Bandwidth Limitations Of Class D Amps??

Audiogon member Kijanki said :
Some people argue that class D, as good as it is for the money, cannot really compete with best traditional SS amp. I can see that, especially with still limited bandwidth...
What are the bandwidth limitations of class D amps, and is it audible?
Bf85b117 4590 4be7 a756 d05f898cdaf9mitch4t


I've never noticed anything with my Spectron Musician III Mk2. It's bandwidth is 100 kHz:

Most Class D amplifier designs require a filter at the output to remove the high frequency switching noise that is present in the signal. Various designs use different types of filters with varying cutoff frequencies. I doubt that any are designed to remove frequencies under 20KHz, but Class D amps are generally more frequency limited at the high end than conventional output stages can be. Whether this is audible can be a subjective question. Some engineers feel that in order to avoid introducing any phase shift to the audible high frequencies requires an amp to have a flat bandwidth response extending to many multiples beyond the audible frequency limit.

* Certain legacy class D amps, for example the ones based on ICEpower modules, tend to take an excruciating amount of time to break in... Sometimes more than 1200 hours. Until they are stabilized, both bottom ends and treble ends can sound foreshortened, wooden, or even harsh.

* After break in, I have experienced some legacy class D amps to become very musical, while others have remained wooden or "matter of fact". The frequency extension and musicality of the final result may have to do with the details of the amplifier design, such as power regulation, and input isolation, and some amplifier designers may have been more successful than others in working around the idiosyncrasies of the now legacy ICEpower technology.

* The new generation of class D modules may constitute substantial engineering departures from their predecessors, and there is no evidence yet that they suffer of inherent bandwidth limitations.

any date for that Kijanki's post?
Because he seems to be using a Rowland 102 - class-D - power amplifier himself these days!

I cannot see why a class-D would have any inherent bandwidth limitations. If the class-D amp is correctly designed for 20-20KHz bandwidth I cannot see why, with today's level of technology - why it would be bandwidth limited....
If you look at Stereophile published bench measurements for Class D switching amps (say Bel Canto ref1000m for example) versus traditional class A/B (non switching amps), I think you'll see examples of what Kijanki referred to.

I would expect the switching frequency used/possible given the current state of the technology to be the technical limitation with switching amps. With current technology, effects appear to be measurable within upper limits of human hearing, ie 20Khz.

All amps have limitations. But with non-switching amps, the limitations are due to other design factors. Still, I believe many are capable of measuring flatter to well beyond 20Khz.

Of course what is measured may or may not be an indication of what is actually heard, especially when it relates to the extremes of what humans are able to hear. The upper limits of what humans can hear is probably still the weak point for Class D switching amps. How weak or how much it matters is debatable I suppose, much like what is the effects of CD redbook sampling parameters. Newer CLass D amps using newer technology might push switching frequency up higher than what was possible even just a few years ago. Have not seen measurements on those yet so not sure if the is the case (yet).

Bombaywalla, Kijanki's post was a couple of days ago on 3-15-13. His entire post is below:
03-15-13: Kijanki
Some people argue that class D, as good as it is for the money, cannot really compete with best traditional SS amp. I can see that, especially with still limited bandwidth, but most of the weight is in the power supply that can be definitely class D (SMPS) since it is only holding steady voltage (much easier task than driving complex load with music signal). Rowland does it so why not the others? I think it is for three reasons:

1. Design itself is much more difficult than just
2. SMPS have bad rap from crude cheap computer applications
3. Audiophiles still believe that it has to be heavy to
sound good.

Third point is very important - why to design complicated light power supply when market believes that heavy=quality.
Kijanki (System | Threads | Answers | This Thread)
The Thread This Post Appeared In Is In This Link
Realize that switching frequency spec of a switching amp would seem to set a clear theoretical limit to what is possible with that design in terms of frequency response. But alone it does not determine sound quality, even as measured. It may determine an absolute limit in theory, but the reality is many factors affect amp performance, switching amp or not. Specs are informative but alone do not tell the whole story. Only our ears can do that and each story will probably be a bit different as a result, especially when it is differences in the fine details of what is heard that is in question.
My Gilmore Raptors are -3db at 60khz
"My Gilmore Raptors are -3db at 60khz"

Arh, are those Class D amps an end to end proprietary design by the maker or do they incorporate another company's Class D amp modules?

ANything more to back up the published specs?
-3db means there will be phase shift problems all the way down to 10khz.
Why some say the upper mids and top end is hard could mean the the filtering is too high and letting the nasty switching noise intrude.
Why some say the upper mids and top end is bland and opaque could mean the filtering is too low and effecting the harmonic content of the upper mids and highs.

This is why you'll see class d manufactures bringing out different versions or mk 1 2 or 3 They change the filtering to the above to either or a combo of both, which I think sounds the worst.
Once technology get more advanced and they invent components whith much higher switching frequencies so they then can use the filter so the effects and phase shifts of them are well out of the audible range then Class D will compete with the good of linear amps, until then it is comprimised, but great for subwoofers.

Cheers George

You might be correct regarding the theoretical weak aspects of Class D but my experience is that they ARE competitive with other amps, not just good for subwoofers. Many users and reputable reviewers would agree.

All designs have flaws/imperfections, switching and otherwise. Gotta look at the whole picture, not just one aspect of one particular design.

It's like when Ralph from Atmasphere focuses on downside of negative feedback and says it can never sound like music. Many SS (and tube) amp owners might disagree, though his theoretical argument seems sound. One aspect of design theory only tells a part of the story though. THere are many factors that go into a good design. Most engineers think their way is best, and few do everything exactly the same, so I do not hold that against a good engineer that has done their homework.

Switching frequency just makes that aspect of Class D design very straightforward to quantify in comparison to many design aspects of non switching amps.

That is an advantage I would say in terms of being able to clearly identify an actual performance bottleneck specification with CLass D, similar to how sampling resolution defines limits of digital media.

Do non switching amps have any parameter that is similarly definitive? Not sure. That make comparisons hard and more subjective, not necessarily a good thing!

I do agree that as switching frequencies get higher, issues will diminish further. That bodes even better for the future of CLass D and seems to be happening already with newer Class D modules I have read about, like Abletec for example, however I have not seen any reviews from pubs or individuals on those yet, so the story is still out.
After listening to them and also measuring them, they do not come up to the sound quality or measurments of good linear amps in the upper mids to trebble.
I conceed they are great value for the money. But if one is into great sound and is an audio buff this is then where they fall down.
In the future when the switching frequency can be much higher then I can see even I will change over to them.
But at the moment I will stick with linear until that happens, and it will happen as technology advances with the high speed switching frequency.

Cheers George
"After listening to them and also measuring them..."

Hi George, which are the class D amps that you have listened to and measured?

Thanks, G.
They all have the same characteristics with their mandatory switching filters on their outputs. And the ones that have smp power supplies have even another problem to deal with, regarding noise.
I'm not going to point the finger at any brands as that would be against forum policies. And I have been warned a couple of years back for it.
All you have to do is listen to them against good linear amps on a revealing system. Then look at any graph/tests done with power output vs thd in the higher audio frequencies and see how it rises very quickly vs lower frequencies, look if you can get it current output in the upper frequencies vs the lower frequencies. Then also if you get get them the phase shift from 10khz to 20khz. And then for good measure the noise figures in the upper frequencies vs lower.

" Class D " and the virtues of, a guru roundtable discussion setup by Absolute Sounds

Cheers George

You did not answer gc's question or address my points.
Uhrn, George.... I apologize, but I develop stomach cramps whenever attempting to cope with the kafkian use of dangling pronouns... So, wouldn't you mind disambiguating a bit, and churn out some honest to Gaia makes and models, so to alleviate my psycho-somatic angsts?

May I also suggest you rest for a spell your cherished passion for ancient historiography of class D lore -- admittedly of the most guruesque kind: leave your oscillators and spectrum analyzers at home, go out in the field, and treat yourself to some listening of brand new class D amps!

You might find the experience, uhrn... well.... "Veritas'bly" uplifting *grins!*
Obviously there is a vested interset in a Class D amp and some are not open to what the best amp makers are saying.
The only ones that have bought out models of class D along side of their flagship linears amps are the ones who want to make money out of them. (It's called keeping the shareholders happy)because they are very cheap to produce compared to great linear designs.
It's too early yet to consider them in the hi end, like I have said I will be the first one to change over to them when the future technology for the switching frequency becomes much higher than it is at present, then they should be comparable or even beat a good linear designs.

Cheers George
Obviously there is a vested interset in a Class D amp and some are not open to what the best amp makers are saying

I don't think GC builds & sells class D amps in his spare time... really.
Just as an idea: listen to a couple. I did. I wasn't enthused by the sound. OTOH it wasn't really bad, just not as good as my amps (which also cost 11x more).
No I don't anyone here builds or sell class d amps either.
What I said and meant was
"there is a vested interset in a Class D amp"
Which means "A personal stake or involvement" in that amp.

Cheers george
Hmmm.... Excellent suggestion George... May be I should really invest in some leading creators of novel class D power conversion modules and amps. My ears tell me it would be a "sound" investment... Unfortunately my wife, source of all financial advice, and always to be heeded in such momentous matters, won't "hear" about it... She insists that only blue chips shall do.

Bottomline... I shall undoubtedly remain poor, but at least audiophilically happy.

I shall now leave you to your fascinating conjectures, whilst you pour over ancient tomes of obsolete lore and alchemic apparata.
"Which means "A personal stake or involvement" in that amp."

So by this logic, we can only discuss things we have no involvement with? That sounds like fun!

By this logic, nobody would discuss anything anywhere because they have a "personal stake". That would include George and his favorite amps as well.


My only personal stake in CLass D is I tried one, liked it and have not had any need to look further.

I look forward to buying even better ones possibly even for less down the road if needed as the technology continues to improve.

I have lots of years of experience and have heard a lot of gear at all price points and of many designs over the years. That is how I get my frame of reference for what sounds good or not. I also look at specs and bench measures and choose my poison.

ANyone who can read can determine for themselves if they think Class D is competitive in the high end already these days. I am willing to bet that as time goes on, more and more people will continue to chose to go that route and be very happy they did.


Maybe I missed it in your posts, but have you actually sat down with a Class D in your system? I assume that you do have an extremely high-end system that you allude to as one in which you can discern the difference between the Class D amps and linear amps.

Have you read any reviews by Wayne Donnelly, Senior Reviewer and Classical Music Editor for "Enjoy The Music"? His reference amps, and they have been for quite some time, are the Spectron Musician III Mk2 Class D amps.

I chose my Spectron amp to replace, to name a few, my Krell, Pass, Karan and Burmester amps. But to be honest, I only have about $65k-$70k in my system, I don't have a high-end system.

" But to be honest, I only have about $65k-$70k in my system, I don't have a high-end system."

That's funny!

If Class D amps are even in the discussion for a system in this price range, high end or not, I would say that's another good indicator that they have arrived.

FWIW I have ~$10K invested in my main system, and slightly less in other gear I use concurrently in various rooms. Most people I know would probably label it "high end" if they cared. That is what I payed, not retail, but I am very value conscience and have bought a lot but not all of my current stuff used or discounted via vendor trade-in programs, etc. I will say I think I know and like quality sound when I hear it and leave it at that without getting into details of what that means to me. My Class D amps list for $6K new, FWIW and represent my single biggest monetary investment in my system, though my speakers would list for about the same amount..
At the NY Audio show recently, a few rooms had very high end class D amps that were superb. Other than a no-limit budget and ability to use a crane to move class A amps, my experience has been where top class D amps are just as good as some very expensive class a/b amps i have used.

they also work better if you want to go "digital" straight from DAC to amp with passive pre-amp
To echo Lapentas comments, I have reasons to believe that the stereotypical artifacts at the outer ends of the audible bandwitdh that were frequently associated with a number of older class D amps may not be necessarily reflected in recent designs based on much evolved power conversion modules, like the Ncore NC1200 by Hypex.

Case in point is the Merrill Veritas monoblock that I have just reviewed for Positive Feedback on issue No. 68:

Once the amps were properly broken in with 1,000 hours of active operations, I did not detect any musical bandwidth limitations, nor any stereotypical class D artifacts with Veritas.

Of course, there are other amps based on Ncore technology (such as Mola-mola Kaluga and ATSAH by Acoustic Imagery) , and other new class D modules other than the Ncore family (such as the Pascal modules). The examination of more examples of new class D amps should confirm or disprove my hypothesis of class D designs having now grown up into general musical adulthood.