, , ,

Between A/D and D/A: the power of pragmatism

ADC and DAC. The first converts an analogue audio signal to digital, the second reverses the process: digital audio is converted to analogue.

We hear this process more often than we might realise. The analogue inputs found on the DEQX PreMate+, Devialet Experts and the KEF LS50 Wireless each digitise the incoming signal upon entry, necessary for when applying DSP-powered loudspeaker and/or room correction. In other words, analogue goes in and analogue comes out but in between the audio signal does time in the digital domain,

Many audiophiles profess an allergy to such conversion. Causing blocked noses and sneezing is a fear of sonic homogenisation: for some, their idealism tells them that analogue is the one true signal path and that digital conversion robs their listening experience of fidelity.

How to reconcile an ‘analogue or die’ approach to the playback chain when many modern vinyl records are pressed from hi-res digital masters? And if this A/D and D/A gets a pass, why not further down the chain?

Taking a recording from A to D in the studio and then D to A before being cut to wax might not be the qualitative corroder many claim it to be. After all, dynamic range differences between vinyl records pressed from these hi-res digital masters can still be heard when played on a reasonably resolving hifi system.

Closer to home, in last month’s Global Feedback column, I invited readers to vote for their preference on two needledrops spilling from two very different cartridges fitted to a Pioneer PLX-500 turntable. The vinyl of choice, The The’s Infected from 1986, was (presumably) pressed from an analogue master.

Phono pre-amplification and A to D conversion in this scenario were executed by the Pioneer’s own internal circuitry. Subsequent D/A conversion would then come from any would-be downloader’s DAC. Any source jitter and/or DAC output stage colouration would be hidden in plain sight.

However, my thinking at the time was that if an ADC were to homogenise the source material, the Pioneer’s cheap-as-chips implementation would more likely do so than a high-end variant like Ayre Acoustic’s QA-9, PS Audio’s NPC, Devialet’s Expert 200 or the Mytek Brooklyn ADC (picture above).

Ahead of posting the reader poll, I expected those with warmer, more forgiving sounding DACs to prefer the direct and in-your-face presentation of the Ortofon Nightclub MKII and those with leaner/meaner/cleaner-sounding DACs to prefer the softer treble (roll-off?) of the Pioneer’s (as yet unindentified) stock cartridge.

The results? Needledrop A (the Pioneer’s factory-fitted cartridge) was preferred by twice as many voters as needledrop B (the Ortofon Nightclub MKII).

By now it’s probably apparent that my intent was not to ascertain which cartridge offered subjectively better performance but that an audible difference between the two cartridges (and their alignment, tracking force, VTA etc) could easily be heard, even after the application of A to D and D to A conversion processes.

The digital domain is not a tunnel from which all entering vehicles emerge as Smart cars. Perhaps a light dusting of solidified exhaust emissions fixes itself to the paintwork but the car’s structure – engine, transmission, body – remains unaltered.

This line of thinking is consistent with Monty Montgomery’s lab-based ADC to DAC conversion as seen here:

Where this story takes a more interesting turn is when a third party DAC’s analogue output is used to feed an existing ADC/DAC signal path. Armchair prognosticators might claim there being little point to hooking up a Chord Mojo’s 3.5mm analogue output via Zu Mission ‘breakout’ cable to the twin RCA analogue inputs of the KEF LS50 Wireless, especially as the latter moves the audio signal into the digital domain in order to apply DSP and hand-off the left channel’s signal to the left-hand speaker via Ethernet cable, but…

In real-world listening tests, the Mojo’s clean-and-lean personality is maintained by the KEF LS50 Wireless’ signal path. Different (to the KEF actives being fed digitally via USB)? Yes. Better? The jury is still out. What we’re most likely hearing is the Mojo’s analogue output stage colouration, as maintained by the KEF’s own A/D and D/A internal process. We don’t hear the KEF DAC’s own colouration because it hides in plain sight.

Just as a phono pick-up’s audible personality can be maintained even with time spent in the digital domain, so too can an upstream DAC’s. Now that’s interesting.

Here’s the kicker: without KEF’s application of A/D conversion to the LS50 Wireless’ analogue entry point, there would also be no subwoofer integration (including x-over point specification and volume levelling), no signal tuning according to placement and no ‘time correction’.

No A/D greeting analogue inputs on the Devialets would mean no SAM (arguably a strong selling point) and on DEQX’s digital pre-amplifiers, no low-latency group delay correction.

But let’s assume my experiments are in some way faulty and Monty Montgomery’s assertion of full signal reconstruction is wrong. Let’s assume a little SQ is lost to ones and zeroes. Would such a tiny qualitative digital step backwards not be dwarfed by the gains to our loudspeakers’ (in-room) performance provided by digital signal processing (DSP)? This is where analogue idealism falters and digital pragmatism ultimately wins out.

Written by John H. Darko

John H. Darko

John is the editor of DAR, from whose ad revenues he derives an income. He is also an occasional contributor to 6moons and AudioStream and currently resides in Berlin, Germany.

Twitter: DarkoAudio
Instagram: DarkoAudio
Facebook: DAR


    • If they are going to offer MQA, all it eould take is a firmware update. They just need enough confirmation from their consumers that it is a desirable feature to add, in order to make it worth their while to do so.

      • I spoke to Kef technical support for about 20 minutes after receiving my speakers and they told me Roon and MQA are coming to the LS 50 wireless.

        • That’s fantastic news! It’s nice to know that a company is actually listening to what people are wanting from a product, and are able to upgrade it with firmware when the development is done. No doubt, partly added to their development plans at the recommendation of people like John Darko. I guess I’d better start saving for a pair! Thanks for the update.

  1. Hi John, like the old saying goes with HiFi – listen and buy what you like the sound of. In your case the message being technology puritanical views vs the reality of what things sound like.

    But I guess the sticky point here (and this will be a lively thread I am sure).. Those that have spent $1k + on the uber DAC with stuggle to stomach the thought of the el cheapy chip in the KEF (unless the chip were as good as the one in the upstream DAC).. And as few have active speakers, then most HiFi users will not have any perceived or speculative knowledge of the benefits of near speaker amplification matched to drivers and DSP/digital cross overs.

    Great read and insights as always sir

  2. Agreed. 100%

    My first true realization of the benefits of DSP was from a CEDIA course on ‘How to Listen’. I already knew from experience how to listen, but what was effectively demonstrated was the effects of proper channel level adjustment, time alignment of speakers, and even further the effect of signal delay to the speakers after time alignment was completed. Each adjustment made a clearly demonstrable difference (read improvement) to what you could hear from a movie soundtrack.

    This has also been my own experience of tweaking my own home systems for music and theatre, plus several car systems with time alignment and EQ adjustments. The most important thing I’ve found from playing with various DSP’s is that time alignment is key so the sound reaches your ears at essentially the same time. After that, room correction for reflections and absorbtion is helpful. Only after these two things have been effectively dialed in should you do any other sorts of EQ adjustments.

    Another great article and experiment. Thanks John!

  3. 1st-order vs. 2nd-order and lower effects. I think you’re spot on that speaker/room correction is a far more severe issue than whatever minor losses may be incurred by A/D conversion if the corrective DPS is properly done. It’s like tweaking with pucks and cones without considering where in the room the speakers and chair are positioning to minimize room modes.

    It all does *something* but some effects are of a very different magnitude than others. If one can make five large steps forward but to do so must accept a small step back, who cares? That’s still 4 and a half steps forward -:)

    As you said, it’s here where ideals/religion clash with pragmatism and the real world. It’s good to see it written about again and again!

  4. Perhaps a better way to judge ADC-DACs chain is by daisy chaining about 20 setups such as the Mtyek Brooklyn ADC and DACs together to hear what the cumulative effects of many passes does to the sound. Which one sounds closest to a straight wire in the end?

    IMO, In a hobby such as vinyl records, using digital files for cutting lacquers makes no sense. Might as well get a DAC and save yourself the additional coloration that vinyl adds. How can they make the argument that vinyl sounds better when there’s a digital process in the middle? Thus those records have less value.

  5. I’m with you 99% on this John. I just need to hear them for myself for that last 1%.

    I think some of us get so deep into the weeds we lose sight of the forest.

  6. Very nice article.
    In my opinion the same “logic incoherence” happens with D/A and passive speakers too, in the sense that DAC manufacturers are so worried about ringing and filters, and 99% of passive speakers (those not “time coherent”) mess with timing in their passive crossovers.

  7. I agree that the benefits of DSP are massive and worth the cost, if any at all, of an extra ADDA conversion.

    However, you should also note that that this is fundamentally incompatible with full MQA decode. MQA hardware buyers will be locked-in to passive speaker architectures without DSP crossovers, and must also forego any DSP EQ for fixing room issues.

    • David, if the DSP where taking place after the MQA decoding then wouldn’t that give you ability to run active speakers?

        • I don’t know if I’d concur with that. The file may be unfolded but the de-blurring hasn’t yet taken place. Wouldn’t any operation to the data corrupt this process?

          • Nope. The deblur is the execute in the digital domain, essentially with a filter. No reason why DSP cannot follow BEFORE conversion to analogue.

  8. Armchair Opinions (those made without actually experimenting) will always be made. Try to move on from those insecure remarks others make and experiment when you can. Here’s the daunting reality for me. In order to make an informed opinion I would need at least three turntables with “pure” signals and three dacs all similarly priced. Five sources each, would, of course, be better. Example: pro-ject debut carbon dc and ifi micro dsd. rega rp6 and gungnir. 2016 vpi scout and vega auralic or yggdrasil. A difficult experiment to be sure, but people don’t even do the experiment with the debut carbon and micro dsd yet will get so indignant about beliefs other than their own, or the all-telling “math” that proves everything. Try to keep an open mind, and if you haven’t had the opportunity to experiment, maybe listen and read more and shout less.

  9. Really… dynamic range as a primary benchmark for vinyl? Audiophiles learned that was a crock as primary audio quality measurement with the first Telarc digital LP recordings in the early 80’s. Lots of dynamic range and musically inedible. I keep mine around as a reminder of hype and hubris.

    • Except its *not* the primary benchmark for vinyl (here) but one dominant benchmark for studio mastering.

  10. I’ve been a recording engineer for thirty years and my take is that different converters sound different It’s whatever flavor you like.
    Also of the sound you hear many times has to do with the quality of the analog circuitry in the output stages given bit rates that are the same. Just as Studer analog decks sounded different than Ampexes or MCI’s.

  11. Agree with you, John.
    People who have listened to quality A/D >D/A conversion generally can’t tell the difference between the original and the reconstruction when played back on the same equipment.
    The idea that all systems like the DEQX or LS50 active are inferior to “traditional” playback is just another audiophile superstition.

  12. I don’t know. When done well I believe adding digital room correction will improve the sound in a room. But every dac has it’s own sound. This means no dac is entirely neutral. A very good dac will decode the signal almost perfectly. A less great dac will always degrade the sound a little more or add something more. Daisy chaining a very good dac like Aqua La Scala in front won’t upgrade the sound of a less great dac. You’ll end up with the bit of colouring of both. As long as we hear differences between dacs daisy chaining dacs isn’t the best idea. At least that’s what I think. I would connect digitally to the KEF’s.

  13. I agree with the argument
    Friends of analogue just ignore time and frequency Issues in speakers and rooms – mostly because there not a darn thing you can do about it in the analogue domain so why pay it any attention.

    For me personally an even more interesting question is do we know enough to get the best out of these DSP systems.

  14. I am running a PDX Level 2 DAC (you know the one John!) into both a classic SS amp – a Rega Elex-R and also into the analogue inputs of a Devialet Expert 120.
    The Dev of course does the whole ADC then DAC then out to the ADH (another acronym) amplifier section. So the million dollar question is, does the essential qualities of the PDX DAC remain unharmed after the Dev has finished with it? In a word Yes.
    All that beautiful detailed, R2R slightly tube warmed goodness is all there and is actually rendered clearer and more vibrantly through the Devialet due to its vanishingly low distortion.
    I was running a Behringer DEQ1924

  15. (cont) I was running a Behringer DEQ2496 which is a much less “hifi” oriented unit that also did an ADC to DAC internally. I could not hear when it was switched in line. If put on a set of headphones with no music and put my amplifier (Rotel) up full bore, I could detect it but any changes would have in reality been buried so far below the noise floor that in practical use this would have been inaudible. The DEQ2496 does all sorts of unmentional evil including parametric equilisation but was a benign factor in my audio chain.

  16. I find the resurgence of vinyl amusing. Practically all of today’s music is produced with a digital chain in the studio, so tons of ADC and probably also DACs already in use before music even hits the consumer. Back in the day when I was in school I took an acoustic electrical engineering class, in one of the labs we were to measure a turntable. I remember being surprised just how terrible in particular the channel separation (below 25dB) was. Digital has come a long way, why anybody would not try to retain digital all the way to the very last stages of your setup is beyond me.

    • Exactly! Converting to digital any place in the vinyl cutting chain makes no sense. Especially for diehard analog purists that vinyl people are. I’m surprised that anyone would buy that? Yet Analog Planet reviews it like it’s better than just playing the digital file with a DAC?

KIH #41 – Art vs. Science

AudioQuest NightOwl Carbon: additive-free headphone listening