Flat Frequency Response/Different Sound?

[IAtaman]

But, these effects are well understood by both amplifier and speaker designers, and competent ones are sure to take appropriate steps to ensure we as consumers don’t run into trouble when we wire them together.

I think that is a fair point, and it is safe to assume design engineers have not forgotten to consider what happens when the load reactance increases, and that we should not be worried the amp will explode just because the phase response of the headphone we have connected to its output is a bit wonky. I also suspect the effect of that on distortion and frequency response would be limited for some amps, but maybe not so limited for others? As you nicely explained, amps usually overshoot with reactive loads, so maybe that effects tonality? If we took two amps people claim sound different, measure distortion and output levels across the frequency range for a number of different headphones some of which have low phase angles within 20-20K range, and close this topic once and for all, that would be great in my opinion.

For what tests you'd like to see, I actually gave this some thought and had a running list, but the more I went into the details of the data Amir publishes, the more I was able to figure out what I wanted to know. For example, I wanted to see THD% vs Frequency at two output power levels; at 0.05mW and max rated power, but then we already have THD vs power graphs so that would probably be redundant. Might still be useful to have see such a graph though if it is easy to do with the AP equipment, as at least it would make the arguments about about amplifier tonality harder to argue. I don't know what exactly are the mechanisms that are suggested by those who claim amps have tonality so would be good to get some input from that side as well.

One thing is not clear however, I hear people talk about transient power, especially in relation to low end, and I am not sure if any of the measurements in the current measurement suite cover that. If not, would be good to have one to get an idea whether amps behave differently at different output power levels for transients.

Just to share, I wanted to see if I can find out what would be the THD% for AK371 at normal listening levels for A30Pro and RME ADI-2, both of which are very realistic scenarios I believe, especially for this forum, given how highly Amir rates these products. If my graph readings are not incorrect THD% for A30Pro is around -85db and around -81db for RME (HPA side only). Very respectable, but not 120db. I don't know if this makes them transparent or not but 85db is less than 16 bits, so maybe cassette people were right all along

 

[Cars-N-Cans]

I think that is a fair point, and it is safe to assume design engineers have not forgotten to consider what happens when the load reactance increases, and that we should not be worried the amp will explode just because the phase response of the headphone we have connected to its output is a bit wonky. I also suspect the effect of that on distortion and frequency response would be limited for some amps, but maybe not so limited for others? As you nicely explained, amps usually overshoot with reactive loads, so maybe that effects tonality? If we took two amps people claim sound different, measure distortion and output levels across the frequency range for a number of different headphones some of which have low phase angles within 20-20K range, and close this topic once and for all, that would be great in my opinion.

For what tests you'd like to see, I actually gave this some thought and had a running list, but the more I went into the details of the data Amir publishes, the more I was able to figure out what I wanted to know. For example, I wanted to see THD% vs Frequency at two output power levels; at 0.05mW and max rated power, but then we already have THD vs power graphs so that would probably be redundant. Might still be useful to have see such a graph though if it is easy to do with the AP equipment, as at least it would make the arguments about about amplifier tonality harder to argue. I don't know what exactly are the mechanisms that are suggested by those who claim amps have tonality so would be good to get some input from that side as well.

One thing is not clear however, I hear people talk about transient power, especially in relation to low end, and I am not sure if any of the measurements in the current measurement suite cover that. If not, would be good to have one to get an idea whether amps behave differently at different output power levels for transients.

Just to share, I wanted to see if I can find out what would be the THD% for AK371 at normal listening levels for A30Pro and RME ADI-2, both of which are very realistic scenarios I believe, especially for this forum, given how highly Amir rates these products. If my graph readings are not incorrect THD% for A30Pro is around -85db and around -81db for RME (HPA side only). Very respectable, but not 120db. I don't know if this makes them transparent or not but 85db is less than 16 bits, so maybe cassette people were right all along

They would be interesting tests to see run. Given how much loser headphone amp design is I suspect there will indeed be impacts on at least the FR, and even potentially distortion depending on their output topology. The most relevant test for that will be the THD+N% vs. output voltage. The less sensitive it is, the more the curves will just stack on top of each other, with the reverse being true for increasing sensitivity. Quick example:

Now, to complete the picture, we would also want to run a THD+N% vs frequency as you allude to in some form, and at chosen fixed power outputs would be a sensible solution. Alternatively, we could choose two impedances, namely 33 ohms and 300 ohms, and do distortion sweeps to clipping at various frequencies as is done with amplifiers:

With these two measurements, you have pretty much fully characterize the amplifier, or at least enough to have confidence in if changing impedances will impact its FR or distortion at various frequencies and outputs. There is enough data to be able to do some reasonable interpolation. Some complications, though, are that the measurement will be impacted by the gain settings, so the full suite of measurements could get quite large at the end due to the need to repeat for all possible configurations.

For the transient power, this is more commonly known as the peak power at 1% THD for speaker amps:

But, given that many headphone amplifiers are already becoming speaker amplifiers in disguise, I don't think its necessary. Back in the day when you only had milliwatts to work with, it was a concern. But when some of these can push out things like 10W of power and 30VP-P of output, the bigger concern would be hearing safety and survival of IEMs if the thing accidentally was run at max volume by the source. Its a concern with low power things like dongles and portable devices with integral amplifiers that are very anemic, but most stand-alone headphone amps should not be a concern since we already know power will likely be available in abundance to begin with.

 

[Cars-N-Cans]

An additional point about the "transient power" is that some people could be thinking about slew rate, which is not an issue with amplifiers. If it can make rated power all the way up to 20 kHz and beyond, then such concerns are a nonpoint since it already has more slew rate than it could ever need. Audio has a 1/f spectrum, so the higher the frequency, the lower the output requirements in terms of power. But, if an amplifier is able to generate a peak of 40 volts of output at 20 kHz into its rated load (either 4 or 8 ohms) with little distortion, then its slew rate will be at least 5 V/μs. However in practice playing music it probably will only ever need a fraction of that due to the bulk of the spectral content mainly being at low frequencies and people not typically running their amplifier at 100% rated output.

 

[antcollinet]

It really isn't sufficient as there is audible differences between amps on certain speakers.

It really isn't sufficient because it doesn't show IRL performance on complex loads, shifting phases, impedance dives and bumps.

Measuring into R load is steady state. Not real.
Some of you seem to like analogies. Here is one; it's like measuring a race cars performance only on acceleration on the bench, not spring load, shock absorbers, wheel geometry and downforce.

So again the question

Which complex load would you test with. Because whichever you test with will only be valid for that load. For equallly complex but different loads it will be no more valid than a resistive load.

And you sure as hell aren't going to be able to test with all complex loads.

Simple fact is a knowledge of the output current capability and frequency response into a resistive load, together with the output impedence gives you the capability to calculate the response into any complex load if you know the characteristics of that load.

The THD/Frequency/voltage/power curves referenced above by @Cars-N-Cans give you more confidence that the amp behaves well under a range of conditions.

Doing more consumes significant time and effort with little value.

 

[IAtaman]

An additional point about the "transient power" is that some people could be thinking about slew rate, which is not an issue with amplifiers.

I dont think so, but I am not sure. I think they were talking about power rails' ability to keep up rather than opamps ability to swing.

 

[antcollinet]

I dont think so, but I am not sure. I think they were talking about power rails' ability to keep up rather than opamps ability to swing.

If the power rails can't keep up, then the op amps can't swing.

IE in order to demonstrate "speed" which can be nothing more than the ability to deliver high freqency at high power (or flat freqency response at full power), all parts of the amp need to deliver. Input stage, intermediate/gain stages, feedback circuits, output stage, power supply.

 

[Cars-N-Cans]

I dont think so, but I am not sure. I think they were talking about power rails' ability to keep up rather than opamps ability to swing.

Ah, ok. If there is a lot of ripple on the rails, it can cause more distortion as the amplifier has to reject it, and its PSRR decreases with increasing frequency. Also it will naturally lead to clipping as the voltage falls since there will be no more headroom for the amplifier's needed voltage swing. Maybe that's what they are thinking of. The usual way of dealing with this in the audiophilia world is to use a gigantic toroidal transformer or two and row after row of electrolytic capacitors, but class-D amplifiers nowadays almost always have well-regulated electronic power supplies instead of unregulated linear ones. This is esp. true with energy efficiency regulations moving to also regulate power factor as well, and linear supplies with oodles of capacitors have abysmal power factors. People will point to the "lack of headroom" with regulated supplies, but when the amplifier already puts out 600 or 800 watts, that is the headroom.

 

[IAtaman]

The usual way of dealing with this in the audiophilia world is to use a gigantic toroidal transformer or two and row after row of electrolytic capacitors,

When I was in middle school decades ago, I used to buy these books with circuit drawings and components lists, would go and buy those components from electronic components suppliers, build my own pcb with UV mask sprays and iron-III-cloride which would stain every thing it touched, and would connect these large toroidal transformers to large bridge rectifiers and 2 rows of 2x10uF caps which would cause the lights to dim due to high inrush currents when you powered them on

But yes, exactly, I think one of the arguments I heard was indeed the linear power supplies would have enough power stored in the filter caps to handle transients well whereas smps devices used in cheaper amps would not be able to - not sure whether it is a fair arguement or not.

 

[antcollinet]

When I was in middle school decades ago, I used to buy these books with circuit drawings and components lists, would go and buy those components from electronic components suppliers, build my own pcb with UV mask sprays and iron-III-cloride which would stain every thing it touched, and would connect these large toroidal transformers to large bridge rectifiers and 2 rows of 2x10uF caps which would cause the lights to dim due to high inrush currents when you powered them on

But yes, exactly, I think one of the arguments I heard was indeed the linear power supplies would have enough power stored in the filter caps to handle transients well whereas smps devices used in cheaper amps would not be able to - not sure whether it is a fair arguement or not.

Whether or not it is a fair argument doesn't really matter, and it will depend on the specifics of each design. It is possible to design a linear PSU with less "peak" capability than switch mode power supply with the same continuous rating.

In any case the test for it in an amplifier is covered by "can the amp deliver 20Khz at full power"

If it can, then the capablity is there to fully deliver any audible transient within the power limits of the amp. That test shows you whether the PSU is able to deliver the power for transients - as well as alll the other electronics in the amp being able to do so.

This is related to the fact that the impulse response of a system is calculable from the frequency response of that system.

 

[Cars-N-Cans]

When I was in middle school decades ago, I used to buy these books with circuit drawings and components lists, would go and buy those components from electronic components suppliers, build my own pcb with UV mask sprays and iron-III-cloride which would stain every thing it touched, and would connect these large toroidal transformers to large bridge rectifiers and 2 rows of 2x10uF caps which would cause the lights to dim due to high inrush currents when you powered them on

But yes, exactly, I think one of the arguments I heard was indeed the linear power supplies would have enough power stored in the filter caps to handle transients well whereas smps devices used in cheaper amps would not be able to - not sure whether it is a fair arguement or not.

I'm in the same boat lol. My 40W desktop amplifier that I built some time back has a modest electronic power supply, but I was too lazy to bother with any inrush limiting, instead just opting to use a big, chunky bridge rectifier and power switch. The only thing really between the capacitors and the wall outlet is the switch, the common-mode filter, and the fuse. Amp comes on, lights go off... Its surprising the filter caps put up with that sort of abuse. Its only for a cycle or two of AC, but still something like 30 or 35 amps of current when it switches on. And yeah, that ferric chloride really makes a mess, and is permanent for anything it comes in contact with. I think I still have a tub of the solution laying around. Probably has turned into crystals by now.

And speaking of which, I think the issue or argument with the SMPS is down to it being regulated. With linear supplies, the rail voltage will float up to 1.41 x the RMS output voltage of the transformer since the caps can charge to the peak voltage. Under load, they will sag down to the RMS voltage, and below depending on how much capacitance there is. I think most of us already know the rest of that story. Conversely with an SMPS, instead of relying on the caps floating up to the power rails, its better to just choose the voltage you want at the outset. Additionally it doesn't really make sense to store power when the thing can run up into the 100's of kHz instead of 100 or 120 Hz. The caps are just to smooth the output waveform from the SMPS' transformer and high speed rectifiers or whatever is being used ahead of the caps. Since it can also regulate in real time and has the ability to control both its output voltage and current, I would say from a design perspective the SMPS is better since its well defined. In terms of output power, would one rather have 600 watts for 50 milliseconds, or 600 watts for as long as the wall socket its plugged into has power? Maybe it really is an unfair argument in disguise, but from a performance perspective I think the answer is obvious. The only time its really an issue is with the cheapie Amazon amps that just use some random-ass power brick. There god only knows what its peak power output is, if it will result in spontaneous home combustion, or an extra 120/240V zing for your headphones down the road. Much like DSP in super-cheap active speakers and headphones, the economy and performance provided by switching power supplies is often abused in the name of maximizing profit. Reminds me of when HR at one of my jobs that was super-safety conscious contracted some random Chinese company to make company logo'd small USB power supplies to include in gift bags to employees. I took one apart. This is what it was inside. Seriously.

They also had them made up for QHSE as well with their logo on the supplies. That one was fucking MINT! I think I used it once. It worked. I didn't die. But, I was sure to stay well away from anything referenced to ground

 

[Cars-N-Cans]

But, still in all seriousness the power supply itself is probably something that should also be considered since it impacts more than just performance, and is often noted by Amir to his credit when something lacks any safety regulatory markings.

 

[IAtaman]

I'm in the same boat lol. My 40W desktop amplifier that I built some time back has a modest electronic power supply, but I was too lazy to bother with any inrush limiting, instead just opting to use a big, chunky bridge rectifier and power switch. The only thing really between the capacitors and the wall outlet is the switch, the common-mode filter, and the fuse. Amp comes on, lights go off... Its surprising the filter caps put up with that sort of abuse. Its only for a cycle or two of AC, but still something like 30 or 35 amps of current when it switches on. And yeah, that ferric chloride really makes a mess, and is permanent for anything it comes in contact with. I think I still have a tub of the solution laying around. Probably has turned into crystals by now.

At some point in time, I found this website called Elliot's Sound Products, where there was all kinds of audio projects and I always wanted to build his soft start circuit he had but never got around to it.

Holy moly that is scary! Also a big fan of bigclive as well!

I think you are right, it boils down to regulation and probably my confusion is caused by the fact that I do not have any hands on experience with smps's beyond a high level understanding of how they operate.

 

[nas]

all of them. (Dunno maybe there are some extreme distorting tube amps where youll hear the distortion)

I have tried my DIY spakers with Dayton PS-95 fullranges with some amps on "everything 0" settings - they sound DIFFERENT. I cannot say definitely if this is group delay, or some EQ is going on, or clipping or anything. I don't know the exact terms to everything my ears hear. Even at 0.1W listening levels - they sound very different.

Of course most of the amps are not on the level of this community - Topping MX3, aliexpress TPA 3116amps, philips micro hifi, some old (pre hdmi) denon and yamaha AVRs, Denon 2500H, and so on. The best sounding for my liking and also I think with the least distortion were TPA3116 based class D boards. As a bonus they have the smoothest and softest bass freq reproduction. From my "golden" ears I could guess best of them are SINAD 70dB range at most. So, if I got -45dB on this test, my distortion limit is SINAD 45dB????

 

[Cars-N-Cans]

I have tried my DIY spakers with Dayton PS-95 fullranges with some amps on "everything 0" settings - they sound DIFFERENT. I cannot say definitely if this is group delay, or some EQ is going on, or clipping or anything. I don't know the exact terms to everything my ears hear. Even at 0.1W listening levels - they sound very different.

Of course most of the amps are not on the level of this community - Topping MX3, aliexpress TPA 3116amps, philips micro hifi, some old (pre hdmi) denon and yamaha AVRs, Denon 2500H, and so on. The best sounding for my liking and also I think with the least distortion were TPA3116 based class D boards. As a bonus they have the smoothest and softest bass freq reproduction. From my "golden" ears I could guess best of them are SINAD 70dB range at most. So, if I got -45dB on this test, my distortion limit is SINAD 45dB????

Keep in mind anything with tone controls will have some "EQ" whether you like it or not, esp. with old-school Baxandall type controls. No matter where they are set to, its still not "flat" most of the time due to how imprecise potentiometers are. There is also the issue of the DC offset in linear amplifiers. If this is just a simple capacitor in the inverting side to roll the gain off to unity at DC, it can cause the low frequency response to roll off quite a bit further up in the bass region in some simpler designs. The only amplifiers that I have that do sound identical are always ones that are essentially flat from DC out to 40-50 kHz. The output is just gain * input. This applies to things like PCs and laptops as well. Those are never flat, either, esp. if they have some gimmick for improving the sound. Cellphones often do this as well. Its surprising how often things are simply just not flat. Really the only way to be sure is to measure if there is anything more than a power switch, and maybe a gain switch. Even though I don't often agree with subjectivist audiophiles, one thing I can agree on is getting as many controls off of stand-alone amplifiers as possible so they just amplify and that's it. But, doing measurements on the output side of power amplifiers carries a lot of risks, so its not really a trivial task to verify it directly if you are not familiar with making measurements. Lots of ways to burn things out or get incorrect readings, even if you have some experience. Good DVMs can be pressed into service to measure the RMS output voltage to verify, but it still requires some form of dummy load, and a source with a flat response that can provide a pure test tone. And don't even remotely expect tube amps to all sound the same due to all the shortcomings they have. Only solid state amps can really be expected to be invariant with respect to what speakers they are hooked to.

As for the 45 dB result from the Klippel listening tests, that's been my experience as well from doing all sorts of listening tests in the past. Distortion that is more than around 40-50 dB below the audio is generally inaudible. There are a number of mechanisms that limit how far down you can hear. In terms of amplifiers, getting them to have a SINAD of around 85 dB or better is fairly easy. The part that is often the issue is the frequency response. That part the ear is very good at detecting if there are differences. Same thing with relative amplification. The amplifiers have to be level matched before they can be compared, preferably by using a test tone and measuring the output voltage to verify the two signal chains are identical with respect to output level. I can't stress that enough as your ear does not have a flat response, and this is actually readily audible with test tones. Even in the flattest speakers and most compliant headphones the perceived SPL level of a pure tone will vary with frequency, and those variations will depend on amplitude. The reasons why are beyond the scope of this discussion, but the important phrase of the day is Fletcher-Munson Curves and these will be omnipresent in any sort of listening tests you do:

Flipping the curves upside-down provides conceptual "FR" curves for roughly visualizing the response of the ears analogous to headphones:

Those are built into you, and will be present no matter what you hear, whether it be sound around you, or sound from your audio equipment, and impact the timbre of music. This is why old-school "Loudness" controls existed. The idea was that they would make music sound "fuller" at low volumes. In reality due to speakers of the time being garbage, esp. in cars, the loudness was always on as the tone controls didn't have enough authority on their own to get anything resembling neutral tonality. But long story short: Didn't match the levels? They will sound different. Did a really bad job or didn't bother at all? They will sound REALLY different! I'd bet scores of audiophile tomes on this and that have spawned from the effect alone. Even Amir often comments on this in his listening tests and videos on topics related to comparative listening. They are worth watching.

 

[nas]

Keep in mind anything with tone controls will have some "EQ" whether you like it or not, esp. with old-school Baxandall type controls. No matter where they are set to, its still not "flat" most of the time due to how imprecise potentiometers are. There is also the issue of the DC offset in linear amplifiers. If this is just a simple capacitor in the inverting side to roll the gain off to unity at DC, it can cause the low frequency response to roll off quite a bit further up in the bass region in some simpler designs. The only amplifiers that I have that do sound identical are always ones that are essentially flat from DC out to 40-50 kHz. The output is just gain * input. This applies to things like PCs and laptops as well. Those are never flat, either, esp. if they have some gimmick for improving the sound. Cellphones often do this as well. Its surprising how often things are simply just not flat. Really the only way to be sure is to measure if there is anything more than a power switch, and maybe a gain switch. Even though I don't often agree with subjectivist audiophiles, one thing I can agree on is getting as many controls off of stand-alone amplifiers as possible so they just amplify and that's it. But, doing measurements on the output side of power amplifiers carries a lot of risks, so its not really a trivial task to verify it directly if you are not familiar with making measurements. Lots of ways to burn things out or get incorrect readings, even if you have some experience. Good DVMs can be pressed into service to measure the RMS output voltage to verify, but it still requires some form of dummy load, and a source with a flat response that can provide a pure test tone. And don't even remotely expect tube amps to all sound the same due to all the shortcomings they have. Only solid state amps can really be expected to be invariant with respect to what speakers they are hooked to.

As for the 45 dB result from the Klippel listening tests, that's been my experience as well from doing all sorts of listening tests in the past. Distortion that is more than around 40-50 dB below the audio is generally inaudible. There are a number of mechanisms that limit how far down you can hear. In terms of amplifiers, getting them to have a SINAD of around 85 dB or better is fairly easy. The part that is often the issue is the frequency response. That part the ear is very good at detecting if there are differences. Same thing with relative amplification. The amplifiers have to be level matched before they can be compared, preferably by using a test tone and measuring the output voltage to verify the two signal chains are identical with respect to output level. I can't stress that enough as your ear does not have a flat response, and this is actually readily audible with test tones. Even in the flattest speakers and most compliant headphones the perceived SPL level of a pure tone will vary with frequency, and those variations will depend on amplitude. The reasons why are beyond the scope of this discussion, but the important phrase of the day is Fletcher-Munson Curves and these will be omnipresent in any sort of listening tests you do:

View attachment 264491

Flipping the curves upside-down provides conceptual "FR" curves for roughly visualizing the response of the ears analogous to headphones:

View attachment 264492

Those are built into you, and will be present no matter what you hear, whether it be sound around you, or sound from your audio equipment, and impact the timbre of music. This is why old-school "Loudness" controls existed. The idea was that they would make music sound "fuller" at low volumes. In reality due to speakers of the time being garbage, esp. in cars, the loudness was always on as the tone controls didn't have enough authority on their own to get anything resembling neutral tonality. But long story short: Didn't match the levels? They will sound different. Did a really bad job or didn't bother at all? They will sound REALLY different! I'd bet scores of audiophile tomes on this and that have spawned from the effect alone. Even Amir often comments on this in his listening tests and videos on topics related to comparative listening. They are worth watching.

Of course I did not do any objective measurements. The most I can do - measure with speaker->microphone setup, and not outside, as it is kinda cold on this part of year. Electric amp output measurement is out of my understanding and capabilities.

You emphasized tone controls: some if not most those (rubbish) amps were/are so bad for my ears, that there is no dial or potentiometer to fix tone imbalance. I know this graph, for my personal ears ~1-2kHz is the most annoying. Of course I played with most of their abysmal EQ settings, and sometimes use EQ on computer, but... there is no fix for garbage things. Most are just bad.
For last 2+ years I keep source constant - PC with Apple 3.5mm dongle DAC.

My question still holds: which amp SINAD, THD+N my ear can tolerate as fully transparent based on this listening test? Can we use this test for such conclusions?

p.s. sorry for hijacking thread...

 

[HarmonicTHD]

Of course I did not do any objective measurements. The most I can do - measure with speaker->microphone setup, and not outside, as it is kinda cold on this part of year. Electric amp output measurement is out of my understanding and capabilities.

You emphasized tone controls: some if not most those (rubbish) amps were/are so bad for my ears, that there is no dial or potentiometer to fix tone imbalance. I know this graph, for my personal ears ~1-2kHz is the most annoying. Of course I played with most of their abysmal EQ settings, and sometimes use EQ on computer, but... there is no fix for garbage things. Most are just bad.
For last 2+ years I keep source constant - PC with Apple 3.5mm dongle DAC.

My question still holds: which amp SINAD, THD+N my ear can tolerate as fully transparent based on this listening test? Can we use this test for such conclusions?

p.s. sorry for hijacking thread...

It will certainly give you a very good estimate. Assume you would be reaching an excellent 60dB SINAD from the tests. DACs or AMPs with 120dB are 1000x lower than your best effort. Even AVRs easily reach 90dB so they exhibit ca 32x less SINAD than your best effort. There are of course difference whether you listen to music (harder) vs test tones (a bit easier, especially for trained listeners).

The topic has also been debated here.

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

In comparison even good speakers get closer to ca. 40 to 50dB THD (see 86 and 96dB SPL tests), that‘s why one should focus on good speakers (and room) and don’t fret about electronics too much.

 

[IAtaman]

It will certainly give you a very good estimate. Assume you would be reaching an excellent 60dB SINAD from the tests. DACs or AMPs with 120dB are 1000x lower than your best effort.

I am not disputing what you are saying, you might probably be right overall, I don't know. I am quite certain though that characterization of an amp that has best case SINAD 120db as 1000x lower noise and distortion than what you can hear is not correct, simply due to the fact that 120db is best case SINAD, or SINAD measured at one output power and frequency, and even the best measuring amps can do a lot worse in real life conditions, again not claiming worse measurement means audible.

Take A30Pro for example, which is listed as 121DB at amplifier SINAD list. If you are using it to listen to AK371 on normal levels then, If I am doing the math correctly, you will be using 0.03mW power. At that power output A30PRo THD+N% is -85 db, excluding noise, not -120db. Difference is still around 18 times, so probably still transparent from that perspective, nevertheless the difference is still 2 levels of magnitude lesser than what calling it a 120db amp implies.

 

[HarmonicTHD]

I am not disputing what you are saying, you might probably be right overall, I don't know. I am quite certain though that characterization of an amp that has best case SINAD 120db as 1000x lower noise and distortion than what you can hear is not correct, simply due to the fact that 120db is best case SINAD, or SINAD measured at one output power and frequency, and even the best measuring amps can do a lot worse in real life conditions, again not claiming worse measurement means audible.

Take A30Pro for example, which is listed as 121DB at amplifier SINAD list. If you are using it to listen to AK371 on normal levels then, If I am doing the math correctly, you will be using 0.03mW power. At that power output A30PRo THD+N% is -85 db, excluding noise, not -120db. Difference is still around 18 times, so probably still transparent from that perspective, nevertheless the difference is still 2 levels of magnitude lesser than what calling it a 120db amp implies.

View attachment 264517

Then take the SINAD at the Power and Frequency you think is more appropriate. The concept is the same and the data is there. Also take the test for whatever frequency you think is appropriate (a better test is probably the Distort software from @pkane ) so your are comparing apples with apples and you consider Munson-Fletcher (because in general our hearing is quite sensitive around 1kHz plus minus).

 

[Cars-N-Cans]

Of course I did not do any objective measurements. The most I can do - measure with speaker->microphone setup, and not outside, as it is kinda cold on this part of year. Electric amp output measurement is out of my understanding and capabilities.

You emphasized tone controls: some if not most those (rubbish) amps were/are so bad for my ears, that there is no dial or potentiometer to fix tone imbalance. I know this graph, for my personal ears ~1-2kHz is the most annoying. Of course I played with most of their abysmal EQ settings, and sometimes use EQ on computer, but... there is no fix for garbage things. Most are just bad.
For last 2+ years I keep source constant - PC with Apple 3.5mm dongle DAC.

My question still holds: which amp SINAD, THD+N my ear can tolerate as fully transparent based on this listening test? Can we use this test for such conclusions?

p.s. sorry for hijacking thread...

As for which are “tolerable”? If you want to be anal, don’t accept anything with less than a SINAD of 85 dB. That will have you covered with respect to distortion. As far as hiss and hum, shoot for a SINAD of 100 dB or better. In reality those numbers can be a good 15 dB or 20 dB less, but since modern designs can get there and we care about objective performance, it’s probably worth it.

As far as the rubbish amps, a vey common problem also is channel imbalance as well. If you have that or any roll-lots I’m FR at the ends of the response, forget it. While you don’t have to go outside or anything to measure, having a simple tone generator (can be a cellphone, just has to play 1 kHz) and a good true RMS DVM will suffice since all you are looking for is both amps, or both channels, to show the same number at 1 kHz.

 

[nas]

It will certainly give you a very good estimate. Assume you would be reaching an excellent 60dB SINAD from the tests. DACs or AMPs with 120dB are 1000x lower than your best effort. Even AVRs easily reach 90dB so they exhibit ca 32x less SINAD than your best effort. There are of course difference whether you listen to music (harder) vs test tones (a bit easier, especially for trained listeners).

The topic has also been debated here.

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

In comparison even good speakers get closer to ca. 40 to 50dB THD (see 86 and 96dB SPL tests), that‘s why one should focus on good speakers (and room) and don’t fret about electronics too much.

Yes, I definitely focus on speakers, on DIY side of things. I am on 500 USD integrated amp market - so want the best bang for my buck. The good thing is that to my personal ears cheap TDA3116 and similar stuff sounds really well.

As for which are “tolerable”? If you want to be anal, don’t accept anything with less than a SINAD of 85 dB. That will have you covered with respect to distortion. As far as hiss and hum, shoot for a SINAD of 100 dB or better. In reality those numbers can be a good 15 dB or 20 dB less, but since modern designs can get there and we care about objective performance, it’s probably worth it.

As far as the rubbish amps, a vey common problem also is channel imbalance as well. If you have that or any roll-lots I’m FR at the ends of the response, forget it. While you don’t have to go outside or anything to measure, having a simple tone generator (can be a cellphone, just has to play 1 kHz) and a good true RMS DVM will suffice since all you are looking for is both amps, or both channels, to show the same number at 1 kHz.

Hiss and hum... Are we talking about the sounds which amp creates itself when it is on but no music playing? How much should I worry about them if listening for speakers on sub 3W levels in real house where life is going on (not on dedicated listening room)??? Is it audible with let's say SINAD 85dB amp? SINAD 100 dB is out of my budget, as I have to also look for some usability, esthetics, electricity usage on standby, WAF and other real world stuff.