AVI - Class D development

[oldric_naubhoff]

Exceptions would be the TacT, Lyngdorf and Sony S-Master Pro series, and as the wikepedia article states, these are less appealing from a design POV due to the inability to effectively implement global feedback schemes to lower distortion; instead each part of the audio chain needs to be corrected individiually/locally via DSP; there is also the problem of the amplfiers high output impedance due inablitiy to implement global feedback.

this is totally incorrect. the only reason why class D measures well is because they use high levels of negative feedback. therefore such amps have very low output impedance, as a bonus. however, you'll never get an open loop class D amp that will measure well. even a SET next to such an amp will be a benchmark impossible to beat.

It has a massive linear power supply, but nonetheless can deliver a tested 170 watts per channel x 7 channels, all channels driven into 8ohms at < 0.01% distortion, so is no slouch on the measurments side of things, and does it all whilst remaining exceptionally cool running - no mean feat.

yeah. I'd like to see how the measurements change at 0.5W or into high frequencies. it's well known truth that amps that solely rely on loop negative feedback to get decent standard THD measurement (at 1kHz) don't measure quite well at low power levels and into high frequencies (if the open loop gain is very narrow).

 

[drummerman]

http://www.primare.net/assets/_managed/cms/files/PrimareUFPD2.pdf

regards

 

[BenLaw]

Interesting link, thanks dm.

 

[nopiano]

Well, the Devialet D-premier is probably the best amp I've ever heard, and that incorporates a large slug of Class D, so I need no persuading it can be amazing.

http://www.devialet.com/technology.php

I also like what I've heard from Primare, though it has not found universal favour in WHF. I'm pretty sure it represents the future.

Don't know what is in a cell phone, but can well imagine some car audio is Class D and I suspect our old Sony DAV-S550 cinema system was Class D (though Sony called it s-master or some such!) because it is tiny and cool to touch.

Whether AVI can deliver on the potential I don't know.

 

[busb]

Exceptions would be the TacT, Lyngdorf and Sony S-Master Pro series, and as the wikepedia article states, these are less appealing from a design POV due to the inability to effectively implement global feedback schemes to lower distortion; instead each part of the audio chain needs to be corrected individiually/locally via DSP; there is also the problem of the amplfiers high output impedance due inablitiy to implement global feedback.

this is totally incorrect. the only reason why class D measures well is because they use high levels of negative feedback. therefore such amps have very low output impedance, as a bonus. however, you'll never get an open loop class D amp that will measure well. even a SET next to such an amp will be a benchmark impossible to beat.

It has a massive linear power supply, but nonetheless can deliver a tested 170 watts per channel x 7 channels, all channels driven into 8ohms at < 0.01% distortion, so is no slouch on the measurments side of things, and does it all whilst remaining exceptionally cool running - no mean feat.

yeah. I'd like to see how the measurements change at 0.5W or into high frequencies. it's well known truth that amps that solely rely on loop negative feedback to get decent standard THD measurement (at 1kHz) don't measure quite well at low power levels and into high frequencies (if the open loop gain is very narrow).

Normally large amounts of neg feedback would be frowned apon but that feedback is applied to the modulated signal rather than the in-band output directly. As for measuring THD at 1kHz, the only thing that's going to reveal is whether a broken amp has been successfully repaired or not rather than equating THD with SQ. As for HF distortion, as long as those out-of-band products don't mix down into the audio spectrum, they are surely insignificant? THD will always rise when measured at low levels due to noise contributions overtaking harmonic content in amplitude.

What I can't understand is why anyone would go to the trouble of either designing or using off-the-shelf class D modules powered from an inefficient linear PSU rather than using switch mode technology. Both class D & switch mode supplies have far more to go wrong due to their innate complexity - that's my issue with such technology: reliability.

 

[JMacMan]

Exceptions would be the TacT, Lyngdorf and Sony S-Master Pro series, and as the wikepedia article states, these are less appealing from a design POV due to the inability to effectively implement global feedback schemes to lower distortion; instead each part of the audio chain needs to be corrected individiually/locally via DSP; there is also the problem of the amplfiers high output impedance due inablitiy to implement global feedback.

this is totally incorrect. the only reason why class D measures well is because they use high levels of negative feedback. therefore such amps have very low output impedance, as a bonus. however, you'll never get an open loop class D amp that will measure well. even a SET next to such an amp will be a benchmark impossible to beat.

It has a massive linear power supply, but nonetheless can deliver a tested 170 watts per channel x 7 channels, all channels driven into 8ohms at < 0.01% distortion, so is no slouch on the measurments side of things, and does it all whilst remaining exceptionally cool running - no mean feat.

yeah. I'd like to see how the measurements change at 0.5W or into high frequencies. it's well known truth that amps that solely rely on loop negative feedback to get decent standard THD measurement (at 1kHz) don't measure quite well at low power levels and into high frequencies (if the open loop gain is very narrow).

Being more specific, I can only speak on technical grounds of the TA-DA9000ES Sony S-Master Pro amp, which I own. Sony state that it does not use any negative feedback - hence my stating it as such.

Quote:

"Superb open-loop performance. Traditional amplifiers typically generate substantial distortion in "open-loop" mode. That's why analog amps use Negative Feedback (NFB). Unfortunately, NFB exposes the signal to Transient Intermodulation Distortion and other dynamic problems. In contrast, the Sony S-Master Pro amplifier achieves excellent fidelity without any negative feedback at all! Distortion remains low without any sacrifice in transient and dynamic characteristics. Music comes alive."

Here is a link to the full Sony White paper on the design.

http://www.docs.sony.com/release/ES_STR_05_Final.pdf

JMac
[/list]

 

[JMacMan]

Exceptions would be the TacT, Lyngdorf and Sony S-Master Pro series, and as the wikepedia article states, these are less appealing from a design POV due to the inability to effectively implement global feedback schemes to lower distortion; instead each part of the audio chain needs to be corrected individiually/locally via DSP; there is also the problem of the amplfiers high output impedance due inablitiy to implement global feedback.

this is totally incorrect. the only reason why class D measures well is because they use high levels of negative feedback. therefore such amps have very low output impedance, as a bonus. however, you'll never get an open loop class D amp that will measure well. even a SET next to such an amp will be a benchmark impossible to beat.

It has a massive linear power supply, but nonetheless can deliver a tested 170 watts per channel x 7 channels, all channels driven into 8ohms at < 0.01% distortion, so is no slouch on the measurments side of things, and does it all whilst remaining exceptionally cool running - no mean feat.

yeah. I'd like to see how the measurements change at 0.5W or into high frequencies. it's well known truth that amps that solely rely on loop negative feedback to get decent standard THD measurement (at 1kHz) don't measure quite well at low power levels and into high frequencies (if the open loop gain is very narrow).

Normally large amounts of neg feedback would be frowned apon but that feedback is applied to the modulated signal rather than the in-band output directly. As for measuring THD at 1kHz, the only thing that's going to reveal is whether a broken amp has been successfully repaired or not rather than equating THD with SQ. As for HF distortion, as long as those out-of-band products don't mix down into the audio spectrum, they are surely insignificant? THD will always rise when measured at low levels due to noise contributions overtaking harmonic content in amplitude.

What I can't understand is why anyone would go to the trouble of either designing or using off-the-shelf class D modules powered from an inefficient linear PSU rather than using switch mode technology. Both class D & switch mode supplies have far more to go wrong due to their innate complexity - that's my issue with such technology: reliability.

It's good that there are members here with significant technical knowledge/understanding who can explain some of the finer, more in-depth nature of things, as I find it very interesting, so appreciate your comments and insight.

My personal approach to HiFi has come from the other end though, i.e. as a trained classical musician and music lover, who likes to know how things work on a reasonably in-depth level, with respect to HiFi and other interests such as Photography, as knowledge is empowerment when it comes to consumer buying decisions generally, but I'd not generally describe myself as being an in-depth technical expert with respect to amplifier technology &/or electronics design, despite being interested.

I'm probably more interested overall and all things considered, in the end result as to how a system reproduces music, more than the finely detailed method of getting there, albeit I'm stil interested in that to a degree; just not as much as some.

As an example, my new active speakers feature(?) Class AB amps on the mid/tweeter, and Class D ICE power on the bass - which is of interest to me, along with other design points/details - but I couldn't tell you what the distortion figures are as the manufacturer doesn't quote them, and I have no test reviews with that information; suffice it to say they reproduce music to a level amongst the best speakers I've ever heard, which on balance, is as much information as I personally need to know, along with the design details, to justify the cost of admission.

A solution that offers both emotional and intellectual satisfaction perhaps?

Others may well be able to argue the pros and cons of the design specifications far better than me, and it's great to hear about those details as I find it interesting as I say, especially as my knowledge is limited compared to some, and it's another very valid area of this hobby, but ultimately for me, with my musical rather than technical background, it's the end musical results that count the most rather than the method of getting there, speaking personally, as far as my involvement in the hobby of HiFi is concerned.

Others of course, come at it from different angles, such as the technology, so it's great to read in-depth comments on that side of it - much appreciated.

John...

 

[drummerman]

Class A, AB or whatever derivative of it are archaic in time line, as are line frequency power supplies. They have been around longer than my grandmother and if I'd tell her she is inefficient or worse, lossy, I'd get whacked around the earyholes.

Every so often someone like Douglas Self comes along and puts a twist on it but the shortcomings are design inherent. They are still the preferred method for most audio specific designs though and there are reasons for it though they are being addressed slowly.

As to complexity and reliability, this will slowly improve no doubt. Amplifiers are traditionally expected to work for years even decades without to many issues whereas a cheap dvd player, phone or whathaveyou can seize to operate after a couple of years and most people wouldn't blink twice. It it was that an unsurmountable problem, companies such as B&O, Primare etc probably wouldnt sell it but I guess only time will tell.

In 'high end' audio, there is another issue though ... . Often, perceived value is associated with weight. If your multi hundred watt Krell, Musical Fidelity or whatever 40 kilogram arc welder now comes in a chocolate selection sized little box weighing the same as 2 litres of milk ... 5 or 10 grand suddenly perhaps doesn't seem to be such 'great value' any more ...

Someone mentioned Devialet. They have made the jump with their own solution and still manage to charge a small hatchback. Perhaps all is not lost ...

regards

 

[andyjm]

Both class D & switch mode supplies have far more to go wrong due to their innate complexity - that's my issue with such technology: reliability.

High end hifi is generally years behind the curve in the application of current technology. It is a conservative cottage industry.

To refer to class D and switch mode as complex or unreliable is very wide of the mark. Anyone viewing this forum is siting behind a computer with north of a billion transistors all working happily away, almost certainly powered by a SMPS and probably with a class D amp on board. Hundreds of millions of smartphones are being charged as I type by wallwart SMPS, and have class D amps for their speakers.

Now it may be true that if all you know is linear supplies and class AB amps that class D and SMPS are complex - but unreliable they are not, and in comparison to modern electronics they are extremely simple.

A class AB amp slowing cooking its internals is no more likely to be reliable than a class D amp running cool.

 

[manicm]

Funny, all this talk about class D, and no-one mentions NAD's two class D integrated amps. Do I detect a hint of snobbery here?

 

[busb]

Both class D & switch mode supplies have far more to go wrong due to their innate complexity - that's my issue with such technology: reliability.

High end hifi is generally years behind the curve in the application of current technology. It is a conservative cottage industry.

To refer to class D and switch mode as complex or unreliable is very wide of the mark. Anyone viewing this forum is siting behind a computer with north of a billion transistors all working happily away, almost certainly powered by a SMPS and probably with a class D amp on board. Hundreds of millions of smartphones are being charged as I type by wallwart SMPS, and have class D amps for their speakers.

Now it may be true that if all you know is linear supplies and class AB amps that class D and SMPS are complex - but unreliable they are not, and in comparison to modern electronics they are extremely simple.

A class AB amp slowing cooking its internals is no more likely to be reliable than a class D amp running cool.

I hope I'm wrong having invested well over £1k in a Primare class D power amp!

In the last 40 years of repairing electronic test equipment, wiring looms have all but disappeared, the brief flirt with wire wrapping came & went. Valves went early in my career & a mixture of valves & transistors was dropped quickly due to poor reliability caused by heat cooking the transistors. As a general observation, modern electronics is pretty reliable with the use of leadless solder's impact yet to be ascertained. Although electronics is more reliable, switch-mode PSUs don't fail gracefully but tend to burn out tracks leading to module level replacement or scrapping the whole unit. Transistor count in itself isn't a great indicator of increasing unreliability because large scale integration of ICs means fewer devices that are fairly reliable to start with that translates into fewer pins to be soldered.

However, the greater the number of devices still translates into a greater number of potential failures (even if this seems to contradict what I've said above. The observation about heat increasing failures is well–made though. My A34.2's case is unventilated so the build-up of dust is unlikely to be an issue. Although the power amp isn't exactly light at 10.5kg, it's lighter than my Primare I21 integrated by 3kgs for the same case size.

 

[Frank Harvey]

I hope I'm wrong having invested well over £1k in a Primare class D power amp!

In the last 40 years of repairing electronic test equipment, wiring looms have all but disappeared, the brief flirt with wire wrapping came & went. Valves went early in my career & a mixture of valves & transistors was dropped quickly due to poor reliability caused by heat cooking the transistors. As a general observation, modern electronics is pretty reliable with the use of leadless solder's impact yet to be ascertained. Although electronics is more reliable, switch-mode PSUs don't fail gracefully but tend to burn out tracks leading to module level replacement or scrapping the whole unit. Transistor count in itself isn't a great indicator of increasing unreliability because large scale integration of ICs means fewer devices that are fairly reliable to start with that translates into fewer pins to be soldered.

I have no idea how the Primare Class D amps measure, but the ones I've heard so far sound great

The only Class D amplifiers I know of that have come back for repair/replacement have been subwoofer amplifiers, but I'm assuming that is because they're probably cheaply made in comparison to hi-fi amplifiers as they've been amplifiers from sub £500 subwoofers.

 

[busb]

Cheers Dave

Good news. I have a slightly jaundiced view of electronics having serviced so much of it. I'm glad to say that my current job entails fixing new stuff that has been built incorrectly - usually wrong components or wiring.