How does RF noise get into a DAC, and why does it make the sound brighter?

Scope. This page reports what Rob Watts has said in public about how RF noise reaches a DAC (or the rest of an audio system) and why he says it sounds brighter. The items cited run from a May 2014 Head-Fi post to an interview published in October 2026 (recording date not stated; its transcript places it before May 2026) Head-Fi, May 2014HiFi Buys 55:16.

The mechanism he describes

In a May 2014 post in the Hugo thread, he says that for Hugo's digital inputs, "RF noise inter modulates with the analogue parts, creating noise floor modulation, which makes it sound harder and brighter" Head-Fi, May 2014. In January 2016, in the same thread, he gives the longer version: out-of-band RF noise makes the sound brighter because audio electronics are non-linear at RF frequencies, so random RF noise intermodulates with the music and creates products inside the audio bandwidth, noise that "pumps up and down" with the signal Head-Fi, Jan 2016. In that post he says the brain cannot separate this noise from the music and, because noise sounds bright like hiss, it "adds an edge to the perceived timbre of the instrument" Head-Fi, Jan 2016. In November 2024 (Watts Up thread) he locates the pick-up at a differential-pair input stage, where random RF noise produces the intermodulation that creates noise floor modulation Head-Fi, Nov 2024.

He illustrates it with a saxophone. In June 2014 (Hugo thread) he says that noise floor modulation has the timbre of hiss, so the brain "lumps the timbre together, the rich sound of the sax becomes brighter and less like a sax" Head-Fi, Jun 2014. The same illustration recurs in a podcast published June 2020 (recording date not stated) Part-Time Audiophile 16:54, a March 2024 seminar MFG Audio Reviews 10:02, a podcast recorded in October 2025 (published March 2026) ecoustics 59:38, and interviews published in May 2026 and October 2026 (recording dates not stated) Audio Force 10:37HiFi Buys 55:16.

He separates random RF noise from switching noise. In October 2021 (TT2 thread) he says switching noise is easy to filter and "does not add to noise floor modulation at all", whereas removing random RF noise down to picovolt levels was "a complete nightmare" Head-Fi, Oct 2021. In a February 2023 Q&A he says a switching regulator's PWM signal "actually doesn't create any sound quality problems" because its products are discrete harmonics outside the audio band, and that "it's the random RF noise that is creating your noise floor modulation" Soundnews 30:41.

How the RF gets in, by route

His summary for Dave. In June 2016 (Dave thread) he says airborne RF is not a problem for Dave because of its machined aluminum case, and "No RF can get in or out - except via the mains supply and I/O's" Head-Fi, Jun 2016. In January 2018 he repeats that airborne RF inducing currents in Dave's ground plane is "difficult to imagine" given the chassis Head-Fi, Jan 2018.

The mains. In June 2014 (Hugo thread) he recounts that in the 1980s he found "the RF characteristics of the mains cable made a big difference", which led him to design a 12-stage RF mains filter Head-Fi, Jun 2014. In a podcast recorded in October 2025 and published in March 2026 he describes his 1980s system sounding worse when the kitchen fluorescent light was on, "brighter and harder and more aggressive", and attributes that to random RF noise on the mains from plasma discharge ecoustics 57:52ecoustics 58:44.

The source and ground currents. In June 2014 (Hugo) he says the benefit of optical is that it isolates the DAC from "the sources' ground injected RF noise" Head-Fi, Jun 2014. In October 2014 (Hugo) he says USB cables can change the sound through "the amount of RF noise that gets injected into the ground plane of Hugo" Head-Fi, Oct 2014. In July 2018 (Qutest thread) he says coax has no galvanic isolation and "will always be the brighter sounding input", and that on Hugo 2 with headphones optical sounds the same as USB because there is no ground loop from the source Head-Fi, Jul 2018. In February 2024 (M Scaler thread) he says the mechanism is current flowing into the DAC ground plane, that a loop is needed for current to flow, and that with battery operation and optical on the M Scaler there is no loop Head-Fi, Feb 2024. In a March 2024 seminar he describes this as common-mode RF current flowing in a loop from the mains through the source, the cable ground and the DAC MFG Audio Reviews 18:03. In October 2022 (TT2 thread) he says that where USB sounds different from optical it is because RF currents enter through the 2pF of coupling capacitance in the TT2 galvanic isolator Head-Fi, Oct 2022.

He hedges on how general this is. In May 2016 (Dave thread) he says he would be surprised if Dave were sensitive to RF from the source via the mains, "but that's not saying it isn't 100%" Head-Fi, May 2016. In April 2016 (Dave thread) he says his finding that the source made no difference "could be very system dependent" Head-Fi, Apr 2016. In May 2017 (Hugo 2 thread) he says that with headphones and a mains-powered laptop he could not reliably tell optical from USB, while driving a power amp so that a ground loop formed he could hear small differences Head-Fi, May 2017.

Cables and their placement. In April 2016 (Dave thread) he says a difference he first heard between two USB cables disappeared once he stopped the headphone cable touching the USB cable; he concluded the headphone cable "was picking up RF noise from the larger audiophile USB cable" Head-Fi, Apr 2016. In October 2021 (Watts Up thread, replying about Mojo) he says "Many audiophile USB cables are engineered to increase RF noise, making them brighter sounding" Head-Fi, Oct 2021, and that USB cables can only change the sound by influencing the random RF noise level, so a brighter sound is "definitely a degradation", while adding "YMMV" Head-Fi, Oct 2021.

The DAC itself and its inputs. In January 2016 (Hugo thread) he calls a DAC "a wonderful source of random RF noise" Head-Fi, Jan 2016. In April 2024 (M Scaler thread) he says USB is by far the worst input because it needs a processor to buffer and decode, which creates "two orders of magnitudes more random RF noise", with Toslink receivers having the lowest power consumption Head-Fi, Apr 2024. In the ecoustics podcast (recorded October 2025) he says each piece of processing inside a DAC creates RF noise that must be isolated from the analog parts ecoustics 1:02:33.

High-frequency noise in the recording. In May and June 2014 (Hugo thread) he says high-rate recordings carry ADC noise-shaper noise that creates noise floor modulation in the analog stages Head-Fi, Jun 2014, and of DXD recordings he says "I hear this with DXD recordings, a brightness that sounds just like noise floor modulation" Head-Fi, May 2014. In November 2015 (Dave thread) he says Dave's 60 kHz HF filter removes ADC noise-shaper noise from 192k, DXD and 384k recordings and makes DXD "much smoother and darker" Head-Fi, Nov 2015; in January 2016 (Hugo thread) he adds that the filter was designed to stop that noise creating noise floor modulation within Dave Head-Fi, Jan 2016.

Other causes of the same sound. He also attributes a brighter sound to low-frequency jitter from an external clock's PLL (January 2018, Watts Up thread): "these skirts sound exactly like noise floor modulation - so it sounds brighter" Head-Fi, Jan 2018. In February 2022 (Mojo 2 thread) he says floating-point DSP implementations create audible noise floor modulation that "adds an artificial brightness and hardness to the sound" Head-Fi, Feb 2022. These are not RF routes; they are other sources he ties to the same audible result.

What it sounds like, by amount

• 2014, Hugo. Random RF noise making it "harder and bright", and "if it gets really bad, the treble will sound grainy and aggressive" Head-Fi, Oct 2014. Reducing it gives a sound that is smoother with better instrument separation and focus Head-Fi, Jun 2014.
• 2015 to 2016, Dave. In November 2015 he says the brain perceives noise floor modulation as brightness, so reducing it makes it sound "smoother and darker" Head-Fi, Nov 2015. In August 2016 he gives a rule of thumb, which he labels as his own "tentative conclusions": around -120 dB (he says typical class D) there is "considerable hardness and glare", at -140 dB grain in the treble, and below -160 dB a much smoother sound Head-Fi, Aug 2016.
• 2017 to 2018. In December 2017 (Dave thread) he says "more noise modulation always sounds brighter" Head-Fi, Dec 2017. In March 2018 (Qutest thread): "In large amounts it shows up as grain, glare and hardness; in small amounts it adds brightness and suppresses timbre variation" Head-Fi, Mar 2018. In a November 2018 open-day talk (uploaded December 2018) he gives class D amplifier noise floor modulation as roughly -115 dB at full output falling to about -130 dB and says it makes the treble sound grainy, while reducing it sounds "warmer and darker" Pursuit Perfect System 11:05. Those figures come; check the recording.
• 2023. In a CanJam NYC talk in late February 2023 (show dates unverified; video published March 2023) he says large levels give "a grainy treble" and smaller levels change the perception of refinement or smoothness and timbre variation The Headphone Show 44:50.
• 2024 to 2026. In March 2024 he says noise floor modulation well below measurable limits is audible, and that this is why cable changes can be heard in terms of brightness with zero measurable difference MFG Audio Reviews 15:45. In an interview published May 2026 (recording date not stated) he says noise floor modulation from power supplies "makes things sound artificially bright" Audio Force 11:27.

Brighter does not always mean RF

He says brightness alone does not tell you the cause. In March 2016 (Watts Up thread) he asks whether a brighter sound is due to "improved transparency, which normally sounds brighter" or to increased noise floor modulation, "which actually sounds very similar", and says it can be almost impossible to decide by AB testing Head-Fi, Mar 2016. In July 2016 (Dave thread) he says of a new Dave unit that sounded brighter that "for certain when you improve transparency things do sound brighter" and that noise floor modulation also sounds brighter and gives a false impression of transparency; his test was a busy track, where true transparency does not fall apart Head-Fi, Jul 2016. In July 2017 (Hugo 2 thread) he says "More refinement equates to a warmer presentation; more transparency gives a much brighter presentation" Head-Fi, Jul 2017. In an interview published October 2026 (recording date not stated) he says that with noise floor modulation both the bright and the dark instruments become brighter, whereas a real improvement makes bright instruments brighter and dark instruments darker HiFi Buys 48:54. In a 2024 talk and a podcast published October 2026 (recording date not stated) he also warns that a change can seem an improvement only because it balances another fault, for example adding distortion to spice up a system that is too warm The Source AV 24:37Part-Time Audiophile 6:40.

For bit-perfect sources into his DACs his statements are narrower and firmer. In May 2017 (Hugo 2 thread): "assuming the source is bit perfect, then if it sounds brighter it most definitely is worse" Head-Fi, May 2017. In May 2018 (Dave thread) he says that for bit-perfect data via USB "the only thing that can possibly modify SQ is RF noise" Head-Fi, May 2018. In December 2017 (Dave thread) and January 2018 (Hugo 2 thread) he advises choosing the warmer source option even if it makes the system too warm, and correcting tonal balance elsewhere Head-Fi, Dec 2017Head-Fi, Jan 2018. In January 2018 (Hugo 2 thread) he says the most likely explanation for a brighter source is RF noise from a Blu-ray player, and "Whichever source sounds warmer is actually technically better" Head-Fi, Jan 2018. In May 2021 (Hugo 2 thread) he calls it a rule of thumb that "the warmer sound is the better, more accurate performance", and says "It is only noise floor modulation that can affect sound quality with bit perfect sources" Head-Fi, May 2021. In the ecoustics podcast, describing Quartet power-supply development, he says "If it sounds warmer, it's better. You've got less noise floor modulation" ecoustics 1:11:58.

On taste, in August 2018 (Hugo 2 thread) he says that if someone prefers the brighter sound from more noise floor modulation, "that's their taste and preference. But it's not accurate" Head-Fi, Aug 2018. In March 2024 he says some people prefer it because it "artificially enhances the sound" MFG Audio Reviews 13:01.

Changes over time

His own USB versus optical listening. In May 2014 (Hugo) he says he prefers optical because it "normally sounds smoother and warmer" Head-Fi, May 2014. In July 2016 (Dave thread) he says "Before galvanic isolation, optical was so much better than USB", which places the 2014 preference before USB was isolated, and that USB "has zero jitter/timing issues as the clock comes from Dave" Head-Fi, Jul 2016. In August 2016 he calls the USB-optical difference "in my bottom scale of SQ differences" and says that when just enjoying music "they all work identically from the musicality POV" Head-Fi, Aug 2016. In January 2018 (Hugo 2 thread) he says that before Hugo 2 he could not tell from listening whether USB's extra brightness was transparency from the common clock or more RF, and that hearing no difference on Hugo 2 settled it Head-Fi, Jan 2018. In January 2018 (Dave thread) he says of his earlier impression of Dave's brighter USB, "But I have changed my mind on this now", attributing it to a small RF increase through the isolators Head-Fi, Jan 2018, and the next day he writes "I used to think that USB had the edge on optical - it's slightly brighter - but I was wrong about that, the brightness is not better timing as the clock comes from Dave, but worse RF noise pick-up" Head-Fi, Jan 2018. In February 2019 (Qutest thread) he writes "Technically USB is better from the clock POV; but in practice I have had optical sound identical to USB, so that proves that my DPLL is capable of removing all source jitter completely" Head-Fi, Feb 2019; after the 2018 reversal he still credited USB with a technical clocking advantage while hearing no difference. In March 2022 (Dave thread) he writes "You have discovered a post where I made a mistake. In 2016 with Dave I did indeed prefer the brightness of USB", and says the extra brightness was RF through USB, with an unmeasurable increase in noise floor modulation Head-Fi, Mar 2022. In October 2022 (TT2 thread) he says "I too used to prefer USB than optical. But I realised later that I was making a mistake" and "there is no technical benefit to using USB over optical when using my DACs" Head-Fi, Oct 2022; that differs from his February 2019 statement that USB is technically better from the clock point of view, and both are reported here as he gave them. In November 2024 (Watts Up thread) he says the same about the 2Qute: USB "was brighter with 2 Qute", he wrongly credited USB clocking, and it "turned out to be noise floor modulation" Head-Fi, Nov 2024.

Audibility threshold. In 2014 (Hugo) he says he has had noise floor changing at -200 dB and it was audible Head-Fi, Jun 2014. In August 2016 (Dave) he gives -200 dB as his rule of thumb for what one can hear, adding "and perhaps even lower" Head-Fi, Aug 2016, and in May 2018 he says he has seen -200 dB audible in simulation Head-Fi, May 2018. In October 2021 (TT2 thread) he says his recent research and listening tests indicate that "even noise floor modulation as low as -350dB is perceivable under single blind conditions" Head-Fi, Oct 2021. The 2021 figure is lower than the earlier ones, which he had already said might not be the limit; beyond "simulation" (2018) and "single blind" listening (2021), the cited items do not describe how either figure was reached.

Product differences he states. In a Passion for Sound interview (recording date not stated; part 1 published December 2019, part 4 January 2020) he says Hugo 1 did not sound the same as optical over USB, and that adding RF filters to Hugo 2's USB path gave audible benefits that did not show up as any change in measurable noise floor modulation Passion for Sound 12:11. In March 2022 (Dave thread) he says the Qute DAC was sensitive to its power supply, and that after he redesigned the regulation and RF filtering in Qutest, a car battery and the supplied supply sounded the same Head-Fi, Mar 2022.

What he has not said

This archive does not show him saying that all brightness is caused by RF noise; he names other causes and says transparency can also sound brighter Head-Fi, Mar 2016Head-Fi, Jul 2017Head-Fi, Jan 2018. His strongest claim is limited to bit-perfect sources into his own DACs Head-Fi, May 2018Head-Fi, May 2021. They do not show him saying every DAC, source or cable will change with RF noise; he calls his results system dependent and gives "YMMV" caveats Head-Fi, Apr 2016Head-Fi, Oct 2021.

Quotes

“RF noise inter modulates with the analogue parts, creating noise floor modulation, which makes it sound harder and brighter.”2014-05-13 Head-Fi, May 2014

“Problem is that many listeners hear the brightness as more detail resolution, and so think it sounds better - but that's another story.”2016-01-04 Head-Fi, Jan 2016

“is this brightness down to improved transparency, which normally sounds brighter, and gives better detail resolution - or is it down to increased noise floor modulation”2016-03-29 Head-Fi, Mar 2016

“Now if somebody prefers the brighter sound from more noise floor modulation, then fine - that's their taste and preference. But it's not accurate.”2018-08-26 Head-Fi, Aug 2018

Dated positions

Caveats and open points

• Some Head-Fi posts have scraper gaps (for example Head-Fi, Jan 2018 and Head-Fi, Jan 2018); no quote here spans a gap.
• His statements are about his own products and test systems and he calls results system dependent; they are not general findings about other DACs.
• He gives -200 dB (2014 to 2018, with 'perhaps even lower' in 2016) and -350 dB (2021) as audibility figures; the cited items say only that these came from simulation and single-blind listening.
• Head-Fi, Oct 2021 (October 2021) carries a date estimated from its post ID.

Topics: RF noise: mechanisms and audibility, Tonal character as a diagnostic, Noise floor modulation