Class A “class D killer” amplifier with THD less than -120dB

[HarmonicTHD]

Neurochrome Mod686. Up to 114dB SINAD and 126dB dynamic range.
Admittedly AB.

Modulus-686

Audiophile composite amplifier based on the LM3886. Just as the BPA200, it uses six LM3886 in bridge/parallel. It uses error correction to ensure the lowest distortion possible.

neurochrome.com

 

[sq225917]

Nice work pma, a useful little amp for sensitive speakers or smaller rooms.

Have you though about making a Headphone amp version of the circuit?

 

[antennaguru]

IME Class D amplifiers require inefficient loudspeakers. That way the idle noise is less audible.

I have compared various Class D modules on both 86 dB and 95 dB efficient loudspeakers, and the 9 dB higher efficiency loudspeakers means 8 times more idle noise coming out of the loudspeakers with no music.

Class D at idle is not that terribly efficient either, as there is a good amount of heat while playing nothing. No, not as much as with Class A, but not so little as to ignore thermal management - which many Class D modules unfortunately do take too lightly.

My personal preference is Class AB, where there is reasonable efficiency and I feel better performance across the whole audio spectrum.

 

[Shazb0t]

IME Class D amplifiers require inefficient loudspeakers. That way the idle noise is less audible.

I have compared various Class D modules on both 86 dB and 95 dB efficient loudspeakers, and the 9 dB higher efficiency loudspeakers means 8 times more idle noise coming out of the loudspeakers with no music.

Class D at idle is not that terribly efficient either, as there is a good amount of heat while playing nothing. No, not as much as with Class A, but not so little as to ignore thermal management - which many Class D modules unfortunately do take too lightly.

My personal preference is Class AB, where there is reasonable efficiency and I feel better performance across the whole audio spectrum.

Wow

 

[jasonhanjk]

Class D at idle is not that terribly efficient either, as there is a good amount of heat while playing nothing. No, not as much as with Class A, but not so little as to ignore thermal management - which many Class D modules unfortunately do take too lightly.

My personal preference is Class AB, where there is reasonable efficiency and I feel better performance across the whole audio spectrum.

Do buy from companies that design it properly and you'll get what you want.

 

[HarmonicTHD]

IME Class D amplifiers require inefficient loudspeakers. That way the idle noise is less audible.

I have compared various Class D modules on both 86 dB and 95 dB efficient loudspeakers, and the 9 dB higher efficiency loudspeakers means 8 times more idle noise coming out of the loudspeakers with no music.

Class D at idle is not that terribly efficient either, as there is a good amount of heat while playing nothing. No, not as much as with Class A, but not so little as to ignore thermal management - which many Class D modules unfortunately do take too lightly.

My personal preference is Class AB, where there is reasonable efficiency and I feel better performance across the whole audio spectrum.

You know modern, well engineered class D can be pretty low noise too (eg. Dyn Range 130dB, 11uV)?

1ET400A - PURIFI

Single-channel, analog-input Class D amplifier module Negligible THD and IMD Extraordinary low noise Load-invariant response Exceptionally clean clipping Low losses & high efficiency Easy to integrate Designed in Denmark KEY SPECIFICATIONS

purifi-audio.com

 

[DanielT]

Interesting, well done PMA! If you ever want to sell it, I'll be happy to buy it. Would be perfect to drive my 8 Ohm, sensitivity 109 dB, compression drivers.

Eighteen Sound - Professional loudspeakers

www.eighteensound.it

 

[pma]

My E1DA Cosmos ADC by @IVX

Cosmos ADC | index

e1dashz.wixsite.com

is on the way from France to Prague. I will post new measurements after it arrives, as my current setup is not good enough to make a reliable measurement of this ultra-low distortion amplifier .

 

[pma]

Some measurements of this amplifier with E1DA Cosmos ADC:

 

[Drengur]

Well done!

 

[Gorgonzola]

Distortion spectrum. Please again forget the THD+N number, +N comes from the measuring system and not from the amp. Unfortunately THD is a measuring system limit as well. Please note very low level of mains related components, though it is supplied from a conventional bridge+capacitors linear supply.

View attachment 201792

I'm not a technical person at all, but having spend several decades looking at distortion measurements in the popular hi-fi press and -- very importantly -- listening to the comments and preference of audiophiles, (most of them subjectivists), I have established my own conjecture and strong believe.

My believe is that most of all it is the harmonic distortion spectrum determines enthusiasts SQ tastes. Specifically I suspect the following with regards to the spectrum:

1. Higher order distortions sound bad, (harsh, grainy, shrill)
2. 2nd, and to a lessor extent, 3rd order products sound pleasant, (i.e. more than merely benign), to many listeners. Hence to preference many have for vacuum tube amplification
3. 2nd and 3rd order harmonic tend to mask the nasty higher order harmonics in one's equipment, not only in the same component but also up & downstream components. Hence the strategy of putting a tube preamp ahead of a S/S power amp, etc.
4. 2nd/3rd order harmonics also tend to mask distortions and excessive brightness in recordings.

(Don't get me wrong: I'm not advocating for components that have high, 2nd/3rd order harmonics. Personally I prefer lowest overall distortion -- at least in case of above average recordings.)

I also really like the idea of measuring the HD spectra at several frequencies along the audible range versus the standard practice of measuring at this or that single frequency.

 

[pma]

Can you do measurements for 0dBFS?

I do not understand your question. "0dBFS" that you read in the plots is related to maximum input signal of the ADC converter with the input divider used. Maximum clean output of the amplifier is there where you see the sudden change in the distortion plot. In the plot below with load 6.8ohm such point is at 20W. These numbers/parameters are not directly related.

 

[DonH56]

Note "dB" is the logarithm of a ratio (10*log10(P2/P1) or 20*log10(V2/V1), etc.) The unit dBFS is dB relative to full-scale which is generally applied to things like ADCs and DACs that have a well-defined full-scale (when all bits are "1", the largest number it can handle). 0 dBFS (or just 0 dB) for an amplifier is rather nebulous since there is not a well-defined "full-scale" value. It could be referenced to maximum output, but that has no clear definition. Is it the output at the knee of the distortion curve, just before it starts to rise steeply, at the 1% THD point as used by many reviewers? Or at the maximum power and distortion as defined by the manufacturer's data sheet, the actual maximum output when hard clipping, something else??? For an amplifier, or any analog gain stage, usually one just looks at the THD sweep such as @pma provided.

What most do is look at the lowest THD number just before the curve takes off, which is around -120 dB THD at 20 W for @pma's amplifier.

HTH - Don

 

[pma]

I am asking you because in most commercial tests you can see THD + N measurements made for 1kHz and 0dB.

This applies to measurements of DACs or ADCs, not to amplifiers. You are confusing the issue.

 

[pma]

I forgot to show measurement of CCIF IMD 19+20kHz (one of the most important distortion measurements, at least to me). So here it is, measured with 96kHz E1DA sampling. Please note also very low harmonic distortion components at 38kHz and 40kHz, excellent high frequency linearity of the amplifier. And as the signal rms level is +3dBr, all the audio in band distortion components are below -120dBr. Again I am able measure this only after buying the E1DA Cosmos ADC (designed by @IVX ). His design gave us semi-professionals a great tool at very affordable price.

.... and a triple tone

 

[Ra1zel]

IME Class D amplifiers require inefficient loudspeakers. That way the idle noise is less audible.

Class D at idle is not that terribly efficient either, as there is a good amount of heat while playing nothing.

Uh huh

 

[fpitas]

IME Class D amplifiers require inefficient loudspeakers. That way the idle noise is less audible.

I have compared various Class D modules on both 86 dB and 95 dB efficient loudspeakers, and the 9 dB higher efficiency loudspeakers means 8 times more idle noise coming out of the loudspeakers with no music.

Class D at idle is not that terribly efficient either, as there is a good amount of heat while playing nothing. No, not as much as with Class A, but not so little as to ignore thermal management - which many Class D modules unfortunately do take too lightly.

My personal preference is Class AB, where there is reasonable efficiency and I feel better performance across the whole audio spectrum.

Maybe some ancient class D amps are as you describe. A friend runs his horns (110dB/!W/!M CDs) directly from Hypex UcD180 modules. No audible noise.

 

[voodooless]

Maybe some ancient class D amps are as you describe. A friend runs his horns (110dB/!W/!M CDs) directly from Hypex UcD180 modules. No audible noise.

Next year, that will be two decades old technology! When will people stop the misinformation

 

[DanielT]

Maybe some ancient class D amps are as you describe. A friend runs his horns (110dB/!W/!M CDs) directly from Hypex UcD180 modules. No audible noise.

Why so much power?

180 W i 4 Ohm, 120 W i 8 Ohm. Speaker "110dB/!W/!M"

UcD180 OEM | H01U000502

The UcD180 OEM amplifier module is the extended version of the Low-profile UcD amplifier module, a Class-D amplifier module meticulously crafted for a multitude of applications. This versatile module seamlessly integrates cutting-edge technology with a m…

www.hypex.nl

 

[fpitas]

Why so much power?

Lol I asked that. The smallest one he could get his hands on. I was skeptical, but it works. He might use a watt peak at full blast. Probably less.