True dat. Just like an 8w tube SE amp is not the same as the 50 watt/ch 3116d2. All I'm saying is those who dismiss class D don't have an excuse not to try it out at that price. Unless one has a preconceived notion of certain class of amplifiers...
Duc
Several years ago I was starting to get into home theater. That subsequently led to me getting back into audio too - after years of a grueling consulting career. After reading a couple reviews of the, then new Pioneer SC-xx series line of receivers, I knew I had to try one out. I ended up picking up the baby of the line, the SC-05 because BB was closing them out for newer models and I got it basically for a song. No pun intended. 
Long story short.... By far this IMO is one of the sweetest receivers I have ever heard. The sound is incredibly smooth, without the clinical, and fatiguing affect often associated with Class D. And I'm still using it with a pair of very, very inexpensive Energy RC towers. When I switch on the Pure Direct mode of stereo listening, I am simply transported and no longer listening to my system, but rather hearing the music. While it's obviously not a DIY solution, the B&O ICEamp modules used in the SC line perform some kind of magic and just let the music flow right out - in the smoothest and most unconstrained way ever. I've been sold on Class D ever since.
Since then I've built several class D amps just for the sheer joy of getting back into the hobby, but none sound near as good as my SC-05. I'm a believer! The future is here.
Rick
Long story short.... By far this IMO is one of the sweetest receivers I have ever heard. The sound is incredibly smooth, without the clinical, and fatiguing affect often associated with Class D. And I'm still using it with a pair of very, very inexpensive Energy RC towers. When I switch on the Pure Direct mode of stereo listening, I am simply transported and no longer listening to my system, but rather hearing the music. While it's obviously not a DIY solution, the B&O ICEamp modules used in the SC line perform some kind of magic and just let the music flow right out - in the smoothest and most unconstrained way ever. I've been sold on Class D ever since.
Since then I've built several class D amps just for the sheer joy of getting back into the hobby, but none sound near as good as my SC-05. I'm a believer! The future is here.
Rick
Interesting article in the Journal of the Audio Engineering Society -- pinning some of the Class-D blame on the output filter.
Further to the Anthem -- JA posted this chart of a 1kHz 100W FFT in the measurements section:
The chart below is the ML #53 @150W --
Ok. the ML53 is an "Interleaved PWM" amplifier -- the manufacturer doesn't want it known as "Class-D"
Both have issues. The reviews remind me of the stuff I used to edit for a bunch of Wall Street stock and bond analysts when we didn't want an issuer to be PO'd. Tread ever so lightly on the manufacturers' toes.
I wonder whether PSRR and reverse-PSRR have been examined. I could do it here if anyone wants to lend me their amp.
Further to the Anthem -- JA posted this chart of a 1kHz 100W FFT in the measurements section:
The chart below is the ML #53 @150W --
Ok. the ML53 is an "Interleaved PWM" amplifier -- the manufacturer doesn't want it known as "Class-D"
Both have issues. The reviews remind me of the stuff I used to edit for a bunch of Wall Street stock and bond analysts when we didn't want an issuer to be PO'd. Tread ever so lightly on the manufacturers' toes.
I wonder whether PSRR and reverse-PSRR have been examined. I could do it here if anyone wants to lend me their amp.
Member
Joined 2009
Paid Member
The only real problems I see that are hindering class D at this time are the smps and amplifier RF emissions and the need to properly filter the out of band high frequency remnants that are corrupting the signal. Once that is well understood and implemented we will have the equivalent of our class AB amplifiers in sound quality. Until then we will continue to hear the complaints about poor frequency response bandwidth and noise with the designs.
What are you guys talking about with the bandwidth limit? 20 khz is tops for range and modulation is at 1.2 MHz. Nyquist is at 40 kHz. Are you referring to slew rate? If driver can't respond faster than 20 kHz, why do you need risetime faster than 25 microseconds? I contend that dominant ps noise is 60 Hz hum and this is much lower in smps than in big iron linear supplies.
Hi,
Nyquest certainly is not stupid, the problem is in the interpretation.
There is no Acquisitions in other areas, including cardiology that are based on Nyquest for the F # of sampling. on periodic signals can be applied Nyquest (then, counting on the kindness of the interpolator filter after). Unfortunately, the music signal is not a sine wave.
The publicity of Nyquest was launched when the 3 giants (Sony.. Etc) have decided to launch on the market, this "icing" of CD (very big business) instead of vinyl records.
Fortunately, most people can not hear the huge errors caused by the chain of reading (DAC) and the interpolation filter. independent of other problems, including the lack synchronization with the music track (this is not easy to explain)
regards
Roberto
Nyquest certainly is not stupid, the problem is in the interpretation.
There is no Acquisitions in other areas, including cardiology that are based on Nyquest for the F # of sampling. on periodic signals can be applied Nyquest (then, counting on the kindness of the interpolator filter after). Unfortunately, the music signal is not a sine wave.
The publicity of Nyquest was launched when the 3 giants (Sony.. Etc) have decided to launch on the market, this "icing" of CD (very big business) instead of vinyl records.
Fortunately, most people can not hear the huge errors caused by the chain of reading (DAC) and the interpolation filter. independent of other problems, including the lack synchronization with the music track (this is not easy to explain)
regards
Roberto
Member
Joined 2009
Paid Member
I understood that there are two main technical challenges
a) distortion from dead-time, needed to avoid shoot-through currents. This has been 'solved' as far as I know through better (faster) switching, better drivers and there are techniques akin to Class AB biassing at the cross-over
b) distortion from output filter due to load-dependent behaviour. I thought this has been 'solved' by taking the filter inside a global feedback loop.
a) distortion from dead-time, needed to avoid shoot-through currents. This has been 'solved' as far as I know through better (faster) switching, better drivers and there are techniques akin to Class AB biassing at the cross-over
b) distortion from output filter due to load-dependent behaviour. I thought this has been 'solved' by taking the filter inside a global feedback loop.
Magically agree on your point also (A), in fact, my last job, in an attempt to make it very close to the class AB, (not recognizable as class D) I have to 150mA (equivalent current bias). This is a huge value when compared to 20mA (max) of competitors in good D Class.
This was used for a special form of dead-time (controlled by the temperature also).
I think that is a big mistake if we try to translate the class D from PA sector, in the audiophile sector, using the same concepts of efficiency.
Last edited:
The JAES article "Switching Power Audio Amplifiers with High Immunity to the Demodulation Filter Effects" (JAES, December 2012) discusses a method of including the filter in one of multiple loops. Perhaps someone from ST Micro would make it more generally available.
b) is something that very few people do. Once TI start using output filter inclusive feedback in their chip amps then so much the better.
I know the industry as a whole is moving towards filter free operation, but this is only really a good idea when short wires to the loudspeaker are used and with low power devices. Certainly, with the voice coil inductance operating as the filter to remove the carrier, there wouldn't be the usual losses involved within the output filter. However I am sure that in such designs, the output filter wouldn't be the major cause of distortion either.
The TPA3118 that's quoted in this thread and that has gained something of a reputation, isn't the best that TI have to offer and is by no means anything special imo. A two channel amplifier built around one of TIs PWM drivers + an output stage like the TAS5614A beats it hands down in terms of technical performance, but these aren't as cheap and easy to come by.
Compared to 5 years ago TI are definitely moving closer and closer towards higher performance with their class D offerings. If discussing their purepath devices, such as the 5614, it would seem that the next logical step would be to include the output filter within the feedback loop, but this might be a risky endeavor concerning stability with third party designs. Then again that's what evaluation modules are for, to help you learn how to use the new parts.
Having said that, I'm still waiting for them to release single chip class D amplifiers, such as the TAS5607A that have the performance of the 6514.
I know the industry as a whole is moving towards filter free operation, but this is only really a good idea when short wires to the loudspeaker are used and with low power devices. Certainly, with the voice coil inductance operating as the filter to remove the carrier, there wouldn't be the usual losses involved within the output filter. However I am sure that in such designs, the output filter wouldn't be the major cause of distortion either.
The TPA3118 that's quoted in this thread and that has gained something of a reputation, isn't the best that TI have to offer and is by no means anything special imo. A two channel amplifier built around one of TIs PWM drivers + an output stage like the TAS5614A beats it hands down in terms of technical performance, but these aren't as cheap and easy to come by.
Compared to 5 years ago TI are definitely moving closer and closer towards higher performance with their class D offerings. If discussing their purepath devices, such as the 5614, it would seem that the next logical step would be to include the output filter within the feedback loop, but this might be a risky endeavor concerning stability with third party designs. Then again that's what evaluation modules are for, to help you learn how to use the new parts.
Having said that, I'm still waiting for them to release single chip class D amplifiers, such as the TAS5607A that have the performance of the 6514.
Class D has precisely nothing to do with DSP. In fact, most class D amplifiers are entirely analog, not "digital amplifiers" as they are often mistakenly called.
The main difference in Class D relative to A and A/B is that in Class D, pulse widths are controlled by the output devices (which transistors can do quite precisely and linearly) and in A or A/B amplifiers, direct current is controlled by the output devices (and none of the power devices have particularly linear current/voltage control curves).
Thanks for posting this. Fits what I hear when listening to Class D. BTW I heard the Devialet D-Premier (the hifi press raves about it) three months ago at an exhibition, if this is the "future" it´s not for me.
Granted those are high levels, the harmonic content is still remarkably nasty. It reminds me - 100x scaled -of a Mullard 3-3 that didn't last long in my system.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.