Phoenix- See the white paper.
Bruno- How cool! Two questions:
1) Any chance you'll make it easy to set it up as a VCCS as in the top-end of the LS1?
2) My dream product from you would be a Socratic controls theory course. Any chance of that entering the Hypex product line?
Bruno- How cool! Two questions:
1) Any chance you'll make it easy to set it up as a VCCS as in the top-end of the LS1?
2) My dream product from you would be a Socratic controls theory course. Any chance of that entering the Hypex product line?
20dB
The NC1200 is available as demo units for sufficiently high-profile OEM customers and shortly samples will be available for sale. The progress of the NC400 module is occasionally reported here. Work on other modules still has to start.
You'll have to wrap your own external loop around it. That's how it was done in the LS1.
Difficult. I never formally studied the topic so I always develop my own methods from scratch when I need them. The result is very idiosyncratic and very specific to whatever it is I'm doing (and on two occasions radically new). If you want to teach a course you have to take a more generic approach. I think I'd make a bad teacher.
Bruno,
Though the reference point of the feedback loop at the speaker return in UCD and nCore is to be the more important aspect of the performance, the current streaming through the boards, to and from, the ground connection still seems to have an effect.
I get much more bass control and impact in the whole spectrum when I ALSO connect speaker return from the speaker terminal on the amp enclosure directly to the ground between the supply caps (speaker return still goes through the board return connection as well). It seem wrong to do it this way, but directing the all the return current through the board alone seems to eat the dynamics and grunt. Would you think its be better to make a short low R jumper from speaker return on the board to the ground connection on the board instead?
best
Though the reference point of the feedback loop at the speaker return in UCD and nCore is to be the more important aspect of the performance, the current streaming through the boards, to and from, the ground connection still seems to have an effect.
I get much more bass control and impact in the whole spectrum when I ALSO connect speaker return from the speaker terminal on the amp enclosure directly to the ground between the supply caps (speaker return still goes through the board return connection as well). It seem wrong to do it this way, but directing the all the return current through the board alone seems to eat the dynamics and grunt. Would you think its be better to make a short low R jumper from speaker return on the board to the ground connection on the board instead?
best
In one way it wouldn't be a bad idea if there were two connections on the board that could be used as return points for NFB (could theoretically be done very easily just by adding two small resistors to the board) that those could use who want to take feedback at the output terminals of an amp or a the speaker terminals directly.
But this would also be dangerous because fools are too inventive sometimes ........
Regards
Charles
But this would also be dangerous because fools are too inventive sometimes ........
Regards
Charles
this is stange.
let`s see what bruno have to say about it.
what module did you try that?
i think it can`t be done at ucd2k but only the half bridge ucd`s
Last edited:
Its UCD100OEM modules; they are basically bare UCD modulators with a higher gain setting and have pin connectors so they´re a mess to connect and experiment with regarding their connections 
Also its a linear supply based on a 500VA toroid offering +/- 35 VDC, meaning quite low DCR, which may be a factor when it comes to the effect of playing with the ground connection...
Yeah, bridget output topology normally means that the amp doesn´t send the return current to ground, but instead the + and - rail acts as each others return (through the output stage of course).
Also its a linear supply based on a 500VA toroid offering +/- 35 VDC, meaning quite low DCR, which may be a factor when it comes to the effect of playing with the ground connection...
Yeah, bridget output topology normally means that the amp doesn´t send the return current to ground, but instead the + and - rail acts as each others return (through the output stage of course).
Last edited:
Bruno, you have tried to describe the listeningimpressions from the NCore's
and what I have found is that they can stand up to some of the best solid-
states (ML33) and that there is no comparison to the ucd-series. The most
interesting comparison for me is the ucd's. If it's not to much to ask for I
would appreciate if you could try to describe the main differences between
them. Do they differ in tonality , musical involvement, stagepresentation,
purity of singel notes, the abbility to resolve for example bassdrum from other
bassnotes, or is there something else that make them so much better. Maybe
all of the above ? Trying to describe the experience of sound in words is'nt
always that easy but I would shure appreciate it, if you would give it a try.
and what I have found is that they can stand up to some of the best solid-
states (ML33) and that there is no comparison to the ucd-series. The most
interesting comparison for me is the ucd's. If it's not to much to ask for I
would appreciate if you could try to describe the main differences between
them. Do they differ in tonality , musical involvement, stagepresentation,
purity of singel notes, the abbility to resolve for example bassdrum from other
bassnotes, or is there something else that make them so much better. Maybe
all of the above ? Trying to describe the experience of sound in words is'nt
always that easy but I would shure appreciate it, if you would give it a try
.Yes. I've been mulling this over with respect to low power nCore projects and I'm not sure I see a value proposition for it beyond it being an interesting project, though. Existing op amps and buffers offer drive capabilities up to 600mA peak or so in the audio frequency range with costs in the range of USD 1.50 to 7.50 per channel, linearity and noise floors are equal to or better than current generation nCore, and---I suspect---lower quiescent dissipation. From the standpoint of getting stuff done throwing down somewhere between a half an 8-SOIC to an 8-SOIC + 8-PSOP seems hard to argue with at that price point. The lowest power point nCore where seems interesting to me is improving upon existing higher order class D parts with drive capabilities of an amp or a few such as the ADAU1592. nCore might also have a performance edge over op amp + Darlington class AB designs but I'd be surprised if the additional complexity offers value from a production engineering standpoint---more of a "because it's fun" thing.
Bruno
Having just found this thread, and read through it and the associated white paper and spec sheet on the Hypex site...
...I've a complementary set of questions which I've haven't found the answers for (someone please send me in the right direction if I've missed them)
- what bits of the NC technology/advances are being taken over to the revised UcD modules?
- which bits of the NC technology will be exclusive to the NC modules?
- ie. what isn't being taken over to the UcD modules?
An associated question is therefore:
- apart from the unique circuit and implementation goodies you are implementing (eg. discrete front end), why will we need specific NC modules at all? Won't the UcD180/400/700 effectively become NC180/400/700?
Thanks
Chris
Having just found this thread, and read through it and the associated white paper and spec sheet on the Hypex site...
(Great story by the way: boy meets girl, boy falls in love with girl, they have a misunderstanding and stray apart, then meet up again and it all ends up happily ever after. Ok, maybe that was a different thread)
...I've a complementary set of questions which I've haven't found the answers for (someone please send me in the right direction if I've missed them)
- what bits of the NC technology/advances are being taken over to the revised UcD modules?
- which bits of the NC technology will be exclusive to the NC modules?
- ie. what isn't being taken over to the UcD modules?
An associated question is therefore:
- apart from the unique circuit and implementation goodies you are implementing (eg. discrete front end), why will we need specific NC modules at all? Won't the UcD180/400/700 effectively become NC180/400/700?
Thanks
Chris
@Juhleren, I've never heard a sonic improvement resulting from deviating from correct theory. I have heard euphonic effects occasionally, but those lose their appeal once you become accustomed to the sound and understand the "trick".
@dahlberg. The top end of Ncore comes across as much more precise with no hint of colouration or "time smear". This in turn results in a much more believable 3d stereo image that remains incredibly stable during complicated passages. That was one of UcD's strong points already so go figure. For bass reproduction this means that bass-heavy material doesn't muddle the rest of the spectrum. In terms of detail it allows you equally to hear the whole choir or individual voices, depending on how you listen. Some amplifiers zoom in on the details while losing perspective (a lot of solid state amps do this) or give a fairly complete but blurred picture (like a lot of valve amps). I'd started thinking it's a compromise or balancing act but even a cursory listen to the NC1200 belies this..
@stefano I wouldn't see the point. You can make line level buffers with unmeasurable distortion (-140dB and better) using standard techniques. Adding a further high order control loop would result in distortion levels that are truly academic and limited by the feedback resistors...
@Chrispa The non-Ncore improvements that are migrating into the UcD range are:
1) Improved gate drivers. We get slightly lower open-loop THD than the old modules (about half) and yet idling losses come down. We're talking about a 30% improvement for 700-style power stages and 40% for 400-style ones.
2) Improved large-signal behaviour. The old UcD modulators were designed without knowledge of how to model the large signal linearity.
What Ncore adds on top of this is 20dB more loop gain. At some point you'll see UcDxxx and NCxxx module pairings where the THD graphs clearly have the exact same shape except that the NC version is ten times lower (not the noise of course).
@dahlberg. The top end of Ncore comes across as much more precise with no hint of colouration or "time smear". This in turn results in a much more believable 3d stereo image that remains incredibly stable during complicated passages. That was one of UcD's strong points already so go figure. For bass reproduction this means that bass-heavy material doesn't muddle the rest of the spectrum. In terms of detail it allows you equally to hear the whole choir or individual voices, depending on how you listen. Some amplifiers zoom in on the details while losing perspective (a lot of solid state amps do this) or give a fairly complete but blurred picture (like a lot of valve amps). I'd started thinking it's a compromise or balancing act but even a cursory listen to the NC1200 belies this..
@stefano I wouldn't see the point. You can make line level buffers with unmeasurable distortion (-140dB and better) using standard techniques. Adding a further high order control loop would result in distortion levels that are truly academic and limited by the feedback resistors...
@Chrispa The non-Ncore improvements that are migrating into the UcD range are:
1) Improved gate drivers. We get slightly lower open-loop THD than the old modules (about half) and yet idling losses come down. We're talking about a 30% improvement for 700-style power stages and 40% for 400-style ones.
2) Improved large-signal behaviour. The old UcD modulators were designed without knowledge of how to model the large signal linearity.
What Ncore adds on top of this is 20dB more loop gain. At some point you'll see UcDxxx and NCxxx module pairings where the THD graphs clearly have the exact same shape except that the NC version is ten times lower (not the noise of course).
Last edited:
So you would say that the NC is an improvment over the ucd-series in aspect of maintaining perspective as well ? I asked about musical involvment and this would as I see it be a big part of just that aspect. I'm also guessing that your decision to go for the discrete inputstage has something to with (musical involvment) this as well.
Thanks for your response. Bengt Dahlberg
Last edited:
Ncore is an overall improvement over UcD on all counts.
The choice for a discrete input stage stems from the fact that I wanted it to be better than what one can do with the best IC op amps. In fact I want the buffer to be absolutely inaudible. That way, there is no reason for anyone to want to replace it by something else.
The choice for a discrete input stage stems from the fact that I wanted it to be better than what one can do with the best IC op amps. In fact I want the buffer to be absolutely inaudible. That way, there is no reason for anyone to want to replace it by something else.
According to the installation instructions the speaker return wire is to be connected to the speaker return connection on the am boards "for best performance" as this point works as the reference for the feedback. Connecting speaker return to the PSU ground is therefore suboptimal regarding the feedback operation as I got it
cheers,
Hmmm, correct theory here means control theory I guess
The problem IMHO in being primarily theory driven is that there usually are diverging theories to adhere to, especially regarding design choices. For me empirical results serves to validate whether a theory based choice seems to pay off in a specific case, and although the feedback theory applied in the design obviously is very crucial, i can not figure out how it actually accounts for the current flow to and form the PSU ground, and the effect of the electrical properties of that connection. My results points at lowering the resistance further still seem to affect the performance, hence my question. Lowering the resistance between speaker return and ground connection on the board shouldn´t mess with the feedback reference, no? -Then what is there to loose?
The way I have done it (quick and dirtyish) by making two current paths form the speaker terminal to the PSU ground is of course not the right way to do it, and I would be lying if there wasn´t any audible downsides, but it just serves to clarify the effect of making a much lower ground return with much lower resistance.
regards,
According to the installation instructions the speaker return wire is to be connected to the speaker return connection on the am boards "for best performance" as this point works as the reference for the feedback. Connecting speaker return to the PSU ground is therefore suboptimal regarding the feedback operation as I got it
cheers,
i will give it a try to my 700`s to see for my self.
- Status
- Not open for further replies.