Without having studied this thread and Zetex/Diodes available information in detail, there is a point, which was probably not considered:
Slew Induced Distortion. (Whatever the significance of this type of distortion is).
Any of today's average linear (non time- or level discrete) power amplifier is quite inimmune to that, but there are a class of amplifiers (higher order loops, call it NDFL, Sigma Delta, PID or whatever, which are more or less equivalent and some FF error correction schemes as well), which can fail to handle such pathological signal conditions correctly.
At least for a TIM100 signal and at least much worse as a test with static sines within the audible range would suggest ...
Before we hear the argument, that these are all artifical signals, it is a well known fact, that e.g. vinyl playback includes strong pulse noises outside the audible band, which can make such designs to "lose countenance" and reveal the disadvantages of aggressive error correction.
So guys, is there any hint, insight, information, that besides the nice THD and DNR and bit resolution figures, the Zetex design (with the feedback from the power stage) behaves well under the above mentioned artificial and real world conditions ?
As good as a well-behaved "bread-and-butter-blameless" Class A(B) design, providing about the same static linearity figures and the same idle power of 50W per channel (NAD M2) ?
Slew Induced Distortion. (Whatever the significance of this type of distortion is).
Any of today's average linear (non time- or level discrete) power amplifier is quite inimmune to that, but there are a class of amplifiers (higher order loops, call it NDFL, Sigma Delta, PID or whatever, which are more or less equivalent and some FF error correction schemes as well), which can fail to handle such pathological signal conditions correctly.
At least for a TIM100 signal and at least much worse as a test with static sines within the audible range would suggest ...
Before we hear the argument, that these are all artifical signals, it is a well known fact, that e.g. vinyl playback includes strong pulse noises outside the audible band, which can make such designs to "lose countenance" and reveal the disadvantages of aggressive error correction.
So guys, is there any hint, insight, information, that besides the nice THD and DNR and bit resolution figures, the Zetex design (with the feedback from the power stage) behaves well under the above mentioned artificial and real world conditions ?
As good as a well-behaved "bread-and-butter-blameless" Class A(B) design, providing about the same static linearity figures and the same idle power of 50W per channel (NAD M2) ?
Any class-D amp with bandwidth limiting in the front end is pretty much immune to any "slew-induced distortion" by construction, there's just no mechanism for it to happen.
This would certainly apply to the DDFA as well; any amplifier with the kind of ridiculously low 19/20kHz IMD figures that this has is extremely unlikely to have any slew-rate problems anyway.
Ian:
You may be right, that this design isn't prone to "hard" IMD effects by design in the basic modulator without feedback and power stage "prediction", but with the "Digital Feedback" from the power stage applied this is certainly not the case.
Please take a look at the following link:
NAD M2 Direct Digital integrated amplifier Measurements | Stereophile.com
The 19/20 kHz IMD is somewhere around 0.04%, which is ridicoulously high.
Please consider, that these figures are two orders of magnitude away from any recent/average DAC performance.
Please take a look at the high order uneven IMD products !
I never saw a recent linear amp/or DAC having such broad IMD harmonics.
From what I see here I would expect a TIM100 in the range of 0.1% from this amp at the same power, which is very high.
Many Japanese linear amplifiers of the mid 60ies were already better than this.
The NAD M2 Master Series Amp is very likely the best commercial implementation of Zetex DDFA available, so I would expect the best possible performance ...
Please also look at the idle dissipation of the M2, which is unbelievable 50 W per channel.
So again: Why choose this proprietary technology over a cheap "blameless Class A(B)" + DAC ? Besides the nice marketing blah ...
You may be right, that this design isn't prone to "hard" IMD effects by design in the basic modulator without feedback and power stage "prediction", but with the "Digital Feedback" from the power stage applied this is certainly not the case.
Please take a look at the following link:
NAD M2 Direct Digital integrated amplifier Measurements | Stereophile.com
The 19/20 kHz IMD is somewhere around 0.04%, which is ridicoulously high.
Please consider, that these figures are two orders of magnitude away from any recent/average DAC performance.
Please take a look at the high order uneven IMD products !
I never saw a recent linear amp/or DAC having such broad IMD harmonics.
From what I see here I would expect a TIM100 in the range of 0.1% from this amp at the same power, which is very high.
Many Japanese linear amplifiers of the mid 60ies were already better than this.
The NAD M2 Master Series Amp is very likely the best commercial implementation of Zetex DDFA available, so I would expect the best possible performance ...
Please also look at the idle dissipation of the M2, which is unbelievable 50 W per channel.
So again: Why choose this proprietary technology over a cheap "blameless Class A(B)" + DAC ? Besides the nice marketing blah ...
This has all been discussed before -- if you look at the DDFA as just a power amp it seems complex, but the real point is to replace the complete signal chain from digital audio in to speaker out with performance similar to a high-end DAC -- not just IMD, but low-level linearity. Many functions such as eq/crossover/limiting/driver protection can be done a lot better in the digital domain than in analogue.
The "digital feedback" from the power stage -- which is actually a noise-shaping ADC clocked at 108MHz -- is probably the best way of closing the feedback loop with very low latency, to allow effective noise shaping and distortion reduction over the audio band.
I suspect the real objections to this type of approach come from those who don't understand how it all works, or don't like the thought of everything being done digitally with no room for analogue tweaking, or it all being proprietary, or all of the above. For those with no such axes to grind, the reviews of the M2 suggest that the design did achieve its objectives.
Anyone who thinks 0.04% IMD at 19/20kHz is "ridiculously high" should ask themselves what kind of speaker they plan to use for listening, never mind the distortion thresholds of the ear ;-)
Sorry but specifications are useless....
Is like a class AB pushpull with FB ie 0.0000 HD can sound better then SE class A
and don't thinks that class D sound better then tube amp...is cheaper and sound good but not better...of course you must add the right speakers.
is better 100db speakers with 1Watt tube amp then 85db with 200watt is you ask me...Open baffle...
Is like a class AB pushpull with FB ie 0.0000 HD can sound better then SE class A
and don't thinks that class D sound better then tube amp...is cheaper and sound good but not better...of course you must add the right speakers.
is better 100db speakers with 1Watt tube amp then 85db with 200watt is you ask me...Open baffle...
Last edited:
dBs and Watts you mention in your last sentence are specifications too.
So I guess only the specifications we don't understand are useless?
Now for a digital system having a power amp with real USB async input is fantastic news !
no more dac ,i/v output stage ,cable ,pot,pre....
Hopes are that price will go down and tech do better ....
One day will be a four channel with dps xover and we can build all DIY speaker we wants.....(2ways)
no more dac ,i/v output stage ,cable ,pot,pre....
Hopes are that price will go down and tech do better ....
One day will be a four channel with dps xover and we can build all DIY speaker we wants.....(2ways)
Specs help engineers to control the quality of their development efforts and help production departments to check, if the product actually works as originally designed. They also help customers to judge products without exclusively relying on belief, ears, looks or plain luck ...
Look at a TacT Millenium and its little brothers from 2004 and later, are you sure the NAD can do anything better, besides missing async USB and perhaps price ?
Did you ever hear about the great Devialet amp ?
Ian:
The IMD could also come from the fact, that the (Iron or Ferrite) output inductors are not enclosed within feedback ? I just try to help here ...
The low level linearity on the NAD/DDFA is only necessary, because it doesn't allow analog volume control/adjustment to the sensitivity of speakers.
Using a 120dB DAC and a coarse stepped analog volume control with min 20dB range, in front of an analog amp gives a comparable/better effective low level linearity of 140dB !
Add a XMOS USB solution and you get all the DSP functionality and a main processor as well, at no extra cost, no XILINX FPGA, and much better overall experience.
So what again was the main advantage of a DDFA amp ?
Oh, I fogot it:
It does not need to handle weak analog signals within a noisy (switched power) environment, as long as you stay digital.
How do you again play a SACD or vinyl over such an amplifier ... ?
ADC ? Aliasing with async power stage ?
Another thought:
What was again the lowest achievalble peak jitter of a sampled feedback system running at a discrete clock of xMHz and heavy group delay modulation (power stage switching speed modulated by power supply ripple and output signal) error source within this loop ?
Look at a TacT Millenium and its little brothers from 2004 and later, are you sure the NAD can do anything better, besides missing async USB and perhaps price ?
Did you ever hear about the great Devialet amp ?
Ian:
The IMD could also come from the fact, that the (Iron or Ferrite) output inductors are not enclosed within feedback ? I just try to help here ...
The low level linearity on the NAD/DDFA is only necessary, because it doesn't allow analog volume control/adjustment to the sensitivity of speakers.
Using a 120dB DAC and a coarse stepped analog volume control with min 20dB range, in front of an analog amp gives a comparable/better effective low level linearity of 140dB !
Add a XMOS USB solution and you get all the DSP functionality and a main processor as well, at no extra cost, no XILINX FPGA, and much better overall experience.
So what again was the main advantage of a DDFA amp ?
Oh, I fogot it:
It does not need to handle weak analog signals within a noisy (switched power) environment, as long as you stay digital.
How do you again play a SACD or vinyl over such an amplifier ... ?
ADC ? Aliasing with async power stage ?
Another thought:
What was again the lowest achievalble peak jitter of a sampled feedback system running at a discrete clock of xMHz and heavy group delay modulation (power stage switching speed modulated by power supply ripple and output signal) error source within this loop ?
Like I said, one reason behind the DDFA was to avoid the analog volume control and amplifier and do everything digitally. This reduces the opportunity for noise to get in -- not entirely, as Bruno pointed out the DAC/feedback nodes are still sensitive, but this small circuit is much easier to keep clean.
The low-level linearity quote was because one of the standard objections to class-D amps is that this is not very good, and DDFA is as good or better than most discrete DACs.
The Xilinx FPGA is used by NAD along with some discrete devices because that's what they had to do before the chipset was ready; the DDFA chipset (modulator + analogue feedback processors) doesn't need any of this, and can do eight single-ended channels including digital crossovers and EQ.
If you want to use analogue inputs then of course an ADC is needed; however the vast majority of music sources are digital nowadays, even among audiophiles.
With a feedback loop like DDFA it's the "reference DAC" which sets the overall performance, errors (including jitter) generated inside the feedback loop are suppressed just like power supply modulation, output stage nonlinearity and so on.
I measured the output jitter of this DAC (DDFA modulator running with a signal) at something under 1ps, including the on-chip oscillator. If this seems too good to be true, our chips for long-haul optical 100G networks have a 100 million gate DSP dissipating 50W on the same chip as multiple 60Gs/s ADCs with about 50fs of jitter -- yes that's femtoseconds
Audio is nowhere near the leading-edge as far as some areas of performance like jitter are concerned...
Hello to all!
Sorry for my english!
I think that Gyula has a lot it right, why?
I was a long time owner of NAD M2, and all was well, but once my son (he plays violin) while listening to guitar music has shown me that NAD M2 has a problem during playback. At some points he had and I feel that music flow somehow been too slow, braked, time slightly deformed, I can not better testify, but when my son would not show it, I had not even noticed. Only after my concentration I could easily hear the problems.
Then I compared NAD M2 directly against my old Sony S-Master TA-DA9000ES, Sony was much smoother and somehow correct temporal in sound, this was for the main argument to NAD for sale.
Can someone tell me if NAD M51 may have the same problems to me?
Sorry for my english!
I think that Gyula has a lot it right, why?
I was a long time owner of NAD M2, and all was well, but once my son (he plays violin) while listening to guitar music has shown me that NAD M2 has a problem during playback. At some points he had and I feel that music flow somehow been too slow, braked, time slightly deformed, I can not better testify, but when my son would not show it, I had not even noticed. Only after my concentration I could easily hear the problems.
Then I compared NAD M2 directly against my old Sony S-Master TA-DA9000ES, Sony was much smoother and somehow correct temporal in sound, this was for the main argument to NAD for sale.
Can someone tell me if NAD M51 may have the same problems to me?
bought by CSR...
DDFA® - Direct Digital Feedback Amplifier Technology
Recently implemented in NAD D 7050
DDFA® - Direct Digital Feedback Amplifier Technology
Recently implemented in NAD D 7050
NAD C390DD Usage
Hi Folks,
I have been living with the NAD C390DD feeding a pair of Aerial Acoustics 10Ts (via Audioquest Volcano BiWired) for over a month and I am really Happy with the sound overall and sound staging.
Can anyone clarify why 2X Upsampling prior to the NAD has such a profound effect? Is it that the Anti Aliasing filter running at a higher rate? Wish the NAD had its own real high quality Upsampler built in? Ah, but which one and can it best be done externally using a Mac Mini as I am using with Pure Music...
Ok, I get the feedback correction sampling theory to a point but still am confused a bit about how it is handling the FET delay and slew rate errors to/from the bridge. Is this truly Integration all fully corrected at this high processing rate?
I would love for someone to tell me where the sonic errors are in what I am hearing! With the 2X Upsampling before the NAD, all I hear is clean and palatable... Of course others may have better ears, but I have over 35+ years of comparing against the best of systems. Yes, the 10Ts are old technology but I wouldn't trade Up (or Down) to anything I have heard... +/- on all.
BTW - Soundstaging is almost as remarkable as the sonic detail that makes me have to listen multiple times to prove to myself... Its scary at times!
I welcome ALL feedback, suggestions and comments! Sound is #1 priority!
-SiCHIPS
Hi Folks,
I have been living with the NAD C390DD feeding a pair of Aerial Acoustics 10Ts (via Audioquest Volcano BiWired) for over a month and I am really Happy with the sound overall and sound staging.
Can anyone clarify why 2X Upsampling prior to the NAD has such a profound effect? Is it that the Anti Aliasing filter running at a higher rate? Wish the NAD had its own real high quality Upsampler built in? Ah, but which one and can it best be done externally using a Mac Mini as I am using with Pure Music...
Ok, I get the feedback correction sampling theory to a point but still am confused a bit about how it is handling the FET delay and slew rate errors to/from the bridge. Is this truly Integration all fully corrected at this high processing rate?
I would love for someone to tell me where the sonic errors are in what I am hearing! With the 2X Upsampling before the NAD, all I hear is clean and palatable... Of course others may have better ears, but I have over 35+ years of comparing against the best of systems. Yes, the 10Ts are old technology but I wouldn't trade Up (or Down) to anything I have heard... +/- on all.
BTW - Soundstaging is almost as remarkable as the sonic detail that makes me have to listen multiple times to prove to myself... Its scary at times!
I welcome ALL feedback, suggestions and comments! Sound is #1 priority!
-SiCHIPS
Folks, I fully appreciate all the info posted on this page regarding the DDFA and all of the math and modeling proof and analysis behind it. Truth is there is still a lot we don't understand, for example, how simple cable designs in specific systems can affect the sound as much as it does! Years of cable testing and evaluation in my systems have confirmed the differences which to my ears is startling to say the least. A Transmission Line is a just that a Transmission Line that can be modeled and deeply analyzed. Or how the ML 332 and 335 sound so different because of the package type for the output drivers -- This one is easier to swallow but again something missed in the design.
On the sound side, it is hard to beat the transparency of a ribbon although its integration into a full range system still leaves me feeling short of my goal. There are many, many really nice and great sounding speakers out there with lots of technology behind them yet still no ideal speaker exist. To me, listening for hours (actually 10+ straight) without getting a headache is very telling! Speakers costing many times that of the 10Ts have not cut the mustard on this front although there are some strong contenders -- Again for ME and my biases that is!
Everything is a massive tradeoff of parameters and finding the right set is a tricky and prejudicial issue. With the massive reduction in AES/EBU cables running around (Transport to D/A to PreAmp to AMP), the NAD has made my decision process that much easier!
I am seriously trying to grasp the "slow" or time deformed mentioned in a prior post and just don't have I quess the ears to hear it. I have been to numerous BSO and Early Music Festivals and likewise performances and just can't get a handle on what this post is all about. Considering the instrument range involved, I cannot understand how an issue would come in place. Yes, Transparency depends very much on the frequency response and its limits but to MY ears a flat response and very well balanced response between the 2 speakers that create the imaging is the main goal. I want to feel "At the Performance"
I started out with the KEF 105s and jumped to the 10Ts for pretty much all of what I just said. I am willing to give up some on the top end as I am getting older and don't hear it as well but I am totally unwilling to give up on that critical smooth Midrange.
Sorry if I sound to be preaching, I don't mean to be... These are MY criteria!
Do It Yourself is an absolutely fantastic approach but I just don't have the time to devote to another hobby as I already have many and just listening and drooling over music is what I can relate to. As a seasoned Digital Design Engineer, I understand and appreciate all the comments made but also totally understand -- As everything in engineering is all about -- Everything is a tradeoff and getting everything as a whole is virtually impossible. It is what keeps us busy and competitive with our peers!
Until the speaker technology catches up, there will still be numerous tradeoff of which we must pick for ourselves what we value the most important and strive for!
Maybe I need to spend Years with the C390DD to find the sonic flaws for myself but until then I will continue to be happy and maybe another system change will happen latter...
-SiCHIPS
On the sound side, it is hard to beat the transparency of a ribbon although its integration into a full range system still leaves me feeling short of my goal. There are many, many really nice and great sounding speakers out there with lots of technology behind them yet still no ideal speaker exist. To me, listening for hours (actually 10+ straight) without getting a headache is very telling! Speakers costing many times that of the 10Ts have not cut the mustard on this front although there are some strong contenders -- Again for ME and my biases that is!
Everything is a massive tradeoff of parameters and finding the right set is a tricky and prejudicial issue. With the massive reduction in AES/EBU cables running around (Transport to D/A to PreAmp to AMP), the NAD has made my decision process that much easier!
I am seriously trying to grasp the "slow" or time deformed mentioned in a prior post and just don't have I quess the ears to hear it. I have been to numerous BSO and Early Music Festivals and likewise performances and just can't get a handle on what this post is all about. Considering the instrument range involved, I cannot understand how an issue would come in place. Yes, Transparency depends very much on the frequency response and its limits but to MY ears a flat response and very well balanced response between the 2 speakers that create the imaging is the main goal. I want to feel "At the Performance"
I started out with the KEF 105s and jumped to the 10Ts for pretty much all of what I just said. I am willing to give up some on the top end as I am getting older and don't hear it as well but I am totally unwilling to give up on that critical smooth Midrange.
Sorry if I sound to be preaching, I don't mean to be... These are MY criteria!
Do It Yourself is an absolutely fantastic approach but I just don't have the time to devote to another hobby as I already have many and just listening and drooling over music is what I can relate to. As a seasoned Digital Design Engineer, I understand and appreciate all the comments made but also totally understand -- As everything in engineering is all about -- Everything is a tradeoff and getting everything as a whole is virtually impossible. It is what keeps us busy and competitive with our peers!
Until the speaker technology catches up, there will still be numerous tradeoff of which we must pick for ourselves what we value the most important and strive for!
Maybe I need to spend Years with the C390DD to find the sonic flaws for myself but until then I will continue to be happy and maybe another system change will happen latter...
-SiCHIPS
NAD has now switched to Hypex nCore. Is this the end of the DDFA technology?
http:// http://nadelectronics.com/products/masters-series/M22-Stereo-Power-Amplifier
http:// http://nadelectronics.com/products/masters-series/M22-Stereo-Power-Amplifier
They didn't switch over completely
I had to write here because I found an amazing synergy using my self built Ncore amp (simple dual SMPS600 & NC400, no modding) and the NAD M51 DAC.
The N51 now is also available in their normal line (D510 I believe?) - this dac is based on DDFA.
Running the NAD M51 balanced into the NC400's brought for me a very calm & peaceful, controlled sound that I didn't have with the Wyred4Sound DAC I was using before. Truly high-end into a pair of B&W 802 diamond speakers.
So if anyone can come up with a DIY based design for this chip, it's very very good!
I had to write here because I found an amazing synergy using my self built Ncore amp (simple dual SMPS600 & NC400, no modding) and the NAD M51 DAC.
The N51 now is also available in their normal line (D510 I believe?) - this dac is based on DDFA.
Running the NAD M51 balanced into the NC400's brought for me a very calm & peaceful, controlled sound that I didn't have with the Wyred4Sound DAC I was using before. Truly high-end into a pair of B&W 802 diamond speakers.
So if anyone can come up with a DIY based design for this chip, it's very very good!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.