Performance of Amps in Active Speakers

[617]

Tom Christiansen offers an LM3886 amp with vanishingly low distortion:
https://neurochrome.com/products/lm3886-done-right

0.0037% distortion for a 1khz tone into 4 ohms at 80 Watts. LF performance might be worse but at nearfield levels an amp like this would be fantastic.

I wouldn't discount the possibility that the chip amps used in active monitors are pretty good, but as always, it would be nice to measure.

@tomchr I assume you know better than me?

 

[dc655321]

D&D is an active speaker, but it's not a studio monitor and it is pricey. Same could be said for Kii and B&O. Then again, some of the Genelec speakers are expensive.

D&D and Kii are targeted at the studio and home user. In fact, studio use was their initial market.
And as you say, Genelec are not chump-change either (cf. Neumann, ATC, Barefoot, etc).

 

[Ron Texas]

D&D and Kii are targeted at the studio and home user. In fact, studio use was their initial market.
And as you say, Genelec are not chump-change either (cf. Neumann, ATC, Barefoot, etc).

There are exceptions and grey lines everywhere.

 

[Cortes]

I own genelec 8331, and it's the best sound I've ever owned, really good. However, their DAC limited to 24/96 and profits of the audio industry make me think the amp&DAC are probably very low level, for sure something very cheap. Anyway, I don't worry to much, since in active territory is the whole package that matters, and in this case it delivers. Of course, I would love to have a better DAC inside that allowed me to experiment with higher rates, DSD, etc.

 

[GrimSurfer]

D&D 8c uses Pascal amps, data sheet with measurements here.

Looks fairly similar to icepower (suspiciously so, in fact).

Thank you, @dc655321.

SINAD of -90.46. Almost certainly inaudible at reference levels.

BTW, this is the combined SINAD that you'd see with a preamp with a SINAD 93.47 hooked up to an amp with a SINAD of 93.47.

Or if we're looking at a DAC, pre-amp, amp combo, the SINAD of each device would need to be -95.2. This is within a whisker of getting full resolution from a 16/44.1 source.

 

[GrimSurfer]

View attachment 36747

That's barely above the ambient noise floor.

The noise floor reading on a mike will be dominated by lower frequencies that are harder to hear according to Fletcher-Munson curves. The ambient noise of middle frequencies (where the ear is most sensitive) will be much, much lower... approaching 0 dB.

Since the SINAD is measured at 1 kHz, in the heart of the ear's most sensitive hearing range, noise and distortion will be well above ambient. As much as 20 dB above the room at this frequency, at reference levels.

 

[q3cpma]

D&D and Kii are targeted at the studio and home user. In fact, studio use was their initial market.
And as you say, Genelec are not chump-change either (cf. Neumann, ATC, Barefoot, etc).

I wouldn't put Neumann and Genelec in the same price bracket as ATC and Barefoot, personally.

 

[dc655321]

I wouldn't put Neumann and Genelec in the same price bracket as ATC and Barefoot, personally.

Isn't a pair of kh310 ~$4500, and kh 420 about $10k?
And Genelec 8351b is ~$8k, 8361 is ~$10k?

Or, did I misread the prices glancing at Sweetwater postings?

 

[JohnYang1997]

My hs7 has lm3886. So no an issue for me. For the price I can't ask for more.

 

[LTig]

Probably there is no official info but from educated guesses made from pictures of guts in reviews of KH420, KH310 and KH120 we can assume that the amps used are TDA7293/7294. KH120 I've inspected myself.
For KH420, see: https://www.fidelity-magazin.de/2015/12/16/neumann-kh-420-messungen/ , where a pic of the amp plate is shown. We can spot 8 pcs. of 15 pin chipamp, 4 on each board. The woofer is running most likely on a bridged-parallel config (4 amps) and the mid and tweeter from 2 amps each, either bridged or paralled depending on the driver impedances -- most likely bridged, though -- , which would match the published power specs. Together this is IMHO a strong indication for TDA7293 being used (and powered from SMPS's).

The K&H O300D (predecessor of Neumann KH310) exclusively uses MC33078 opamps and 4 x TDA7294 [1] (running at +/- 40V) for the power amps (1 each for mid and high, 2 in bridge mode for low). Audio signals pass through 8 to 10 opamps before reaching the TDAs.

Source: schematics of the O300D - sorry I'm not shure whether I'm allowed to publish them.
[1] A note in the schematics hints that TDA7293 may have been used as well.

 

[ernestcarl]

Just as the LSR 3 series are the cheaper brethren of the LSR 7 series, one could probably also consider the Presonus Sceptres as the cheaper/more affordable cousins of Fulcrum Acoustics' reference monitors e.g. RM28ac

I've only found one reviewer of this speaker that actually did a (partial) teardown. The class D amplifier appears to be a Swiss made Anaview ALC0180 -- I'm not certain as to which figures exactly are relevant here for comparison. Additionally, on a separate board housing the inputs and controls, there is the "the 33079 operational amplifier [which] acts as the EQ, the Codec is AK4621, 115dB for dynamic input/output, and the DSP is ATMEL." I presume I don't hear any of the Class D distortion at my usual listening volumes. In fact, I'm much more concerned about the non-linearities in the FR than SINAD.

As far as audible background noise, I would need to put my ear about a foot or closer to hear HF hiss from the horn loaded compression unit. Not even close to the kind of hiss and annoying harshness I hear from LSR305 at close range (which, of course, is a none issue as long as I'm staying at the MLP for the aforementioned ~1.6m away).

 

[August]

I think the internal AMP performance of the active speakers is usually not very good (not related to whether they can be heard), and most of their THD+N is between 50-90, because this is the key to profit. I hope there are more products with built-in NCore.

 

[JohnYang1997]

I think the internal AMP performance of the active speakers is usually not very good (not related to whether they can be heard), and most of their THD+N is between 50-90, because this is the key to profit. I hope there are more products with built-in NCore.

Actually that's because those active speaker designer understand that the amp distortion will be a magnitude or two lower than the speakers.
The performance of classic class AB amp are still considered better against most used class D amps and they are true for current products. The benefit of Class D is higher power output and less heat generated hence less heatsink(which cost money). The shift from class AB to class D is an ongoing process.
From a different perspective that how active speakers work. And why mere lm3886 can still be better where most people think. The speakers are directly driven from the amplifier without crossover which is done before power amplification. The designer can use 4ohm speakers pair with these chip amps to get more power output. And class D amps are generally worse at low impedance load vs class AB because of the output filters(true that introducing low frequency feedback across the filter can help linearity).
So if a designer want a better performed active speaker he may choose class AB. Then if for cost, choose class D. And good performing class D cost considerably more due to complexity and the price of IP. Only when very high power active speakers and very good performance is needed, a designer then will choose better performing class D amps.

 

[mkawa]

Actually that's because those active speaker designer understand that the amp distortion will be a magnitude or two lower than the speakers.
The performance of classic class AB amp are still considered better against most used class D amps and they are true for current products. The benefit of Class D is higher power output and less heat generated hence less heatsink(which cost money). The shift from class AB to class D is an ongoing process.
From a different perspective that how active speakers work. And why mere lm3886 can still be better where most people think. The speakers are directly driven from the amplifier without crossover which is done before power amplification. The designer can use 4ohm speakers pair with these chip amps to get more power output. And class D amps are generally worse at low impedance load vs class AB because of the output filters(true that introducing low frequency feedback across the filter can help linearity).
So if a designer want a better performed active speaker he may choose class AB. Then if for cost, choose class D. And good performing class D cost considerably more due to complexity and the price of IP. Only when very high power active speakers and very good performance is needed, a designer then will choose better performing class D amps.

i've been thinking about this with my loudspeaker setup that is nearly a powered speaker situation due to the need for dsp shaping prior to the final gain stage. in this case i know the dsp stage drops thd to the -80db mark, which we would pooh-pooh around here. however, the final output speaker measurements in ideal conditions (NRC measurements) only measures ~-85db thd, and that is only when perfectly on-axis. (the point of the dsp is actually to do phase correction on a dipolar variant of the latter design).

i've some extensive casual and critical listening and the minor benefits of a few db of distortion in the audio chain is wildly negated by the difference in on vs off-axis response (in this particular case, off-axis response completely disappears. it's a bit freaky).

 

[August]

Actually that's because those active speaker designer understand that the amp distortion will be a magnitude or two lower than the speakers.
The performance of classic class AB amp are still considered better against most used class D amps and they are true for current products. The benefit of Class D is higher power output and less heat generated hence less heatsink(which cost money). The shift from class AB to class D is an ongoing process.
From a different perspective that how active speakers work. And why mere lm3886 can still be better where most people think. The speakers are directly driven from the amplifier without crossover which is done before power amplification. The designer can use 4ohm speakers pair with these chip amps to get more power output. And class D amps are generally worse at low impedance load vs class AB because of the output filters(true that introducing low frequency feedback across the filter can help linearity).
So if a designer want a better performed active speaker he may choose class AB. Then if for cost, choose class D. And good performing class D cost considerably more due to complexity and the price of IP. Only when very high power active speakers and very good performance is needed, a designer then will choose better performing class D amps.

This is also the reason why active speakers from China tend to use TAS5754.

 

[Blumlein 88]

Just for good measure (sorry about the pun), with some LSR 305s playing 95 db at the LP, my Umik 1 gives readings mostly around .5 %THD. That is nearly all 2nd harmonic except at near the cross over point where it is mostly 3rd harmonic.

Same measurement with some Revel F12's same position and loudness, driven by Wyred4Sound ST500 amp, numbers of mostly around .2 % THD again mostly 2nd harmonic except at the upper crossover point where it is mostly 3rd harmonic. Using some other microphones THD drops by about .05% (to around .15 % THD) so maybe the UMIK is limited to some extent for distortion measures. Though I'd need to do some more measurement comparisons to be sure of that.

Harmonics beyond the 3rd were generally .01% in places where it would create the harmonic between 3-5 khz where our hearing is most sensitive. So an amp with THD of -80 db or less isn't likely to color sound beyond the speaker. Same is likely true at THD -70 db depending upon the distribution of the upper harmonics.

 

[watchnerd]

Thd+noise of .09% should be equivalent to an SINAD of 61 db

http://www.sengpielaudio.com/calculator-thd.htm

Now if SINAD of 61 db is good enough for an active studio monitor, why do some think the Crown XLS 1502 is junk at 76 db. Or, why are folks around here celebrating when Bruno's latest pushes a few more db past 100 SINAD?

Really, how much is enough?

Measurement fetishism seems to be on the rise at ASR.

 

[GrimSurfer]

Measurement fetishism seems to be on the rise at ASR.

Fettishism is an issue if the discussion is on inaudible distortion (i.e. Great debate on why SINAD of -130 dB is so much better than -120 dB). After all, if it cannot realistically be heard then an impact on sound quality (SQ) cannot logically be claimed.

When the discussion involves SINAD which is demonstrably audible within the context of reasonable (reference) playback levels and the scientifically established threshold of audibility (Fletcher-Munson), the issue is logically one of sound quality (SQ).

These are not subjective issues. Nor are they relative ones. Objectivity is necessary which means that people have to park their personal interests at the door.

Some may say, therefore, that what I've written supports the view that ASR ratings be based on audibility. The problem is that people don't accept audibility limits when they are shown to exist in gear with a SINAD of "x". So how can anyone say that they will be accepted in the form of a net assessment? They simply can't.

It has been my experience that people who have the most difficulty accepting the audibility thresholds of THD+N are those who:

1. Own gear that knowingly doesn't measure up; or

2. Have subjectivist leanings or have been conditioned to "like what they like" instead of "linking what they know to be accurate"; or

3. Or don't understand/refuse to accept the underlying science that differentiates audibility thresholds; or

4. Are prone to rationalizing away audibility by statements like "x" is nearly as inaudible as "y", which is a conflation of points 1-3 and akin to claims of being half pregnant.

These are not insults or ad hominem attacks. They are the only things that I can logically think of to explain why people selectively apply the science of sound or dismiss it altogether on ASR.

 

[watchnerd]

These are not insults or ad hominem attacks. They are the only things that I can logically think of to explain why people selectively apply the science of sound or dismiss it altogether on ASR.

So we need to keep striving for infinitely low distortion and SNR?

Surely there is a point where we reach diminishing returns and 'problem solved'.

 

[watchnerd]

And, if SNR and distortion are the most important things, what about the point made about top tier external amps vs what one finds in most active monitors (which are often less good than a Bemchmark)?

Do we think the other advantages of active speakers (e.g. crossovers in DSP) offset potentially worse amp specs?

If so, why?

Are improvements in phase and impulse response more important than distortion?