs, it can throttle back an amplifier's output and even trigger the protection circuitry. And the audible differences can be the action of the protection circuitry. Here is a link for his Otala's AES presentation.

http://hifisonix.com/wordpress/wp-content/uploads/2010/10/Measuring-TIM.pdf.pdf

Just my $0.02.... ;)
I believe Otala is credited with discovering or at least bringing to the forefront the notion of TIM. His work is what I based my measurement approach on.

There are lots of factors effecting amplifier sound, some of greater issue than others. That is really a bigger topic. I was just making a very specific point.
 
E

<eargiant

Senior Audioholic
Yup, I mentioned Matti Otala in post #62 and while I don't think he was he one that came up with the idea he ran with it.

Anyone that's curious can listen to my restored Sansui AU-X1 and compare it to the Class D Peachtree Nova 150 I have on hand.
 
GrimSurfer

GrimSurfer

Senior Audioholic
Yup, I mentioned Matti Otala in post #62 and while I don't think he was he one that came up with the idea he ran with it.

Anyone that's curious can listen to my restored Sansui AU-X1 and compare it to the Class D Peachtree Nova 150 I have on hand.
I suspect that a well restored AU-X1 would do very well against a Class D amp such as the Peachtree.
 
highfigh

highfigh

Seriously, I have no life.
It has been shown that the addition of high feedback (all else held constant) reduces THD and some forms of IM dramatically but increase a far more audible TIM or transient intermodulation distortion. Such designs can measure well but sound bad.

I would say that there is no debate about this issue other than to what degree it matters. This phenomena is well known amongst designers but some feel that modern methods of optimizing the overall design can keep TIM at reasonable levels. Research into all of this is severely lacking and whatever difference it does make is bleeding edge. Audible but likely swamped by bigger problems.

You can process wav files with added TIM to get a sense of what it sounds like. If I get a chance to do this I can try to create some.
The reason I asked is that I looked at the specs for the Parasound A21+ and shows that TIM is unmeasurable while THD is higher than I would have expected. I remember when amplifiers tried to get THD as low as possible and it often appeared in the <.001% range. OTOH, it wasn't long before, that we had people coming into the store to ask "How many amps does this put out" and, because Sansui harped about it, "What's the Slew Rate?".
 
GrimSurfer

GrimSurfer

Senior Audioholic
It really has arrived. The LC filter on the output is far from an extraordinary measure. Many of the worlds best Class A and AB also have LC filters on the output stage. They are used to avoid oscillation from external feedback into the feedback loop itself.

Modern Class d amplifiers offer extraordinary advantages over Class AB amplifiers. The ultrasonic noise is nothing more than a measurement artifact. It has no impact on the sound quality of performance of the amplifier anywhere near its operating bandwidth.
Here's an interesting article on Class D amplification:

https://www.eetimes.com/document.asp?doc_id=1274731#
 
P

PENG

Audioholic Slumlord
It has been shown that the addition of high feedback (all else held constant) reduces THD and some forms of IM dramatically but increase a far more audible TIM or transient internodulation distortion. Such designs can measure well but sound bad.

I would say that there is no debate about this issue other than to what degree it matters. This phenomena is well known amongst designers but some feel that modern methods of optimizing the overall design can keep TIM at reasonable levels. Research into all of this is severely lacking and whatever difference it does make is bleeding edge. Audible but likely swamped by bigger problems.

You can process wav files with added TIM to get a sense of what it sounds like. If I get a chance to do this I can try to create some.
I am sure one can do all those things to prove the ill effects of feedback, and present papers. In practice, I am confident the major reputable AVR manufacturers such as D&M, Sony, Onkyo, Pioneer, NAD, Anthem and Yamaha won't go crazy on applying feedback indiscriminately and/or excessively just to get THD down while ignoring the negative impacts of higher order harmonics, IMD and TIM. They are not stupid, they know numbers on paper can get them only so far and so long. I know you are not suggesting the contrary, but are making the point about the ugly side of negative feedback, if not used wisely. And, partly also because of the title of this thread, I thought I would try to calm down people who might potentially become a little nervous and suddenly start hearing IMD and TIM from their low THD spec gear that they never thought were there until now.:D:D
 
P

PENG

Audioholic Slumlord
I would love to read the paper one of the paper on the audibility of TIM, the last sentence in the abstract itself is an interesting read already, it says:

The most sensitive group of listeners could reliably perceive 0.5 % of momentary TIM. Low values of TIM were generally perceived only as slight changes in the tonal character of the sound, and not as distortion. In a number of cases, a preference was found for the slightly distorted sound.
 
GrimSurfer

GrimSurfer

Senior Audioholic
The reason I asked is that I looked at the specs for the Parasound A21+ and shows that TIM is unmeasurable while THD is higher than I would have expected. I remember when amplifiers tried to get THD as low as possible and it often appeared in the <.001% range. OTOH, it wasn't long before, that we had people coming into the store to ask "How many amps does this put out" and, because Sansui harped about it, "What's the Slew Rate?".
Many of these measurements require some interpretation before their significance can be fully appreciated.

A spec of 0.001% THD means less than were I to say that the distortion of an amp lies at -100dB. Things come into even sharper focus when I compare this with human hearing, which (IIRC) performs no better than -120dB detection in ideal circumstances (virgin ears, good health, high concentration, low ambient noise, lab environment).

To that we need to add the type of harmonics. Even order harmonics are part of everyday sound and are often interpreted as a normal part of the fundamental frequency being heard by the human ear. Odd order harmonics are not normal and are interpreted differently.

Now if a competing manufacturer were to say that *their* amp's THD was 0.0001%, it doesn't mean that the noise is ten times lower than the competing unit. It means that the noise is less by a very larger factor because this equates to a THD of -120dB. Not only is this -20dB better, but the nature of the decibel scale means that it has at least "half as half" as much noise.

So the difference between 0.001 and 0.0001 is like the difference between going to the moon and going to Mars. Big difference.

Now think about what this means when the THD is advertised as 0.01 or 0.1 in a tube amp. The sound might be pleasing because of the nature of the noise (even harmonics) but it is still quite noisy and well within the hearing range of retired deep sea divers and artillery officers. And that, my friends, IS noisy (though not quite as bad as the phase distortion of many loudspeakers at their crossover points).

In an earlier post, I provided a link to an EE Times article on noise measurement in Class D amps. I need to read more on the subject, but the thrust of the article is that audible non linearity is a pernicious problem inherent to Class D topologies. The author of the piece designs Class D amps, so this is not an advocacy article written by a Class A or AB designer using misdirection to avoid discussion on gain or switching noise. It's an electrical engineer talking about (what he sees as) an engineering problem.

Much gets buried (or worse, goes unmentioned) by Class D audio amp manufacturers who, with some but not unassailable legitimacy, drop a really impressive THD figures in a spec sheet.

Manufacturers throw figures around in the manner that best suits them. For a variety of reasons, few take the time to explain what those measurements mean or why they may be relevant (or irrelevant) to the consumer.

This is not a disease unique to Class D amp manufacturers, it is an industry wide problem.
 
Last edited:
Irvrobinson

Irvrobinson

Audioholic Spartan
The ultrasonic noise is nothing more than a measurement artifact. It has no impact on the sound quality of performance of the amplifier anywhere near its operating bandwidth.
Then why does the Class D industry universally specify measurements with a 20KHz brick wall filter? Have you measured a Hypex amplifier without a filter, to plot the artifacts?
 
Status
Not open for further replies.
newsletter

  • RBHsound.com
  • BlueJeansCable.com
  • SVS Sound Subwoofers
  • Experience the Martin Logan Montis
Top