Digital amps and digitally active speakers

[Thomas savage]

could one of you more learned folks explain just how digital amplification works , is it any good? As far as I'm aware it's basically a powered dac with volume control in the digital realm?

Also in terms of actives that use this technology like the dynaudio XD range, anyone heard them ... Any thoughts ?

http://www.stereophile.com/content/dynaudio-focus-200-xd-powered-loudspeaker#5XQsCzwK8YCckBL2.97

 

[Cosmik]

In the absence of a reply from a more learned person, I'll chip in with the suggestion that what is really meant by digital amplifier is usually Class D, which is generally used because it is much smaller and more efficient than a conventional amplifier.

I could imagine a variety of 'topologies' that might take in analogue, or a PCM stream, or DSD, and generate the drive signals to the 'digital' on/off output transistors of the amp in slightly different ways. Some schemes might be considered more 'digital' than others, but ultimately they are all pretty much equivalent..?

 

[RayDunzl]

As with other amplifiers, they use the properties of the mysterious and apparently fundamental particle/waves known as electrons as an important and indispensible part of their operation.

So, ultimately, nobody knows how it works. It just does. And differently, than the traditional Class A and A/B amplifiers, which was your question, which I have avoided up to now.

Wait, here is someone trying to explain:

http://www.eetimes.com/document.asp?doc_id=1274757

 

[Thomas savage]

Don't think I'm talking about class D as in mola mola, nord , ncore etc ( this seems diffrent hence the questioning )

http://www.pureaudiostream.com/true-digital-audio-amplifiers/

http://www.stereophile.com/content/pure-digital-amplifiers

http://www.stereophile.com/content/dynaudio-focus-200-xd-powered-loudspeaker#5XQsCzwK8YCckBL2.97

 

[DonH56]

Class D amplifiers can be architected different ways. If you take the PCM stream from a digital source, amplify and filter it, then you have essentially created a "digital" amplifier that is like a delta-sigma DAC with a high-power output stage. The output is one form of a class D amplifier. Most processors these days do everything in the digital domain up to the final power amp, which may be digital (class D) or not.

 

[Thomas savage]

Thanks

 

[amirm]

Class D amplifiers can be architected different ways. If you take the PCM stream from a digital source, amplify and filter it, then you have essentially created a "digital" amplifier that is like a delta-sigma DAC with a high-power output stage. The output is one form of a class D amplifier. Most processors these days do everything in the digital domain up to the final power amp, which may be digital (class D) or not.

And that's exactly what they do. From TI data sheet for this part :

And what is in it:

 

[Cosmik]

And that's exactly what they do. From TI data sheet for this part :

So it's a PCM-to-PWM converter. Presumably the right sort of DSD could drive the power outputs directly..?

Is there any necessity for feedback from the output to the earlier stages? Is the output so linear, and the output impedance so low that the speaker's impedance has negligible effect, making feedback unnecessary? If so, that is a major advance on a normal amplifier topology.

However, looking at the Wikipedia entry, we have:

The actual output of the amplifier is not just dependent on the content of the modulated PWM signal. The power supply voltage directly amplitude-modulates the output voltage, dead time errors make the output impedance non-linear and the output filter has a strongly load-dependent frequency response. An effective way to combat errors, regardless of their source, is negative feedback. A feedback loop including the output stage can be made using a simple integrator. To include the output filter, a PID controller is used, sometimes with additional integrating terms. The need to feed the actual output signal back into the modulator makes the direct generation of PWM from a SPDIF source unattractive.[6] Mitigating the same issues in an amplifier without feedback requires addressing each separately at the source. Power supply modulation can be partially canceled by measuring the supply voltage to adjust signal gain before calculating the PWM[7] and distortion can be reduced by switching faster. The output impedance cannot be controlled other than through feedback.

If feedback is necessary, then it must come back into the modulator stage, and suddenly it all seems a little less straightforward and less 'digital'.

 

[fas42]

If the process is "perfect", including being able to take into account any non-optimum behaviour of the speaker, then feedback won't be necessary. The real world, unfortunately, ...

 

[amirm]

Is there any necessity for feedback from the output to the earlier stages?

Indeed the Wiki stole the words out of my mouth or should I say keyboard. They seem to punt the power supply issue though:

"Power Supply and Decoupling TDAA audio amplifier technology is a direct PCM-to-PWM converter without feedback; this requires good decoupling and a good power supply. The power supply voltage needs to be regulated because the H-bridge is switching the supply level directly to the outputs via the output reconstruction filter. The power supply in a TDAA amplifier can be regarded as a voltage reference in a DAC. Changes in the power supply act as a volume change and, in fact, by having a power supply to the output stage that is varied logarithmic, this can be used as a volume control.

Normally, digital volume control is easier to implement. Texas Instruments digital audio processors, such as the TAS3001 or the TAS3002, can implement volume, bass, and treble control, and can also do parametric equalization completely in the digital domain."

The power supply would need to be switchmode to keep up with the efficiency model of this part. Unfortunately switchmode power supply tend to not have great dynamic response so work needs to be put in there to make them perform. The alternative they suggest is interesting which would involve in an analog to digital converter to measure the supply changes and apply digital volume correction to counter!

The target market for this seems to be mass market amplification with good-enough specs without feedback. Rated distortion is 0.03% for the part which is not great. TI acquired the company which means they saw a large market for it like home-theater-in-a-box than any high-end audio applications.

 

[DonH56]

Not really my field, but I have designed a few higher-frequency class D and similar output stages. As already stated, power supply noise is one of the Achilles Heels of class D outputs in general. You are switching rail-to-rail (high-frequency) pulses so any modulation of the power supply is directly translated to the output. Thaty means very "stiff" supply with broadband decoupling is required. The good news is that the power supply noise spikes are waaay above the audio band.

Feedback is also very tricky. It is hard to close the loop around the output stage and keep everything stable, especially if you try to include the output filter needed to suppress all the high-frequency switching noise. There are some really neat schemes, some of which include feedforward and parallel loops, others use a hybrid amplifier topology mixing class A or AB with class D handling most of the high power.

I have noticed digital and analog volume controllers in various AVRs and amplifiers. There is a phobia against digital in some circles due to the loss of resolution (bits) as you decrease the output level. The counter argument is that you don't need such a high number of levels (bits) as the signal gets softer. However, digitally-controlled (programmable) analog attenuators are used in many audio components.

The dynamic performance and wideband noise for any power supply is typically set by the output decoupling network. Switching supplies, operating at hundreds of kHz or more, can recharge their filter caps much more quickly but one pulse at a time. IME their dynamic response is better than conventional linear supplies but my experience does not really include supplies used for audio power amplifiers. And here is so much variation in design, so many trades in designing for target output current and ripple, that I am not sure there is a general rule of thumb. The high switching rate means you can use smaller capacitors to achieve ripple comparable to a linear supply with much larger capacitors. You do generally need a plethora of small wideband capacitors to suppress switching noise, and power and ground plane layout is critical to control coupled and radiated noise.

FWIWFM - Don

 

[watchnerd]

could one of you more learned folks explain just how digital amplification works , is it any good? As far as I'm aware it's basically a powered dac with volume control in the digital realm?

Also in terms of actives that use this technology like the dynaudio XD range, anyone heard them ... Any thoughts ?

http://www.stereophile.com/content/dynaudio-focus-200-xd-powered-loudspeaker#5XQsCzwK8YCckBL2.97

I have heard them, and compared them to the new Contours.

As @amirm notes, the XD range uses TI PWM modules. I heard the Focus 200 XD vs the Contour 20 powered by Bel Canto 600 monoblocks (nCore-based).

IMHO, it was pretty much a tie and they sounded more similar than different. The Contour 20 has slightly more over-engineered / better drivers than the Focus 200 XD, but the Focus 200 had more uniform response, both on and off axis. The Contour 20 certainly looks fancier. Both, through their respective amps, played quite loud with ease -- as loud as I would ever want, at least.

That being said, I didn't like the digital volume on the Focus XD and built-in Wifi streaming that isn't upgradeable doesn't appeal to me. So I placed an order for the LYD 7, instead, which uses the same family of TI amps, but a more recent version, and has more fancy pants DSP tricks in its digital crossover (and doesn't have the consumer-level Wifi, Bluetooth, etc.).