• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

A List of Low Noise Amplifiers

Tuocsteem

Member
Joined
Sep 17, 2020
Messages
19
Likes
32
I've been struggling with finding a desktop, near-field amplifier that has a low enough noise floor that I can't hear constant hissing from my speakers. In the chance that anyone else has the same problem, here's a collection of amplifiers that are probably quiet enough to be inaudible and some related notes.

The results of my experience with 86dB sensitivity speakers on a desk are as follows: 347uV is incredibly distracting, can hear it clear across the room. 70uV is quiet but still always noticeable. 16uV is approaching inaudible at the listening position- I can clearly hear it with my ear against the tweeter but can only follow the hiss ~1.5 feet before I lose it. I have never had an amplifier that is truly inaudible in a quiet room but it stands to reason that <4uV would be approaching "can't be heard without tool assistance" levels of performance. Therefore my arbitrary determination for this list is that <50uV is worth mentioning and <20 uV is good.

If there are any additional amplifiers that fit these noise requirements please do let me know. I probably won't continually update this post but I might make a spreadsheet if there is enough new info.


Benchmark AHB2 (Power Amp)- 7.1 uVrms A-wtd
  • Has low gain mode (9.2 dB Low Gain, 17.0 dB Mid Gain, 23 dB High Gain)
  • No volume control
  • Not compact

Hypex Nilai500DIY (Amplifier Module)- 27 uVrms un-wtd (High gain), 15 uVrms un-wtd (Mid gain), 9 uVrms un-wtd (Low gain)
  • Has low gain mode (Low Gain 11.86 dB, Mid Gain 21.75 dB, High Gain 27.88dB)
Hypex NC500 OEM (Amplifier Module)- 10 uVrms un-wtd (Unbuffered)
  • Without buffer has low gain of 12.4 dB. Buffer implementation depends on assembler, some have switchable gain.
Hypex NCx500 OEM (Amplifier Module)- 20 uVrms un-wtd (Buffered), 7 uVrms un-wtd (Unbuffered)
  • Without buffer has low gain of 11.5 dB, otherwise 26.8 dB with stock buffer.

Neurochrome Modulus-86 (DIY Power Amp)- 13.7 uVrms A-wtd, 17.2 uVrms un-wtd
  • No finished units for sale. Requires significant assembly including acquiring a case and soldering components.
  • Release spec was 33 uVrms A-wtd, 42 uVrms un-wtd and later changed to new spec. I'm not sure if there's any documentation of the improvements made or third party measurements of the new performance. Source
  • Compact

Purifi 1ET400 (Amplifier Module)- 11.5 uVrms A-wtd (Unbuffered)
  • Without buffer has low gain of 12.8 dB. Buffer implementation depends on assembler, some have switchable gain.
  • Not compact
Purifi 1ET7040SA (Amplifier Module)- 14 uVrms A-wtd (Unbuffered)
  • Without buffer has low gain of 13 dB. Buffer implementation depends on assembler, some have switchable gain.
  • Not compact

Schiit Rekkr (Power Amp)- 4 uVrms A-wtd
  • Manufacturer does NOT provide residual noise spec, the value is computed from SNR and output specs.
  • Fixed low gain of 12 dB
  • Extremely low output (3Wx2 at 4 Ohms)
  • No volume control
  • Compact
Schiit Gjallarhorn (Power Amp) - 12.6 uVrms un-wtd
  • Manufacturer does NOT provide residual noise spec, the value is computed from SNR and output specs.
  • No volume control
  • Compact
Schiit Vidar 2 (Power Amp)- 20.0 uVrms un-wtd
  • Manufacturer does NOT provide residual noise spec, the value is computed from SNR and output specs.
  • No volume control
  • Not compact

Soncoz SPG1 (Integrated Amp)- 9.8 uVrms A-wtd
  • Manufacturer does NOT provide residual noise spec, the value is computed from SNR and output specs.
  • Not compact

Topping LA90/Topping LA90D (Integrated Amp)- 3.4 uVrms A-wtd (high gain), 1.8 uVrms A-wtd (low gain)
  • Has low gain mode (9.4 dB Low Gain, 19.4dB High Gain).
  • Original LA90 is discontinued.
  • Compact
Topping PA5, PA5 II, PA5 II Plus (Integrated Amplifiers)- 16 uVrms A-wtd
  • Potentially serious reliability issues.
  • Original PA5 is discontinued.
  • Compact
Topping PA7, PA7 Plus (Integrated Amplifiers)- 17 uVrms A-wtd
  • Compact
Topping RA3 (Integrated Amp)- 16 uVrms A-wtd
  • No US-based reseller at time of posting, only direct from AliExpress.
  • Not compact

Honorable Mentions:

Hypex NC2K OEM (Amplifier Module)- 20 uVrms un-wtd
  • Has fixed low gain of 20 dB.
Hypex NC1200 OEM (Amplifier Module)- 28 uV un-wtd (buffered), 8 uV un-wtd (unbuffered)
  • Without buffer has low gain of 11.6 dB.
Hypex NC52MP (Amplifier Module)- 35 uV un-wtd
  • Not sure if you can buy this anywhere
Hypex NC122MP, NC250MP, NC252MP, NC500MP (Amplifier Modules)- 40 uV un-wtd
  • NC502MP is slightly worse at 47 uV un-wtd.
  • Not compact
ICEpower 1200AS1, ICEpower 1200AS2 (Amplifier Modules)- 25 uVrms A-wtd
  • Not compact
ICEpower 2000AS1 HV, ICEpower 2000AS2 HV- 38 uVrms A-wtd
  • Not compact
Neurochrome Modulus-286 (DIY Power Amp)- 28.5 uVrms un-wtd, 23.0 uVrms A-wtd (High gain). 22.0 uVrms un-wtd, 17.0 uVrms A-wtd (Low gain)
  • Has sort of low gain mode (20 dB Low gain, 26 dB High gain) with resistor swap during assembly.
  • Compact
Neurochrome Modulus-686 (DIY Power Amp)- 25.0 uVrms un-wtd, 20.0 uVrms A-wtd (High gain). 20.0 uVrms un-wtd, 16.0 uVrms A-wtd (Low gain)
  • Has sort of low gain mode (20 dB Low gain, 26 dB High gain) with resistor swap during assembly.
  • Not compact
Neurochrome LM3886 Done Right (DIY Power Amp)- 29 uVrms A-wtd, 38 uVrms un-wtd
  • Compact
Schiit Aegir (Power Amp)- 31.8 uVrms A-wtd
  • No volume control
  • Not compact
Yamaha A-S2200, A-S3200 (Integrated Amplifiers)- 33 uVrms A-wtd
  • Totally different price category and market. If you are considering one of these you most likely do not need this list.
Yamaha A-S301, A-S501, A-S701, A-S801 (Integrated Amplifiers/Stereo Receivers)- 40 uVrms A-wtd
  • The only members of this list that function as Stereo Receivers. Some of the only modern devices with a bunch of varied inputs AND subwoofer out AND tone controls AND acceptable noise floor. Very safe bet if reliability is a concern.
  • Most specs assume CD input with PURE DIRECT or CD DIRECT AMP mode and SNR loss with tone controls on is not documented. DAC performance is unknown.
  • Not compact

Notes:
Most of this information is available elsewhere on this forum albeit scattered. I'm mainly compiling all of this for my own research and to keep the discussion alive. Credit to this thread for inspiration and particularly @staticV3 for listing a bunch of these already.

All of the info comes direct from manufacturers, whatever number is specified as maximum residual noise. Computed values assume SNR is relative to the highest stated total output.

There are many additional parameters that I find useful (multiple inputs, balanced architecture, amplifier class, idle power consumption etc.) but I'd rather not make this too encyclopedic. Purposely omitted are:
  • Distortion and power specs as I don't think they are relevant for most of the entries on this list.
  • Price as it is variable but suffice it to say only a handful of these devices are "affordable".
  • SNR spec as it is in relation to total power. Generally any device with meaningful output needs >110 dB SNR w.r.t peak output to have a low noise floor. High output devices likely need >120 dB SNR. Add another 10 dB to these requirements if you want noise inaudibility. This is also all relative to speaker sensitivity and listening distance.

Compact vs. Not compact is roughly equivalent to "is it smaller than a textbook?". This has nothing to do with topological spaces. It is important to me that a desktop amplifier actually fits on a desk.

Some of the DIY Hypex and Purifi amps could theoretically be made compact as monoblocks or with separate power supplies but I went with common implementations e.g. VTV to determine that they are all not compact. All of the units listed as "Amplifier Module" either require significant assembly or are only available from assemblers such as VTV, Buckeye, Nord etc. Most are designed as pure power amps with no integrated volume control. Actual performance depends on implementation and, in general, noise levels will be higher than listed here but still acceptable.

I omitted from this list anything that would be classified as vintage or long discontinued. Devices which overwhelmingly failed to make it on the list were 1) Basically every TI, Infineon etc. chip amp (excluding LM-3886), including those that power most desktop amps that are popular on this forum. Some of the ICs are theoretically capable of good noise performance but good implementations are few and far between and 2) Basically every AVR, although I did not search too thoroughly.
 
Last edited:
Benchmark AHB2 (Integrated Amp)- 7.1 uVrms A-wtd
  • Has low gain mode (9.2 dB Low Gain, 17.0 dB Mid Gain, 23 dB High Gain)
  • Not compact
It is a power amp, not an integrated amp.
By my scales, it is quite compact.
 
It is a power amp, not an integrated amp.
Fixed.
By my scales, it is quite compact.
I wish I had one on my desk to compare. Going just by published dimensions, it's about textbook sized and surprisingly tall. Accounting for clearance for cooling and whatnot it seems like it would require a dedicated place not just stuck under a computer monitor or in a mini components stand. The rackmount option is right out, of course.
 
Going just by published dimensions, it's about textbook sized and surprisingly tall.

It's tiny for a 'power' amplifier.

Consider @Kal Rubinson has seen a lot of power amps in his time...

Pictured: Old Krell Master Reference, Kal wheeled out the front for kerbside collection day... (kidding)
1693012393608.jpeg
 
Fixed.

I wish I had one on my desk to compare. Going just by published dimensions, it's about textbook sized and surprisingly tall. Accounting for clearance for cooling and whatnot it seems like it would require a dedicated place not just stuck under a computer monitor or in a mini components stand. The rackmount option is right out, of course.
I have three of them stacked up. Of course, I am comparing against standard power amps and have no experience with nor an interest in "desktop." When compared to amps that are capable of significant power suitable for floorstanding or standmounted full-range speakers, it is decidedly compact. Textbook sized? Sure.
 
Schiit Gjallarhorn (Power Amp) - 34.7 uVrms un-wtd
According to Schiit's Test Report, the Gjallahorn has this much noise:
-8Ω Stereo: 9.2μVrms (A-wt)
-4Ω Stereo: 10.3μVrms (A-wt)
-8Ω Mono: 12.4μVrms (A-wt)

Soncoz SPG1 (Integrated Amp)- 9.8 uVrms A-wtd

For some reason I've never heard of this amp nor seen it discussed but it seems to be real. No third party measurements available.
Here are third party measurements: https://www.l7audiolab.com/f/soncoz-sgp-1/
 
I could hear SMSL A300 hissing from the tweeter but I can't do so with Topping MX3s
 
Is there a way to calculate the noise in uV if the manufacturer does not directly specify? For example, I use JBL CSA-2120 which are listed as "Greater than 100 dB - 1kHz, A-weighted", presumably relative to the 1.4 Vrms nominal input.

JBL sells them for background music at the yoga studio. :D

Regardless, it's effectively silent at nearfield into anything less sensitive than an Altec compression driver, which needed a few dB of L-Pad.
 
It's tiny for a 'power' amplifier.

Consider @Kal Rubinson has seen a lot of power amps in his time...

Pictured: Old Krell Master Reference, Kal wheeled out the front for kerbside collection day... (kidding)
View attachment 307902
... and a mono-block no less. Do the lights dim at the local steel mill when it is switched on?
 
Is there a way to calculate the noise in uV if the manufacturer does not directly specify? For example, I use JBL CSA-2120 which are listed as "Greater than 100 dB - 1kHz, A-weighted", presumably relative to the 1.4 Vrms nominal input.

You can work back from the specified rated power, converted to volts over the specified load to come up with the residual noise floor in volts.

120W at 4 ohms is 21.91V RMS.
100dB is a factor of 100,000 times.
21.91V / 100,000 = 219uV which is 0.2mV.
 
Is there a way to calculate the noise in uV if the manufacturer does not directly specify? For example, I use JBL CSA-2120 which are listed as "Greater than 100 dB - 1kHz, A-weighted", presumably relative to the 1.4 Vrms nominal input.
Here are its specs:
Screenshot_20230826-070731_Drive.png

Rated 8Ω Power is 120W, which equals sqrt(120*8)=31Vrms.
Noise with A-wt is >100dB below that, so 31*10^(-100/20)=.00031

The CSA-2120 has <310μVrms (A-wt) of noise.
 
... and a mono-block no less. Do the lights dim at the local steel mill when it is switched on?

Kal rings the steel mill when he is finished listening, to allow them to fire up their furnaces.
 
According to Schiit's Test Report, the Gjallahorn has this much noise:
-8Ω Stereo: 9.2μVrms (A-wt)
-4Ω Stereo: 10.3μVrms (A-wt)
-8Ω Mono: 12.4μVrms (A-wt)


Here are third party measurements: https://www.l7audiolab.com/f/soncoz-sgp-1/

Fixed both issues and added the Vidar 2 and Aegir to their respective categories. Somehow I calculated the Gjallahorn's noise wrong by simulatenously using the wrong power spec and making a typo on the SNR. I assume the numbers come from the SNR on the APx555 report on the bottom of the product page? I just stuck to a worst case value computed from the published specs.
 
Is there a way to calculate the noise in uV if the manufacturer does not directly specify?

The formula I've been using is sqrt(P*R) * 10^(-SNR/20) * 1e6
With
  • P = Power in Watts from the published power spec
  • R = Resistance in Ohms from the published power spec
  • SNR = SNR at the output level of the power spec, usually full output
  • 1e6 scales from Volts to microVolts
This is equivalent to the answers from @restorer-john and @staticV3

There can be some ambiguity as to which power spec is the power spec. Unless specified, it's usually the pair of values with the highest wattage for stereo speakers (not mono, not combined output to the pair) which tends to be at 8 Ohm and maximum distortion.
 
Notice the SNR is stated specifically for rated 8Ω Power

Below rated 8 ohm power. We don't know what figure they are using. Vague, poorly written specifications from Harman are all we've come to expect these days.

1693031271487.png
 
Last edited:
Back
Top Bottom