DSD and Class D

September 11, 2017
 by Paul McGowan

A common engineering joke around PS Audio is the TLA, a self-referencing acronym of the finest order. TLA, DSD, PCM, oh my, oh my. Sometimes I get a little crazy with all these initials representing technologies.

One of the most common misconceptions is that Class D and DSD are the same. Nothing could be further from the truth.

DSD is PDM (crap – there those TLAs go again). Pulse Density Modulation—sometimes known as 1-bit audio—is about as close to analog as digital audio gets. Instead of PCM (there we go again…) which involves ones and zeros in undecipherable code, DSD bunches greater or fewer numbers of “on” bits together to form music. Take a look at this picture:

The white 1-bit spaces are zeros while the blue 1-bit spaces are ones. Notice when the ones outnumber the zeros the red sine wave goes up in level. White spaces have less energy and the sine wave goes down. This is very different than Class D.

Class D is another clump of letters, PWM (… sigh). Pulse Width Modulation has only one “bit” as well, but instead of many 1-bits of greater or lesser density, as in DSD, Class D keeps that bit on or off for shorter or longer periods within a given frame of time. Here’s an example.

Note how the red shapes get wider as the sine wave rises, narrower as it gets smaller. This is what Class D is.

If you’d like to get a personalized view of this, I produced a video called The Difference between DSD and Class D.

Wrapping our arms around all these terms so we have an inkling of what they mean helps navigate the complex world of high-end audio.

Subscribe to Paul's Posts

13 comments on “DSD and Class D”

  1. A;though I owned, briefly, an example of the first commercial class D amplifier (a Sinclair X10) class D and DSD are both technologies which came ‘after my time’, and I have little familiarity with them. One thing which does strike me immediately about your description is that the supply voltage needs to be absolutely constant, far more so than for analogue devices, because any variations are going to affect the output signal directly. I therefore suspect that you need better PSU’s (another acronym – Power Supply Unit!) with them. I also worry a bit about reactive loads for class D amplifiers. A low output impedance conventional amplifier will soak up the energy fed back, but with these digital technologies it will go to the supply rail, which will potentially increase distortion. I guess I need to do some reading.

  2. There’s money in them thar TLAs, if you play your cards right. Against the odds, MQA the “TLA du jour” seems to be transforming from lead to gold (for Bob Stewart at least). I was looking at Chord the other day, possibly the closest company in Europe to PSA (TLA superieur), who make reference power amps and FGPA (FLA) DACs. Chord are distinctly lacking in TLAs, possibly unnecessary when you name your products Dave, Toby, Hugo, Mojo etc,. The company was started after the founder had a long and successful career designing power supplies for major defence contractors and others. Their topology is Sliding Class AB using proprietary Dynamic Coupling technology, so the obvious acronyms are SLAB and DyC. Maybe not. Seems TLAs are a bad excuse for not coming up with a decent name for something in the first place. Chord also win the prize for the worst acronym-based product name ever, past or future, the QBD76HDSD, it was their flagship DAC (there we go again, you can’t escape the TLA curse), I’m not surprised they switched to naming products after random people.

  3. i am somewhat confused. From the illustration, it seems like DSD/PDM is a one bit sampling of PWM? At least that had been my previous understanding, so maybe my bias is causing me to not see the distinction.

    Of course, how the sampling, or dithering, occurs will impact how the sound is digitized. An analogy for me is back in the old dot matrix printer days where there was no gray capability, but dot patterns approximate gray levels – of course the same density level has multiple patterns so how the patterns are selected and juxtaposed impacts how nice the image appears.

  4. p.s. I can’t believe anyone can confuse Class D and DSD, the main confusion is people thinking the D in Class D is Digital.
    PSA doesn’t help by its flagship model being called a DSD (maybe HDDSDAC would have been better) and I didn’t know I’d owned a PWD until after I sold it. I just thought it was a MkII DAC. Finding PWDMkII+2500=DSD, I went MDAC+.

    1. If you think of the D amplifier generating the output signal by switching on/off at high frequency, that resembles DSD generating the lower level output signal in the DAC…. I think that’s where the “confusion” comes in — but as noted in the post, usually the amplifiers use PWM instead of PDM.

      That’s where my understanding ends and my own confusion begins. I’ve read examples such as, given say a time interval t, a 50% signal would be t/2 on + t/2 off for PWM vs say 8 alternating on/offs for PDM…. but this example just confuses me bec doesn’t that mean that they differ only when you look at how the sampling/discretization intervals work? I don’t think I’ve seen a precise definition differentiating the two — at least not one which I understood.

      1. To be honest, I’ve been using Class D amplifiers for most of the last 20 years. I briefly used Linn, but found it rather dry. The tech behind all Class D I’ve had is beyond my understanding, but Linn provided an explanation of Class D (and its misconception) that makes sense to me:

        “CHAKRA is a linear (continuous time) amplifier technology with a switch mode (discrete time) power supply. There are now several products on the market based on switch mode amplifier technology commonly referred to as ‘Digital’ or ‘Class D’ amplifiers. This attempt to associate the well known attributes of digital systems with these products is, however, at best a slightly misleading premise.
        A traditional amplifier audio circuit features a continuously variable (linear) output stage generating the required output by dissipating as heat the difference between the power supply voltage and the required output voltage. A Class D amplifier uses very high speed switches to send the entire voltage of the power supply, with minimal power loss, as a sequence of pulses to the output, where a filter averages the pulses before the loudspeaker terminals. The frequency, density and duty cycle of these pulses determine the average output voltage. This is exactly how a switch mode power supply works, except that instead of generating the system power supply, the Class D amplifier circuit is generating the final audio output.
        Like SMPS, the concept of Class D amplifier audio circuit technology is not demanding; indeed Class D amplifiers have been around for 30 years, but have failed to penetrate the specialist high fidelity sound equipment market, primarily because they lack the subtlety of well executed linear amplifier technology. However, with the ever-more widespread adoption of multi-channel systems there is pressure on the industry to develop lighter, smaller and cheaper power amplification platforms which in turn has sparked renewed interest and research effort in Class D technology. The prospect of combining Linn’s Switch Mode Power Supply technology with advanced Class D audio circuitry is an appealing one, but can only be considered when the technology is mature enough to offer a performance advantage – in other words, when it meets the standards Linn customers expect from our company.”

      2. I am finding this a bit of an uphill struggle myself. I think this is. more or less, where I have got to. With PWM (pulse width modulation) you have a series of pulses at constant frequency but varying width. With PDM (pulse density modulation) the pulses have constant, and narrow, width but appear with varying frequency in time. The overall effect is rather similar, since they are different ways of representing an on to off ratio which is your signal level, and what appears after the pulses have been put through a low pass filter.

        This is where things star getting a bit hazier. A class D amplifier seems normally to be implemented using PWM, but it can also use PDM by utilisng a Delta Sigma conversion rather than a comparator, and the result seems, to me, at first sight, effectively DSD. They say this results in lower distortion. Delta SIgma and ADC are still very much in the region of my mental cartography marked ‘Here be Dragons’.

        This may not have helped, but at least you know you are not struggling alone

  5. I’ve been watching some old videos from 2013 and 2014 RMAF with debates of PCM versus DSD versus analog tape versus cassettes and some of Michael Fremer’s vinyl lunacy just for laughs with a little Atkinson thrown in for a view from planet Stereophile and all I can say is….THEY’RE ALL NUTS. The only thing they agree on is that no one agrees with anyone else about anything. It’s like watching mice run madly through a maze with no way to get to the cheese. Whenever I took my dogs to Petco they had the same fascination with watching real mice as I have watching these mental mice. They want it so badly they can already taste it but it always eludes them. They sometimes even dream they are eating it.

    Meanwhile I’m trying hard to get interested in listening to recordings again. As anyone who has read my posts has heard I have what I believe is the most difficult sound system to adjust in the world. Right now I’m struggling to make Vladimir Ashkanazy’s recordings of Beethoven’s piano sonatas sound like it’s being played on a Steinway piano. Struggling is hardly the word for it. Maybe there is something wrong with the recording but it is fantastic playing and whether you have the 3 sonata version on one disc or the 6 sonata version on two discs (somehow I’ve got both) this recording is one I highly recommend. Stunning performances. Getting there slowly but not quite there yet. So close I can almost taste it but so far it just doesn’t sound quite right yet. Can I fix it? I never know until I’ve fixed it.

    1. Perhaps the analogy should be sending your beasts to Petco with $50 and their arguing over whether to get a bag of Blue Buffalo or a bag of Canidae. If they had any sense they’d get a half bag of each and enjoy both. They would be impossible because they’re dogmatic.

Leave a Reply

Stop by for a tour:
Mon-Fri, 8:30am-5pm MST

4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

Stop by for a tour:
4865 Sterling Dr.
Boulder, CO 80301

Join the hi-fi family

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram