Can we possibly measure the quality of DACs in active DSP speakers

[abdo123]

Ofcourse but if the amps in some active speakers are in the 70s SINAD wise as read in another thread how bad can the DAC be to have an impact to the overall performance.

well you better have 90-100 SINAD DAC otherwise that 70 will drop to 50 or lower.

 

[GaryH]

well you better have 90-100 SINAD DAC otherwise that 70 will drop to 50 or lower.

Your numbers are way, way off. A DAC and amp each with THD at -90 dB would add an entirely inaudible <0.1 dB THD to a speaker with THD at -70 dB (plug in the numbers into the calculator at the bottom of this page to see for yourself). And that's worst-case with the THD from each component in exact phase with each other (which they are highly unlikely to be). Then there's the fact that hardly any speakers will have a SINAD as high as 70 anyway, reducing the requirements for DAC/amp SINAD to achieve transparency even further.

 

[abdo123]

Your numbers are way, way off. A DAC and amp each with THD at -90 dB would add an entirely inaudible <0.1 dB THD to a speaker with THD at -70 dB (plug in the numbers into the calculator at the bottom of this page to see for yourself). And that's worst-case with the THD from each component in exact phase with each other (which they are highly unlikely to be). Then there's the fact that hardly any speakers will have a SINAD as high as 70 anyway, reducing the requirements for DAC/amp SINAD to achieve transparency even further.

no body cares about Harmonic Distortion, I'm mostly concerned about noise, which will be amplified by the amplifier.

 

[sergeauckland]

This seeming obsession with SINAD gets on my wick.
When 0.1% distortion, i.e -60dB, is inaudible under any sensible programme conditions, why be obsess with SINADS of -100 being considered poor?

The fact that modern devices can achieve -100dB routinely doesn't make it necessary, or that anything less is actually bad.

I'm perfectly happy with anything that manages -80dB as that's already 10x better than it needs to be, which is 10x better than is audible under programme conditions.

This obsession with stupidly high SINAD numbers I think makes us something of a laughing stock with those who actually understand what matters and what doesn't.

S

 

[GaryH]

no body cares about Harmonic Distortion, I'm mostly concerned about noise, which will be amplified by the amplifier.

The noise floor of the vast majority of DACs is way way below audibility when listening to music. The amplication of the amp will not be enough to change this.

 

[AnalogSteph]

AFAIK, many of the higher-priced active DSP monitors actually have really decent converters... IIRC (nearly) TOTL parts for e.g. the Genelec 1237A or Neumann KH750. Things tend to get a bit marginal on the bottom end, with dynamic range on both ends barely exceeding 100 dB(A) and slightly less than ideal anti-alias filters... think CS5341 in the JBL 30x or ADAU1701 converters in others, with converters and DSP running at a mere 44.1 kHz at times. There's definitely some room for improvement left there. I'd shoot for an overall dynamic range (A/D+D/A) of about peak dB SPL, and processing at 48 kHz minimum, ideally more like 96. Modern AKM ADCs' filters even perform best at 192-384 kHz, which may require some SRC first.

I have a DAC from nearly 20 years ago, the Bryston BP25-DA. I just replaced it with a Topping D70s for digital sources. I like the Topping better, but I would place no wager on my ability to consistently pick it in a double blind test. I largely use CDs as my digital source.

Had a quick search, and this model appears to be the first DAC (well, preamp with DAC) I've come across to actually use the CS43122, Crystal's flagship part for a few years in the early 2000s. Neat! Specs are dynamic range 122 dB(A), THD+N -102 dB, filter ripple +/-0.0001 dB, filter stopband -102 dB. Here's a whitepaper on the part. I sense some modding potential... and the DAC part itself might not actually be the limiting factor (a case for some tape loop action?). An old newsletter is telling us that they are apparently using an ASRC to upsample to 96 kHz, much like the Benchmark DAC1 of the day, so the same caveats about staying away from 0 dBFS are very likely to apply. If the DAC performs anywhere near chip spec, you can definitely afford it!

But yeah, connected to a plain CD player you are probably better off with a DAC(-pre) with controllable digital attenuation like the Topping, which I think being AKM-based would even provide about 2 dB of headroom as well.

The fact that modern devices can achieve -100dB routinely doesn't make it necessary, or that anything less is actually bad.

I'd think most of us actually well aware of this. SINAD is a bit of a stupid metric for audio anyway... there are applications like preamp replacement where <-120 dB of the +N part is actually needed but nobody is going to hear the difference between -90 or -120 dB THD, assuming things do not get dramatically worse anywhere else in the audible spectrum. (I mean, plenty of people are happy with stock Behringer *2496 series devices, and they are up to like -70 dB THD by 20 kHz...) It's really more of a metric of engineering quality. Those who may actually require high SINAD like that for some non-audio application in the future (maybe something SDR) will be glad that there were some silly audiophiles who bought the gear to begin with.

 

[abdo123]

The noise floor of the vast majority of DACs is way way below audibility when listening to music. The amplication of the amp will not be enough to change this.

and the vast majority of amplifiers are also distortion limited.

a S/N of 70 dB is very pathetic in 2021.

Anyway if you give me some realistic numbers i will do some math and we can both figure out how much a DAC contributes to total noise.

 

[GaryH]

a S/N of 70 dB is very pathetic in 2021.

Anyway if you give me some realistic numbers i will do some math and we can both figure out how much a DAC contributes to total noise.

Where are you seeing a DAC signal to noise ratio of 70 dB? 'The world's cheapest DAC' for example has a noise floor around -120 dB. Obviously higher quality DACs have even lower noise floors, typically in the -130 to -140s.

 

[abdo123]

Where are you seeing a DAC signal to noise ratio of 70 dB? 'The world's cheapest DAC' for example has a noise floor around -120 dB. Obviously higher quality DACs have even lower noise floors, typically in the -130 to -140s.

Did you just take the number from the FFT window and assume that's the actual noise floor?

the Dynamic range is the actual S/N ratio.

Also my 70dB comment was on amplifiers.

anyway is the world's cheapest DAC your DAC of choice? then what is your amplifier of choice?

 

[GaryH]

Ah yes my mistake, I'd forgotten about that quirk of the FFT graphs. I brought up 'the world's cheapest DAC' to show that even this has decently low noise levels with its 99 dB dynamic range. (Still not sure where you're getting that 70 dB figure from.)

 

[JohnYang1997]

Ah yes my mistake, I'd forgotten about that quirk of the FFT graphs. I brought up 'the world's cheapest DAC' to show that even this has decently low noise levels with its 99 dB dynamic range. (Still not sure where you're getting that 70 dB figure from.)

All the noise from the DAC will get converted and amplified. There has to be some condition when it is audible or not to someone.
If you don't play your dac at max output, you don't get your full SNR basically.
If you have 2uV noise at DAC output, amplifying 20 times is 40uV. If you have 10uV noise at DAC output amplifying 30 times is 300uV. Big difference.

 

[RayDunzl]

Reading all these reviews about transparent DACs and following up with speaker reviews, it just interesting for my naive brain to know what the SINAD of say D&D or Genelec internal DACs are. Can this be done?

In the case of the JBL 3 series, no.

Digits from the ADC go into the STA350BW, I see no DAC output pins, only the Amplifier outputs, since the DAC/DSP/and AMP are all in one chip.

 

[gags11]

In the case of the JBL 3 series, no.

Digits from the ADC go into the STA350BW, I see no DAC output pins, only the Amplifier outputs, since the DAC/DSP/and AMP are all in one chip.

Would it be possible to measure the Amp outputs? Would be interesting to know how the whole system behaves prior to the transducer input.

 

[JohnYang1997]

https://www.audiosciencereview.com/...ds/neumann-kh80dsp-teardown.24116/post-813416
Here's one. The amplifier.

The dac/codec is CS42526. -100dB THD+N much better than the amplifier.

 

[mdsimon2]

All the noise from the DAC will get converted and amplified. There has to be some condition when someone is audible or not.
If you don't play your dac at max output, you don't get your full SNR basically.
If you have 2uV noise at DAC output, amplifying 20 times is 40uV. If you have 10uV noise at DAC output amplifying 30 times is 300uV. Big difference.

100% this. I wish ASR reviews showed absolute noise measurements. In the current state it is a bit of a pain to compare noise performance as the dynamic range measurements are not done at consistent levels.

Michael

 

[GaryH]

There has to be some condition when it is audible or not to someone.

And this is the key, determining what noise level is low enough to be inaudible when listening to music, at a typical listening distance, at a typical volume. Any noise specs below this are purely an engineering exercise with no relevance to the user's aural experience of listening to music with the product.

 

[mdsimon2]

Would it be possible to measure the Amp outputs? Would be interesting to know how the whole system behaves prior to the transducer input.

Yes and it is not pretty. Considering the rather audible tweeter hiss it should come as no surprise that if ranked against discrete amplifiers it would be bottom of the barrel performance.

Michael

 

[JohnYang1997]

100% this. I wish ASR reviews showed absolute noise measurements. In the current state it is a bit of a pain to compare noise performance as the dynamic range measurements are not done at consistent levels.

Michael

You can use IMD graph to determine the noise level (roughly, as AP has its self noise). It's pretty consistent to me. It's quite difficult to measure noise accurately, because of noise modulation and turning off the output when there's no signal.

 

[mdsimon2]

You can use IMD graph to determine the noise level (roughly, as AP has its self noise). It's pretty consistent to me. It's quite difficult to measure noise accurately, because of noise modulation and turning off the output when there's no signal.

I agree that there is enough info in the reviews to compare devices in most cases but it definitely requires a bit of work. Reflecting on it some more we are certainly dealing with an embarrassment of riches, probably best to ignore me . Without ASR it would be much more of crap shoot.

Michael

 

[Katji]

+1
[...]
.. Tearing down a speaker (some quite expensive), voiding the warranty in the process just to get to the SINAD is not worth the bother and the cost... IMHO. My $0.02 ..etc...

Big heavy speakers come in crates. :-s [To spell it out:] Shipping is expensive. The concept here is constrained - overall - to people sending stuff to Amir within the USA.
Summary (including the technical aspect which has been well explained): impractical or absurd.