Bruno
Just some quick questions regarding the N CORE technology.
1. When will the N Core type improvements be fully implemented into the UCD400HG HxR modules and what 'V' number on the board will identify them?
2. In your opinion, how much better will the full N Core amp sound compared to the upgraded UCD?
3. Are you still using B&W 802's in your evaluations?
John
Just some quick questions regarding the N CORE technology.
1. When will the N Core type improvements be fully implemented into the UCD400HG HxR modules and what 'V' number on the board will identify them?
2. In your opinion, how much better will the full N Core amp sound compared to the upgraded UCD?
3. Are you still using B&W 802's in your evaluations?
John
1. I think we'll suddenly spring the news when it's done. It'll be after the Ncore DIY module is out because that one has priority.
2. No contest.
3. I've had to leave them behind at Philips, they weren't mine. So for most of the time I use something homegrown which was good enough for UcD but which has now become the weakest link. I did get the chance to tweak the NC1200 on a pair of 800D's though, with a set of ML33's standing by as a reality check. Big smiles all round, eventually.
2. No contest.
3. I've had to leave them behind at Philips, they weren't mine. So for most of the time I use something homegrown which was good enough for UcD but which has now become the weakest link. I did get the chance to tweak the NC1200 on a pair of 800D's though, with a set of ML33's standing by as a reality check. Big smiles all round, eventually.
Bruno
What a shame...802's not the sort of thing you'd slip into your lunchbox on the last day at the office!!
Last week I had the opportunity to listen to a pair of Eminent Technology LFT-8B lodspeakers in my listening room being driven by my monobloc UCD400HG HxR's with SMPS400A400...............WOW! what a sound. Some of the best loudspeakers I've ever heard and I include the 802's in that list.
Check out... Eminent Technology, Incorporated for details.
The price in the US is approx $1900 and in the UK £2300, so they're a bargain.
The UK importer thought that the UCD's sounded better with the LFT-8B's than all the other amps he had tried on them. Some praise indeed.
I think they have an importer in Denmark and I think you'd not be wasting your time getting a listen.
Regards
John
What a shame...802's not the sort of thing you'd slip into your lunchbox on the last day at the office!!
Last week I had the opportunity to listen to a pair of Eminent Technology LFT-8B lodspeakers in my listening room being driven by my monobloc UCD400HG HxR's with SMPS400A400...............WOW! what a sound. Some of the best loudspeakers I've ever heard and I include the 802's in that list.
Check out... Eminent Technology, Incorporated for details.
The price in the US is approx $1900 and in the UK £2300, so they're a bargain.
The UK importer thought that the UCD's sounded better with the LFT-8B's than all the other amps he had tried on them. Some praise indeed.
I think they have an importer in Denmark and I think you'd not be wasting your time getting a listen.
Regards
John
Oh, dear I'm late to this thread. I can only plead vacation, though thankfully it was an American one rather than a proper European one. 
More seriously, Bruno, congratulations on a great design. This is inspiring! I look forward to reading up on higher order loops.
Sounds like my kind of project! The boards for the "power" amplifiers I've been working on lately are dominated by 100 mil pitch connectors. Well, and bypass caps---prolifigate consumers of space, those.
Completely excessive. My power hogging, ridiculously inefficient dipoles typically run around 400uW RMS per channel with peaks to 40mW. If the Ncore 400 idle dissipation's, say, 6W that'd be an efficiency of, um, -83dB. In comparison, with careful class AB parts selection I can get that up to a whopping -48dB (0.4% for those of you who've failed to take Bruno's suggestion to use dB instead of percent to heart).
More seriously, Bruno, congratulations on a great design. This is inspiring! I look forward to reading up on higher order loops.
Hi again
Bruno,
Do you expect the ncore to be as sensitive to the quality of the PS as the "standard" UCD´s and with the unboard regulators employed are the quality of the buffer supplies of equally great importance to the SQ as the main supplies to the output?
I´m currently playing around with some small UCD modulators and I´m quite impressed I must admit It seems the HF abilities of the PS is very audible through these modules -even more than with my ncc200 amp it seems...
PS. are you aiming on a specific price point with the DIY ncore modules yet; if so any thoughts you like to share here?
cheers,
Joakim
Bruno,
Do you expect the ncore to be as sensitive to the quality of the PS as the "standard" UCD´s and with the unboard regulators employed are the quality of the buffer supplies of equally great importance to the SQ as the main supplies to the output?
I´m currently playing around with some small UCD modulators and I´m quite impressed I must admit
It seems the HF abilities of the PS is very audible through these modules -even more than with my ncc200 amp it seems...PS. are you aiming on a specific price point with the DIY ncore modules yet; if so any thoughts you like to share here?
cheers,
Joakim
It's interesting how opinions vary concerning the sensitivity of amplifiers to the supply quality. Anyhow. There are 2 places where supply quality can come in. One is the main rails, the other the small-signal rails. The diy module uses discrete op amps with about 10 to 20dB better PSRR than the usual miller-feedback circuit. On top of that it's got HxR built in. I think this pretty much solves the sensitivity to the small-signal rails. On the high voltage end, PSRR is essentially loop gain plus an extra 20dB afforded by a forward correction. You're saying you're playing with UcD modulators of your own? Do you have this forward correction in there? I'm not spelling it out but I noticed on some Russian forum that several people know what it is.
Thanks for the immediate and informative reply Bruno.
The UCD modulators are the UCD100OEM which no longer exists on your webpage. I don´t know how the actual modulator compares against the standard DIY st and hg types, but as i got it, it has no input buffer hence it essentially is a modulator with elevated gain (24,5dB) compared to the other implementations of the UCD modulator(12ish dB) , right?
I may be hyper sensitive to the PS compared to others, but to me it seems to make all the difference between unbearable and very pleasant in my case/setup...
The UCD modulators are the UCD100OEM which no longer exists on your webpage. I don´t know how the actual modulator compares against the standard DIY st and hg types, but as i got it, it has no input buffer hence it essentially is a modulator with elevated gain (24,5dB) compared to the other implementations of the UCD modulator(12ish dB) , right?
I may be hyper sensitive to the PS compared to others, but to me it seems to make all the difference between unbearable and very pleasant in my case/setup...
Vote for muscle
Bruno, it's great to see you've leaped ahead with the new circuit. The specs are magnificent, near Halcro levels (perhaps better in some ways) for 1/10(?) the cost.
I'm out of the pro audio game now, but anyway...I vote urgently in favor of no great reduction in power relative to the roughly-UcD700-class prototypes. That is, lower power for alternate nCore modules is fine as long as a solid 300+watts @8 etc. is kept available. The reason is that many of the *best* transducers today can dissipate more power now than even models of just a few years ago: a lower-power amp won't maximize the driver's ability. We are both fans of B&W 802s of various vintages; consider what its drivers can accept. The high linear excursion of individual drivers has changed, and the need for higher peak voltage and current has increased even more for typical passive-crossover speakers like the 802s. I have some passive 2-way ATCs (SCM20SLs) with just one 165mm woofer per enclosure that very cleanly consume all the 200 watts that my (cough) ICE amps can deliver at 6 ohms. Truly, *given a choice,* would you run all or any portion (mid+tweeter, and either individual or series bass drivers) of 802Ds or 800Ds with less than 300 watts on tap?
In a perfect world one could buy any great speaker like B&Ws and ATCs without internal crossovers (i.e. like PA systems) so that one could use purely digital input for crossovers to apply bandpass and delay like your Grimm LS-1 speakers do. Even then I would never want less than 200 watts@8 for each driver. I think that your many mastering friends would desire that their speakers give out before their shiny new nCore amps, just sayin'...
Best wishes!
P.S. For Christmas, a Grimm UC-1?
Bruno, it's great to see you've leaped ahead with the new circuit. The specs are magnificent, near Halcro levels (perhaps better in some ways) for 1/10(?) the cost.
I'm out of the pro audio game now, but anyway...I vote urgently in favor of no great reduction in power relative to the roughly-UcD700-class prototypes. That is, lower power for alternate nCore modules is fine as long as a solid 300+watts @8 etc. is kept available. The reason is that many of the *best* transducers today can dissipate more power now than even models of just a few years ago: a lower-power amp won't maximize the driver's ability. We are both fans of B&W 802s of various vintages; consider what its drivers can accept. The high linear excursion of individual drivers has changed, and the need for higher peak voltage and current has increased even more for typical passive-crossover speakers like the 802s. I have some passive 2-way ATCs (SCM20SLs) with just one 165mm woofer per enclosure that very cleanly consume all the 200 watts that my (cough) ICE amps can deliver at 6 ohms. Truly, *given a choice,* would you run all or any portion (mid+tweeter, and either individual or series bass drivers) of 802Ds or 800Ds with less than 300 watts on tap?
In a perfect world one could buy any great speaker like B&Ws and ATCs without internal crossovers (i.e. like PA systems) so that one could use purely digital input for crossovers to apply bandpass and delay like your Grimm LS-1 speakers do. Even then I would never want less than 200 watts@8 for each driver. I think that your many mastering friends would desire that their speakers give out before their shiny new nCore amps, just sayin'...
Best wishes!
P.S. For Christmas, a Grimm UC-1?
Last edited:
Sam Lord,
Well said. Love those 802s.
"Small power" is cute, workable, simple, fun, educational, challenging etc. etc. or may be coupled with other advantages in some designs, but considered sperately low power itself can't rationally be "ideal" for even many common uses.
Even without passive crossovers and with using some of the most efficient subs: If you run active crossover to two Danley DTS-10 tapped horns they recommend 2000 watts for each. Comparable extension and output from less efficient sealed boxes with EQ requires many times more than that. I could brige two 300-watt modules for each of the two drivers in each Danley DTS-10, so I'd use 8 amp channels for the 2 subs. That would work, but...it would be less than ideal. Then again, there are other amps perhaps more appropriate. My planar dynamics are power-hogs too (partly because they require EQ).
Well said. Love those 802s.
"Small power" is cute, workable, simple, fun, educational, challenging etc. etc. or may be coupled with other advantages in some designs, but considered sperately low power itself can't rationally be "ideal" for even many common uses.
Even without passive crossovers and with using some of the most efficient subs: If you run active crossover to two Danley DTS-10 tapped horns they recommend 2000 watts for each. Comparable extension and output from less efficient sealed boxes with EQ requires many times more than that. I could brige two 300-watt modules for each of the two drivers in each Danley DTS-10, so I'd use 8 amp channels for the 2 subs. That would work, but...it would be less than ideal. Then again, there are other amps perhaps more appropriate. My planar dynamics are power-hogs too (partly because they require EQ).
Cyclecamper,
As said before in this thread, for your substation-needs and unlimited bass-adventures, other amps are probably more relevant as this ncore is said to be an audiophile class-d amp.
300W @ 8R would be nice, but just because some newer speakers can deal with more power and larger currents, it doesn´t mean that one necessarily needs that power on tab. Of course if the increased power capabilities are obtained through heavier coils and thereby less efficiency, one might need more, but that is due to the fact that the speakers are flawed by design... Again what could be gained through higher sensitivity in speakers is much more expensively gained through more power in the amp department....
Maximizing the ncore´s current capabilities is the important factor here
Maximizing the ncore´s current capabilities is the important factor here
+1 for that..
It's not a question of absolute need, but maximizing the performance per cost. There is a reason that the best active monitors use 200+w channels, just look at internal amps in ATC, B&W, Quested, K+H, PMC,...shall I continue?
Now, the ncore does actually stay super clean until it clips hard, so you have excellent sound until that point. Class AB designs all have increasing distortion products with power and freq way below clipping. But you didn't make that argument...
Nobody is talking about inferior drivers here.
Drivers above a certain sensitivity distort too much. At this stage in driver evolution, beyond around 91dB sensitivity in woofers and around 94B in tweeters you fight a losing battle, IMO. Maybe there is a woofer a couple dB more efficient than B&W's latest 8" and 10" models and just as good, but it would surprise me.
Since the ncore's 2ohm current performance is a big improvement over the UcD700, I think any current vs. voltage performance imbalance is gone. No, it's voltage and current, IMO....Maximizing the ncore´s current capabilities is the important factor here
Respectfully
That sounds great Bruno
Sam, the b&W 800´s featuring 2x10" and 2x8" woofers both have a very average sensitivity about 90dB (and dips to 3,1ohm and 3,5 ohm respectively) which means the woofers roughly puts out 87dB/8ohm which IS NOT impressive and probably relates to high moving mass. I don´t see why higher sensitivity necessarily should increase distortion if its gained by keeping mass low rather than using oversize coils and very thick membranes. Of course very thin and soft membranes and very low linear operation of the coil is not good, but my Audio Technology C10KAP woofers are spec´ed to be 90dB/W and they blend very well in my speakers with and an undamped focal tc120TD5 tweeter (that´s around 93,5dB spl from a single 8ohm woofer!!)
I´m not saying that the B&W´s got bad woofers at all, its just the mentality; they probably expect their costumers to use mega power amps and choose to scarify efficiency somewhat for increased power specs without actually gaining anything other than increased costs for their costumers. B&W are NOT the worst at all. The studio monitors you mention are probably alot worse, but i don´t know them and frankly couldn´t care less -I have never heard a so called studio monitor that impressed me, but I haven´t looked for one either
If one can´t live with a healthy 200W/8ohm with good current capabilities under "normal" living conditions, ones speakers are just too heavy
cheers,
Sam, the b&W 800´s featuring 2x10" and 2x8" woofers both have a very average sensitivity about 90dB (and dips to 3,1ohm and 3,5 ohm respectively) which means the woofers roughly puts out 87dB/8ohm which IS NOT impressive and probably relates to high moving mass. I don´t see why higher sensitivity necessarily should increase distortion if its gained by keeping mass low rather than using oversize coils and very thick membranes. Of course very thin and soft membranes and very low linear operation of the coil is not good, but my Audio Technology C10KAP woofers are spec´ed to be 90dB/W and they blend very well in my speakers with and an undamped focal tc120TD5 tweeter (that´s around 93,5dB spl from a single 8ohm woofer!!)
I´m not saying that the B&W´s got bad woofers at all, its just the mentality; they probably expect their costumers to use mega power amps and choose to scarify efficiency somewhat for increased power specs without actually gaining anything other than increased costs for their costumers. B&W are NOT the worst at all. The studio monitors you mention are probably alot worse, but i don´t know them and frankly couldn´t care less -I have never heard a so called studio monitor that impressed me, but I haven´t looked for one either
If one can´t live with a healthy 200W/8ohm with good current capabilities under "normal" living conditions, ones speakers are just too heavy
cheers,
Buffer sorrow
Well it seems the simulation model is mostly, but not entirely right. So it's working stably now, at least the one single discrete op amp that I'm testing first.
The buffer is essentially the good old two-opamp prefix that UcD modules use, except that the op amps are discrete. It's tough. I decided against matched JFETs because any pair of unmatched BJT's has a better offset spec than a matched pair of FETs. Anyhow, low input current doesn't mean that AC input impedance is either high or linear, especially if you're going for low noise FETs. You might be surprised how much a jfet input stage can distort if you don't match the impedances at either input. So, no FETs. BJT's draw base current, which could cause a significant DC error when impedances are mismatched. So my first idea was to use a MAT14 quad. Two for the input pair, a third to generate a current reference to compensate the base current.
Strangely enough, even though I've bootstrapped the input pair I get a fair bit of distortion with unmatched input impedances, nearly as much as what the LM4562 does (which would make discrete pointless imho).
Actually that's something you might want to see in measurement. As Douglas Self pointed out several times, noninverting circuits are most op amps' demise. New "super amps" like the 4562 have stupendously low distortion if you either use them inverting or you match the impedances at both inputs exactly. If you don't, distortion is far from unmeasurable. Look at the attached graph. The spectrum is taken with a 8Vrms 10kHz source (0dBr=8V) through a 3k source impedance. Red is the generator alone. Grey is the LM4562. This is not negligible. All of this distortion is caused by Cob modulation of the input transistors. So actually those of us who prefer passive attenuators are pretty much left in the cold.
With the MAT14 I got a result that was slightly better than the LM4562 but not much. I had started wondering if the input capacitance of that transistor isn't just so high that after bootstrapping you're still looking at a fairly nonlinear capacitance. So I stripped the MAT14 and put an unmatched pair of BC846BW in there instead, which has much, much lower capacitance. Bam, distortion gone (blue plot).
So there we are, with bog standard cheapo transistors and tested under obvious worst case conditions I get distortion which is tons lower than the LM4562 under the same conditions, and probably lower than any other op amp, FET or bipolar. On the one hand my choice to drop FETs is vindicated, on the other hand I'm screwed because I can't use my base current compensation trick anymore. I'll have to find another way of getting round that one. The BC846B isn't particularly low-noise but I suppose the 2SC4117 will be equally fine distortion-wise.
To be continued...
Well it seems the simulation model is mostly, but not entirely right. So it's working stably now, at least the one single discrete op amp that I'm testing first.
The buffer is essentially the good old two-opamp prefix that UcD modules use, except that the op amps are discrete. It's tough. I decided against matched JFETs because any pair of unmatched BJT's has a better offset spec than a matched pair of FETs. Anyhow, low input current doesn't mean that AC input impedance is either high or linear, especially if you're going for low noise FETs. You might be surprised how much a jfet input stage can distort if you don't match the impedances at either input. So, no FETs. BJT's draw base current, which could cause a significant DC error when impedances are mismatched. So my first idea was to use a MAT14 quad. Two for the input pair, a third to generate a current reference to compensate the base current.
Strangely enough, even though I've bootstrapped the input pair I get a fair bit of distortion with unmatched input impedances, nearly as much as what the LM4562 does (which would make discrete pointless imho).
Actually that's something you might want to see in measurement. As Douglas Self pointed out several times, noninverting circuits are most op amps' demise. New "super amps" like the 4562 have stupendously low distortion if you either use them inverting or you match the impedances at both inputs exactly. If you don't, distortion is far from unmeasurable. Look at the attached graph. The spectrum is taken with a 8Vrms 10kHz source (0dBr=8V) through a 3k source impedance. Red is the generator alone. Grey is the LM4562. This is not negligible. All of this distortion is caused by Cob modulation of the input transistors. So actually those of us who prefer passive attenuators are pretty much left in the cold.
With the MAT14 I got a result that was slightly better than the LM4562 but not much. I had started wondering if the input capacitance of that transistor isn't just so high that after bootstrapping you're still looking at a fairly nonlinear capacitance. So I stripped the MAT14 and put an unmatched pair of BC846BW in there instead, which has much, much lower capacitance. Bam, distortion gone (blue plot).
So there we are, with bog standard cheapo transistors and tested under obvious worst case conditions I get distortion which is tons lower than the LM4562 under the same conditions, and probably lower than any other op amp, FET or bipolar. On the one hand my choice to drop FETs is vindicated, on the other hand I'm screwed because I can't use my base current compensation trick anymore. I'll have to find another way of getting round that one. The BC846B isn't particularly low-noise but I suppose the 2SC4117 will be equally fine distortion-wise.
To be continued...
Attachments
We used "high-test" (old name for premium petrol) MAT02EHs to very good effect in our tweaked, all-bipolar, more-or-less "blameless" AB design of many years ago. OK, there was an AD712 in the DC offset circuit. All diff input stage resistors needed close matching and cost$$. Class D efficient it was not...
I've been gone a few months from here, sorry to repeat others' comments earlier--I didn't read the entire thread. BUT, please what OEMs?? I haven't seen a *consumer* amp available in America with all-Hypex inside since the demise of Exodus! OK, Meridian has a super unit but $$$. Where are the others? NAD Master series? Everybody else (CI, Hexateq, and...?) uses UcD with a heavy old PS, no SMPS. I'd be happy to buy a muscular finished amp if I could find one; I'm too busy and lazy any more to DIY. (Sorry, why am I here?) I DO understand the cannibalization argument for leaving the high power modules to OEMs, that is a very real concern.
Once again, congratulations on this achievement!
I've been gone a few months from here, sorry to repeat others' comments earlier--I didn't read the entire thread. BUT, please what OEMs?? I haven't seen a *consumer* amp available in America with all-Hypex inside since the demise of Exodus! OK, Meridian has a super unit but $$$. Where are the others? NAD Master series? Everybody else (CI, Hexateq, and...?) uses UcD with a heavy old PS, no SMPS. I'd be happy to buy a muscular finished amp if I could find one; I'm too busy and lazy any more to DIY. (Sorry, why am I here?
) I DO understand the cannibalization argument for leaving the high power modules to OEMs, that is a very real concern.Once again, congratulations on this achievement!
Last edited:
- Status
- Not open for further replies.