WAF. Watts And Franks. Rob and John. The Odd Couple of high-end audio – friends, whose professional relationship stretches to twenty years.
Through the haze of arrival jet lag that weighed heavy during the Munich press launch for Dave – Chord Electronics’ forthcoming statement D/A converter – I recall Franks grinning his way through a short anecdote about how Watts once unknowingly bought his (Franks’) parents’ house in Kent.
Several months earlier at CES, I’d learnt from Watts that he (now) lives in Wales where he works remotely from Chord Electronic’s Maidstone headquarters, a converted pump house that backs on to the River Medway.
Before designing digital audio products for Chord Electronics, Watts played consultant to silicon chip manufacturers and invented Class Z digital power amplifier technology.
WTA. If reading up on Watts’ Transient-aligned Algorithm bends your brain, you’re not alone. DAC design is complicated business, especially when it comes to FPGA code. I’m capable of explaining only the essence of Watts’ approach.
I sat down with the man himself – flanked by Franks of course – during their visit to Sydney in March. In doing his level best to accommodate my rudimentary understanding of matters technical, Watts stressed the importance of time domain accuracy. The rationale here is that our brains are super-sensitive to any inaccuracies in when sounds start and end, errors with which is often what leads some D/A converters to sound dry and ‘tinny’.
Redbook/CD-quality audio is sampled at 44.1kHz. That’s 44100 times per second. Dividing a second into 44100 chunks gives us 0.00002267573sec/sample. In other words, Redbook samples are taken at intervals of 22 microseconds (2 x 10^-6 seconds). According to Watts, this falls short of the human brain’s ‘sampling’ speed (or inter-aural delay) of 4 microseconds and therefore Redbook data doesn’t contain sufficient timing accuracy. What if the transient strikes between two Redbook sample moments?
Watts’ filter inserts data points between the original samples in order to keep the brain up to speed with “important perceptual cues” for which proper timing is fundamental. Central to Watts’ thinking is that the brain contributes as much (if not more) to auditory perception than our ears alone and by reconstructing the original waveform a more natural (and less edgy/harsh) sound presents.
And if Watts’ interpolation filter sounds a lot like simple up-sampling, it is.
BYO. However, as with anything audio, it ain’t what you do but the way that you do it.
Implementation falls not to an off the shelf chip hosting its manufacturers code but to an FPGA with Watts’ own. The answer to ‘why FPGA?’ is found in the number of taps required for the FIR filter’s proper reconstruction of the missing bits. Think of taps as memory slots. According to the Whittaker-Shannon sampling theory, an infinite supply of taps will allow for reconstruction perfection of a bandwidth limited signal.
Back on earth, DAC designers work with a far lower tap count. Off the shelf silicon brings anything between 100 and 256 into play. A number that proved woefully short for Watts’ needs. He says his core code was ready to run a good while before the longer tap length possibilities of FPGAs dropped to pricing levels low enough meet the commercial reality of DAC design.
“It is due to this short tap length and the filter algorithm employed that generates the transient timing errors. These errors turned out to be very audible. Going from 256 taps to 2048 taps gave a massive improvement in sound quality – much smoother, more focused sound quality, with an incredibly deep and precise sound stage,” reads the Watts-penned, website-hosted theory.
As soon as John Franks found FPGAs affordable enough for his company’s converters, they were loaded with Watts’ WTA code, encrypted to prevent reverse engineering, and dropped into Chord Electronics’ D/A converters. The first WTA model, the DAC64, ran with 1000 taps. The QuteHD followed with 10,000 taps. Then came Dave’s forerunner the QBD76 before 2014 gave us the QuteEx and Hugo with 26,000 taps each.
Tapping out. Watts’ special sauce also extends to a noise shaper designed to tackle noise-floor modulation, where the noise floor moves to the music and messes with the brain’s ability to properly separate sounds. Here our technical discussion pulls away from my own tech know-how. WTA moves to WTF. Checking out…
According to Watts, “Noise floor modulation makes the resulting sound hard, bright and aggressive; it degrades instrument separation and focus. By reducing noise floor modulation we see an improved sense of focus, smoothness and refinement – it sounds much more natural”.
The one word I keep coming back to when describing the original Hugo (reviewed here) is filigreed: “Ornamental work of fine (typically gold or silver) wire formed into delicate tracery,” says Google. The Chord portable has an uncanny ability to follow every last detail to its natural conclusion. I liken this trait to the motorcycle mounted cameramen who provide visual coverage of the Tour de France; they’re able to weave amongst the peloton before flooring it to follow breakaway riders. Few events are missed down on the ground whilst the helicopter coverage (of lesser DACs) doesn’t nail the aspect of ‘being there’ quite as well. Viewed from overhead one must also contend with the occasional visual and signal obfuscation of trees and hillsides.
If all this sounds a little too abstract for readers who’d prefer for more straight forward descriptors, think of it this way: the Hugo’s laser-guided precision never once pricks the mind with needles and pins; abundant clarity is given to spaces between players. Congealed this DAC is not. Not even slightly. The trade off is a minor hit to acoustic mass, most notable in a leaner low end. Partnering the Hugo with a reedier-sounding amplifier like the Clones Audio 25i might not be the best foot forward.
Not only did the Hugo defy expectations of its handheld form, it stands tall amongst full-sized rivals that don’t arrive with the Hugo’s battery power for off-grid playback, robust headphone staging with switchable cross-feed circuit for closer audible proximity to in-room loudspeaker listening, aptX Bluetooth for playback hand-off to non-audiophile pals and a smartdevice-compatible USB port for proper up-n-go portability.
Bonus features notwithstanding (and nine months on from my original review) I’d be hard-pressed to recommend a better-sounding DAC at or below its US$2.5K marker. That’s why the Chord Hugo very nearly summited the Darko DAC Index to sit amongst far costlier company. Recent SQ-elevating firmware revisions to PS Audio’s DirectStream have put more daylight between it and the Chord Electronics portable. The Resonessence Labs INVICTA Mirus sounds more muscular and sinuous than the more intellectually agreeable Hugo, which appeals to the mind more than the body.
I’m not alone in my deep appreciation of the latter’s super-sharp ROI. Accolades spilling forth for the Hugo have been numerous from forum folk and reviewers alike – a conSENSus that’s not as common as we might hope for. Even the traditionally hard-to-please Naim-focussed flat earthers over at PFM went bonkers for Watts and Franks’ handheld unit.
Criticisms are few: narrow socket clearance, even after a chassis revision, meant one had to exercise a little more care than usual when choosing partnering cables; learning the interface’s colour-coded volume control, source selection and sample-rate detection took time; the coaxial input offered a more robust, fleshed out feel than going USB direct making a USB converter almost essential for maximal Hugo-juicing.
Incidentally, Rob Watts told me that he listens to his own Hugo connected via toslink.
Talking of which, to head-fi-ers who drop the best part of $2K on an Astell&Kern DAP only to outsource D/A conversion and headphone drive to the Hugo I cannot relate. If all that’s required is a transport, why not go simply lasso an iPod Touch to the Hugo over USB for far fewer dollars down?
Whatever your digital transport choice, there’s also the issue of in-pocket bulk that comes from ‘bricking it’: rubber-strapping one’s DAP/iDevice to the Hugo. Here we defer to the cliche that says live and let live. Time spent on Tokyo trains readily reveals the lengths to which many a Japanese listener will go in order to maximise sound quality on the (often long) morning and evening commute. In this context, a DAP strapped to the back of a Hugo is child’s play. This is after all how Hugo got its name: where I go, Hugo (“You go”). Ya feel?
Just because I find DAP/DAC/amp combos too unwieldy out in the street doesn’t mean that you will. Moreover, you don’t need a review to tell you that the Hugo won’t fit in a front jeans pocket or that with cable connectors working both ends of the Hugo the head- or feet- first conundrum is unsolvable.
At the other end of the usage spectrum, would be buyers wanting to deploy the Hugo in DAC-only mode in their home hifi rack appealed to Chord Electronics for a model with balanced outputs and/or full-size (Type B) USB sockets to keep existing USB cables in play.
A new model freed from the design brief restriction of portability could see a larger chassis connote greater audiophile seriousness and more internal real-estate for Chord’s Kent-based engineers to apply changes that would (fingers crossed) result in an even better sounding converter.
Between two CES events that theory became reality. The Chord Hugo TT was formally launched at CES in January of this year. When I caught up with Rob Watts and John Franks in Vegas it was abundantly clear they were looking to capitalise on momentum kickstarted by the original model.
You probably know by now that TT stands for Table Top. At 2950GBP/US$5000 the Hugo TT’s entry fee is double that of its predecessor. At 3kg, product mass increases nine fold – a red card to pocketability and portability, as was intended.
Off-grid listening remains in tact with the battery seeing trickle charge from a switch-mode wall wart – unplug it and the Hugo TT continues as if nothing has happened. It’s a chicken with its head cut off!
Battery life might be irrelevant on the TT but that didn’t stop Franks and co. from doubling its size and adding 10,000,000uF worth of low output impedance super-capacitors, both implemented to assist with the unit’s transient responsiveness and overall dynamics.
TT’s FPGA chip remains a Xilinx Spartan 6 accommodating a 26K tap (WTA) filter length. Per January’s press release, the TT’s FPGA “has the same specification and measured performance as its mobile sibling”. However, the analogue output stage did see some (unspecified) renovations.
Headphone output specifications remain the same: 35mW into 600 ohms, 70mW into 300 ohms, 320mW into 56 ohms, 600mW into 32 ohms and 720mW into 8 ohms. On paper, the those numbers don’t look like much. Tell that to Beyerdynamic’s 600Ohm T1 – whose performance at the hands of the TT is nothing short of stunning, especially with electronic music. As with Vinnie Rossi’s RWA and LIO, the low output impedance associated with battery- and ultracap-power translates to high current delivery and – here – a downright righteous listening experience: clean and incisive that takes us down-to-the-bone with detail delivery without the X-ray chill so often (emotionally) associated with neutrality.
Table top doesn’t mean the user remains deskbound. The supplied aluminium-shelled remote allows for basic control from across the room with the TT’s rudimentary front LED showing input selection and volume level. Strike up two more plus points for the newcomer.
Out back, twin type B USB sockets: ‘SD’ for smartdevices, ‘HD’ for Mac or PC. Both asynchronous but the latter being galvanically isolated up to 384kHz might be the reason why I noted no obvious performance disparity between it and the coaxial input as I did with the original model. A USB-S/PDIF converter remained surplus to requirements, even with the TT’s coaxial input being of the more audiophile-approved BNC variety.
With the TT deployed in DAC mode in a two-channel rig that comprised KEF LS50 standmounts powered by the Vinnie Rossi LIO, the spirit of transient detail extrapolation continued. More obvious here was shape and thrust in the bottom end. Not an over-emphasis but superior space clearance around bass notes allowing them to step more easily from the darkness and recede again. Think of the way a golf-course greenskeeper plies his trade: he doesn’t enlarge the green; it’s his edge-manicuring skills that cause easier ocular separation from the fairway.
In this sense, the TT demands of the listener some acclimating patience. Jumping in at the deep end with conclusions from the get-go could lead to talk of a U-shaped presentation – this is absolutely not the case. Rather, it’s Rob Watts’ code unearthing more detail in a more athletic fashion than one might have heard previously. Like the Hugo before it, the TT sounds nothing like its rivals, which also means that not everyone will fall for its charms hook, line and sinker.
Then there’s soundstage depth. It’s one area in which the Hugo TT convincingly bests the Resonessence Labs INVICTA (US$4995). Listeners with leaner sounding systems will prefer the Canadian. Got plenty of acoustic mass going on already? Hugo TT is your go to guy. That is unless you need the INVICTA’s SDcard slot and HDMI-routed OSD. Either will be suffice for the head-fi-er committed to simultaneously connected A/B headphone comparisons. Compared to the Hugo, the TT doubles the number of 6.4mm sockets to two and halves the 3.5mm holes to one.
For more seasoned readers, the $2.5K question is how well does the little fella compare sonically with the big guy? The short answer is ‘pretty darn well’. The longer answer involved several runs through some Air, Jenny Hval, The Bad Plus and Tom Waits as well as more extended A/B sessions with all manner of material. Once I felt I had a pretty good grasp on the differences I drafted in a mate to sit in on a final back to back to confirm findings.
Both Hugo and Hugo TT share the same house sound of top-end detail delivered with tenderness and well above average ambient extension. On front-to-back player positioning I was unable to separate the two. Similarly, the TT regularly called up F-words: filigree and finesse.
Both units are superbly talented in exposing textural information but the TT has the definite edge with tonal colour saturation. It also gives a firmer hand to drama and poise, kick-punching more adroitly, especially down below. An additional spoonful of acoustic mass handed to Tom Waits’ throat by the TT means we hear more grit and gravel. (Thanks for the loan!).
Looking back at the original Hugo from the TT’s elevation, junior sounds a little paler, more diluted and less rhythmically excitable. You pays more, you gets a little more. The TT is the better sounding DAC but the original Hugo at half the price remains the far keener value proposition.
Running the TT up the hill against the also FPGA-d PS Audio DirectStream (reviewed here) led to less clean cut results. Next to the DirectStream’s more seductive manner, the Chord unit sounds tighter, crisper and a few degrees cooler – altogether more refreshing! The Hugo TT offers the punchier bass but the differences here probably won’t be sufficient for DirectStream owners to switch their allegiances overnight, especially as its firmware upgradeability makes it a moving target. Besides, if you were drawn to the DirectStream because of its warmer hues you’ll probably find the Hugo TT a little too overt with clinical incision.
That said, with lesser quality source material like the original master of Kate Bush’s Hounds of Love the TT drops in some much needed effervescence and cleaves instruments apart for a more immersive experience. Running a beta of the Yale OS on the PS Audio decoder lends it a smidgen more generosity when fleshing out erstwhile reedier sounds. Central to Chord’s counter attack is its application of ‘vivace’ when driving the rhythms of Aphex Twin’s “Diskhat ALL Prepared1mixed [snr2mix]” and the strings players on Kate Bush’s “Cloudbusting”, about which my auditioning sidekick said “You could see their shoulders move”. His preference fell to Ted Smith’s design but mine to Watts and Franks’…just.
In oversimplified terms the PS audio is warmer, more relaxed. With richer amps it might prove too much of a good thing. The DirectStream’s talents are more immediate than the TT’s, which require more time for a reviewer to peg with similar accuracy. Actually, you don’t uncover them, they find you – but only if you’re committed to playing deep into the fourth quarter.
On sonics alone we’ll call the DS/TT duel a draw; a matter of taste if you will. However, the more compact Brit arrives a full US$1000 cheaper than the American but with robust headphone amplification and battery power already in tow. The Hugo TT wants not your fancy power cord.
Lastly, a challenge to throne rights. Would the Hugo TT best the Aqua La Scala MKII (€4890). Kristen Hersh’s string version of “A Loon” and Moritz Von Oswald’s “Sounding Line 6” were added to the playlist for this do-or-die. The TT went first: laundered and speckless with impeccably-motivated rhythm.
With the La Scala stepping in, the difference was immediate and unmistakable, the Aqua acing the challenger on drama, separation and (especially) scale. If the Hugo TT lays it all out visually in front of the listener in 16:9, the Aqua pumps the vertical and to give us 16:12. It makes the KEFs sound MUCH bigger whilst simultaneously keeping a keen eye on teasing out subtleties that sound smaller in Hugo hands. The latter’s bass drum skin delineation also sounded more obviously crisp-fried.
The TT vs. La Scala delta was more pronounced than the comparisons that preceded it. Next to the La Scala, the tabletop Hugo emotionally cowers. No King is deposed but his courts-men have been re-shuffled.
DSD listeners take note: you’ll not find yourself adequately served by the Italian. It’s PCM1704 chips handle only PCM, which looks positively quaint next to the TT’s handling of up to 24bit/192kHz PCM over toslink and 32bit/384kHz PCM over its BNC coaxial and USB HD. Note: DSD is ‘decimated’ to PCM prior to FPGA processing.
What we have in the TT is an all-rounder. A magnificent DAC at the US$5K marker that effectively supplies expectation-defying headphone staging free of charge. Not only does its supercap/battery power complement Vinnie Rossi’s world view by taking all circuitry off the mains for zippier current responsiveness, but its audibly tender incisiveness plays proper counterpoint to the LIO’s tubed-up pre-amplifier stage.
Chord’s Hugo TT is a decoder for detail freaks who demand exposure to every last nuance without the compromise of emotional distance. It’s one of the few DACs that I’ve heard at any price that seemingly takes musical insight to a cellular level whilst keeping sight of the bigger picture required to properly move the listener.
Further information: Chord Electronics
Thank you to Radiance AV for supplying the review unit.