Hypex Ncore

Status
Not open for further replies.
Re ADC latency, one of the other linked papers actually shows a working amp with ADC feedback. Theoretically you can live with latency of up to a whole switching cycle, in practice you need a lot less because of the extra phase shift incurred in antialias and noise filtering. But still, some of the new tricks I've worked out over the past two years have staggering implications as to how much delay and phase shift you work around without affecting loop dynamics at all.
Just gave that a read... congrats on successfully making the ADC-feedback system work, I dismissed the idea as too much trouble to implement. Well done. I'll switch to stating that it's way too complicated for an audio amp. :D

The "DSD DAC" paper was a really interesting read also, especially the DSD/shift register/scrambler trick which is beautifully simple.
 
Oh dear. I was a bit disappointed to find that after my posting a link to the "all amplifiers" paper (which makes the argument in quite some detail) there was still some uncertainty left in some readers' eyes but that could be due to the fact that a lengthy paper isn't necessarily a clear answer. Presuming of course that people actually read it before piping up...
But it's not necessary that people get miffed about it on my behalf ;)

Re ADC latency, one of the other linked papers actually shows a working amp with ADC feedback. Theoretically you can live with latency of up to a whole switching cycle, in practice you need a lot less because of the extra phase shift incurred in antialias and noise filtering. But still, some of the new tricks I've worked out over the past two years have staggering implications as to how much delay and phase shift you work around without affecting loop dynamics at all.

Re Zetex: Zetex is essentially the "PEDEC" system, in that it starts off with a "perfect PWM" signal generated in the small-signal domain (i.e. without the nonidealities of the power stage) which is then subtracted from the actual power stage output to form the error signal which is subsequently integrated and used to add variable delay to the switch transitions. The only thing Zetex does that makes it look "more digital" is to convert the integrated error signal to digital and do the variable delay digitally (block diagram pages 51-52 of http://www.hypex.nl/docs/papers/AES124BP.pdf). I've not yet worked out the reasoning behind this added complexity. Performance, though not bad, is not in line with the overhead.

I don't think Zetex's circuit is as simple as you're saying -- the integrating error feedback and subsequent PWM ADC forms part of a noise shaping feedback loop back into a digital noise shaper running at >100MHz (not just moving the edge delays), with the same equivalent ADC sampling rate so the feedback delay is very small (equivalent to very high oversampling ratio on the PWM waveforms). There is no front-end DAC as such to add its own noise and distortion, though as you say the "perfect reference DAC" sets the performance -- this is pretty good though, certainly better than most standalone DACs (including many that would be used to drive an nCore amp :)

The key is to think of it not just as a power amp but a complete system with digital signal processing, including preamp/filter/EQ/dynamic limiting/speaker monitoring -- the real reason for going digital is so that all this can be done starting from a digital input signal with negligible added noise and distortion, at a low cost. Once you've decided to do a chip the "overhead" is essentially nil, adding other features is pretty much free.

If you just want a power amp with an analogue input and nothing else then the complexity doesn't make sense, as you say.

(disclaimer -- I was involved with the low-jitter driver/clock circuit used as the reference DAC, which has in the region of 1ps jitter :)
 
The trouble with purely subjectivist reviews is that you're essentially at the mercy of the reviewer's taste.
Not to mention that the reviewer's speakers and room/acoustics come into the consideration of which amp he will label as 'musical'. He might find a coloration/distortion that fits his frequency response or his problem with high gain reflections better and that has in reality very little meaning for others.

Excellent post by the way.
 
I presume that by "other inputs" you mean additional measurements. After all, nobody would seriously argue that an on-axis magnitude response fully describes what a speaker does. How a speaker sounds is fully determined by the sound field that speaker produces in your listening room in response to an electrical signal (I hope that isn't controversial). If a speaker does something sonically that you couldn't guess from the measurements it just means there's something you forgot to measure.

Frequency response measurements tell most of the story provided that you don't just measure on axis. Of course a large collection of responses taken at various angles becomes difficult to read but luckily you can get a good indication by averaging the response over all angles (horizontal and vertical) to get an idea of what sort of energy it's putting into the room (called the power response). If that doesn't look too funny you know that the room response has some chance of integrating well with the direct sound. Also interesting are responses taken at the angles that bounce back at the listener as first reflections.

If the on-axis response is flat, if the power response is smooth (it may have a slope or a gentle shelf) and if distortion across frequency doesn't do any ridiculous things you can be pretty much sure the speaker is going to give you years of very convincing playback. That doesn't mean that speakers who don't tick all the boxes will sound bad. Frequency response deviations and nonlinear distortion can sound really nice. That is what most sound engineering (mixing and mastering) is about: adding distortion, EQ'ing, compressing and whatnot. It is sometimes thought that audio objectivists presume that all distortions sound ugly. And since that's clearly not true, audio objectivism is therefore supposedly false. Well, no objectivist has ever claimed distortion (linear or nonlinear) necessarily sounds ugly.

They can sound really really nice. But that doesn't mean correct. Just like some amplifiers with loads of distortion add a lot of musicality that wasn't originally there. The trouble with purely subjectivist reviews is that you're essentially at the mercy of the reviewer's taste. If he just runs after that thing which sounds just that bit nicer than the previous thing he liked, he's locking himself into a positive feedback loop searching for the ultimate ear-candy. It's reasonable to expect that the occasional visit to unamplified live sound should keep this in check, but that presumes that the people who record music don't go out and buy the same high-end loudspeakers that the reviewers have just praised all the way to the heavens. But that's precisely what they do, particularly classical music studios and mastering engineers. Unless measurements get given their proper place as a design and evaluation tool, just the way they are in any engineering endeavour, there is no end in sight. Probably the only reason why people think not relying on measurements is somehow acceptable is because it's easy to confuse audio with music. I hate to break it but music is art, and audio is engineering. I don't like engineering to get in the way of art, which is why I think it should be undetectable.

From an audio quality point of view the ideal situation would be where reviews simply established to what degree the product fulfils its requirements. If it changes the sound it's bad, if it doesn't it's good. The longer I'm in this business the clearer it's become to me that measurements and sound go hand in hand provided you strive for the least colouration. If magazines actually reviewed along those lines you would be surprised how rapidly the audio industry would converge on seriously good equipment, both technically and sonically. Really good audio would just turn into a commodity.

But there lies the rub. Audio magazines are not in the business of evaluating kit. They're in the business of selling paper, which they do by expedient of printing things on that paper in order to make it more attractive. Reasonably objective magazines, so far, have either failed or moved into lighter prose and more sparse data. What all of this means is that it isn't in audio journalists' interest to tow the sonically neutral line. You just can't keep filling magazines with reviews of amplifiers that become ever and ever more transparent (and hence identical).

This is not an indictment. It's just logical.


I read their press release (which embarrassingly implied the amplifier is digital) and it didn't mention Hypex by name so I suppose that's how they want it. Still, when you bump into it there'll be no mistaking.


I'm not sure what you're trying to say. There are probably exceedingly creative ways of boiling water, but that doesn't make all of them worthy of pursuit, bar as an art project perhaps. The "first watt" philosophy is predicated on an outright lie and a fantastically illogical one at that: that products with good performance all the way up to clipping automatically perform worse at 1W than those designed to be good only up to 1W. The minimalist mode of thinking is only partly valid when one refuses to use global feedback. Indeed, in those conditions the fewer stages, the better. In spite of this, even the most arduous minimalist will know where adding a cascode does serious miracles. They may feel vindicated knowing that even "aircraft hangar" amp designers these days have succumbed to minimalism.

Bruno thanks for your thorough and very logically argued response.

I don´t think at all that you are controversial in any way when stating that the sound of a speaker is determined by the in room sound field it creates in response to an electrical signal. I fully agree, and that was also greatly what I had in mind writing my short comment.

Where I don´t follow you all the way, is in how I read you to suggest we should recognize measurements. It is not that I would like to argue for audio subjectivism as opposed to audio-objectivism, but there may be more into it than that dichotomy.

Let me try to explain what I mean by this.
I am currently doing an ethnography on science. The particular science is in applied physics on an engineering university, so they shouldn´t be too far off regarding your expressed position on audio engineering as applying natural scientific representations of the world in design and thereby the role of measurements, no?

What is very apparent to the theoretician scientists who´s work I currently follow, is that the data their work rely on is in no way purely "objective" as it is the outcome of a measuring process which means it is both technically and intellectually constructed. By this I do not mean to imply that the data is false, but elevating it above criticism is not good practice for them. Then again they have to use data, so what else could they do than to rely on the data they can get, although they know it isn´t perfect or anywhere near being objective.

What I learn from these observations is to be cautious about how we rely on data and measurements as they in fact are constructed and therefore not only a purification of a greater whole, but also an interpretation of what they are designed to represent.

If referring to measurements as "objective" I argue that we are missing the greater picture of how technology (herein also our measurement tools and methods) have developed and matured to be replaced by new methods and technology over and over again throughout history. Recognizing the limitations of our technology and measurement tools is how we strive for developing not just better interpretations of the measurements we produce and rely on in design and science, but also better technology. If we emphasize the role of our current measurement technology to be elevated above criticism, we are in fact driving a technocratic process blindfolded. That´s why I think its important to open our eyes and ears too, to listen and value (also subjectively) what we make of our technology :)

Do you really believe that what we currently can technically measure is the last there will ever be to say about how well a loudspeaker or any other technical artifact performs? No of course not, I guess that you are pragmatical about your tools too, and that is probably greatly contributing to how you continue to perfect your designs over and over again. At least I would be surprised if we will not see some great advances in measuring technology that will influence how we understand and objectify what we work with!

Other positions than the dichotomy you put forth (audio- objectivism and subjectivism) could be audio-constructivism and relativism, which are putcomes of the philosophy of science debates that my above explanation has been greatly inspired from. Personally I consider myself as all, and none of these positions at once (probably mostly the last one if I have to choose). I think it is important to acknowledge and be reflexive about that all positions have their respective positive and negative aspects -just to be symmetrical in the argumentation at least :)

Your analysis on the audio review magazine industry is spot on BTW. Of course they sell glossy papers and that is what they either become good at or stop doing. That said I don´t think that purely "objective" reviews would capture all of what audio gear is about. Some subjective input is probably necessary as after all, we all have different tastes and products with no sound of their own is utopia (I haven´t heard the ncore yet of course ;) ) as I see it though we should strive for it as I read you to suggest. I think that anyone who reads a review has to be able to interpret the reviewer. That of course implies that one should know the reviewer and know how to relate to his or her taste (not trivial of course!).

kind regards,
 
......Just like some amplifiers with loads of distortion add a lot of musicality that wasn't originally there.....

Essentially agreed on what Bruno puts up, but I can't stand the term "musicality" in relation with playback equipment.
For me "musicality" only occurs at the other side of the microphones; a performance can be, or can be percepted being "musical"; the performer(s) can subjectively being judged "musical".
Playback equipment can therefore only strive for a reproduction which does the least harm of what happened at the performance side. When the playback sound is found to add a lot of "musicality" there actually is a lot of coloration coming in which should not be there; I guess that's what Bruno means.
 
@iand, I know how the Zetex works in much better detail than I showed in that post, but I don't think the detailed operation detracts from the basic idea that it's comparing the power stage output with the output of what is essentially a small-signal 1-bit DAC and feeding back the integrated error into the digital domain through an ADC.

Of course that needs to add further loop gain to shape the quantisation noise of the correction (which is determined by the system clock as well). But still the amplifier essentially ends up reproducing the spectrum found at the 1-bit DAC "reference". All of it. Just try adding an extra, externally generated analogue signal. It gets amplified. The circuit cannot differentiate between the intended signal and any you add externally. So the "reference" is truly a DAC. That's why it has to be this good.

This arrangement does have its merits, mind you. Since the reference contains roughly the same HF as the output, the error signal is essentially ripple free. This eliminates "ripple aliasing distortion" (i.e. the gain change the ripple would otherwise cause in the modulation process). But you get the same in an ordinary PEDEC controller.

@juhleren, I think I'll be in trouble (wife waiting in bed) if I stay up late enough to give your reply the consideration it deserves tonight. So tomorrow :)

Essentially agreed on what Bruno puts up, but I can't stand the term "musicality" in relation with playback equipment.
For me "musicality" only occurs at the other side of the microphones; a performance can be, or can be percepted being "musical"; the performer(s) can subjectively being judged "musical".
Playback equipment can therefore only strive for a reproduction which does the least harm of what happened at the performance side. When the playback sound is found to add a lot of "musicality" there actually is a lot of coloration coming in which should not be there; I guess that's what Bruno means.

Let's call it "pleasantness" then. Indeed, to call euphonic distortion "musicality" may be giving it too much credit.
 
Last edited:
I think I'll be in trouble (wife waiting in bed) if I stay up late enough to give your reply the consideration it deserves tonight. So tomorrow :)
I get it it's her who drew this then...
 

Attachments

  • 386341_2762567107032_1343486314_2984764_1937026868_n.jpg
    386341_2762567107032_1343486314_2984764_1937026868_n.jpg
    14.9 KB · Views: 789
@iand, I know how the Zetex works in much better detail than I showed in that post, but I don't think the detailed operation detracts from the basic idea that it's comparing the power stage output with the output of what is essentially a small-signal 1-bit DAC and feeding back the integrated error into the digital domain through an ADC.

Of course that needs to add further loop gain to shape the quantisation noise of the correction (which is determined by the system clock as well). But still the amplifier essentially ends up reproducing the spectrum found at the 1-bit DAC "reference". All of it. Just try adding an extra, externally generated analogue signal. It gets amplified. The circuit cannot differentiate between the intended signal and any you add externally. So the "reference" is truly a DAC. That's why it has to be this good.

This arrangement does have its merits, mind you. Since the reference contains roughly the same HF as the output, the error signal is essentially ripple free. This eliminates "ripple aliasing distortion" (i.e. the gain change the ripple would otherwise cause in the modulation process). But you get the same in an ordinary PEDEC controller.

I don't think we're disagreeing about how the Zetex DDFA works or its advantages -- it was just that you said you couldn't understand why the complexity was needed, so I was pointing out that it wasn't just a power amp.

As you say the DAC has to be pretty damn good, but at least there are few other error sources apart from this, you could say the power amp function is pretty much distortion-free on top of this, since the reference DAC output and the PA output are essentially identical -- looking at the results, the performance of the entire DDFA circuit is almost the same as the reference DAC alone. And you can use it as a multichannel solution for active speakers including crossovers and EQ.

Processing the error feedback signal and the feedforward signals digitally also opens up many possibilities for effective current/power limiting and amplifier and speaker protection (including dynamic EQ and driver excursion limiting) as well as "ideal" (distortion-free) gain limiting/clip prevention (using digital lookahead) and so on -- it's possible to do many things like this with digital algorithms that are impossible with analogue circuits.

Of course this is really moving into the realms of multichannel DSP where many amplifier designers fear to tread -- and where many audiophiles would say we shouldn't go because it ruins the sound, ignoring that many of the same techniques are used in the studio before they ever get their precious recordings...

Ian

P.S. I'm not trying to plug the Zetex solution, but speaking as an engineer it seems an elegant solution to building a large part of the sound reproduction chain at relatively low cost and with great consistency and flexibility. Digital in and speaker(s) out has its advantages :)

P.P.S. But it's even more overkill for a single-channel power amp only...
 
I'd be happy to find that Meridian thinks the NCORE is a confirmation from Hypex.
based on interviews I've read, Bob Stuart seems one of the most down to earth guys in the business. I'm curious if it goes into an active system.

...

some Googling later:
News | whathifi.com

The speaker uses a single 8.5cm aluminium-cone wideband drive unit powered by a 100W wide-range super-linear low-feedback amplifier, this covering frequencies down to 200Hz.

At that point the downward-firing, relfex-port-tuned 14cm aluminium cone bass driver, powered by a 150W Class D amp rolls in.
 
Hi Bruno,
I am quite surprised about your definition of digital and analogue.
Since you pointed to this paper, I hope you are willing to discuss it, even if it is not a specific ncore topic.
Until I read your AES convention paper 353 I always said a digital
signal would be any signal where we use a digital protocol (using only certain discrete values) for coding the information.
But you are stating: 'What makes signal digital is whether the recipient interprets it as such.'

I am really wondering why a digital information protocol should flip to analogue just by being fed into an analogue recipient.

Examples:
If we use a traditional triangle modulator for our PWM, then the information is coded in a duty cycle that allows any value from 0 - 1. No quantization, every value is possible, like we are used to it from analogue world.
Yup, intuitively I would call this an analogue signal.

Instead of an analogue triangle PWM modulator we could use a uC to generate a PWM. Different from an analogue triangle modulator such a PWM is not able to deliver all duty cycles, but will show quantization steps, depending on the chosen resolution.
Lets have our digital protocol like this:
2 bit resolution and we can have the following duty cycles: 0 1/3 2/3 1
PWM frequency may be any typical value used for class D PWM, i.e. 400kHz.
If we use this protocol for coding music and feed it by a half bridge to an analogue speaker ==> IMHO it would still remain a signal with digital information coding. Why should I suddenly call it analogue? Just because I feed it to an analogue transducer?

Erhm, well of course a 2 bit resolution might not be most fortunate - but I am definitely expecting that DSPs which can do such a PWM in real time with 16bit...24bit resolution and even can handle a fast post filter feedback loop will become cheap within some years. Why should I call such a system analogue?
Of course such a system will still have to handle all the intruding analogue headache like unstable supply, limited switching speed and dead time adjustments as well as non linearities of the output filter, output impedance, back EMF ...
But the information coding would be digital. Should we really call it analogue just because we have chosen a digital protocol that is suitable for an analogue transducer?

Coming from the reasoning above your definition really feels counter intuitive for me. Why did you define it as you did? I am sure you had a reason, but this reason is not obvious to me.


Bye
Markus
 
On a somewhat different matter than pure tech, here's my sketchupped "artists" impression of a case. As you can see, I'm not that much of an artist.

Turns out that a mate of mine has a CNC router (diy) that "should" be able to do aluminium and he was actually happy to test it out somewhere in the (not so near) future.

Still a lot to be done, but I'm aiming for a kind of Linn Klimax look. Really cool. Sort of a sidewalk-tile (stoeptegel??) meets fondleslab meets klimax-esque thing(y).

Try searching google images for the linn klimax. I especially like the "roofed" connector section at the back. Sort of a cross reference to Linn (who I think started it) and Devialet (yeah right!). Don;t forget to add KLIMAX to the query in google if you have your filters set to none (NSFW results). W in this case may also be wife (not that I have one, but hey gf is almost wife sometimes).
 

Attachments

  • case1.gif
    case1.gif
    98.6 KB · Views: 734
  • case2.gif
    case2.gif
    68.1 KB · Views: 716
  • case3.jpg
    case3.jpg
    62.6 KB · Views: 700
Last edited:
so you're basically talking about a DSP controlling the switchers and an A/D converter providing the digital feedback information? I would indeed call that a digital amp. but no matter how I look at it I can't see any way to avoid the A/D part, since one must take analog feedback information back into the digital domain in order to keep the amp digital.

wouldn't that need a very precise ADC with the added complications (jitter like distortion etc)?

looking naively at it... a performance ADC is anything but simple and in order for the idea to be effective it would need to be fast too. could we compensate for A/D conversion errors in the digital part of it? hm...
 
Wow, you engineering knuckleheads sure know how to put a damper on an exciting thread :)

Here I was all excited to read about impressions of the NCore from early adopters, followed by enthusiastic responses about a potential case design. Now you guys go off into some goofy, boring diatribe about digital theory as if anyone on a DIY audio forum would be interested in that nonsense. he he he

In all seriousness, as much as you guys want to debate quantum physics, could you start another thread and leave this one to those of us who use a computer for email and the internet and a calculator to balance our checkbook.

Every time I get an email notice that someone has responded to this thread I get all excited thinking I am going to read someone's experience with their new NCore build or there is NCore enclosure news. Sure enough when I get here, you pocket-protector guys have hijacked this thread again :xeye:

I'll be patient until you guys get this out of your system and then we can get back to the important stuff - how does the NCore sound and how are the NCore amp builds coming along.
 
Wow, you engineering knuckleheads sure know how to put a damper on an exciting thread :)

Here I was all excited to read about impressions of the NCore from early adopters, followed by enthusiastic responses about a potential case design. Now you guys go off into some goofy, boring diatribe about digital theory as if anyone on a DIY audio forum would be interested in that nonsense. he he he

In all seriousness, as much as you guys want to debate quantum physics, could you start another thread and leave this one to those of us who use a computer for email and the internet and a calculator to balance our checkbook.

Every time I get an email notice that someone has responded to this thread I get all excited thinking I am going to read someone's experience with their new NCore build or there is NCore enclosure news. Sure enough when I get here, you pocket-protector guys have hijacked this thread again :xeye:

I'll be patient until you guys get this out of your system and then we can get back to the important stuff - how does the NCore sound and how are the NCore amp builds coming along.

Hear! Hear! - from one of the "latent" readers of this thread.
 
Status
Not open for further replies.