Just how legit is this person's blind test results?

[Purité Audio]

Unfortunately just that, interested enough to spend on equipment but not interested enough to learn how things actually work.
Keith

 

[symphara]

transparently measured amps

What does this mean to you? Is there just a SINAD limit, or a more complex set of metrics that an amplifier needs to clear, in order to be considered transparent?

 

[Blumlein 88]

What does this mean to you? Is there just a SINAD limit, or a more complex set of metrics that an amplifier needs to clear, in order to be considered transparent?

We have more than one thread on this very topic. No, for the 11 millionth time, it is not just SINAD. Pardon my sensitivity to the idea SINAD tells all.

Here is one.

When is a DAC or Amp audibly transparent?

At what SINAD and other metrics can we consider a DAC or Amp audibly transparent? I know we want a well engineered device, but at what threshold can you no longer hear a difference (even for very trained ears)?

www.audiosciencereview.com

And some more.

Audibility thresholds of amp and DAC measurements

Introduction A recurring theme on ASR is whether or not the various measured qualities of the devices are audible. In this post, I'll present some clear and visual thresholds for when those imperfections can be considered a potentially audible concern. I will not explain the basics of amp/DAC...

www.audiosciencereview.com

What matters about SINAD measurements

First may I commend this site and its founder\admin for a fine service to several audio communities, ranging from newbies to audiophiles to professionals. I myself am in the latter category – a professional audio engineer since 1958 (my first paying gig after graduating RCA Institutes, and...

www.audiosciencereview.com

 

[symphara]

We have more than one thread on this very topic. No, for the 11 millionth time, it is not just SINAD. Pardon my sensitivity to the idea SINAD tells all.

Here is one.

When is a DAC or Amp audibly transparent?

At what SINAD and other metrics can we consider a DAC or Amp audibly transparent? I know we want a well engineered device, but at what threshold can you no longer hear a difference (even for very trained ears)?

www.audiosciencereview.com

And some more.

Audibility thresholds of amp and DAC measurements

Introduction A recurring theme on ASR is whether or not the various measured qualities of the devices are audible. In this post, I'll present some clear and visual thresholds for when those imperfections can be considered a potentially audible concern. I will not explain the basics of amp/DAC...

www.audiosciencereview.com

What matters about SINAD measurements

First may I commend this site and its founder\admin for a fine service to several audio communities, ranging from newbies to audiophiles to professionals. I myself am in the latter category – a professional audio engineer since 1958 (my first paying gig after graduating RCA Institutes, and...

www.audiosciencereview.com

I had seen those posts (your first "some more" is the same as the first one btw), I just wasn't sure what it means to you.

So we have:

Lenient
Dynamic range, linearity: 96 dB
THD, IMD: -66 dBFS / 0.05%
Noise: -85 dBFS / 0.005%
SINAD: 85 dB
Crosstalk: -60 dBFS
Frequency response: ±0.5 dB/octave
Channel balance: 1 dB
Output impedance: 2 ohms

Strict
Dynamic range, linearity, SINAD: 120 dB
THD, IMD, noise, crosstalk, jitter: -120 dBFS / 0.0001%
Frequency response, channel balance: ±0.1 dB
Output impedance: 0.16 ohms

"Strict" I think it's mostly out of the question.

I wonder how many >100W (8Ω) amplifiers are there, that pass "lenient"? I was looking at some Stereophile measurements for powerful amps and none seem to pass the "lenient" criteria.

If the criteria is very narrow, you can absolutely be correct but this is a "technically correct" kind of thing, with no much applicability.

Otherwise, the normal stuff most people buy, especially for big speakers, doesn't seem to be "transparent". So you're correct, but so is everyone else...

 

[jae]

On a side note, we should have polarity checks in reviews for headphones in particular (there are a number that are wired wrong from factory, some hifiman and moondrop for example), I wonder if any devices have issues.

 

[Blumlein 88]

I had seen those posts (your first "some more" is the same as the first one btw), I just wasn't sure what it means to you.

So we have:

Lenient
Dynamic range, linearity: 96 dB
THD, IMD: -66 dBFS / 0.05%
Noise: -85 dBFS / 0.005%
SINAD: 85 dB
Crosstalk: -60 dBFS
Frequency response: ±0.5 dB/octave
Channel balance: 1 dB
Output impedance: 2 ohms

Strict
Dynamic range, linearity, SINAD: 120 dB
THD, IMD, noise, crosstalk, jitter: -120 dBFS / 0.0001%
Frequency response, channel balance: ±0.1 dB
Output impedance: 0.16 ohms

"Strict" I think it's mostly out of the question.

I wonder how many >100W (8Ω) amplifiers are there, that pass "lenient"? I was looking at some Stereophile measurements for powerful amps and none seem to pass the "lenient" criteria.

If the criteria is very narrow, you can absolutely be correct but this is a "technically correct" kind of thing, with no much applicability.

Otherwise, the normal stuff most people buy, especially for big speakers, doesn't seem to be "transparent". So you're correct, but so is everyone else...

There may be a handful that meet strict or come very close. Plenty pass lenient. Not all, but plenty. You need to add some provisions for covering 20hz to 20 khz and all the way up to rated power.

 

[jae]

I wonder how many >100W (8Ω) amplifiers are there, that pass "lenient"? I was looking at some Stereophile measurements for powerful amps and none seem to pass the "lenient" criteria.

Amplifier SINAD list ASR + 3rd parties

List Name,THD+N (5W, 4 ohms),SINAD (dB) (5W, 4 ohms),Power (W) (1% THD+N, 4 ohms),Price (USD),Source Topping LA90 (stereo, bypass, low gain),0.00009%,121,90,800,<a href="https://www.l7audiolab.com/f/topping-la90/">L7 Audio Lab</a>,Suggestions, corrections and more info here: Topping LA90 Discre...

docs.google.com

 

[rdenney]

Standards for amplifiers playing into speakers or headphones into an actual listening space are less demanding than for upstream components that have not yet been amplified. I think the "lenient" and "strict" standards above are for line-level components, not amps, but I may be remembering that incorrectly.

Speakers that maintain distortion levels better than 1% (-40dB) across the spectrum are rare enough, and rooms are a whole other thing.

If I calibrate the "lenient" range to my own hearing, I'd have to add about 20-40 dB. I cannot hear harmonic distortion worse than about -36dB when playing actual music while listening in headphones. Other kinds of distortion and noise are much easier to hear--I can hear IMD at much lower levels if the IMD products are in my hearing range. I can hear background noise higher than -80dB only when it is amplified by a very powerful amplifier, and then only when I put my ear up to the tweeter.

Distortions can be detected at much lower levels by applying special listening techniques, such as gain-riding the tails of reverberation or severely amplifying background noise, or listening to test tones. I've heard all sorts of noise artifacts in extremely amplified background noise, but that amplification is at levels that are far beyond what could be tolerated by the people or the equipment if playing an actual signal, even a quiet one. I can hear harmonic distortion at much lower levels when listening to pure sine waves, vs. when listening to actual music. These are not strategies relevant to whether amps are functionally transparent.

Every listener is different, of course, but considering my own hearing and experience, I would be surprised if more than a fraction of a percent of trained, young listeners could consistently detect distortions more strict than the lenient category above with music played at listenable levels into transducers. And even if they could, it wouldn't be "ridiculously easy" for them--it would be a very subtle effect that would demand lots of quick back and forth.

There are lots of amps that do better than 85 dB SINAD, which is close to the average on the big amplifier test list, but that was probably not the case 40+ years ago. But I suspect that even 40 years ago, any decent amp was better than 60 dB SINAD, and unless it exhibited particularly severe IMD with mixing products at an unfortunate frequency, or hearable background noise, it would still perform as functionally transparent when playing actual music at listenable levels into reasonable loads. Unless it was driven outside its linear range, of course. Coloration (changes in frequency response into specific loads) are another fault that affects amps when driving difficult loads, but even then I think it's mostly that they lack the power to fill up the waveform at all frequencies even when playing into a load that is close to a short.

Rick "thinking reliable power output into complex loads has the greatest effect on transparency" Denney

 

[symphara]

Amplifier SINAD list ASR + 3rd parties

List Name,THD+N (5W, 4 ohms),SINAD (dB) (5W, 4 ohms),Power (W) (1% THD+N, 4 ohms),Price (USD),Source Topping LA90 (stereo, bypass, low gain),0.00009%,121,90,800,<a href="https://www.l7audiolab.com/f/topping-la90/">L7 Audio Lab</a>,Suggestions, corrections and more info here: Topping LA90 Discre...

docs.google.com

This is just a SINAD list, it doesn't cover the criteria. Or do you have a different position, that a specific SINAD threshold is all an amp needs to basically be in kern(amps)?

There may be a handful that meet strict or come very close. Plenty pass lenient. Not all, but plenty. You need to add some provisions for covering 20hz to 20 khz and all the way up to rated power.

Not the ones I've seen so far and certainly nothing I personally own, but I will keep looking.

My impression is that it's a very weak claim. This is a by-product of the complex and restrictive criteria needed for it to be correct.

 

[JRS]

I believe we need to revise the allowable output impedances as the 2 ohms (lenient) was wrt headphones with 33 ohm impedance.

Ignoring the effects of running out of current and the attendant voltage sag, and assuming a nominal impedance of 8 ohms with a minima of 4,
we get 20 log [8/10]/[4/6] = 20 log [1.2] = 1.58dB

If we take use the strict value of 0.2 ohms we get 20 log [8/8.2]/[4/4.2] = 20 log [ 2*4.2/8.2] = 20 log [1.024] = 0.21 dB, which seems like a fair variation.

Of course were we to drop it to SOTA amps such as Purifi or Hypex, the output impedances are in the micro-ohms at 1kHz and can be considered load invariant. Again with simple voltage divider assumptions.

 

[jae]

This is just a SINAD list, it doesn't cover the criteria. Or do you have a different position, that a specific SINAD threshold is all an amp needs to basically be in kern(amps)?


Not the ones I've seen so far and certainly nothing I personally own, but I will keep looking.

My impression is that it's a very weak claim. This is a by-product of the complex and restrictive criteria needed for it to be correct.

You can click on the "source" link for each amplifier for the rest of the measurements from the original reviews, most of the sources using AP analysers will have a variety of measurements and you will see that all the top measuring stuff usually satisfies most if not all the "lenient" criteria throughout most of its operational range.

 

[symphara]

You can click on the "source" link for each amplifier for the rest of the measurements from the original reviews, most of the sources using AP analysers will have a variety of measurements and you will see that all the top measuring stuff usually satisfies most if not all the "lenient" criteria throughout most of its operational range.

Wait, what?

For "transparency", I think they should satisfy the "strict" criteria. The "lenient" stuff is a "maybe, depending on the listener" thing, if you look at the fine print.

Now you're telling me that not all "maybe"s are required, and we just go with "most", or "some"?

 

[jae]

Wait, what?

For "transparency", I think they should satisfy the "strict" criteria. The "lenient" stuff is a "maybe, depending on the listener" thing, if you look at the fine print.

Now you're telling me that not all "maybe"s are required, and we just go with "most", or "some"?

Perhaps it was rhetorical one, but I was simply responding to your question: "I wonder how many >100W (8Ω) amplifiers are there, that pass 'lenient'?".

Based on whatever the "lenient" criteria was quoted above, there are plenty on the list that satisfy and exceed that criteria.

Ignoring the 120 dB requirement, many even satisfy the other criteria for "strict": a 0.1 dB FR/channel balance, low output impedance etc. are not concerns in most of the new well-engineered amplifiers. I would say it ceases to be a "depending on the listener" thing from a practical point of view if it can meet those criteria and can make the 85-100+ dB mark, which a number do.

As for amplifiers specifically with >120 dB SINAD, there are virtually none measured.

 

[Blumlein 88]

Wait, what?

For "transparency", I think they should satisfy the "strict" criteria. The "lenient" stuff is a "maybe, depending on the listener" thing, if you look at the fine print.

Now you're telling me that not all "maybe"s are required, and we just go with "most", or "some"?

You can slice and dice the requirements, but this is always going to be a gray area. There is no dead clean and clear set of criteria unless they are all far beyond reproach. There are a small number of amps that can do that. You have THD all below -120 db as good enough, but possibly -80 db is going to be enough for 98% of people under conditions of music listening or maybe 99%. A wide somewhat gray area.

Without looking, I think the Purifi amps meet these criteria in the lenient case, Benchmark, Mola Mola and some of the Hypex units. What are you seeing they don't meet?

 

[symphara]

You can slice and dice the requirements, but this is always going to be a gray area. There is no dead clean and clear set of criteria unless they are all far beyond reproach. There are a small number of amps that can do that. You have THD all below -120 db as good enough, but possibly -80 db is going to be enough for 98% of people under conditions of music listening or maybe 99%. A wide somewhat gray area.

That's fine, I personally have no issue with someone postulating that amps passing the "strict" criteria are indistinguishable by any human observer. It sounds eminently reasonable. But I don't think it applies to much, if anything.

Then you get into this grey area, that's also fine, but I don't think anyone can act as if someone swore in the ASR church if there's some claim of hearing differences in the grey area. Not saying there are myself, in this particular example that started the thread - just to be clear. But you have plenty of members who went directly for snark and insults, as always the case, which unfortunately has been building a particular reputation for this website.

Without looking, I think the Purifi amps meet these criteria in the lenient case, Benchmark, Mola Mola and some of the Hypex units. What are you seeing they don't meet?

Those are good, yes.

Most amplifiers on the market don't seem to have this level of detail available at all, so there's nothing to be said. I've been looking at the usual stuff I find at dealer auditions - Bryston, Accuphase, Rotel (Michi), Krell, Naim - it's usually the case they fail even some of the "lenient" criteria listed above.

Sometimes technical reviews are strange. Looking at the Lyngdorf TDAI-3400, I found two articles right here. The first shows it's a basket-case of bad measurements. The second finds it's rather good because of its EQ and therefore is recommended...

 

[KSTR]

As for amplifiers specifically with >120 dB SINAD, there are virtually none measured.

That's because there is a physical thermal noise brickwall which is limiting SINAD.
It depends on:

• amplifier gain
• test level (output voltage)
• source impedance and residual noise
• measurement bandwidth

Typical amps have 26dB gain (20x) and are measured at 5W here which is 4.5Vrms for a nominal 4Ohm load. Source impedance often in the 50Ohms range. Measurement bandwidth is 20Hz to 20kHz.

Assuming a noiseless source and a noise- and distortionless amplifier this gives a 124dB brickwall:
50hms source resistance (at room temperature) with 20x of gain creates 2.6uVrms of noise in a 20Hz...20kHz bandwidth.
4.5V / 2.6uV = 1,730,000
20*log10(1,730,000) = 124dB

 

[Blumlein 88]

That's because there is a physical thermal noise brickwall which is limiting SINAD.
It depends on:

• amplifier gain
• test level (output voltage)
• source impedance and residual noise
• measurement bandwidth

Typical amps have 26dB gain (20x) and are measured at 5W here which is 4.5Vrms for a nominal 4Ohm load. Source impedance often in the 50Ohms range. Measurement bandwidth is 20Hz to 20kHz.

Assuming a noiseless source and a noise- and distortionless amplifier this gives a 124dB brickwall:
50hms source resistance (at room temperature) with 20x of gain creates 2.6uVrms of noise in a 20Hz...20kHz bandwidth.
4.5V / 2.6uV = 1,730,000
20*log10(1,730,000) = 124dB

This is why I posted thd not thd+n.

 

[Blumlein 88]

That's fine, I personally have no issue with someone postulating that amps passing the "strict" criteria are indistinguishable by any human observer. It sounds eminently reasonable. But I don't think it applies to much, if anything.

Then you get into this grey area, that's also fine, but I don't think anyone can act as if someone swore in the ASR church if there's some claim of hearing differences in the grey area. Not saying there are myself, in this particular example that started the thread - just to be clear. But you have plenty of members who went directly for snark and insults, as always the case, which unfortunately has been building a particular reputation for this website.


Those are good, yes.

Most amplifiers on the market don't seem to have this level of detail available at all, so there's nothing to be said. I've been looking at the usual stuff I find at dealer auditions - Bryston, Accuphase, Rotel (Michi), Krell, Naim - it's usually the case they fail even some of the "lenient" criteria listed above.

Sometimes technical reviews are strange. Looking at the Lyngdorf TDAI-3400, I found two articles right here. The first shows it's a basket-case of bad measurements. The second finds it's rather good because of its EQ and therefore is recommended...

What about this Benchmark doesn't clear the lenient limits and come close to the strict ones? It has been measured here, and they have AP measures in the owners manual.

Benchmark AHB2 Power Amplifier

Benchmark's new AHB2 Power Amplifier - "The Quietest, Cleanest Audio Amplifier on the Planet"

benchmarkmedia.com

 

[Galliardist]

Lenient spec:
Dynamic range, linearity: 96 dB
THD, IMD: -66 dBFS / 0.05%
Noise: -85 dBFS / 0.005%
SINAD: 85 dB
Crosstalk: -60 dBFS
Frequency response: ±0.5 dB/octave
Channel balance: 1 dB
Output impedance: 2 ohms

What I can find quickly for low gain figures for the tested amps - only what I can read, I have not tried to calculate anything or guess from graphs

ASR review (obvious numbers)
Magni 3 Heresy
Dynamic range, linearity:
THD, IMD:
Noise: SNR 122
SINAD: 119
Crosstalk:
Frequency response: flat
Channel balance: less than 1dB
Output impedance: 1.3ohm

Magni 3+
Dynamic range, linearity:
THD, IMD:
Noise: SNR 119
SINAD: 103
Crosstalk:
Frequency response: flat
Channel balance: ~1dB
Output impedance:1.3ohm

Notes: the review refers to the possibility of channel imbalance from the volume control with either model, same control

Manufacturer specs (again, I'm not trying to calculate other figures) - these are from other sites, models replaced on company website
Magni 3 Heresy
Dynamic range, linearity:
THD, IMD: not given/Less than 0.0006% (-105dB) at 1V RMS into 32 ohms, CCIF
Noise: SNR Greater than 115dB, A weighted, referenced to 1V RMS
SINAD:
Crosstalk: Less than -70dB, 20 Hz-20 kHz in either gain
Frequency response: 20Hz-20Khz, +/-0.04dB
Channel balance:
Output impedance:Less than 0.25 ohms at either gain

Magni 3+
Dynamic range, linearity:
THD, IMD: unknown/Less than 0.002% (-95dB) at 1V RMS into 32 ohms, CCIF
Noise: SNR Greater than 115dB, A weighted, referenced to 1V RMS
SINAD:
Crosstalk: Less than -70dB, 20 Hz-20 kHz in either gain
Frequency response: 20Hz-20Khz, +/-0.04dB
Channel balance:
Output impedance:Less than 0.2 ohms at either gain

Because they have replaced these models, the measurement report that would have been on the website is not available to me.

Another thing of note is the out of spec output impedence in Amir's test results, though it remains within the lenient specification.

While some figures are missing, everything I can find suggests that apart from the volume control, measurements show these amps within the lenient spec for these two in low gain.
The part 1 test in high gain has issues, but the part 2 test is OK if the amps are to specification or similar to the devices Amir tested, unless someone can show poor linearity.

This is no substitute for measurements on the actual equipment. What was published there has no suggestion of anything different, though.

It's worth remembering that the lenient spec is just a guideline. We do not have strict research criteria for the lenient threshold. As such, the Part 2 test is, strictly speaking, an "allowable" result - though one that has decades of empirical research stacked against it, and a test that does not look to be valid.

If that lenient spec is some kind of semi-official benchmark here, it would be nice if reviews gave the measurements listed in an obvious way. I guess though that it's just another thread at the end of the day.

 

[symphara]

What about this Benchmark doesn't clear the lenient limits and come close to the strict ones? It has been measured here, and they have AP measures in the owners manual.

Benchmark AHB2 Power Amplifier

Benchmark's new AHB2 Power Amplifier - "The Quietest, Cleanest Audio Amplifier on the Planet"

benchmarkmedia.com

I didn't say it doesn't. But we made tremendous progress from "all amps sound the same" to "look, there's Benchmark and this $13000 Mola Mola".