Noise floor modulation

Noise shaping, accuracy and conversion hardware

Answers on this topic

• How does RF noise get into a DAC, and why does it make the sound brighter?
• What does Rob mean by depth, and what in a DAC improves or collapses it?
• What does Rob think of Class D amplifiers, and how do his pulse array power amps differ from them?
• What is noise floor modulation, how does Rob measure it, and why does he say it is audible below what can be measured?
• Why might a brighter or more detailed change actually be a step backward?

Everything he said on this topic (209)

One-line summaries in our words, oldest first. Each link opens the original.

• 2014-05-16Watts credits Hugo's 26,368 WTA taps for its timing, says more taps should help further, and suggests hi-res can underperform red-book because ADC out-of-band noise causes noise floor modulation. Head-Fi, May 2014
• 2014-06-01Watts says higher rates improve timing but admit ADC noise-shaper HF noise causing noise floor modulation; Hugo's 26k-tap filter shrinks the timing gain; he considers a PCM HF filter. Head-Fi, Jun 2014
• 2014-06-09Watts recounts 1980s mains cable experiments that showed RF noise mattered, leading to a 12-stage RF mains filter, and explains RF-induced noise floor modulation, audible even at -200dB. Head-Fi, Jun 2014
• 2014-06-22Watts re-tests optical vs coax on Hugo, finds optical warmer with better separation, and explains how RF-induced noise floor modulation harms separation and timbre. Head-Fi, Jun 2014
• 2015-06-27Watts defines noise floor modulation, says it is audible at immeasurable levels, and shows APx555 plots of Dave with an unchanged noise floor near -180dB with and without signal. Head-Fi, Jun 2015
• 2015-11-21Watts explains Dave's DSD filtering, how out-of-band noise causes noise floor modulation heard as brightness, the 60 kHz HF filter helping even CD, and PCM+ versus DSD+ modes. Head-Fi, Nov 2015
• 2015-12-18Watts says Dave's HF filter is a 60 kHz cutoff set of 16FS-to-256FS WTA coefficients meant to cut ADC noise in hi-res recordings, and it unexpectedly improves 44.1 kHz too. Head-Fi, Dec 2015
• 2016-01-04Watts explains that source differences heard with Hugo came from RF noise (noise floor modulation heard as brightness) and correlated ground currents degrading depth, which Dave's galvanic isolation removes. Head-Fi, Jan 2016
• 2016-01-18Watts explains that out-of-band RF noise intermodulates into noise floor modulation that sounds bright, so his DACs are limited to 50 or 60 kHz; Dave's HF filter targets DXD noise. Head-Fi, Jan 2016
• 2016-01-18Watts says Dave's filter targeted ADC noise shaper noise in 176.4 kHz and above recordings, yet audibly warms redbook; he suspects lower noise floor modulation and keeps it on. Head-Fi, Jan 2016
• 2016-01-29Watts shows an APx555 plot with a -180 dB noise floor identical with no signal and at 2.5 V, and says noise floor modulation makes sound brighter and harder. article, Jan 2016
• 2016-03-04Watts explains the pulse array modulator and seven benefits (constant switching, jitter immunity, signal-unrelated currents, 104 MHz noise shaping), why Dave has 20 elements, and Mojo versus Dave THD. Head-Fi, Mar 2016
• 2016-03-13Watts explains AES17 dynamic range, gives Dave's 2.6 uV residual (-128.4 dB re 6.8 V, -117.7 dB re 2 V), and says its unchanging noise shows no noise floor modulation. Head-Fi, Mar 2016
• 2016-04-03Watts explains Class D's timing and noise floor modulation problems, then linear-amp issues: input-stage NFM, metal-contact non-linearity, feedback, and phase intermodulation, solvable by analog noise shaping. Head-Fi, Apr 2016
• 2016-05-13Watts doubts DAVE is sensitive to source RF via the mains and explains how RF intermodulating in analog electronics causes noise floor modulation that sounds harder and brighter. Head-Fi, May 2016
• 2016-05-14Watts clarifies that DAVE's below-0.5 milliohm supply is a discrete linear reference circuit, not the switcher, with 0.22 uV noise to avoid measurable noise floor modulation. Head-Fi, May 2016
• 2016-07-18Watts explains how source RF couples into DAVE through the USB isolator's 2 pF capacitance and ground plane, raising noise floor modulation, and why a battery-powered laptop avoids it. Head-Fi, Jul 2016
• 2016-07-24Watts says DAVE's harmonics are below -150 dB and noise floor -180 dB, unchanged with no signal or 2.5 volts output; he says noise floor modulation sounds brighter and harder. Hifistatement 22
• 2016-08-05Watts describes DAVE's power system: multistage RF filtering, a 3 milliohm output-stage supply and a discrete reference under 0.1 milliohm; he says noise floor modulation is audible to -200 dB. Head-Fi, Aug 2016
• 2017-03-15Watts says white engages WTA2 at 256FS; the second-order HF filter (-3 dB at 40 kHz) removes ADC noise-shaper noise in 96/192 files, preventing noise floor modulation, sounding warmer. Head-Fi, Mar 2017
• 2017-03-25Watts suggests an Audacity FFT to check hi-res files: true 96/192 shows rising ADC noise-shaper noise, which he filters out to reduce noise floor modulation; upsampled files differ. Head-Fi, Mar 2017
• 2017-10-13Watts says standard ADC noise shapers manage 140 dB, hopes Davina recordings will convey depth, and says ADCs have heavy noise floor modulation, audible even at -200 dB. miceblue425 38:19
• 2017-10-13Watts says Davina tests at zero noise floor modulation versus about 4 to 5 dB in standard ADCs, and that the APx555 uses four ADCs and a notch filter to avoid it. miceblue425 39:02
• 2017-11-29Watts explains that 2 GHz ferrites improve depth by blocking signal-correlated ground bounce from FPGA logic transitions, while random RF causes noise floor modulation and brightness. Head-Fi, Nov 2017
• 2017-12-02Watts calls the reference voltage the key internal supply, explains 2Qute's upgraded regulation and RF filtering, his jitter-removing DPLL, and pulse array jitter tolerance (DAVE 128 to 128.5 dB). Head-Fi, Dec 2017
• 2017-12-05Watts lists distortion, small-signal nonlinearity and noise floor modulation as transparency factors; measuring with a minus 60 dB signal, most DACs reach about minus 60 dB, his about minus 175. RMAF 9:53
• 2017-12-05Watts says conventional DACs' noise rises with signal, minus 160 to minus 140 dB, Class D far worse, and that inaudible-to-measurement NFM is heard as lost refinement and smoothness. RMAF 10:38
• 2017-12-05Watts says mains cables change sound via RF noise intermodulating in analog electronics, so random noise pumps with the signal, confusing timbre perception and making dark instruments sound brighter. RMAF 11:24
• 2017-12-05Watts says his NFM work began hearing mains cables through high-PSRR preamps; intermodulation pumps the noise floor, and even unmeasurable NFM affects refinement and smoothness. RMAF 23:57
• 2017-12-05Watts says chips cancel reference noise with a differential structure, but it cancels only at zero, not full scale, making it a big source of noise floor modulation. RMAF 31:09
• 2017-12-05Watts says the DSD modulator creates noise floor modulation in the digital domain, with DSD64 at about 110 to 126 dB versus the 350 dB his DACs need. RMAF 36:37
• 2017-12-05Watts says pulse array's lower noise avoids input stage overload and NFM, Hugo 2 and DAVE use an analog noise shaper amplifier, and audiophile design swung from measurement to listening. RMAF 46:31
• 2017-12-05Watts says he listens only after thorough measurement, and shows DAVE with noise floor modulation eliminated at 2.5 volts and harmonics below minus 150 and minus 155 dB. RMAF 48:25
• 2017-12-05Watts says the HF filter is for HD recordings, whose ADC out-of-band noise shaper noise adds noise floor modulation; filtering it sounds noticeably warmer. Head-Fi, Dec 2017
• 2017-12-18Watts says sampling images cause transient timing errors and their out-of-band noise overloads analog stages causing noise floor modulation, which none of his DACs from Mojo to DAVE measurably show. Head-Fi, Dec 2017
• 2018-01-09Watts says the HF filter matters because ADC noise shaper noise above about 25 kHz causes noise floor modulation; Davina's noise will rise only above 768 kHz. Head-Fi, Jan 2018
• 2018-02-17Watts shares Hugo 2 measurement slides from his CanJam NYC talk, explaining low output impedance, load-independent distortion, no measurable noise floor modulation or jitter, and APx555 methods. Head-Fi, Feb 2018
• 2018-02-23Watts says putting this FPGA in Dave would inject signal-correlated noise from 528 DSP cores into the ground plane, peaking at the smallest signals and degrading depth perception. MajorHiFi 20:02
• 2018-03-11Watts says switchers beat linear regulators on Hugo 1 through lower RF noise, which causes noise floor modulation heard as brightness, and advises trying a USB battery bank first. Head-Fi, Mar 2018
• 2018-03-22Watts explains the 2.75dB WTA overload margin, delays audio in JRiver to keep 1M taps with video, describes visible noise floor modulation, and gives DAVE and Hugo 2 level limits. Head-Fi, Mar 2018
• 2018-04-30Watts says the SYS2722 cannot measure Dave (-118dB versus -128dB on an APx555), calls NFM a top-3 SQ indicator, and uses Verilog simulation to verify -350dB and -300.99dB performance. Head-Fi, Apr 2018
• 2018-05-05Watts explains 1-bit DSD noise shapers are signal-dependent and unstable, unlike N-bit designs; Davina's N-bit pulse array ADC has 400 dB noise shaper performance and signal-independent activity. Head-Fi, May 2018
• 2018-05-17Watts says with bit perfect USB data only RF noise can change the sound, by creating noise floor modulation that sounds brighter; jitter has no influence on his DACs. Head-Fi, May 2018
• 2018-05-18Watts says the DPLL's buffered regenerated clock eliminates source jitter (Hugo 2 USB/optical tests showed no change), and explains why signal-related noise sounds brighter, audible even at -200 dB. Head-Fi, May 2018
• 2018-05-27Watts says Stereophile's AP cannot measure Dave because ADC noise floor modulation interferes; the APx555 uses a notched second ADC for residuals, showing Dave's noise floor near -180dB. Head-Fi, May 2018
• 2018-08-26Watts says power banks beat mains on RF noise (10 MHz to 10 GHz), linear supplies lack RF filtering, RF causes audible noise floor modulation, and batteries avoid ground loops. Head-Fi, Aug 2018
• 2018-12-12Watts says all DACs create noise; a fixed noise is unimportant, like background air conditioning, but the problem is noise that changes with the signal. Pursuit Perfect System 9:36
• 2018-12-12Watts defines noise floor modulation as noise varying with signal, says all DACs have it, it is audible, and large amounts, as in Class D amplifiers, make treble grainy. Pursuit Perfect System 10:18
• 2018-12-12Watts says a Class D amplifier's noise floor moves from minus 115 dB at full output to about minus 130 dB; reducing NFM sounds warmer, and unmeasurable levels remain audible. Pursuit Perfect System 11:05
• 2018-12-12Watts says a discrete FPGA-based DAC measures and sounds better, the pulse array eliminates noise floor modulation by constant switching, and a DSD modulator switches most when reproducing zero. Pursuit Perfect System 15:43
• 2018-12-12Watts says a DSD output's switching varies with signal; each switch makes glitch noise, plus N/P FET speed asymmetry, so signal-dependent glitch noise creates distortion and noise floor modulation. Pursuit Perfect System 16:35
• 2018-12-12Watts shows DAVE's noise floor at minus 179 dB with or without signal, zero measurable though not zero NFM; a good conventional DAC goes from minus 160 to 140 dB. Pursuit Perfect System 18:40
• 2018-12-12Watts says other manufacturers avoid discussing noise floor modulation because they either do not see it as a sound problem or know they cannot fix it without a pulse array. Pursuit Perfect System 19:25
• 2018-12-12Watts says discrete parts allow a very low-noise, low-impedance reference to avoid noise floor modulation and distortion, and that RF noise, a big NFM source, needs filtering throughout. Pursuit Perfect System 34:07
• 2018-12-12Watts says RF-induced NFM sounds brighter and is mistaken for transparency but becomes tiring within an hour; he then states TT2 gives 9.3 V RMS and five amps of current. Pursuit Perfect System 38:55
• 2018-12-12Watts says the jitter test shows no jitter components, removing source and internal jitter, and that he signs off no design until it shows no measurable noise floor modulation. Pursuit Perfect System 41:18
• 2018-12-13Watts says he judges improvements by checking timbre, space, detail and separation separately, and that RF noise pumping with the signal sounds bright and etched, faking transparency. Pursuit Perfect System 25:00
• 2018-12-16Watts explains why a DAC's reference supply needs zero noise and impedance, says chip DACs hide poor references, and describes TT2's DAVE-standard discrete reference and its warmer sound. Head-Fi, Dec 2018
• 2019-03-04Watts says removing harmonic distortion while ignoring noise floor modulation sounds bright and analytical, while eliminating noise floor modulation restores warmth, and shares TT2 measurements. Head-Fi, Mar 2019
• 2019-05-11Watts says chip DACs suffer substrate noise, NFM and jitter, while his discrete DACs measure -120dB signals to +/-0.001 dB and resolve -186 dB within 1dB. Head-Fi, May 2019
• 2019-06-16Watts shows by a zero-crossing example why higher rates improve transient timing, says 705/768 kHz is needed, and says hi-res ADC noise causes noise floor modulation the HF filter reduces. Head-Fi, Jun 2019
• 2019-06-16Watts offers three reasons green sounds warmer on RBCD, favoring the HF filter retuning the 256FS-to-2048FS stage from 160 kHz to 40 kHz, lowering noise floor modulation. Head-Fi, Jun 2019
• 2019-11-25Watts says few DAC designers understand sampling theory or noise floor modulation, warns clip-on ferrites worsen the isolated M scaler, and says floating point creates noise floor modulation. Head-Fi, Nov 2019
• 2019-11-27Watts explains 2 GHz loop currents through the DAC ground plane, why clip-on ferrites stopped helping after his M scaler treatments, and why linear PSUs degrade sound. Head-Fi, Nov 2019
• 2019-12-16Watts says PWM switches at a constant rate whatever the signal, so glitch energy becomes fixed noise instead of signal-dependent noise that creates noise floor modulation and distortion. Passion for Sound 5:29
• 2020-01-02Watts says his DACs still have noise floor modulation, just none measurable, and immeasurable modulation can still be audible; he begins with a saxophone example. Passion for Sound 7:31
• 2020-01-02Watts says noise rising with a saxophone's level cannot be separated from it by the brain, adding bright noise that makes a warm saxophone sound artificially bright. Passion for Sound 8:07
• 2020-01-02Watts says modulation noise from saxophone and piano is spectrally identical, giving the brain four signals instead of two, merging the instruments and degrading separation and focus. Passion for Sound 8:51
• 2020-01-02Watts says reducing measurable noise floor modulation makes timbre richer, smoother and warmer and separation more solid, and that all DACs other than his create measurable modulation. Passion for Sound 9:36
• 2020-01-02Watts says noise floor modulation tests need identical settings with only the level reduced, and his plots show no measured modulation in any product from Mojo to DAVE. Passion for Sound 10:05
• 2020-01-02Watts says they will still have residual noise floor modulation, which might still be audible. Passion for Sound 10:24
• 2020-01-02Watts says optical is the best input because it carries no RF noise; RF entering non-linear analog electronics intermodulates with the audio, spreading noise across the bandwidth. Passion for Sound 10:34
• 2020-01-02Watts says the result is a lump of noise in the audio band linked both to the audio signal and to the random RF noise. Passion for Sound 11:15
• 2020-01-02Watts says RF causes noise floor modulation; adding RF filters to Hugo 2's USB made it sound like optical, unlike Hugo, with no measurable change, so unmeasurable modulation matters. Passion for Sound 12:11
• 2020-04-09Watts says mains conditioners neglect RF, a single-blind test tied a new HF filter's warmth to noise floor modulation, and DAVE is about 13-bit accurate, M scaler over 16. Head-Fi, Apr 2020
• 2020-04-09Watts says the new HF filter is a 4th-order 50kHz digital filter of two bi-quads, tuned for a small 20k boost, and its warmth comes from reduced out-of-band noise. Head-Fi, Apr 2020
• 2020-05-04Watts says USB matters through RF noise and signal-correlated ground currents, that his tests find noise floor modulation at -180 to -200 dB audible, and Chord USB filtering suffices. Head-Fi, May 2020
• 2020-06-06Watts says it signals more RF injected into the DAC, raising noise floor modulation that falsely seems more detailed and increases listening fatigue. needs context Head-Fi, Jun 2020
• 2020-06-25Watts explains noise floor modulation as noise rising and falling with the signal, which is measurable, substantially changes sound and makes warm-sounding instruments sound bright. Part-Time Audiophile 16:11
• 2020-06-25Watts says modulated white noise brightens a saxophone, valves have local feedback, high overload margin and good RF linearity, and RF noise intermodulating with the signal causes noise floor modulation. Part-Time Audiophile 16:54
• 2020-06-25Watts says tubes innately suffer less noise floor modulation so sound warmer and smoother, but add distortion such as second harmonic, and transformers give high output impedance. Part-Time Audiophile 17:40
• 2020-06-25Watts says tubes' poorer damping factor adds low-frequency output and warmth, and semiconductor designs can avoid noise floor modulation with care over RF noise, open-loop nonlinearity, layout and decoupling. Part-Time Audiophile 18:22
• 2020-06-25Watts says bipolar or MOSFET designs can be made free of noise floor modulation, but it is harder and requires knowing what you are doing. Part-Time Audiophile 19:02
• 2020-09-30Watts says his DACs use switching regulators, which sounded better than linear ones from Hugo onward, and that random RF noise, not fixed switching noise, causes noise floor modulation. Head-Fi, Sep 2020
• 2020-10-19Watts says a poster's USB versus optical difference on Dave fits RF raising noise floor modulation, though he heard only a tiny change, and says he has heard NFM below -180 dB. Head-Fi, Oct 2020
• 2021-01-25Watts says the treble extension and tighter bass heard with an Sbooster supply are noise floor modulation from more RF, and SMPS RF filters give softer, warmer, less fatiguing sound. Head-Fi, Jan 2021
• 2021-03-20Watts says earth voltage matters only when the source drives current across the DAC ground plane, whose impedance creates voltage drops that the analog stages turn into noise floor modulation. Head-Fi, Mar 2021
• 2021-04-03Watts says supply ferrites are fine, any trace of noise floor modulation seems audible so he expects desktop two-box, and stock cables with a battery M scaler matched Wave Stream. Head-Fi, Apr 2021
• 2021-05-22Watts says all DSD speeds have problems; DSD256 beats DSD64 but not WTA PCM, and two-level modulators cause noise floor modulation that needs at least 3-bit modulation. Head-Fi, May 2021
• 2021-05-31Watts says Hugo 2's USB sensitivity depends on mains loops, advises optical and avoiding DDCs in mains systems, and explains RF-induced noise floor modulation as brightness. Head-Fi, May 2021
• 2021-06-26Watts says a linear PSU only degrades or matches TT2's stock unit, since random mains RF raises noise floor modulation and brightness, and regrets not adding more RF filtering. Head-Fi, Jun 2021
• 2021-07-19Watts says ADC interpolator tests showed noise floor modulation audible near -360 dB, explains M scaler ferrites and Wave cables' solid ferrites, and suggests battery power plus optical for isolation. Head-Fi, Jul 2021
• 2021-08-31Watts says Stereophile's poor Dave -90.31dB result did not reproduce on his APx555, calls the SYS2722 inadequate, and shows Dave with zero measurable noise floor modulation and jitter. Head-Fi, Aug 2021
• 2021-10-04Watts describes SPICE RF filter work showing random RF noise, unlike switching noise, causes noise floor modulation audible to -350dB; linear PSUs pass it, SMPS and batteries do not. Head-Fi, Oct 2021
• 2021-11-09Watts says the cited plots were not made on an APx555 and shows a 2.5 V versus no-output overlay with the noise floor at -176 dB, independent of signal. Head-Fi, Nov 2021
• 2021-11-09Watts says the poster's plots are not noise floor modulation tests, which overlay 0 dBFS and -999 dBFS with identical APx555 settings; only the APx555's quad ADC can do this. Head-Fi, Nov 2021
• 2022-01-31Watts names three issues separating his DACs from others: small-signal accuracy, important for depth and detail, and noise floor modulation, which other DACs show heavily; he begins defining it. Darko.Audio 6:56
• 2022-01-31Watts defines noise floor modulation as noise rising with signal level, says the brain cannot separate it from the instruments, and says it makes the sound artificially bright. Darko.Audio 7:45
• 2022-01-31Watts says noise floor modulation impairs instrument separation and timbre; his DACs including Mojo 2 and Mojo show none measurably, unlike all others; the third factor is transients. Darko.Audio 8:26
• 2022-01-31Watts says conversion uses discrete flip-flops, references and resistors, separating digital noise from the analog, giving very low distortion, and is the only way to eliminate measurable noise floor modulation. Darko.Audio 12:12
• 2022-01-31Watts says that test sounds harder, brighter and less deep despite flat response, because floating-point arithmetic creates very low but audible noise floor modulation, making conventional DSP sound bright. Darko.Audio 20:36
• 2022-01-31Watts says measuring with an ADC directly is useless because its noise floor modulation exceeds his DACs'; the APx555 instead notches the signal, limited to a single tone. Head-Fi, Jan 2022
• 2022-01-31Watts presents Mojo 2: coupling capacitor removed, new noise shaper, and a UHD DSP EQ he says passes -301dB amplitude, phase and NFM transparency tests that 64-bit DSP fails. Head-Fi, Jan 2022
• 2022-02-09Watts says switching noise from DC-DC converters is easily filtered while random mains RF passes through linear supplies, so power banks are fine and linear PSU warmth is NFM. Head-Fi, Feb 2022
• 2022-02-12Watts says SMPS tones do not cause NFM unlike random RF, recalls switching regulators beating linear in Hugo, and prefers power banks for isolation from RF ground currents. Head-Fi, Feb 2022
• 2022-03-14Watts says ADCs have large NFM, so measuring it needs a distortion-free analog notch before the ADC, as the APx555 does; Stereophile's measurements do not do this. Head-Fi, Mar 2022
• 2022-03-19Watts says small fixed noise is harmless, like concert-hall air conditioning, but noise rising with the music is noise floor modulation, which the brain cannot separate from the instrument. MFG Audio Reviews 5:54
• 2022-03-19Watts says the brain fuses modulated noise with the instrument, so a warm saxophone sounds bright and hard, and noise from two instruments mixes and confuses the brain. MFG Audio Reviews 6:45
• 2022-03-19Watts says noise floor modulation blurs sounds together and makes warm things bright, and is audible well below measurable levels, so he publishes claims about measurable noise floor modulation. MFG Audio Reviews 7:30
• 2022-03-19Watts says his products have no measurable noise floor modulation but some unmeasurable amount that is still audible, and his research found the subjective limit much smaller than expected. MFG Audio Reviews 8:20
• 2022-03-19Watts says the FPGA is not the DAC: discrete flip-flops and resistors do the conversion, avoiding the substrate noise in silicon DAC chips that causes noise floor modulation. MFG Audio Reviews 13:14
• 2022-03-19Watts says floating point DSP inherently creates noise floor modulation, about -300 dB for 64-bit, which is audible; he has heard noise floor modulation at -400 dB. MFG Audio Reviews 34:48
• 2022-03-19Watts shows no noise floor modulation (same -177 dB floor with signal on or off), low distortion, and a perfectly reproduced 16-bit small signal with left and right superimposed. MFG Audio Reviews 45:02
• 2022-03-19Watts calls DSD a nightmare: 10 microseconds of phase shift errors, noise floor modulation from 130 dB to 115 dB at DSD64, and too few pulses, so it sounds soft. MFG Audio Reviews 53:59
• 2022-03-21Watts warns about placebo in listening tests and explains the HF filter reduces noise floor modulation while WTA2 (16FS to 256FS) improves transient timing, neither changing in-band response. Head-Fi, Mar 2022
• 2022-03-28Watts says USB cannot outperform optical apart from optical's 192k limit, and aims for future DACs immune to source and PSU RF, which cause noise floor modulation. Head-Fi, Mar 2022
• 2022-03-29Watts explains why valve output or input stages would worsen noise floor modulation and RF decoupling, recounts a zero ohm resistor problem, and attributes valve warmth to distortions. Head-Fi, Mar 2022
• 2022-05-28Watts attributes a 1 kHz measurement artifact to APx555 ground currents over USB, discusses ground loops and optical as reference, and gives settings for noise floor modulation tests. Head-Fi, May 2022
• 2022-09-07Watts says unidentified blind tests kill sensitivity, DAVE has no measurable noise floor modulation, and explains setting up the AP test correctly with -999dB and fixed input levels. Head-Fi, Sep 2022
• 2022-09-14Watts recounts RF filters warming the sound since the 1980s, and says a null result with Mojo 2's flat UHD DSP HF filter shows the effect is noise floor modulation. Head-Fi, Sep 2022
• 2022-09-18Watts says the filters remove noise above 20 kHz; old 96 and 192 kHz recordings have noise above 24 kHz, and filtering sounds warmer with less noise floor modulation. The Source AV 38:10
• 2022-10-12Watts says noise floor modulation is controlled by isolating the digital and analog domains and by the conversion architecture, the pulse array inside DAVE. Upscale Audio 2:50
• 2022-10-12Watts explains that in a normal DAC component switching depends on the signal, each switch adds a noise glitch, and the changing noise produces noise floor modulation. Upscale Audio 3:28
• 2023-02-04Watts says noise floor modulation affects timbre perception, and that emotional involvement depends on variation, in space and in timbre, such as bright and dark instruments heard together. Dusty Audio 29:12
• 2023-02-04Watts says conventional DACs measure about a -160 dB noise floor with a -60 dB signal but the noise rises with signal level; the aim is a static noise floor. Dusty Audio 31:39
• 2023-02-04Watts says noise that pumps with the signal cannot be separated by the brain from the instrument, so a saxophone with modulated noise sounds brighter. Dusty Audio 32:28
• 2023-02-04Watts says with two instruments the brain cannot assign modulated noise to either, blurring them and degrading instrument separation and focus. Dusty Audio 33:06
• 2023-02-04Watts says Chord DACs from Mojo 2 to Dave have no measurable noise floor modulation, all others do, and that any amount, however small, is subjectively important. Dusty Audio 33:58
• 2023-02-04Watts says noise floor changes at -350 dB still impair instrument separation and timbre; they show only in digital simulation, never with real test equipment, yet remain audible. Dusty Audio 34:43
• 2023-02-05Watts says improving distortion at 10 MHz in the digital domain has a big subjective impact, down to aliasing and noise floor modulation, and he is working on transient accuracy. Soundnews 23:46
• 2023-02-05Watts says regulator PWM intermodulates into discrete harmonics outside the audio band, but random RF noise creates in-band products and noise floor modulation; switch-mode supplies' filters help. Soundnews 30:41
• 2023-02-05Watts says in a power amplifier the ripple grows with output, so noise from 50 Hz to 10 kHz pumps with the music signal, which is very audible. Soundnews 34:22
• 2023-03-26Watts says random RF intermodulates in analog electronics into noise down to DC that pumps with the signal, while a regular 500 kHz component's products stay outside the audio band. Eric's Hi-Fi 16:50
• 2023-03-26Watts says he hears -350 dB errors found in digital simulation that analog test gear cannot measure, and claims no measurable noise floor modulation in any of his DACs. Eric's Hi-Fi 31:00
• 2023-03-26Watts says residual unmeasurable noise floor modulation is audible as warmth and smoothness changes, explaining power supply differences and why USB cables or optical versus USB sound different. Eric's Hi-Fi 32:01
• 2023-03-28Watts says DSD's 1-bit limit denies the noise shaper the overload it needs, causing internal overloads or instability and hence noise floor modulation. The Headphone Show 20:39
• 2023-03-28Watts says DSD64 has 115 to 125 dB of noise floor modulation, though he hears it at -350 dB; DSD cannot dither its quantizer and has linearity problems. The Headphone Show 21:26
• 2023-03-28Watts says unitary DACs have signal-dependent switching activity causing noise floor modulation, noise shapers limited to 64 or 128FS (6 MHz), and transient resolution of hundreds of nanoseconds. The Headphone Show 28:14
• 2023-03-28Watts says noise floor modulation causes grainy treble or reduced smoothness and timbre variation even below measurable levels; pulse array, invented in 1995, aimed to match DSD256 without its problems. The Headphone Show 44:50
• 2023-05-27Watts says reducing RF noise lowers perceived volume by one or two dB, since noise floor modulation sounds brighter, and is convinced a linear supply on TT2 adds RF noise. Head-Fi, May 2023
• 2023-07-30Watts says 64-bit floating point creates noise floor modulation, unlike his noise-shaped fixed point in Mojo 2's UHD DSP, and suggests Class H for efficient amplification. Head-Fi, Jul 2023
• 2023-08-30Watts says DAVE has RF filters on its outputs and a pre-amp adds random RF noise and noise floor modulation, so the most transparent pre-amp is none. Head-Fi, Aug 2023
• 2023-10-01Watts says 64-bit floating-point EQ has noise floor modulation around -300 dB that is audible as brightness, and proposes a +1 dB EQ with -1 dB volume test. The Source AV 10:40
• 2023-11-20Watts says 64-bit floating point creates audible noise floor modulation, and Mojo 2's EQ runs at 16FS with noise-shaped nodes in 104 bits. Head-Fi, Nov 2023
• 2024-02-15Watts says NFM sounds etched and brighter but degrades focus, the DPLL handles source PSU noise via optical, and battery plus optical on the M Scaler removes the ground loop. Head-Fi, Feb 2024
• 2024-03-20Watts explains that random RF noise intermodulates with the audio signal, producing random in-band noise that pumps up and down with the signal, which he calls noise floor modulation. MFG Audio Reviews 8:15
• 2024-03-20Watts says random RF noise must be removed, while fixed-frequency switching noise matters less because its products sit around the carrier; with his filters fitted his system always sounded good. MFG Audio Reviews 9:00
• 2024-03-20Watts finishes noting random RF noise was eliminated and turns to the subjective consequences of noise floor modulation, starting a rich-sounding saxophone example. MFG Audio Reviews 9:45
• 2024-03-20Watts says noise floor modulation adds noise that tracks the saxophone, which the brain cannot separate from it, so the saxophone's timbre sounds brighter than it should. MFG Audio Reviews 10:02
• 2024-03-20Watts says with saxophone and piano together noise floor modulation adds noise tracking each instrument plus a mixed product, confusing the brain's job of separating sounds into entities. MFG Audio Reviews 10:47
• 2024-03-20Watts says the mixed noise degrades instrument separation and focus, defining focus as a quiet instrument keeping its presence when a loud instrument comes in. MFG Audio Reviews 11:36
• 2024-03-20Watts says with noise floor modulation quiet instruments are swamped by loud ones; with none, quiet instruments stay audible, as with live jazz without a PA. MFG Audio Reviews 12:30
• 2024-03-20Watts says reducing random RF noise improves separation and focus and sounds quite a lot warmer and softer; some prefer noise floor modulation because it sounds etched and artificially enhanced. MFG Audio Reviews 13:01
• 2024-03-20Watts says a live cello images slightly rounded, while noise floor modulation artificially sharpens imaging and sounds crystalline and brighter, mistaken for transparency; removing it allows louder listening. MFG Audio Reviews 13:57
• 2024-03-20Watts says removing noise floor modulation keeps quiet instruments in focus, giving real transparency, and that it causes listening fatigue after about half an hour as the brain struggles. MFG Audio Reviews 14:45
• 2024-03-20Watts begins asking how much noise floor modulation can be heard; the sentence is cut off in the transcript. MFG Audio Reviews 15:25
• 2024-03-20Watts says however small noise floor modulation is, it can still be heard. MFG Audio Reviews 15:34
• 2024-03-20Watts says noise floor modulation hundreds of dB down is audible, shown by simulating errors digitally and listening, which is why cables can change brightness with zero measurable difference. MFG Audio Reviews 15:45
• 2024-03-20Watts says he does not know how impossibly small errors are perceived, and guesses the brain uses an autocorrelation function that over long periods resolves infinitely small errors in noise. MFG Audio Reviews 16:30
• 2024-03-20Watts says ground planes have finite impedance at GHz, so RF current creates voltage drops that cause intermodulation and noise floor modulation; this current flows in a mains-source-DAC loop. MFG Audio Reviews 18:03
• 2024-03-20Watts summarizes that linear supplies create distortion, lack RF filters (so noise floor modulation, measurable or not), add signal-related hum harmonics, and toroidal transformers buzz and radiate signal-dependent magnetic fields. MFG Audio Reviews 37:31
• 2024-04-15Watts shares CanJam slides: optical sounds warmer from less noise floor modulation, USB's etched sound is mistaken for transparency, and with isolated sources USB and BNC can match optical. Head-Fi, Apr 2024
• 2024-05-10Watts defines noise floor modulation as noise pumping with the signal, says it sounds artificially bright yet seems more transparent, is audible below measurable limits, and explains why digital cables sound different. Reactcore 0
• 2024-07-12Watts defines noise floor modulation as noise rising with signal and says noise pumping with a saxophone sounds bright, like off-channel FM hiss. HIFICLUB 32:03
• 2024-07-12Watts says the brain cannot separate the modulated noise from the instrument, so a saxophone sounds too bright, and two instruments' noise modulation mixes together. HIFICLUB 32:49
• 2024-07-12Watts says noise floor modulation degrades timbre and the brain's ability to separate instruments, making the brain work harder. HIFICLUB 33:42
• 2024-07-12Watts says all DACs create noise floor modulation; the pulse array was invented to improve small-signal resolution and eliminate measurable noise floor modulation, though some unmeasurable remains. HIFICLUB 34:29
• 2024-07-12Watts says reducing remaining unmeasurable noise floor modulation has been a struggle on the Ultima DAC, to launch maybe next year or the year after. HIFICLUB 35:12
• 2024-07-12Watts says DSD's switching activity varies with signal, causing noise floor modulation, and its 64 times noise shapers reach only 130 dB, so minus 301 dB is impossible. HIFICLUB 39:59
• 2024-07-12Watts says he has found no lower limit to audible noise floor modulation, below minus 300 dB, so making DACs independent of source and mains takes much more work. HIFICLUB 48:04
• 2024-10-25Watts says balanced DACs still use ground-referenced I to V converters, balanced complexity raises noise floor modulation, and common mode rejection fails at RF. Head-Fi, Oct 2024
• 2024-11-30Watts says substrate noise matters less in op-amps than DAC chips, and explains noise floor modulation as RF intermodulation in the differential input stage, reduced by local decoupling. Head-Fi, Nov 2024
• 2025-02-17Watts says transparent EQ needs at least '108 bits, at 104 bits' and aggressive noise shaping, not floating point, which creates audible noise floor modulation heard as unnatural brightness. audio-2G 15:44
• 2025-02-18Watts says the HF filter fails a null test but not perceptually, citing a decimation-filter test showing its sound change comes from less out-of-band noise and noise floor modulation. Head-Fi, Feb 2025
• 2025-03-23Watts says pre-Mojo 2 HF filters were second order with about -0.9dB at 20kHz, while Mojo 2's UHD DSP allows an aggressive filter with no drop, reducing noise floor modulation. Head-Fi, Mar 2025
• 2026-03-17Watts says floating point creates noise floor modulation, which is perceivable however small, using 'perceivable' rather than 'audible' since the brain uses it to make sense of music. ecoustics 48:39
• 2026-03-17Watts says random RF intermodulates with the music in analog circuits, so the noise floor pumps up and down with the signal, and flat random noise sounds bright. ecoustics 58:44
• 2026-03-17Watts says noise tracking a saxophone makes it sound bright instead of rich and smooth, because the brain cannot separate the distortion from the instrument's timbre. ecoustics 59:38
• 2026-03-17Watts says with piano and saxophone together, intermodulation produces noise mixing both instruments, making it harder for the brain to separate them. ecoustics 1:00:25
• 2026-03-17Watts says reducing RF noise improves timbre and instrument separation, and that noise floor modulation however small damages separation because the brain uses correlation. ecoustics 1:01:20
• 2026-03-17Watts says analog components see that noise, creating noise floor modulation that makes the sound harder and brighter. ecoustics 1:08:06
• 2026-03-17Watts says in listening tests the mains supply sounded better than a 24-volt lithium-ion battery, so he repeated the test carefully, focusing on noise floor modulation. ecoustics 1:11:12
• 2026-03-24Watts says noise floor modulation, noise pumping with the signal and caused by random RF noise, is subjectively huge: rich, dark instruments become brighter, so RF must be filtered. Audio Force 24:57
• 2026-05-03Watts says mains power brings RF noise, which creates noise floor modulation, noise rising with signal; this bright white noise pumping with a saxophone cannot be separated by the brain. Audio Force 10:37
• 2026-05-03Watts says the brain compensates for harmonic distortion the ear itself creates, but noise floor modulation is not harmonically related: 20 Hz to 20 kHz noise changes with signal, brightening instruments. Audio Force 12:18
• 2026-05-03Watts says noise floor modulation makes a saxophone brighter, and a saxophone and piano together create joint modulation noise, so the brain cannot separate the two instruments. Audio Force 13:05
• 2026-05-03Watts says noise floor modulation degrades instrument separation and causes listening fatigue, and is perceivable however small, even when impossible to measure. Audio Force 13:50
• 2026-05-03Watts says Quartet (transcribed 'Cortax') has no measurable noise floor modulation while all other DACs, at any price, do, so separation and timbre suffer less with Chord DACs. Audio Force 14:34
• 2026-10-02Watts says these chip mechanisms create signal-correlated noise and thus noise floor modulation, while a discrete DAC fed pulses from an FPGA avoids substrate noise and uses linear parts. HiFi Buys 31:32
• 2026-10-02Watts says R2R glitch energy makes noise vary with the signal, and some delta-sigma DACs share this because their low-bit conversion creates signal-dependent noise. HiFi Buys 40:29
• 2026-10-02Watts says real improvement makes bright instruments brighter and dark ones darker, while noise floor modulation mistaken for transparency makes every instrument brighter, revealing a distortion. HiFi Buys 48:54
• 2026-10-02Watts says it is when noise varies with the signal that it becomes a major issue. HiFi Buys 52:21
• 2026-10-02Watts explains that ideally the noise floor stays static as the signal rises and falls, but in digital it rises and falls with the signal: noise floor modulation. HiFi Buys 52:42
• 2026-10-02Watts says NFM shows when the signal is switched on and off at fixed gain, Audio Precision autoscaling off; high-performance DACs show 2 to 3 dB. HiFi Buys 53:19
• 2026-10-02Watts says a couple of dB change in a background noise that is inaudible on sensible headphones cannot be heard as noise, yet it makes a big difference. HiFi Buys 54:03
• 2026-10-02Watts says hearing is a construct the brain creates, separating instruments, placing them in space and perceiving timbre, which is why noise floor modulation matters. HiFi Buys 54:29
• 2026-10-02Watts says background hiss is bright, so noise modulating with a saxophone is perceived as extra brightness, making a rich, smooth saxophone sound brighter. HiFi Buys 55:16
• 2026-10-02Watts says a saxophone and trumpet each create their own noise floor modulation, so the noise floor becomes a hybrid of both that confuses the brain. HiFi Buys 55:56
• 2026-10-02Watts says this noise floor modulation degrades the ability to discriminate the saxophone and trumpet as separate entities. HiFi Buys 56:43
• 2026-10-02Watts says NFM of a couple of dB on a -160 dB noise floor affects timbre and instrument separation; in listening tests you cannot hear it but you perceive it. HiFi Buys 56:59
• 2026-10-02Watts says he has perceived noise floor modulation at -350 dB, impossible to hear, as degraded instrument separation or too-bright timbre. HiFi Buys 57:34
• 2026-10-02Watts says a saxophone with its own microphone and amplifier is fine, but once instruments are mixed together through a PA the quality is lost. needs context HiFi Buys 1:06:30
• 2026-10-02Watts says a chip DAC's bandgap reference, decoupled by a capacitor, pumps with current draw, carries RF noise and modulates dynamically, unseen in static measurements. HiFi Buys 1:10:51
• 2026-10-02Watts says in R2R and DSD DACs switching activity depends on the music, creating signal-dependent noise and NFM, so conversion needs constant switching activity. HiFi Buys 1:11:40
• 2026-10-02Watts says the pulse array, an array of PWM signals, has constant glitch energy: half the elements on changes output noise by only half a dB, unlike R2R or delta-sigma. HiFi Buys 1:12:47
• 2026-10-02Watts says other DACs show a huge change in measured noise, there are many ways to create noise floor modulation, and eliminating it is a major headache. HiFi Buys 1:13:34
• 2026-10-02Watts says DSD sounds natural as distortion vanishes into noise, but he wanted its benefits without its problems: a faster noise shaper and no signal-dependent switching causing NFM. HiFi Buys 1:42:15
• 2026-10-02Watts says an array of PWM schemes, rising edges canceling falling edges, made switching signal-independent, cut jitter sensitivity and let the noise shaper run at 104 MHz. HiFi Buys 1:43:13
• 2026-10-11Watts says his direct interpolation filter avoids this; 64-bit floating point creates small but audible noise floor modulation, compounded by FFT and IFFT, requiring much higher precision. Distorsiuni Interioare 36:27