Class D amplifier with digital inputs

[josedgm]

Hi guys.

This is my first post. But I've been reading your reviews and forums for many years. You all should be proud of this community. Most of you are very serious, respectful and most importantly you are willing to share your knowledge.

I have a crazy idea in mind, and I'd like to know your opinions.

A regular Minidsp system is like this
Source -> [Minidsp DSP -> DAC ] -> Power amplifier.

I know the Minidsp DAC is not bad, but it is not the best, and instead of expending in a DSP Digi and a DAC, what if I use the I2S outputs on the minidsp to feed a Class D amp digitally.

Most Class D amplifier turns analog input signal into a PWM to feed the power stage. But there are some that take a I2S digital input and turn this into the PWM. So there won't be a need for DAC. I think that the best DAC is no DAC: not only it is cheaper, but also it is purer.

What do you think? Have any of you tried something like this? How good are those digital input class D amplifiers in general?

Best regards,

Jose

 

[sergeauckland]

As PWM is analogue, whether the incoming digital signal is I2S or S-PDIF, there will still need to be a DAC in the amplifier, so I'm not sure what benefit there would be.
S.

 

[VintageFlanker]

This is my first post. But I've been reading your reviews and forums for many years. You all should be proud of this community. Most of you are very serious, respectful and most importantly you are willing to share your knowledge.

Welcome!!


I think you've been misleaded about class D being strictly Digital. What you're describing there would be Full Digital Amplifier instead.

I think that the best DAC is no DAC: not only it is cheaper, but also it is purer.

I see no reason for this. There's nothing telling us that the DA conversion would be any better that way than many good DACs.

A regular Minidsp system is like this
Source -> [Minidsp DSP -> DAC ] -> Power amplifier.

Sure but which MiniDSP product are we talking about?

 

[MakeMineVinyl]

As noted earlier, class D amplifiers are essentially analog. Sure, the signal could be converted digitally to PWM, but there's still a conversion stage, and what you propose seems like something which would not yield any audible benefit.

 

[josedgm]

Hi

Sure but which MiniDSP product are we talking about?

I'm talking about the 2x4 HD

Sure, the signal could be converted digitally to PWM

Yes, you are correct. There is a conversion stage from analog in to PWM. Usually is like this
Source -> [minidsp: DSP -> DAC] --analog-->[amplifier: signal conditioning -> PWM -> POWER -> LowPass filter] -> speakers

What I'm talking is since some class D amplifiers use I2S digital input, there won't be any need for the DAC stage and feed the PWM converter directly from the DSP.

Source -> [minidsp: DSP] --I2S-->[amplifier: PWM -> POWER -> LowPass filter] -> speakers

No DAC means no added distortion from the DAC and less $ to spent.

Also, the analog signal conditioning at the input stage in the power amplifier won't be needed. (this can help to reduce humming and other kind of noises).

I'm thinking on something like TAS5760M

 

[sergeauckland]

How do you feed a PWM converter directly from DSP without a DAC process? There still has to be a DAC whatever it's called.

S

 

[josedgm]

Using the I2S outputs directly from the DSP (they are in a connector at the MiniDSP board). Almost all DSP have these outputs that then feed the DAC IC.

The plan here is to eliminate the DAC and simplify the amplifier.

 

[NTK]

An article by Bruno Putzeys explaining there aren't really "digital" class D amps, and making a DAC with output capability of a power amplifier is not a good idea.
https://www.hypex.nl/img/upload/doc/an_wp/WP_All_amps_are_analogue.pdf

 

[sergeauckland]

I repeat, how do you feed an analogue amplifier with a digital I2S signal without a DAC?

S

 

[Koeitje]

The D doesn't stand for digital

 

[josedgm]

I repeat, how do you feed an analogue amplifier with a digital I2S signal without a DAC?

I2S is a standard to communicate ICs. The data stream from the DSP in the digital domain is serialized. Usually the DAC is connected to the DSP via I2S.

But he plan is to connect an class D amplificar that recevies this I2S stream directly to its PWM converter (without any analog conditioning). The only analog stage will be the output.

 

[josedgm]

The D doesn't stand for digital

some are pure digital

I mean, they don't have an analog input. Those digital amps can receive I2S streams (pure digital) from whatever source (a DSP) and feed their PWM (like other class D)

 

[Blumlein 88]

The old Tact amplifiers actually worked this way, and were available with only digital inputs. The DA conversion was the power output stage. Wadia has also made amps that work in this manner. Otherwise class D isn't digital as others have said.

 

[Koeitje]

some are pure digital

I mean, they don't have an analog input. Those digital amps can receive I2S streams (pure digital) from whatever source (a DSP) and feed their PWM (like other class D)

Are there any high-end class D amplifiers that allow that? I don't believe NCore en Purifi do, and the rest of them just get outperformed. If you goal is the theoretical best you are better off just getting a MiniDSP SHD with NCore or Purifi amplification.

 

[voodooless]

No DAC means no added distortion from the DAC and less $ to spent.

Find me one I2S input class D amp that can do better than a cheap DAC chip and a respectable analog Class D design? Distortion in DAC’s is never an issue.

Also, the analog signal conditioning at the input stage in the power amplifier won't be needed. (this can help to reduce humming and other kind of noises).

Yes it will! It is essential for a low distortion design. You need these feedback designs to get decent performance.

I'm thinking on something like TAS5760M

SINAD of about 70.. so much for less distortion.

 

[josedgm]

Thank you all for your replies

Yes it will!

Good/expensive amps that take analog signals use differential inputs to avoid humming and other noises get amplified at the power stage. But, please remember, since the DSP and the IC AMP chip will be talking serialized digital data via I2S, the analog signal conditioning for the amp is not needed. This kind of communication is way more noise resistant than analog signals.

SINAD of about 70

And according to the datasheet it is 105.

Have any of you listened/tested/measured a TAS5760M or similar?

 

[boXem]

I think there is a basic misunderstanding in your reasoning. All good class D amplifiers do the PWM conversion in the analog domain.
You are not the first to have the idea to do this conversion in the digital domain. Issue is that it's inherently non linear. Bruno Putzeys patented a way to do this conversion linearly. It is applied in the Mola-Mola DAC. Since it becomes terribly complicated to include the feedback from the output filter in the conversion in the digital domain, it's not worth the effort for an amplifier unless having a dedicated chip. Dedicated chip that the people at Axygn have done.
More generally, thinking that less components = less distortion is wrong. In many cases, adding components is needed to lower distortion. Feedback in example.

 

[voodooless]

Good/expensive amps that take analog signals use differential inputs to avoid humming and other noises get amplified at the power stage. But, please remember, since the DSP and the IC AMP chip will be talking serialized digital data via I2S, the analog signal conditioning for the amp is not needed. This kind of communication is way more noise resistant than analog signals.

Most proper designs are balanced from start to finish. Not for low noise, but to achieve higher output voltage and cancel out some distortions. The signal conditioning is mostly the feedback design and modulator. These are way easier to make in the analog domain.

And according to the datasheet it is 105.

Where did you find that: https://www.ti.com/lit/ds/slos772f/slos772f.pdf SNR is not SINAD!

Have any of you listened/tested/measured a TAS5760M or similar?

That is rather unimportant, we’re talking objective performance here, and looking at that, it’s clearly inferior.

The fallacy here is thinking that a class D amplifier is a digital device in the first place. It's not! They are (for the most part) purely analogue, so feeding them with a digital signal really doesn't solve any issues. None of these "full digital" design reach state of art performance at the moment.

 

[dougi]

Even Lyngdorf have given up "true digital" in their power amps it seems. The SDA2400 has a digital input, but it is a normal class D PWM amp. amp

They still use true digital in their integrateds but they measure worse. TDAI by Amirm

 

[EB1000]

Theoreticallly, having an integrated amp in which the digital audio signal is upscaled to 32bit @768kHz and converted to a Three level PWM to drive an GaN based full bridge power stage, where the volume control is implemented by digitally controlling the DC rail voltage of the power supply while feeding a full scale unatinuated digital signal to the PWM modulator, should be the best possible class d implementation. Too bad no such amp exists...