Is Chord's digital volume control transparent, and does it lose resolution at low volume?

Practical answer. Watts advises using the volume control built into your Chord device rather than a pre-amp or passive attenuator Head-Fi, Jun 2014Head-Fi, May 2018Head-Fi, Mar 2020. Of pre-amps he said in 2018 that they "do add a softness and bloom to the bass (due to coupling capacitors) which some prefer. I don't", so on that point he states his own preference Head-Fi, May 2018. Against passive attenuators he gave technical reasons: distortion and a softer, poorly defined bass from cable dielectric absorption (2014), and added noise and distortion and degraded small-signal linearity, "thus upsetting depth and detail resolution" (2020) Head-Fi, Jun 2014Head-Fi, Mar 2020. From 2014 to 2024 Watts describes his volume controls as digital and transparent, from "the only way of making it completely transparent" (Hugo, 2014) to "my 100% transparent volume controls" (2024) Head-Fi, Nov 2014Head-Fi, Mar 2016Head-Fi, Sep 2023Head-Fi, Jan 2024. On low volume, he said in 2014 that Hugo's control moves in exact dB steps, so "using IEM's with a low volume setting won't make any difference" Head-Fi, May 2014. Of Dave he wrote in 2017 that because the digital path is better than 350 dB, "a major change in volume will have no effect on detail resolution and (more critically) depth perception" Head-Fi, Oct 2017. In November 2024 he also advised not letting the source alter volume: "always send bit perfect data" Head-Fi, Nov 2024.

Why he says so. He holds that analog volume controls add noise and small-signal non-linearity, which he says harms depth (2016) Head-Fi, Mar 2016. The digital problem he identifies is the truncation after multiplying by the volume coefficient (24 bits times 24 bits gives 48); in 2020 he said dither "isn't perfect" and "The only way of completely maintaining full transparency when truncating is to use 11th order noise shaping, and this only at 705/768 kHz" Head-Fi, Mar 2020.

By product

• Hugo (2014 to 2015). In 2015 he said the volume sits in the WTA filter core with 51-bit internal accuracy, that "the truncation is done via noise shaping and dithering", adding "just a fixed noise at -180dB" Head-Fi, Jan 2015. He gave the range as +18 dB down to -70 dB (2015) and advised feeding Hugo straight into power amps rather than through a pre-amp or passive control (2014) Head-Fi, Mar 2015Head-Fi, Jun 2014.
• Dave (2016 to 2021). In 2016: "don't worry about the volume control changing sound quality - it doesn't" Head-Fi, Mar 2016. The same year he gave maximum attenuation as -75 dB and said a 24-bit signal there (-219 dB) is still 131 dB above the truncation noise of about -350 dB Head-Fi, Mar 2016. In 2021 he said the volume output is "simply not truncated - nor is it dithered - but aggressively noise shaped", reproducing a -301 dB test signal to better than +/-0.001 dB and within +/-0.001 degrees Head-Fi, Jul 2021.
• Hugo 2 (2018 to 2020). In 2018: "volume is implemented in Hugo 2 exactly the same as Dave" Head-Fi, Jan 2018. He said it runs at a minimum of 705/768 kHz with very aggressive noise shaping, keeping added distortion and noise below -350 dB Head-Fi, Feb 2018.
• Qutest (2018 to 2023). No user volume control (2018, 2020) Head-Fi, Mar 2020Head-Fi, Jan 2018. Its 1, 2 or 3 V output setting is done digitally, "using Hugo 2's volume control" (2018), which he called more transparent than analog (2019); in 2023 he said there are "No changes in the analogue domain at all" Head-Fi, Jan 2024Head-Fi, Jan 2018Head-Fi, Jan 2019Head-Fi, Sep 2023. Software volume in the player, he said in 2018, "will not be quite as good" as Chord's, though he suspected it beats an analog control Head-Fi, Feb 2018Head-Fi, Feb 2018. In 2020 he said Roon volume beats a pre-amp or passive control, and to "keep your source bit perfect (apart from when you have to do volume control with Qutest)" Head-Fi, Mar 2020.
• Hugo TT2 (2018 to 2020). In DAC mode the outputs are fixed and volume is disabled; otherwise you use low gain mode or the volume control (2018) Head-Fi, May 2018. DAC mode equals low gain mode with volume fixed at -5L (2020) Head-Fi, Jul 2020.
• Mojo 2 (2022). The volume is the last stage of the DSP core; with EQ and cross-feed off it receives bit-perfect data from the first WTA filter Head-Fi, Feb 2022. He used the 104-bit DSP core for volume, saying a transparent volume control must reproduce a -301 dB signal perfectly Head-Fi, Mar 2022. There are no analog gain changes; he raised the steps from the original Mojo's 96 to 128 for very efficient IEMs Head-Fi, Dec 2022. In a 2022 talk he gave the range as -108 dB to +18 dB MFG Audio Reviews 39:06.

What he does not claim

Asked in 2018 whether TT2's volume control is bit perfect: "So no not bit perfect; but within the audio bandwidth no distortion or noise added (to a -350dB accuracy)" Head-Fi, Oct 2018. In 2021 he said no volume control can be bit perfect, and calling his "absolutely lossless" would be "technically inaccurate", since his evidence is subjective and he was researching how far noise-shaped truncation can go before it stops being transparent Head-Fi, Aug 2021. By contrast, re-quantizing to 24 bits with triangular dither is "certainly not audibly transparent" (2019) Head-Fi, Mar 2019.

He gave no measured analog-output figures at each volume setting for any product, and nothing on Poly's volume. His case rests on the -301 dB digital-domain test and listening Head-Fi, Aug 2021Head-Fi, Mar 2022.

Quotes

“a major change in volume will have no effect on detail resolution and (more critically) depth perception”2017-10-25 Head-Fi, Oct 2017

“Always use Hugo 2's volume control - adding a passive will increase noise and distortion, and more importantly degrade small signal linearity, thus upsetting depth and detail resolution.”2020-03-12 Head-Fi, Mar 2020

“So no not bit perfect; but within the audio bandwidth no distortion or noise added (to a -350dB accuracy).”2018-10-27 Head-Fi, Oct 2018

“I can change the attenuation without any SQ degradation to detail resolution, timbre reproduction or soundstage depth perception.”2023-09-24 Head-Fi, Sep 2023

Dated positions

Caveats and open points

• Nearly all of the evidence is Watts describing his own design and listening results; no independent measurements are in this archive.
• The Mojo 2 volume range of -108 dB to +18 dB comes; check the recording.
• On the original Mojo this archive has only its step count (96); nothing was found on Poly's volume handling.
• He describes the truncation method differently by product and date (noise shaping and dithering for Hugo in 2015; noise shaped, not dithered, for Dave in 2021); this archive does not show these as a change of view.
• His 2024 advice to keep the source bit perfect sits alongside his 2020 acceptance of software volume for Qutest, which has no volume control of its own.

Topics: Volume, gain and output level, Re-quantization, truncation, dither and word length