Audibility thresholds of amp and DAC measurements

[Serge Smirnoff]

Really? Ok, my test tone is 43100/3 Hz. I add third harmonic 50dB down. That aliases to 1 kHz. You can't hear it?

I should have added: -50dB with music signal (or prog. simulated noise). This estimation is preliminary and based on measurements of portable players, where Chord Hugo 2 has df = -34dB and also measurements of various loopback tests from this page (they have source recordings, which I can measure), where the best one shows -70dB. I should note that in the latter case different test music and shorter excerpt is used, so with my test material the figure can be a bit lower by 5-7dB.

 

[j_j]

Try a female vocalist, a-capella, with low frequency noise added under her voice.

Good luck with that.

 

[Serge Smirnoff]

Try a female vocalist, a-capella, with low frequency noise added under her voice.

Good luck with that.

Is "Tom's Diner" by Suzanne Vega is suitable for the test? Can you specify the required noise in more detail or give the guide how to prepare it?

 

[j_j]

Is "Tom's Diner" by Suzanne Vega is suitable for the test? Can you specify the required noise in more detail or give the guide how to prepare it?

Tom's Diner has a different property, it creates a BMLD (binaural masking level depression) situation so it can confuse two issues.

 

[odyo]

Dacs and amps have different measurements how that works ? Let's say i'm using modi 3 + heresy.

modi 3 specs:
THD:
Modi 3: <0.001%, 20Hz-20KHz, at max output
IMD:
Modi 3: <0.001%, CCIR
S/N: >104dB, referenced to 1.5VRMS, unweighted
Crosstalk: -80dB, 20-20kHz

Heresy specs:
THD+N:
Low Gain: Less than 0.0002% (-113dB) at 1V RMS into 32 ohms
IMD:
Low Gain: Less than 0.0006% (-105dB) at 1V RMS into 32 ohms, CCIF
SNR:
Low Gain: Greater than 115dB, A weighted, referenced to 1V RMS
Crosstalk: Less than -70dB, 20 Hz-20 kHz in either gain

Are these bottleneck each other ? or how it works ?

 

[rwortman]

Distortions add through out the chain. Bottleneck isn't the operative word. Every device adds it's own distortion to the signal.

 

[IowAudio]

So if I use a SMSL M100 DAC to say feed my Marantz SR4021 stereo receiver that has lower specs wouldn't the Marantz be the limiting factor in sound quality? Could be the best DAC in the world but you feed the analog signal to a lesser amp its only gonna get so good right? I ask because it seems like many people think these little $100 DACs marketed based on super high sampling rates are cheap easy solutions that make their whole system "audiophile" quality. Not to say they don't improve the sound of a home theater receiver BUT a home theater receiver isn't going to resolve the abilities of even say the M100 DAC.

SMSL M100 - THD+N=0.0005%, Dynamic Range=114db, SNR=106
Marantz SR4021 THD=0.08%, SNR=95db

 

[rwortman]

Audiophile magazine writers and many readers obsess over impossible to quantify and often times imaginary qualities of sound equipment. They talk a bunch about "synergies" and "system matching" which is almost always based on nonsense. Obsessing over measured spec's isn't really that much better. Operated within its operating envelope that receiver is going to be transparent or pretty close to it as will most decently designed amplifiers that don't have audible hiss or hum. I would guess that 90+% of the DAC's measured here would be similarly difficult to pick out in a blind listening test. 95db SNR is just about all CD spec audio can do and a 16/44 file downsampled from high rez is audibly the same as the high rez file, if not all the time, close to it. Get an amp that can drive your speakers without audible noise and buy a DAC you like the looks of from a reputable manufacturer and go listen to some music. If you don't like they way your system sounds, buy better speakers (or headphones), EQ your system to your room, or have someone that knows what the are doing treat your room. Chasing sound improvements by swapping DAC's around seems like a waste of money to me. I honestly would not bet on my ability to tell the best DAC on Amir's list from a Chromecast Audio in a blind test.

 

[BillG]

"Obsessing over measured spec's isn't really that much better..."

For the enlightened consumer, it's about the value proposition a product represents - when a $100USD DAC can outperform a $15,000USD one, it indicates that something is seriously amiss in high end audio.

Most of us here are familiar with the thresholds of audibility, and we don't concern ourselves with that much once it's been achieved. However, if I can find a less expensive product that performs the function I require very well, you damn right I'm interested!

For the scientist and engineers here, it's their job and passion to constantly push the envelope, regardless of whether it's audible or not.

 

[rwortman]

I care a lot about value. I just don't think its only about how many decimal points to the right of the zero the distortion number is. For an audio engineer to passionately push the envelope to achieve something no one can hear seems a waste of resources. Kind of like Mike Moffat and his obsession with closed loop math in his digital filters. I guess some engineers designing DAC's are competing for better specs, it's up to you if that's of top importance. DAC's have been a solved problem for quite a while. If I was a young man and wanted to use my engineering talent to further the state of the art in audio, I would not be designing DAC chips or stand alone DAC's in the box. There are more challenging and more audible boundaries that need pushing. BTW, my post was to the question of whether connecting a cheap DAC with excellent spec's to a receiver with less impressive numbers and calling it "audiophile quality" was foolishness. It wasn't a judgement about the whole forum. It isn't foolishness, and good luck finding an amp that measures as well as the best DAC's on the list. They are unicorns.

 

[IowAudio]

I agree, numbers don't always equate to good or bad sound quality. I prefer to just listen to what ever gear and let my ears make the decision. I was asking just to reassure my assumptions when it comes to comparing numerical specs. Many people I've chatted with on the matter tend to think a $100 32bit/384khz DAC is a magic audiophile bullet for their system. I know high sampling rates are mostly marketing BS to lure in ignorant customers. I know regardless of specs and price many DACs or amps sound different good or bad. I tend to find better source material, well recorded and produced music, better speakers, and room treatment help a lot. I know this has been discussed a thousand times. Thanks for your reply's.

 

[BillG]

"good luck finding an amp that measures as well as the best DAC's on the list. They are unicorns..."

Amps are a different scenario electrically, as I'm sure you're already aware of, and I was referring to DACs simply as an illustration.

As for wasted resources, science/engineering types generally don't consider that in the pursuit of perfection (I'm a scientist-engineer in another field) - that's the job of business management.

Back to amps, we've now at least three tiers of relatively inexpensive modules that are demonstrably transparent, which at the analog end of the processing chain is sufficient (for now): #1 Purifi, #2 Hypex, and #3 ICEPower. I suspect some of Texas Instruments' PurePath HD modules may comprise a fourth tier, but we've not seen any (?) of those independently analysed here. Benchmark's amp may beat them all - I'd have to go back and look to verify that. However, I didn't include it as they're not a ready made solution that can dropped into a case by the average DIYer.

 

[j_j]

Back to amps, we've now at least three tiers of relatively inexpensive modules that are demonstrably transparent, which at the analog end of the processing chain is sufficient (for now): #1 Purifi, #2 Hypex, and #3 ICEPower. I suspect some of Texas Instruments' PurePath HD modules may comprise a fourth tier, but we've not seen any (?) of those independently analysed here. Benchmark's amp may beat them all - I'd have to go back and look to verify that. However, I didn't include it as they're not a ready made solution that can dropped into a case by the average DIYer.

Don't leave out Pascal!

 

[RichB]

Don't leave out Pascal!

Some amplifier implementations, including Pascal, lack of FTC power and distortion from 20 Hz to 20 kHz.
There seems to be an increasing number of manufacturers with only 1 kHz specifications that can provide better power and distortion specifications.

ATI provides full bandwidth specifications for their NCore amps, others do not.

- Rich

 

[NDRQ]

Some amplifier implementations, including Pascal, lack of FTC power and distortion from 20 Hz to 20 kHz.
There seems to be an increasing number of manufacturers with only 1 kHz specifications that can provide better power and distortion specifications.

ATI provides full bandwidth specifications for their NCore amps, others do not.

- Rich

Actually many multitone test here proves that if the 1khz test shows good performance, the mutlitone test will be usually in the same good league.
So even if there is only 1khz data, that is a good indicator for performance.

IMO the biggest problem with manufacturers is that they are usually giving data, that not even close to the actual mesurements.

 

[RichB]

Actually many multitone test here proves that if the 1khz test shows good performance, the mutlitone test will be usually in the same good league.
So even if there is only 1khz data, that is a good indicator for performance.

IMO the biggest problem with manufacturers is that they are usually giving data, that not even close to the actual mesurements.

It's only a problem when it is a problem

Any of these manufacturers are capable of providing 20 Hz to 20 kHz FTC measurements. @amirm can so can they.
Since we do not listen to 1 kHz tones, these specs are provided to inflate power ratings.

Here is the ASR review of the Devialet Expert 200 Amplifier:
https://www.audiosciencereview.com/...-200-amplifier-dac-and-streamer-review.12286/

This 200 WPC amplifier would be FTC rated at 10 watts.

- Rich

 

[j_j]

It's only a problem when it is a problem

Any of these manufacturers are capable of providing 20 Hz to 20 kHz FTC measurements. @amirm can so can they.
Since we do not listen to 1 kHz tones, these specs are provided to inflate power ratings.

Here is the ASR review of the Devialet Expert 200 Amplifier:
https://www.audiosciencereview.com/...-200-amplifier-dac-and-streamer-review.12286/

This 200 WPC amplifier would be FTC rated at 10 watts.

- Rich

Interestingly, I have the full specs for the Pascals. There's nothing sinister there. I got them from their web site, too. Has it changed?

 

[RichB]

Interestingly, I have the full specs for the Pascals. There's nothing sinister there. I got them from their web site, too. Has it changed?

When researching D-Sonic Amps, I found only 1 kHz specification.

- Rich

 

[pozz]

Came back to this thread after a long while. Thanks again @flipflop for all the work and updates.

 

[gwarlek]

It's not covered in the post, but Multitone measurement thresholds would be the same as for other FFT measurements? Also does the amount of "grass" compared to clean multitones actually matter as long as it's below the audible threshold?