ASIC Miner Silencers: What the Physics Says, What We Actually Claim

Search for an ASIC miner silencer and you will find products promising to make a 75-plus-decibel machine apartment-friendly, with confidence no acoustics lab would sign. We sell silencers too — 3D-printed expansion-chamber mufflers designed in Montreal, Quebec — and here is the first thing we will tell you about them: we have never measured their acoustic performance, so we publish no dB figure at all. What we have is a textbook transmission-loss model and a candid account of what it predicts, including the unflattering parts. This post explains where miner noise comes from, what a duct silencer physically can and cannot do, and how to pick from our silencer line without believing anything we have not earned the right to claim.

Where miner noise actually comes from

A miner is loud for one reason: it moves a huge volume of air through a tight, hot box using small fans at high speed. That noise reaches your ear through three mechanisms, each with different fixes.

The blade-pass tone. Every time a fan blade sweeps past the strut or grille behind it, it launches a pressure pulse. The pulses repeat at the blade-pass frequency: shaft speed in revolutions per second times blade count — a seven-blade fan at 6,000 RPM puts its fundamental at 700 Hz. That hard, whining tone is narrow-band: its energy concentrates at one frequency and its harmonics, which matters enormously for silencer design.

Broadband turbulence. The whoosh underneath the whine: air separating off blade tips, scrubbing heatsink fins, tumbling through the chassis, smeared across the whole spectrum. It rises steeply with airflow velocity, which is why a miner at a reduced power target is dramatically quieter. No single-frequency trick touches broadband noise; you slow the air, absorb the sound, or move it somewhere it does not matter.

Structure-borne vibration. The miner shakes, the shelf shakes, the floor shakes, and a room two walls away hums. This path bypasses the air entirely, so no duct-mounted device can address it — isolation feet and a heavy, decoupled surface are the tools. Our ASIC noise reduction guide covers all three paths and how to diagnose which dominates your installation.

What a duct silencer can and cannot do

In-duct silencers come in three textbook flavours, and understanding them tells you what any vendor’s product — ours included — can plausibly deliver.

The reactive expansion chamber

The classic car-muffler principle. Sound travelling down a duct hits a sudden enlargement in cross-section, and the impedance mismatch reflects part of the acoustic energy back toward the source. Attenuation is governed by the area ratio between chamber and duct — the bigger the step up and down, the more energy reflected — and varies cyclically with frequency depending on chamber length. Two consequences follow from the math: a chamber only slightly larger than its duct does almost nothing, and even a good chamber is a moderate, frequency-dependent attenuator, not a wall. Automotive mufflers get big reductions by cascading chambers and adding absorption, at a back-pressure cost a mining fan cannot tolerate.

The absorptive silencer

Line a duct with fibrous or open-cell material and broadband high-frequency energy converts to heat in the pores. This is how commercial HVAC silencers work and it is the right tool for the turbulent whoosh — but it needs physical length, does little at low frequencies, and the lining sits in a dusty, warm airstream where it loads with debris. Our printed silencers are purely reactive; a single printed body has no lining. Because our designs are open source, you are free to add absorptive material yourself — a legitimate modification, and one reason we publish the files.

The quarter-wave notch

A closed side branch of length L cancels sound at the frequency whose quarter wavelength equals L — a deep, narrow notch, with almost no effect elsewhere. It is the precision instrument of the family: nearly useless against broadband noise, potentially very effective against a single tone like the blade-pass whine. The catch is tuning. The branch’s effective acoustic length is longer than its physical bore, because air just outside the mouth participates in the resonance — the “end correction” — and getting this wrong shifts the notch off the target tone, where it does nothing.

“Modelled, not measured”: why we publish no dB number

Now the uncomfortable part, stated plainly. Our silencers have never been on an acoustic test bench — no microphone, no before-and-after measurement of a running miner. Every acoustic statement we make comes from the idealised transmission-loss model of the geometry, and an idealised model ignores flow noise, wall transmission, and everything else separating a formula from a room with a miner in it.

Here is what the model says, published as engineering output and explicitly not as a marketing claim:

• The 6-inch expansion chamber computes to roughly 1–4 dB of idealised transmission loss, varying with frequency.
• The 8-inch chamber computes to roughly 0.14 dB — effectively nothing, because an 8-inch duct on the same plenum leaves almost no area mismatch to reflect against. We could have buried that number; it is on the product page instead.
• The v2 quarter-wave branch targets a nominal 750 Hz, shifted to roughly 653–675 Hz by the mouth end-correction in the model. Our first tuning was wrong: the v1 “752 Hz” figure was computed from the bore alone, and the as-built v1 column actually resonated around 460–510 Hz. We found the error, retuned, and said so. Final tuning waits on a measured blade-pass tone from fans in service — until then, the notch may sit off-target.

If those numbers sound modest, that is because honest numbers for a compact reactive silencer are modest. A single chamber that fits a printer bed and passes an ASIC’s full airflow was never going to deliver hotel-room silence, and any acoustics textbook will tell you so.

The claims problem in this market

Shop this category and you will see noise-reduction percentages, dramatic dB deltas, and “silent mining” language on products in the same physical class as ours. We will not name anyone. But a percentage noise reduction is close to meaningless without the measurement distance, frequency weighting, load condition, and before/after methodology — and those details are almost never published. Decibels are logarithmic; a big percentage could describe several very different measurements or none at all. A big number with no methodology is copywriting, not acoustics. Our answer is not a bigger number: it is a public bench-measurements page where our claims and their provenance live in the open, and product pages that print their least flattering model outputs.

Our line, and the honest pick-list

The silencer family covers the common fan configurations — all single-piece ASA prints, all reactive, and every caveat above applies to each.

The line embodies a real trade-off. A 6-inch exit restricts more but creates the area mismatch that gives the chamber whatever attenuation it has; an 8-inch exit breathes better and reflects almost nothing. That is why we describe the 8-inch v2 as flow-first with a tonal notch, not as a quieter product — its chamber contributes ~0.14 dB in the model, so essentially all its acoustic work happens at one frequency, in the branch.

So the honest recommendations are simple. If noise is your priority: the 6-inch v2 — and if your existing run is 8-inch duct, feed it through the 8-to-6 reducer rather than buying the 8-inch v2. That keeps the meaningful expansion ratio and the tuned branch. If airflow is your priority — long runs, shared plenums, hot rooms — take the 8-inch v2, knowing you are buying flow with a single-frequency notch. If you want a plain exhaust connection with no acoustic pretensions, the standard dual-120 shroud exists for exactly that; match duct diameter to your run with the ventilation guide. And because the designs are open source, none of this is a walled garden: print them, modify them, sell them if you like. Publishing the files alongside the caveats is what an engineering company does.

What actually silences a miner most

A silencer is one layer of a system, and not the biggest one. The largest noise reduction available to almost every home miner is free: firmware power and fan tuning. Fan noise climbs brutally with speed, so a lower power target does more than any printed part we sell — see where machines and power modes land in our miner noise levels comparison. After that: a better room; ducting the exhaust out of the living space (what shrouds are for — moving the noise source matters more than muffling it); vibration isolation; and only then a silencer to shave the duct path. Our quiet home hashcenter build walks that stack in order, and the miner noise planner sets expectations for your room and distance before you spend anything.

Measure before you order

No D-Central shroud or silencer carries a guaranteed-fit claim, deliberately: our interface dimensions are modelled against reference CAD and gaskets, and the industry disagrees with itself — the dual-120 screw pattern is documented at both 105.0 and 105.4 mm, fan pitch at both 120.0 and 121.0 mm. Our slotted holes span those disputes, but the machine on your shelf outranks every drawing. Verify fan size and count, screw pattern, fan pitch, and true duct diameter before ordering; full interface dimensions are on each product page and in the open files at /3d-models/.

FAQ

How many dB will a D-Central silencer cut from my miner?

We do not know, and we will not pretend to. The idealised model computes roughly 1–4 dB for the 6-inch chamber and ~0.14 dB for the 8-inch, plus a narrow notch on v2 parts; no measurement has been made. When we have bench data it will be published at /bench-measurements/.

Is the 8-inch silencer quieter than the 6-inch?

No — by the physics, the opposite. The 8-inch chamber’s area mismatch is so small its idealised contribution is ~0.14 dB. The 8-inch v2 exists for people who need 8-inch airflow with a tonal notch on top. For noise, take the 6-inch v2 with the 8-to-6 reducer.

Why not just line the silencer with foam?

Absorptive lining genuinely works against broadband noise, but a lined silencer needs a retained, serviceable liner that tolerates dust and heat, and a single-piece printed reactive chamber has no provision for one. The files are open — if you experiment with lining the chamber, we would honestly like to hear your measurements.

Will the v2 notch kill my miner’s whine?

Only if the notch lands on your fans’ actual blade-pass frequency. It is tuned to a nominal 750 Hz (~653–675 Hz with end-correction, per model), and final tuning awaits a measured blade-pass tone. If your fans run at a different speed, the tone moves and the notch may miss it.

D-Central.TECH Universal ASIC Shroud: Dual 120mm Fans to 8" – High-Efficiency Cooling for Antminer, Avalon, Innosilicon | Reduce Noise, Extend Lifespan, Boost Hash Rates $29.99 CAD

Shop Shrouds →