Which input sounds best on a Chord DAC (optical, coax, USB or BNC), and why does Rob prefer optical when it is said to have more jitter?
Short answer
From January 2018 Watts has repeatedly named optical (Toslink) the reference, best-sounding input on his DACs Head-Fi, Jan 2018. Two 2018 replies that say "USB is best" do not mention optical: one names only AES, the other ranks USB against coax and AES Head-Fi, Apr 2018Head-Fi, Sep 2018. In 2022 he wrote "There is no possibility of USB performing better than optical (excepting opticals 192k limit)." Head-Fi, Mar 2022Head-Fi, Mar 2022 He grounds the preference in isolation: he says his DACs remove source jitter, so optical's possibly higher jitter does not matter, while optical's light-only link stops RF noise currents from the source entering the DAC ground plane, which he says causes noise floor modulation and a brighter sound Head-Fi, Jun 2014Head-Fi, Jul 2018MFG Audio Reviews 23:47. USB can equal optical with quiet or battery-powered sources (Qutest, 2018), and coax is the brightest, most noise-sensitive input (Qutest 2018, Mojo 2 2024) Head-Fi, Jul 2018Head-Fi, Aug 2024. From late 2015 to 2016 he rated USB slightly better than optical on DAVE and Hugo TT, which he later called a mistake Head-Fi, Feb 2016Head-Fi, Jul 2016Head-Fi, Jan 2018Head-Fi, Mar 2022. After the reversal the record is less tidy: in February 2019 he still wrote that USB is technically better from the clock point of view while optical sounded identical to him Head-Fi, Feb 2019, whereas in 2022 he called the clocking benefit an illusion and said there is no technical benefit to USB Head-Fi, Mar 2022Head-Fi, Oct 2022.
Current position (2018 to 2024)
Watts treats optical as the benchmark every other input is measured against. In January 2018, on DAVE: "Yes optical is the best input - its really a question of getting the other inputs to match optical." Head-Fi, Jan 2018. In May 2018: "Toslink is the gold standard, and the target to hit with USB" Head-Fi, May 2018. In 2022 he wrote "Optical should always be considered the reference (best) input." and added that USB may equal it or sound brighter with worse depth, depending on source and system Head-Fi, Mar 2022. The same month: "There is no possibility of USB performing better than optical (excepting opticals 192k limit)." Head-Fi, Mar 2022. At CanJam NYC in March 2024 he called optical "the best sounding digital connection" despite its reputation MFG Audio Reviews 22:54, and in August 2024 (Mojo 2 thread) he wrote that "Optical is the reference, as it can't inject RF noise into the DAC ground plane." Head-Fi, Aug 2024. His preference for optical goes back to 2014 on the original Hugo: "My preference has always been optical - it's much smoother, and more natural." Head-Fi, Apr 2014. He still personally ran optical from his Windows PC in November 2023, reporting "better SQ and no SR clicks" Head-Fi, Nov 2023.
Two 2018 replies rank USB without mentioning optical: in April 2018 a DAVE reply opens "USB is best; AES EBU is a joke standard" Head-Fi, Apr 2018, and in about September 2018 (date estimated from the post ID) a TT2 reply says "USB is best against coax or AES" Head-Fi, Sep 2018. The April reply names only AES as the alternative, and the September reply ranks USB against coax and AES. In May 2018 on TT2 he described USB as "very close to optical", sometimes with the impression of better resolution on USB and better depth on optical, "If there is a difference, it's inconsequential" Head-Fi, May 2018.
Why optical, when it is said to have more jitter
Jitter, in his account, is removed by the DAC. In April 2014: "Generally, optical is not liked as it has more jitter than coax,but all my DAC's have been able to remove incoming source jitter." Head-Fi, Apr 2014. In June 2014, on Hugo: "In the case of Hugo, the jitter levels does not matter, as everything gets re-timed against a low jitter clock, so source jitter is completely removed." He added that his AP analyzer "actually measures lower jitter levels on optical than the coaxial inputs" Head-Fi, Jun 2014. In 2015 he reported J-test results on Hugo with no measurable jitter artifacts (below 500 fS) on USB, optical or coax Head-Fi, Jul 2015. In 2019 he said "the jitter from optical is tiny - about 1 nS", and that "my previous AP had lower jitter measurements from optical than from BNC (my current APX555 has slightly better coax now)", and that his DPLL makes it irrelevant Head-Fi, May 2019. In 2018, on Hugo 2: "I have applied 2uS of jitter (that's 1000 times bigger than one normally gets from a source) on the inputs, and there is zero measured effects" Head-Fi, Oct 2018. In 2022, in the DAVE thread, he described the mechanism for SPDIF: the incoming word clock only frequency-locks a new word clock generated from the local 104.25 MHz oscillator, with data read out of a micro buffer, so incoming phase and jitter have no influence Head-Fi, Mar 2022. In an interview published in 2022 (recording date not stated) he said optical "has slightly higher levels of jitter than BNC" and that this is not a problem for Chord DACs AVstore 42:56. At CanJam 2024 he conceded "generally speaking, optical is slightly worse jitter" and then said "Jitter is not an issue with my DACs at all" MFG Audio Reviews 23:47.
What optical removes is ground-borne RF noise. He says that with bit-perfect data the source can change the sound only through RF noise: "The only way SQ can be impaired is via RF noise connected via the ground, and optical does not suffer from this problem." Head-Fi, Jul 2018. RF noise currents in the DAC ground plane, he explains, intermodulate in the analog electronics and produce noise floor modulation, which sounds brighter and harder and degrades instrument separation, timbre and depth Head-Fi, Jun 2014Head-Fi, Oct 2019Head-Fi, May 2021Head-Fi, Mar 2022. Optical is "the perfect galvanic isolation, as the coupling capacitance is almost zero" Head-Fi, Jul 2018. In 2020 he added that an optical receiver is a diode and an amplifier, while USB decoding has "twenty times the power dissipation" plus its own processor clock, so "here optical wins big time" on RF noise injected into the DAC Head-Fi, Aug 2020; in 2024 he put USB at "two orders of magnitudes more random RF noise" Head-Fi, Apr 2024.
Why people dislike optical, in his view. Because it sounds softer: "the hardness can easily be confused with more detail resolution and impact" (2014) Head-Fi, Jun 2014. In 2024: "I think optical has a bad reputation because it sounds rich, warm and naturally rounded due to the reduction in noise floor modulation." Head-Fi, Apr 2024.
By input
USB
- Original Hugo (2014 to 2017). Optical was clearly better; in 2017 he wrote "With Hugo 1 galvanic isolation made huge differences to the SQ" Head-Fi, Jun 2017. One 2014 exception: if the only alternative was letting the PC resample, use USB, because "the difference in SQ in using optical is small compared to the PC re-sampling or sample rate converting" Head-Fi, Sep 2014.
- DAVE and Hugo TT (late 2015 to 2016): USB rated slightly ahead. November 2015: "With Dave the best input (by a tiny margin) is USB, then optical is very close." Head-Fi, Nov 2015. February 2016: "For both tt and Dave, usb is best, closely followed by optical, then last is coax." Head-Fi, Feb 2016. March 2016: "The reason USB is slightly better is because the timing comes from the FPGA" Head-Fi, Mar 2016. In July 2016 he said optical will "eliminate all RF" but is not as good as USB, adding "Before galvanic isolation, optical was so much better than USB." Head-Fi, Jul 2016; in March 2016 he hoped "to close the small gap from optical to the slightly better sounding USB input" Head-Fi, Mar 2016. He still called the gap small, "in my bottom scale of SQ differences" Head-Fi, Aug 2016. The record is not uniform: one day before the February 2016 ranking he told a DAVE owner "Optical will sound the best." Head-Fi, Feb 2016, and in January 2016 he said optical "sounds almost identical to the USB" on DAVE Head-Fi, Jan 2016.
- The reversal (January 2018 onward). After Hugo 2 tests with an MSI laptop that had both outputs, he wrote "I used to think that USB had the edge on optical - it's slightly brighter - but I was wrong about that" and attributed the brightness to "worse RF noise pick-up", since the clock comes from DAVE in either case Head-Fi, Jan 2018Head-Fi, Jan 2018. The record after the reversal is not uniform. In February 2019 (Qutest thread) he wrote "Technically USB is better from the clock POV; but in practice I have had optical sound identical to USB" Head-Fi, Feb 2019. In March 2022 he repeated the correction ("You have discovered a post where I made a mistake.") and called the clocking benefit "an illusion" Head-Fi, Mar 2022. In October 2022, on TT2, he wrote "So there is no technical benefit to using USB over optical when using my DACs." Head-Fi, Oct 2022, which differs from his 2019 remark. In November 2024, looking back at his earlier 2Qute view, he noted "it's so easy to prefer an aberration, error or distortion" Head-Fi, Nov 2024. In January 2018 he described the earlier uncertainty: "I could not tell from listening tests which was the explanation." Head-Fi, Jan 2018.
- Hugo 2 (2017 onward). On headphones with his MSI laptop: "Comparing optical to USB, I struggled to hear any difference at all." Head-Fi, Feb 2017. He limited the finding to that setup; a grounded output, such as a power amp, brought back a small difference Head-Fi, Jul 2017Head-Fi, Dec 2017Head-Fi, May 2018. He credits RF filtering on the USB ground and VBUS for closing the gap Head-Fi, Jan 2018.
- Source dependence (2021 to 2024). In 2021 a powerful new desktop PC sounded better on optical than USB: "So for sure USB can be identical to optical, but with nosier sources, optical will win out." Head-Fi, Apr 2021. He repeated this for the M Scaler in 2022 and 2023 Head-Fi, Feb 2022Head-Fi, Jun 2023 and for Mojo 2 in 2024, where battery-powered sources should equal optical Head-Fi, Feb 2024Head-Fi, Feb 2024. At CanJam 2024 he said a low-powered, battery-operated source is "pretty much guaranteed" to make USB sound the same as optical MFG Audio Reviews 24:40.
- Why isolated USB still falls short. The desktop DACs have galvanic isolation on USB, but he says the isolator's roughly 2pF coupling capacitance lets some RF through, particularly in the GHz range Head-Fi, Mar 2022Head-Fi, Oct 2022Head-Fi, Oct 2021Head-Fi, Jul 2018. At CanJam NYC 2022 (late February; the video was published in March 2022) he said "Mojo and Hugo 2 does not have galvanic isolation" MFG Audio Reviews 56:58; in 2020 he wrote that Hugo 2 "is not galvanically isolated" Head-Fi, May 2020.
Coax, BNC and AES
- Coax. On Qutest in 2018: "Coax has no galvanic isolation, and direct coupling to the grounds, so will always be the brighter sounding input." Head-Fi, Jul 2018, and "But Coax I can't put RF filtering in without degrading lock abilities." Head-Fi, Jul 2018. In 2022: "If you have a choice, always go for optical over coax." Head-Fi, May 2022. In 2024: "So coax is the most sensitive input to common mode RF noise currents" Head-Fi, Aug 2024.
- BNC. In 2015 he said DAVE's BNC/AES result "depends upon the source and cabling" Head-Fi, Nov 2015. In an interview published in 2022 (recording date not stated), discussing sources feeding the M Scaler, he called a non-isolated BNC link "the worst solution because it's not galvanically isolated" AVstore 41:06. In 2023 he ranked a traditional non-isolated BNC below optical and USB: "optical is a much better connection. followed by USB" Soundnews 49:45. The M Scaler's BNC outputs are galvanically isolated, giving "almost optical standard" but not equal, because of the isolator's capacitance AVstore 42:06Soundnews 48:58. In 2024 he kept dual BNC on the Quartet: "But optical doesn't solve the problem because you've still got the mains connected to Dave and the mains connected to the scaler." He chose filtering inside the scaler instead ConvinceMeAudio 10:19. One outlier: in February 2018, in a reply that names only "the M scaler" as the product, he wrote "In my system, I use all inputs, and I have never felt that one is better than another, but YMWV." Head-Fi, Feb 2018.
- AES/EBU. 2016, on DAVE: "this is not the best input IMHO", because it injects source RF noise and sounds brighter Head-Fi, Apr 2016. 2018: "AES EBU is a joke standard, not intended for high data rates" Head-Fi, Apr 2018. On TT2 in 2018 he rated USB above coax or AES Head-Fi, Sep 2018.
- I2S. He sees no benefit: "I2S supplies no sonic or technical benefit" (2022) Head-Fi, May 2022. In 2023 he said his preference "On the source input is always optical" Soundnews 48:58.
Conditions he attaches to optical
- Disconnect every other wired connection to the source, or the isolation is shorted out: "You must remove all wired connections from the source when using optical." Head-Fi, Oct 2022Head-Fi, Nov 2018.
- A mains-powered USB-to-optical converter creates a new RF loop. In 2022: "If you try a USB to optical converter, only use one that does not require an external PSU." He named one candidate while hedging: "I haven't tried it, but Douk Audio U2 USB Converter looks like a good bet to try." Head-Fi, Oct 2022Head-Fi, Jan 2020.
- Optical from a mains-powered source is not automatically perfect. In 2017 he wrote "Just because a source uses optical does not guarantee perfect isolation" and said a battery source over DAVE's isolated USB can isolate better Head-Fi, Oct 2017. Intersecting mains loops can still let source noise in Head-Fi, Feb 2019Head-Fi, May 2021.
- Cable: in 2022 a 5m KabelDirect (as he spelled it) plastic fiber ran at 192kHz and "This cable sounds identical to my 2m Lifatec glass optical fibre" Head-Fi, Oct 2022. Optical tops out at 192k Head-Fi, Mar 2022.
How he tells listeners to choose
Use optical as the benchmark and pick whichever input or source comes closest to it: "the source that gets closer to the sound of optical is the best one" (2018) Head-Fi, May 2018. When in doubt, prefer the warmer result: "The rule of thumb with sources is the warmer sound is the better, more accurate performance" (2021) Head-Fi, May 2021. He attaches "YMMV" or "YMWV" to nearly every source statement Head-Fi, Aug 2018Head-Fi, Jun 2023.
What he has not said: this archive has no input ranking from Watts after November 2024 Head-Fi, Nov 2024, and nothing comparing inputs on the Quartet itself beyond its design choices ConvinceMeAudio 10:19. He describes the audible optical versus USB gap only in words such as "the difference between optical and USB is minute" Head-Fi, Oct 2017. His measurements show the inputs essentially equal: in 2015 on Hugo, identical dynamic range on all three inputs, with optical THD and noise slightly worse (-114.6 dB against -115.5 dB, left channel), which he called "possibly an AP ground issue" Head-Fi, Jul 2015; in 2016 on DAVE, "Using the jitter test you can measure v tiny degradation using optical, but its barely detectable." Head-Fi, Mar 2016; and in 2020 on DAVE, "USB has identical measured noise floor modulation as optical" Head-Fi, Oct 2020.
Quotes
“For both tt and Dave, usb is best, closely followed by optical, then last is coax.”2016-02-24 Head-Fi, Feb 2016
“…I used to think that USB had the edge on optical - it's slightly brighter - but I was wrong about that…”2018-01-12 Head-Fi, Jan 2018
“The only way SQ can be impaired is via RF noise connected via the ground, and optical does not suffer from this problem.”2018-07-29 Head-Fi, Jul 2018
“So for sure USB can be identical to optical, but with nosier sources, optical will win out.”2021-04-07 Head-Fi, Apr 2021
Dated positions
| When | Products | Position |
|---|---|---|
| 2014 to 2015 | Hugo (original) | Optical is his preferred and best input on Hugo, since the DAC re-times source jitter and optical isolates ground-borne RF noise. Head-Fi, Apr 2014Head-Fi, Jun 2014Head-Fi, Jul 2015 |
| 2014-09 | Hugo (original) | USB is better than optical when the alternative is letting the PC resample. Head-Fi, Sep 2014 |
| 2015-11 to 2016-08 | DAVE, Hugo TT (original) | USB is best by a small margin because its timing comes from the FPGA, with optical very close and coax/AES last; a 2016-02 DAVE reply says 'optical will sound the best'. Head-Fi, Nov 2015Head-Fi, Feb 2016Head-Fi, Mar 2016Head-Fi, Jul 2016Head-Fi, Feb 2016 |
| 2017-02 to 2020 | Hugo 2 | USB sounds the same as optical on headphones with his MSI laptop, but a small difference returns when the output is grounded to a power amp. Head-Fi, Feb 2017Head-Fi, Jul 2017Head-Fi, Jul 2018Head-Fi, May 2020 |
| 2018-01 to 2024-11 | DAVE, Hugo TT2, Hugo 2, 2Qute | He reversed the USB-is-better view: USB's extra brightness was RF noise, so optical is the reference and USB can at best equal it (a 2018-05 TT2 post still calls the gap inconsequential, with USB sometimes seeming better resolved; a 2019-02 Qutest post still calls USB technically better from the clock point of view while optical sounds identical to him; by 2022 he called the clocking benefit an illusion). Head-Fi, Jan 2018Head-Fi, Jan 2018Head-Fi, Jan 2018Head-Fi, May 2018Head-Fi, Feb 2019Head-Fi, Mar 2022Head-Fi, Mar 2022Head-Fi, Oct 2022Head-Fi, Nov 2024 |
| 2018 to 2024 | Qutest, Hugo 2, Mojo 2, Hugo M Scaler | Coax and non-isolated BNC are the brightest, most noise-sensitive inputs; choose optical over coax; the M Scaler's isolated BNC is near optical but not equal. Head-Fi, Jul 2018Head-Fi, May 2022Head-Fi, Aug 2024AVstore 41:06AVstore 42:06 |
| 2016 to 2021 | DAVE | AES/EBU is a technically poor standard that injects RF noise and sounds brighter. Head-Fi, Apr 2016Head-Fi, Apr 2018Head-Fi, Jun 2021 |
| 2021 to 2024 | Hugo M Scaler, Mojo 2, DAVE, Hugo TT2 | Whether USB equals optical depends on the source: quiet or battery sources match it, noisy mains-powered PCs sound worse than optical. Head-Fi, Apr 2021Head-Fi, Mar 2022Head-Fi, Jun 2023Head-Fi, Feb 2024MFG Audio Reviews 24:40 |
| 2014 to 2024 | Hugo (original), Hugo 2, DAVE, Qutest, Mojo 2, Hugo M Scaler | Optical may carry slightly more jitter (about 1 nS, though his older analyzer measured less on optical than coax), but his DPLL removes source jitter, so it does not affect the sound. Head-Fi, Jun 2014Head-Fi, May 2019Head-Fi, Oct 2018Head-Fi, Mar 2022Head-Fi, Jul 2018Head-Fi, Feb 2024MFG Audio Reviews 23:47 |
| 2018-01 to 2023-11 | DAVE, Hugo M Scaler | Optical (Toslink) is the best input and the gold standard USB should be made to match; he uses optical himself. Head-Fi, Jan 2018Head-Fi, May 2018Head-Fi, Apr 2019Head-Fi, Nov 2023Soundnews 49:45 |
Caveats and open points
- Most of his optical-versus-USB listening results are tied to specific sources (an MSI laptop with both outputs, a high-power desktop PC) and setups (headphones versus grounded power amps); he qualifies them with YMMV/YMWV.
- His DAVE statements are not uniform: in February 2016 two replies a day apart conflict (Head-Fi, Feb 2016 'Optical will sound the best' versus Head-Fi, Feb 2016 'usb is best'), and two 2018 replies (Head-Fi, Apr 2018, Head-Fi, Sep 2018) say 'USB is best' but name only AES or coax as the alternative, not optical, so they do not show how he ranked USB against optical after January 2018.
- Head-Fi, Sep 2018 and Head-Fi, Apr 2021 carry dates estimated from post IDs.
- Statements about the M Scaler's BNC concern its isolated outputs feeding a DAC, not a source's BNC into a DAC. The February 2018 'M scaler' reply (Head-Fi, Feb 2018) predates the Hugo M Scaler's release, so its product is unclear.
Topics: Choosing and comparing inputs, Ground loops, galvanic isolation, earthing and mains, Jitter, clocks and the DPLL