Integrated Amp discussion...

[AcuDefTechGuy]

Over the din and from a distance, I heard a piano playing and it really sounded like a piano, not reproduced music. As I moved closer, it still sounded like a real piano and when I stood closer to them, it STILL sounded like a real piano!

You mean it doesn’t sound like a real piano in your system?

I have a real piano in my Study room.

And my HT system sounds like a real piano.

 

[GrimSurfer]

I pulled the following from a manufacturer's web site to reinforce what I've thinking and saying wrt Class D amps. I've edited out a word or two to protect the identity of the manufacturer because my argument isn't with them as much as it is with industry wide practices that make it difficult for an audiophile to understand where one piece of technology sits with regard to another. Truth be told, I respect this manufacturer for the products it makes and believe they represent reasonable value for money.

Quote Begins:

"xxxxx had planned from the outset to take advantage of Class D’s high current, low distortion, ruler flat linearity for the output stage—knowing full well it is the input stage that determines sonic character. "

Quote Ends.

In my personal opinion...

Point #1. Planning to do something ("ruler flat") and actually achieving it are different things. Nowhere in the description or specs for the product does it talk about the product's linearity. It never says that the product delivers ruler flat linearity. It only says that the engineer planned to bake this into the design. These are two different things.

Point #2. If I compare the first and second phrases, am I being asked to believe that an amp's current delivery in the output stage doesn't determine its sonic quality? How about distortion in the output stage, doesn't that have any impact on an amp's sonic quality too? Doesn't linearity matter either?

I suspect that it all matters, but would concede that some things might matter more than others. But the manufacturer doesn't explicitly state say anything definitive on the output stage's ability to influence sonic quality.

The prose appears (to me) to be an attempt to shif the customer's attention away from that which they don't want to talk about (D class output stage) to that which they do (Class A input stage). A critical reader cannot be certain of the motivation behind this, but it would be reasonable to think that the product has a much better input stage than output stage. But how can this be if the output stage has all of the ideal attributes listed in the product plan? Well, maybe there's a difference between what was planned and what was achieved. It's a distinct possibility.

Examined in this way, it would appear that my concern over a lack of specs and limited dialogue on Class D linearity issues isn't unreasonable. We're not being given any information that the linearity issue inherent to Class D amplification has been fully resolved, or even resolved to a point where it no longer matters. Instead, were being asked (in this case) to focus on the Class A input stage.

This is something that concerns me as much as when other manufacturers use the same type of misdirection to paper over shortfalls in their A and AB designs. Broadly speaking, it is an issue that Audioholics, as a reviewer of audio equipment, appears to work very hard to resolve through its product tests.

The difference is that we audiophiles (and Audioholics) have had much more time to become familiar with how Class A and AB amps are measured, where the engineering challenges lie, etc. So we know what to look for... most of the time. The critical reviews posted by Audiholics demonstrate this.

But we all haven't had that long to know what to look for in a Class D amp -- Only four years if one accepts this particular Class D manufacturer's contention (stated elsewhere on their web site) that high quality Class D audio amps have only been possible since about 2015.

That doesn't seem to bother some folks as much as me saying that Class D amps may require a little more development to overtake Class AB for audiophile duties. It's not that I'm standing on Appalachian granite as much as the whole industry is currently standing on Louisiana bayou. And here I am being called out for saying Class D isn't there yet instead of receiving convincing data that it is.

So if the industry wants to resolve this issue in everyone's mind (ie. not just those who believe in what they are being told, sort of, and hearing) they need to speak directly to how the various design challenges of Class D have been resolved. They need to develop tests and list appropriate specifications of their products. Then, and only then, can we turn this discussion into something more than a product vs product discussion or Beavis and Butthead (to quote an earlier, and very funny meme that shadyJ posted) cage match.

 

[PENG]

May be we should talk about bottleneck analysis, that probably should be done from the beginning of the recording process to the speaker output in room. Even if we only want to consider electronics, wouldn't it be the mic, the first electronic device the mic(s) is/are connected to, the 2nd, or the 3rd...., instead of just worry about the 1st, 2nd, or output stage performance of preamps, power amps etc., in the "audiophile's" comfortable room?

There are quite a few electronic devices in between too, such as the media players. I would think that some of those devices are also affected by their FR, THD, TIM, IMD, slew rate, input/output impedance etc. so shouldn't they be considered in any bottleneck analysis? If you narrow it down to just the integrated (because of the thread title) amp, then just for example, in Dr. Rich's bottleneck analysis, he identified the volume control LSI IC as one of the weak point of AVRs and some AVPs, but he seemed to have assumed even the cheap integrated amps don't use those LSI chips. That is not the case, as I know the A-S801 (not picking on Yamaha, it just happens to be an example I know of) uses the kind of LSI chip he mentioned for volume control so it seems reasonable to assume there could be other integrated amps that also have similar bottlenecks based on the use of similar LSI ICs for volume control. Imo it is a shame that most people ended up relying on manufacturer's marketing and internet hearsay. That's natural though, as there are not too many facts and science based objective studies in the so called "hifi" audio field.

We do a little better on the psychoacoutics side obviously and fortunately. No surprise there either, audio electronics are not health and safety related, it is mostly a hobby/entertainment thing. So even if there are enough EE/Psycoacoustics PhD/professors who are hifi enthusiasts, or even audiophiles themselves, they won't get much funding or motivation to do such facts and science based studies in this field anyway. So we are stuck with roughly two camps debating forever, the "all well designed amps sound the same if used within their limits, level matched" blablabla one one side and the even well build/designed amps could sound different from arctic cold to Death Valley hot and/or from clinical, analytical to silky smooth and golden warm kind of sound blablabla on the other. Neither group seem to bother considering if there are obvious weaker devices upstream that may just render whatever minor differences introduced by their beloved amps moot. Most do seem admit there are bottleneck downstream, the in room performance of their speakers for example, but that just sort of defies their logic, at least to some extents.

Want to see some marketing hypes? Below is one from the Benchmark media website:

https://benchmarkmedia.com/products/benchmark-ahb2-power-amplifier

It is not an integrated amp, but similar in physical size.

"The AHB2 is the ideal match to the performance limits of High-Resolution-Audio (HRA) formats. Benchmark's AHB2 power amplifier delivers musical details without introducing the masking effects of amplifier noise and distortion. The dynamic range of the AHB2 is 132 dB, making it 10 to 30 dB quieter than some of the very best reference-quality power amplifiers.
The frequency response of the AHB2 extends beyond 200 kHz. With the AHB2, you will discover previously overlooked nuances, dynamics, and detail in your favorite recordings and experience the full potential of the latest high-resolution 24-bit PCM and 1-bit DSD audio formats."

I am still considering buying either that amp, or the Lyngdorf TDAI-2170, but I do believe the advertised superior noise/dynamic specs of the AHB2 won't do a thing to improve sound quality in any of my systems due to bottlenecks up and down stream. If I did buy one of the two, it will most likely be my last amp purchase.

 

[GrimSurfer]

May be we should talk about bottleneck analysis, that probably should be done from the beginning of the recording process to the speaker output in room. Even if we only want to consider electronics, wouldn't it be the mic, the first electronic device the mic(s) is/are connected to, the 2nd, or the 3rd...., instead of just worry about the 1st, 2nd, or output stage performance of preamps, power amps etc., in the "audiophile's" comfortable room?

There are quite a few electronic devices in between too, such as the media players. I would think that some of those devices are also affected by their FR, THD, TIM, IMD, slew rate, input/output impedance etc. so shouldn't they be considered in any bottleneck analysis? If you narrow it down to just the integrated (because of the thread title) amp, then just for example, in Dr. Rich's bottleneck analysis, he identified the volume control LSI IC as one of the weak point of AVRs and some AVPs, but he seemed to have assumed even the cheap integrated amps don't use those LSI chips. That is not the case, as I know the A-S801 (not picking on Yamaha, it just happens to be an example I know of) uses the kind of LSI chip he mentioned for volume control so it seems reasonable to assume there could be other integrated amps that also have similar bottlenecks based on the use of similar LSI ICs for volume control. Imo it is a shame that most people ended up relying on manufacturer's marketing and internet hearsay. That's natural though, as there are not too many facts and science based objective studies in the so called "hifi" audio field.

We do a little better on the psychoacoutics side obviously and fortunately. No surprise there either, audio electronics are not health and safety related, it is mostly a hobby/entertainment thing. So even if there are enough EE/Psycoacoustics PhD/professors who are hifi enthusiasts, or even audiophiles themselves, they won't get much funding or motivation to do such facts and science based studies in this field anyway. So we are stuck with roughly two camps debating forever, the "all well designed amps sound the same if used within their limits, level matched" blablabla one one side and the even well build/designed amps could sound different from arctic cold to Death Valley hot and/or from clinical, analytical to silky smooth and golden warm kind of sound blablabla on the other. Neither group seem to bother considering if there are obvious weaker devices upstream that may just render whatever minor differences introduced by their beloved amps moot. Most do seem admit there are bottleneck downstream, the in room performance of their speakers for example, but that just sort of defies their logic, at least to some extents.

Want to see some marketing hypes? Below is one from the Benchmark media website:

https://benchmarkmedia.com/products/benchmark-ahb2-power-amplifier

It is not an integrated amp, but similar in physical size.

"The AHB2 is the ideal match to the performance limits of High-Resolution-Audio (HRA) formats. Benchmark's AHB2 power amplifier delivers musical details without introducing the masking effects of amplifier noise and distortion. The dynamic range of the AHB2 is 132 dB, making it 10 to 30 dB quieter than some of the very best reference-quality power amplifiers.
The frequency response of the AHB2 extends beyond 200 kHz. With the AHB2, you will discover previously overlooked nuances, dynamics, and detail in your favorite recordings and experience the full potential of the latest high-resolution 24-bit PCM and 1-bit DSD audio formats."

I am still considering buying either that amp, or the Lyngdorf TDAI-2170, but I do believe the advertised superior noise/dynamic specs of the AHB2 won't do a thing to improve sound quality in any of my systems due to bottlenecks up and down stream. If I did buy one of the two, it will most likely be my last amp purchase.

Great post, PENG. Ambiguous wording and normative expressions can be found throughout the industry. I see no reason to exclude any manufacturer from an audiophile's critical eye.

Not to defend Benchmark, and more as a way of confirming my understanding, I have read that amp manufacturers design wideband amps as a means of reducing audible intermodulation distortion. So while we can't hear an amp capable of reproducing sound above 20kHz (ie. beyond 200kHz),there might be some audible effects if such designs keep IM distortion out of the audible range.

Is that your understanding as well?

If so, what does this do to Benchmark's claim, if anything? (I understand your issue was with bottlenecks of connected equipment, but this question sprang to mind as I was reading the quote...)

How might this apply to a Class D with a 20kHz brick wall filter on the output stage?

These are detailed questions, but some audiophiles love to wallow in the tech! (Myself included)

 

[Irvrobinson]

How might this apply to a Class D with a 20kHz brick wall filter on the output stage?

Just for clarity, Class D amplifiers do not have 20KHz brick wall filters on the output stage, they are measured using such filters as specified by the AES-17 specifications. The amplifiers themselves in the field are unfiltered.

https://www.ak.tu-berlin.de/fileadmin/a0135/Unterrichtsmaterial/KT-Labor_WS0809/1_ADDA/aes17.pdf

 

[Out-Of-Phase]

"....but it has been evident for a very long time now that solid-state electronics, particularly linestage preamplifiers and power amplifiers, have reached a point where they are effectively a solved problem such that it is exceeding difficult to tell one from another even in the most exacting A/B comparisons."

 

[GrimSurfer]

Just for clarity, Class D amplifiers do not have 20KHz brick wall filters on the output stage, they are measured using such filters as specified by the AES-17 specifications. The amplifiers themselves in the field are unfiltered.

https://www.ak.tu-berlin.de/fileadmin/a0135/Unterrichtsmaterial/KT-Labor_WS0809/1_ADDA/aes17.pdf

So if they don't have filters, then distortion in the ultrasonic output could lead to audible distortion via intermodulation. If that's a reasonable possibility, then listing a TIM Distortion spec for any Class D amp might be helpful to the consumer.

Is that correct?

 

[Irvrobinson]

So if they don't have filters, then distortion in the ultrasonic output could lead to audible distortion via intermodulation. If that's a reasonable possibility, then listing a TIM Distortion spec for any Class D amp might be helpful to the consumer.

Is that correct?

Not knowing the magnitude of whatever TIM they produce, I can’t say. My concern was merely that I was uncomfortable with designs that were used one way and specified for measurement another way. That looked suspect to me. But finally seeing unfiltered measurements on the audioscience web site has quelled my concerns for the Hypex products at least.

 

[highfigh]

Their theory sounds good (practically too I guess, based your audition experience) though understandably not much details on their website. That means potentially, if not yet, it should be an excellent alternative to some of the excellent class D amps such as Bel Canto's.

What I heard was their system, including speakers-

You mean it doesn’t sound like a real piano in your system?

I have a real piano in my Study room.

And my HT system sounds like a real piano.

I was referring to the way a real piano distributes the sound and how it provides a sense of the size of the space. It was a large hall and a small room doesn't sound the same. The 'sound' of the piano, based on frequency response, amplitude, decay of the notes and dynamics are different when reproduced because the speakers' radiate pattern is so different.

 

[everettT]

So if they don't have filters, then distortion in the ultrasonic output could lead to audible distortion via intermodulation. If that's a reasonable possibility, then listing a TIM Distortion spec for any Class D amp might be helpful to the consumer.

Is that correct?

Most do at the module level, but after a manufacturer implements the modules all bets are off. BOs first run of their ice modules were less than stellar for full range, audible or not

 

[highfigh]

The point that Class D amps are used in live applications and now, in consumer audio was made, but WRT live, at some point, I think they decide that the sound is 'good enough', since it's hardly a perfect listening environment and much of what is said to be lacking won't be audible to the majority of the audience. Also, many speakers for live sound aren't receiving input with 20Hz-20KHz response because it's just not necessary and it would sound overly bright at that SPL.

 

[GrimSurfer]

Not knowing the magnitude of whatever TIM they produce, I can’t say. My concern was merely that I was uncomfortable with designs that were used one way and specified for measurement another way. That looked suspect to me. But finally seeing unfiltered measurements on the audioscience web site has quelled my concerns for the Hypex products at least.

This is certainly a fair comment.

Need to see the specs first. But if they're absent or hidden, harboring suspicions seems prudent.

The point that Class D amps are used in live applications and now, in consumer audio was made, but WRT live, at some point, I think they decide that the sound is 'good enough', since it's hardly a perfect listening environment and much of what is said to be lacking won't be audible to the majority of the audience. Also, many speakers for live sound aren't receiving input with 20Hz-20KHz response because it's just not necessary and it would sound overly bright at that SPL.

That's how I saw things too.

Were I a musician, I wouldn't want to lug a heavy A or AB amp around. Nor would I want to take the chance that my powerful Class A/AB amp would trip circuits at my venue.

Were I in charge of sound at a rock concert, I'd want to maximize my sound output for every bit of electricity drawn. I'd be ambivalent to to lowmor moderate distortion because it likely won't be noticeable in a noisy crowded environment.

So for live music? Sure, a Class D would be the amp to choose. No question.

But one can't legitimately say that Class D's use in live venues makes it the best choice for home listening. So we seem to be in complete agreement, highfigh.

 

[highfigh]

This is certainly a fair comment.

Need to see the specs first. But if they're absent or hidden, harboring suspicions seems prudent.



That's how I saw things too.

Were I a musician, I wouldn't want to lug a heavy A or AB amp around. Nor would I want to take the chance that my powerful Class A/AB amp would trip circuits at my venue.

Were I a sound engineer managing a rock concert, I'd want to manage my electricity useage while maximizing my sound output. I'd be ambivalent to anything up to moderate distortion because it likely won't be noticeable in a noisy environment.

So for live music? Sure, a Class D would be the amp to choose. No question.

But one can't legitimately say that Class D's use in live venues makes it the best choice for home listening. So we seem to be in complete agreement, highfigh.

An extremely large venue will have plenty of power and large sound companies bright their own generators, in a lot of cases but it's cartage that makes light weight attractive. In many cases, rock acts don't even bring their own guitar/bass amps and sometimes, they rent the PA. It's just too expensive to carry everything, unless it's done as a way to maintain total control over every aspect of the show. Getting there for an arena gig and finding out that the available PA & backline amount to speakers on a stick and practice amps IS NOT going to make anyone happy, even if micing the amps would result in good sound and a more manageable stage volume level.

For some bass players, tiny amps work great and I know one who's also an EE- he doesn't just buy some POS and drag it along with him- he looks at how it was designed/built and listens. When he goes out to play, it goes into a bag with his effects and for small gigs, he can go in and out without more than one trip, with no help from anyone else.

 

[GrimSurfer]

Getting there for an arena gig and finding out that the available PA & backline amount to speakers on a stick and practice amps IS NOT going to make anyone happy...

For some bass players, tiny amps work great and I know one who's also an EE- he doesn't just buy some POS and drag it along with him- he looks at how it was designed/built and listens.

LOL. I can see that pi$$ing off the band. It might be a survivable moment if Cat Stevens was sting up, but I'd sure hate to deal with QOTSA in a similar situation. Might lead to a bit of aggro, as a Brit would say.

Bass players are pretty particular, and their role in keeping the band together musically is often overlooked. So I can see why they'd want a little more control over their rig.

 

[<eargiant]

May be we should talk about bottleneck analysis, that probably should be done from the beginning of the recording process to the speaker output in room. Even if we only want to consider electronics, wouldn't it be the mic, the first electronic device the mic(s) is/are connected to, the 2nd, or the 3rd...., instead of just worry about the 1st, 2nd, or output stage performance of preamps, power amps etc., in the "audiophile's" comfortable room?

There are quite a few electronic devices in between too, such as the media players. I would think that some of those devices are also affected by their FR, THD, TIM, IMD, slew rate, input/output impedance etc. so shouldn't they be considered in any bottleneck analysis? If you narrow it down to just the integrated (because of the thread title) amp, then just for example, in Dr. Rich's bottleneck analysis, he identified the volume control LSI IC as one of the weak point of AVRs and some AVPs, but he seemed to have assumed even the cheap integrated amps don't use those LSI chips. That is not the case, as I know the A-S801 (not picking on Yamaha, it just happens to be an example I know of) uses the kind of LSI chip he mentioned for volume control so it seems reasonable to assume there could be other integrated amps that also have similar bottlenecks based on the use of similar LSI ICs for volume control. Imo it is a shame that most people ended up relying on manufacturer's marketing and internet hearsay. That's natural though, as there are not too many facts and science based objective studies in the so called "hifi" audio field.

We do a little better on the psychoacoutics side obviously and fortunately. No surprise there either, audio electronics are not health and safety related, it is mostly a hobby/entertainment thing. So even if there are enough EE/Psycoacoustics PhD/professors who are hifi enthusiasts, or even audiophiles themselves, they won't get much funding or motivation to do such facts and science based studies in this field anyway. So we are stuck with roughly two camps debating forever, the "all well designed amps sound the same if used within their limits, level matched" blablabla one one side and the even well build/designed amps could sound different from arctic cold to Death Valley hot and/or from clinical, analytical to silky smooth and golden warm kind of sound blablabla on the other. Neither group seem to bother considering if there are obvious weaker devices upstream that may just render whatever minor differences introduced by their beloved amps moot. Most do seem admit there are bottleneck downstream, the in room performance of their speakers for example, but that just sort of defies their logic, at least to some extents.

Want to see some marketing hypes? Below is one from the Benchmark media website:

https://benchmarkmedia.com/products/benchmark-ahb2-power-amplifier

It is not an integrated amp, but similar in physical size.

"The AHB2 is the ideal match to the performance limits of High-Resolution-Audio (HRA) formats. Benchmark's AHB2 power amplifier delivers musical details without introducing the masking effects of amplifier noise and distortion. The dynamic range of the AHB2 is 132 dB, making it 10 to 30 dB quieter than some of the very best reference-quality power amplifiers.
The frequency response of the AHB2 extends beyond 200 kHz. With the AHB2, you will discover previously overlooked nuances, dynamics, and detail in your favorite recordings and experience the full potential of the latest high-resolution 24-bit PCM and 1-bit DSD audio formats."

I am still considering buying either that amp, or the Lyngdorf TDAI-2170, but I do believe the advertised superior noise/dynamic specs of the AHB2 won't do a thing to improve sound quality in any of my systems due to bottlenecks up and down stream. If I did buy one of the two, it will most likely be my last amp purchase.

If you get the Benchmark AHB2 I would strongly suggest you also get the DAC2 or DAC3 that it was designed for. Basically for the reason you described. So digital signal>DAC2>AHB2>speakers. I'm reasonably confident you will hear a difference.

IME, the line you bolded in red is not BS. At least not in the case of Benchmark products. I know most here believe that all DACs sound the same but they definitely do not IME. At least not when you get to the level of something like the DAC2. It's really amazing how a unit like that can really take a system to the next level. I've had several DACs in the <$300 range, one in the $700 range and even some of the low cost supposedly giant killers like NWAVGuy's ODAC. Many sounded similar (some cleaner than others) but all very good, but the Benchmark is in a league of it's own. Of course, the rest of the system that follows it right up to the speakers should be at a level that will allow the differences to be heard. A DAC like this is probably a waste on speakers of average performance regardless how they measure. I have never heard my digital files reproduced like that- like they said-
you will discover previously overlooked nuances, dynamics, and detail in your favorite recordings.

As far as the bottleneck in the recording session before the digital file was created, there's not much we can do about that. I think of music reproduction as if I'm trying to extract as much of the recorded content as possible, warts and all. Regardless, I also am one to frequently bring up (don't think I have in this forum though) the limitations of the recording process but mainly when I'm complaining about an old analog recording (say ~ pre-1985) being repackaged into a hi-res container. IMO in order for a file to be truly Hi-Res it must have been recorded in Hi-Res in the first place. But I diverge...

So yeah, my suggestion would be to get the DAC2 or DAC3 first and then get the AHB2 unless you are going to purchase them both at the same time.

 

[John Parks]

I know most here believe that all DACs sound the same but they definitely do not IME.

That'll get you strung up around these parts, partner.

Not to mention that I totally agree with you.

 

[GrimSurfer]

Hey <eargiant!

First off, thanks for doubling my audio bill by raising points about pairing Benchmark devices! Damn!

Second, since you indicated an interest in recording you might enjoy this piece by Bob Katz:

http://www.aes.org/technical/documentDownloads.cfm?docID=65

 

[PENG]

So while we can't hear an amp capable of reproducing sound above 20kHz (ie. beyond 200kHz),there might be some audible effects if such designs keep IM distortion out of the audible range.

Is that your understanding as well?

Not based on the description they provided on their website, and I doubt the two things are related. I think they are saying that their feed forward technique has a lot to do with achieving such low distortions overall, regardless of the nature of the distortions. I read the Stereophile review and JA's measurements that seem to show a little worse then specified but still excellent, probably unmatched by power amps that cost much more.

If so, what does this do to Benchmark's claim, if anything? (I understand your issue was with bottlenecks of connected equipment, but this question sprang to mind as I was reading the quote...)

How might this apply to a Class D with a 20kHz brick wall filter on the output stage?

I thought it was an example of how a particular component, in this case, a power amplifier in the audio signal chain from recording to speaker output in a room, can be so near perfect (say, a straight wire with gain..) in measurable criteria, yet it most likely won't make the differences they claim audible, due to the weaker components up and down stream. As an extreme example, if the source is at MP3 128 kb/s resolution and the player is a $200 Android phone connected to the AHB2 using a 3.5 mm jack to RCA wire, can you hear the differences between the AHB2 and a decent AVR such as a RX-A2080 in a single blind AB test?

I expect someone would say yes, some would say no, and that's the kind of forever debate I referred to.

 

[GrimSurfer]

Here's something I read on the Electrical Engineering forum a while back:

"TIM seems to be exacerbated by amplifiers that don't have much frequency headroom above the highest desired frequency, and a high global feedback ratio."

https://electronics.stackexchange.com/questions/156643/what-is-transient-intermodulation-distortion

The guy who wrote this has a Master's in Electrical Engineering and participates on the EE forum. So there might be something to what he's saying. It certainly appears to make sense from a feedback design perspective.

It also helps explain why so many AB manufacturers produce amps capable of output well beyond the audible range (which puzzled me until I read more about TID). Even my modest AB amp, from the 90s, lists TID as "unmeasurable low" in its spec sheet. Makes sense, given that it is a wideband feedback design (freq response of 0.5-150 kHz and a power bandwidth of 10-100 kHz).

That got me thinking about what might happen when filters are put in place that limit frequency headroom in a Class D amp.

 

[PENG]

Here's something I read on the Electrical Engineering forum a while back:

"TIM seems to be exacerbated by amplifiers that don't have much frequency headroom above the highest desired frequency, and a high global feedback ratio."

https://electronics.stackexchange.com/questions/156643/what-is-transient-intermodulation-distortion

The guy who wrote this has a Master's in Electrical Engineering and participates on the EE forum. So there might be something to what he's saying. It certainly appears to make sense from a feedback design perspective.

It also helps explain why so many AB manufacturers produce amps capable of output well beyond the audible range (which puzzled me until I read more about TID). Even my modest AB amp, from the 90s, lists TID as "unmeasurable low" in its spec sheet. Makes sense, given that it is a wideband feedback design (0.5-150 kHz).

That got me thinking about what might happen when filters are put in place that limit frequency headroom in a Class D amp.

That was negative feedback they referred to. Benchmark uses feed forward, implying they don't use any amount of feedback but I could be wrong. They are low in distortions period, whether they are THD (including crossover distortions), IMD, or TIM distortions.

Benchmark did provide a lot of details about their feed forward error correction approach.

https://benchmarkmedia.com/blogs/application_notes/feed-forward-error-correction

Back to bandwidth, for class AB amps, I believe high slew rate (hence bandwidth potentially) is one of the indicator of less likely to have bad TIM distortions. Another review did show a slew rate consistent with 100 kHz bandwidth, that's not that unusual, in fact seems average for amps at the $3000 price point, but for a 100 W rated amp its more than high enough aside from TIMs that we have measurements for, not surprisingly.

Here's the review if you are interested:
http://www.avmentor.net/reviews/2016/benchmark_dac2hgc_ahb2_2.shtml
and it says the following, about bandwidth/slew rate:
"Finally, the square wave response shows a very good transient behavior, without any particular problem. Slew rate was estimated at 16V/uS and rise time at 3.5mS, a value that corresponds to a conventionally calculated bandwidth of 100kHz."