• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Multiple Amp Questions

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
Thanks for your time in advance for whatever you can answer.

1. Does Amir do any measurement for speaker amps that explains how much hiss I can expect to hear from a passive speaker when playing nothing? Do I look at power vs distortion and look at THD% for 50mW? I look at the graph below me an I have no idea how loud hiss would be coming the speaker. Will it be dead silent in a special room isolated from outside noises? Is this level of performance we can expect from $500 speaker amps?
index.php

2. What is a good speaker power calculator? I'm using this (http://www.hometheaterengineering.com/splcalculator.html).

3. I used a cheap $30 dBa meter from Amazon and pointed it at a speaker while playing music from where my ear normally would be. It read about 65db. Let's add 5db to it in case for peaks I guess, and another 5db if I need to crank up the volume while I walk around the room. 5db more for EQ leeway. 80db. If I sit 2ft from the speaker, for a speaker like LS50 Meta with sensitivity of 85db (2.83V/1m), the calculator says I only need 0.2w. I understand sitting close to a speaker and not listening very loudly puts a huge load off the amp's shoulders, but isn't that insanely low? Like I-could-drive-this-off-a-headphone-amp-maybe low? I tested a Neumann Kh80 and it drew 8.3w constantly both idle and while playing something and the LS50 is supposed to be an inefficient speaker.

4. According to the chart below, the impedance of the speaker can swing all the way up to almost 30ohms. How can we be sure an amp can deliver enough power at 30ohms if the amp reviews test 4 and 8ohms? Is it really just too easy to power them at that point, so there's no point in testing that high of impedance?
index.php

5. I read for headphones that we want output impedance of an amp to be less than 1/30th of the nominal impedance of the headphones. What about speakers? Headphones go to 300ohms, even more so a 30th of that is no big deal. Speakers seem to go very low so a 30th of that is a much smaller number.

6. There is interference coming from my computer that I can hear from my Rokit 6 g2. As I move my mouse I can hear the noise change. If I start a game it gets louder and different. I think it's ground loop. I have a Neumann KH80 here. The power connector to socket is 2 prong. I don't seem to hear this interference on this speaker. From Sound on Sound's article:

The major benefit of being a Class-II device is that there is absolutely no chance of suffering ground-loop hums (or buzzes) when connecting these monitors to a computer interface, even if using unbalanced connections — and that may well be a very welcome feature for many! However, the potential downside is the possibility of having no ground reference in the audio system at all (eg. if using a laptop with a bus-powered interface), and that can reduce the effectiveness of the equipment’s RF screening to result in nasty buzzes due to external interference.

Is this speaker just somehow immune to ground loops?

7. Is my assessment correct? To kill ground loops, DAC needs to have balanced output, and amp needs to have balanced input. Amp doesn't need balanced output to the speaker though. If the speaker is powered, it needs to have balanced input.

8. For Rythmik F12, in order to hook up balanced speakers I have to use an audio interface like Motu Ultralite Mk5 because the sub lacks internal crossover/balanced inputs, correct?

9. The speaker preference rating is for mid/far field listening. I only listen 19-24in away. Do I just look at CEA2034 and Early Reflections graphs and try to come to my own conclusions? The estimated in-room response's gotta be for mid/far field, so it's useless. Right?

10. Overall it seems headphone amps/dacs have improved significantly in the past 5 years. How much have speakers/headphones/speaker amps improved?
 
Last edited:

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,364
Likes
2,723
Location
US East
  1. You can see that at the lower power end of the curve (in this case of the Buckeye 6 ch MC502MP it is when power output is less than ~3 W) the curve rises in a straight line with decreasing power. This means that, below ~3 W, THD+N is all noise, and this noise is constant regardless of output level.

    Pick a point of the curve (the darker red curve). At 1 W, noise is -90 dB. Output voltage for 1 W into 4 ohms is 2 Vrms. The noise voltage is 2*10^(-90/20) = 6.32e-5 Vrms (or 63 uVrms). If your speaker sensitivity is 85 dB SPL (2.83 V/1 m), the SPL of the noise only will be 85 + 20*log10(6.32e-5/2.83) = 85 - 93 = -8 dB SPL (at 1 m). Even with your ear at 1/16 m (6 cm) from the speaker for +24 dB gain, the noise will only be 16 dB SPL, which is inaudible in practically all situations.

  2. These calculator are OK for rough estimates. For a detailed explanation on figuring out how much amplifier power you need, see "Designing a Home Theatre: Part 3":
    https://routledgetextbooks.com/textbooks/9781138921368/home-theatre.php

  3. The 65 dB SPL you measured was an average. I'd add another 10 dB (10X power) for to cover the dynamic peaks. If your average listening volume is only 65 dB SPL, I'd agree you wouldn't need much power. Also note that A-weighting exclude bass frequencies, which can account for a large part of the amplifier power.

  4. Speakers are sensitivity to voltage, not power. High impedance are easy loads to drive, they don't draw much current, and therefore not much power. Worry about the minimum impedance, not maximum.

  5. This information is given as the damping factor of the amplifier. It is usually rated for 8 ohm loads, and should therefore be adjusted to the minimum speaker impedance (e.g. will be half at 4 ohm). If damping factor is above 20, after adjustment and for the full audible frequency range, the amplifier is fine. There is no need for a damping ratio higher than 20. Well designed solid state amps have damping factors (8 ohm) in the hundreds.

  6. Ground loops can be a complicated topic and can be caused by a large variety of reasons. Balanced connections (which really are just floating signals, i.e. not ground referenced) should be able to solve most of them. USB power DAC's connected to computers can sometime be problematic due to inadequate power isolation/filtering. Many people don't encounter problems with single ended connections -- balanced is the "safer" choice.

  7. See above.

  8. Don't really understand your question. Rythmik offers balanced connection option for the F12 (with the A370XLR3 amp). If you want to high pass your main speakers, you'll need to do it with an external device. The Rythmik subs do not have this functionality.

  9. You may safely consider all electronics (for home audio reproduction) are solved problem. That's not to say that you can't go wrong -- some are purposely or otherwise poorly executed/implemented.
 

bravomail

Addicted to Fun and Learning
Joined
Oct 19, 2018
Messages
810
Likes
431
Sorry, I have many questions. Thanks for your time in advance for whatever you can answer.

1. Does Amir do any measurement for speaker amps that explains how much hiss I can expect to hear from a passive speaker when playing nothing?

2. What is a good speaker power calculator? I'm using this (http://www.hometheaterengineering.com/splcalculator.html).

3. I used a cheap $30 dBa meter from Amazon

4. According to the chart below, the impedance of the speaker can swing all the way up to almost 30ohms. How can we be sure an amp can deliver enough power at 30ohms if the amp reviews test 4 and 8ohms?

5. I read for headphones that we want output impedance of an amp to be less than 1/30th of the nominal impedance of the headphones. What about speakers? Headphones go to 300ohms, even more so a 30th of that is no big deal. Speakers seem to go very low so a 30th of that is a much smaller number.

6. There is interference coming from my computer that I can hear from my Rokit 6 g2. As I move my mouse I can hear the noise change. If I start a game it gets louder and different. I think it's ground loop.

7. Is my assessment correct? To kill ground loops, DAC needs to have balanced output, and amp needs to have balanced input. Amp doesn't need balanced output to the speaker though. If the speaker is powered, it needs to have balanced input.

8. For Rythmik F12, in order to hook up balanced speakers I have to use an audio interface like Motu Ultralite Mk5 because the sub lacks internal crossover/balanced inputs, correct?

9. The speaker preference rating is for mid/far field listening. I only listen 19-24in away. Do I just look at CEA2034 and Early Reflections graphs and try to come to my own conclusions? The estimated in-room response's gotta be for mid/far field, so it's useless. Right?

10. Overall it seems headphone amps/dacs have improved significantly in the past 5 years. How much have speakers/headphones/speaker amps improved?
1. Accept hiss as an inevitable. Use insensitive speakers or weaker amps to reduce it.
2. I never measured my room or did any speaker amp power calculations. But my experience tells me in most cases, for regular bookshelf speakers, regular rooms - u don't need that much power. Those tiny chip ClassD amps work wonders. 100W is enough for me. I certainly won't go for 400 or 1000W. Borderline is 20W - to accommodate impedance swings. Or it could be 5W at which Amir measures. :)
4. U gotta watch for those impedance dips, not highs - not all amps and their PSUs can handle high current.
5. U do wanna keep ur amp output impedance as low as possible, but going 10 times lower than 2Ohm is impractical. So the amp designers found a way to compensate for it, and measure it as "damping force".
6-7. It's a ground loop or some other EMI issue. U can try using optical output from ur PC (buy motherbord with it), then balanced from DAC to AMP. I would start with using simple power filters/strips for PC and ur other equipment. They do work wonders and supress EMI very effectively.
8. No idea on ur sub
9. Near field experience is very different. Bass-subbass will be much higher for example.
10. There was much progress made in last 5 years. Look for TI TPA3255 chip amps, Hypex nCore and Purifi EVAL1. There still will be hiss even with those amps - u r too close to ur speakers :)
 
OP
BoredErica

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
  1. You can see that at the lower power end of the curve (in this case of the Buckeye 6 ch MC502MP it is when power output is less than ~3 W) the curve rises in a straight line with decreasing power. This means that, below ~3 W, THD+N is all noise, and this noise is constant regardless of output level.

    Pick a point of the curve (the darker red curve). At 1 W, noise is -90 dB. Output voltage for 1 W into 4 ohms is 2 Vrms. The noise voltage is 2*10^(-90/20) = 6.32e-5 Vrms (or 63 uVrms). If your speaker sensitivity is 85 dB SPL (2.83 V/1 m), the SPL of the noise only will be 85 + 20*log10(6.32e-5/2.83) = 85 - 93 = -8 dB SPL (at 1 m). Even with your ear at 1/16 m (6 cm) from the speaker for +24 dB gain, the noise will only be 16 dB SPL, which is inaudible in practically all situations.

  2. These calculator are OK for rough estimates. For a detailed explanation on figuring out how much amplifier power you need, see "Designing a Home Theatre: Part 3":
    https://routledgetextbooks.com/textbooks/9781138921368/home-theatre.php

  3. The 65 dB SPL you measured was an average. I'd add another 10 dB (10X power) for to cover the dynamic peaks. If your average listening volume is only 65 dB SPL, I'd agree you wouldn't need much power. Also note that A-weighting exclude bass frequencies, which can account for a large part of the amplifier power.

  4. Speakers are sensitivity to voltage, not power. High impedance are easy loads to drive, they don't draw much current, and therefore not much power. Worry about the minimum impedance, not maximum.

  5. This information is given as the damping factor of the amplifier. It is usually rated for 8 ohm loads, and should therefore be adjusted to the minimum speaker impedance (e.g. will be half at 4 ohm). If damping factor is above 20, after adjustment and for the full audible frequency range, the amplifier is fine. There is no need for a damping ratio higher than 20. Well designed solid state amps have damping factors (8 ohm) in the hundreds.

  6. Ground loops can be a complicated topic and can be caused by a large variety of reasons. Balanced connections (which really are just floating signals, i.e. not ground referenced) should be able to solve most of them. USB power DAC's connected to computers can sometime be problematic due to inadequate power isolation/filtering. Many people don't encounter problems with single ended connections -- balanced is the "safer" choice.

  7. See above.

  8. Don't really understand your question. Rythmik offers balanced connection option for the F12 (with the A370XLR3 amp). If you want to high pass your main speakers, you'll need to do it with an external device. The Rythmik subs do not have this functionality.

  9. You may safely consider all electronics (for home audio reproduction) are solved problem. That's not to say that you can't go wrong -- some are purposely or otherwise poorly executed/implemented.
Thank you for the reply!
1. That 1W figure assumes the speaker draws 1W when idle, right? The two speakers I've ever measured the power draw of (Neumann KH80/Rokit 2 g6) drew 8.3-14.3w of power idle. Would a passive speaker like LS50 Meta really draw almost no power (<0.25w or less) when nothing is playing? If so why do I see such (relatively) high power draw from the wall from my active speakers?

However this question seems academic because drawing more power in this case pulling 8-20w *decreases* THD+N. In fact it's so low THD+N for a chunk of it at 8-20w is *lower* than Noise at 1w? On top of that, if worst case is true and a passive speaker can actually pull almost no electricity while idle, with the way the noise curve is sloping up towards upper left, it'll never reach anywhere close to audibility when nothing is playing, even with ear pressed to tweeter in a quiet studio environment it seems.

2. 65db average + 10db peaks + 5db EQ + 5db for moving around the room = 85db. You mentioned bass frequencies with an A weighted db meter. Wouldn't that be irrelevant if I the content played for testing loudness covers all the frequencies? If the mids and highs play at up to 75db, wouldn't bass only play a up to 75db, assuming my content was balanced and didn't have bass that was far louder than anything else?

9. It doesn't seem that way to me. Speakers can always be flatter, or have better and better directivity unless they're already perfect omnidirectionally. Or, they can be cheaper. If I look at the speaker preference rating, of the ones scoring higher than 7.8 or so under $1000 per speaker, there aren't *that* many choices which are confirmed to have no audible hiss nearfield (or are passive) and look good. If more and more speakers could just put more effort into doing well, we'd have more options and more enclosure styles to choose from.

In terms of speaker amps, the one I'd like to LS50 Meta are $500. I wish we could lower that price down to $250 or improve that THD+N figures to be closer to headphone amps. (The latter is more improvement for improvement's sake now I have your response on just how inaudible the noise from these nice amps are.) I'm assuming then, all the active speakers with hiss I can hear nearfield with amp turned all the way down just have incredibly garbage THD+N figures, 50db+ worse than the Buckeye? That seems unfathomable.

-
During the time you wrote your reply I added a question, and what was question 9 is now question 10.
1. Accept hiss as an inevitable. Use insensitive speakers or weaker amps to reduce it.
2. I never measured my room or did any speaker amp power calculations. But my experience tells me in most cases, for regular bookshelf speakers, regular rooms - u don't need that much power. Those tiny chip ClassD amps work wonders. 100W is enough for me. I certainly won't go for 400 or 1000W. Borderline is 20W - to accommodate impedance swings. Or it could be 5W at which Amir measures. :)
4. U gotta watch for those impedance dips, not highs - not all amps and their PSUs can handle high current.
5. U do wanna keep ur amp output impedance as low as possible, but going 10 times lower than 2Ohm is impractical. So the amp designers found a way to compensate for it, and measure it as "damping force".
6-7. It's a ground loop or some other EMI issue. U can try using optical output from ur PC (buy motherbord with it), then balanced from DAC to AMP. I would start with using simple power filters/strips for PC and ur other equipment. They do work wonders and supress EMI very effectively.
8. No idea on ur sub
9. Near field experience is very different. Bass-subbass will be much higher for example.
10. There was much progress made in last 5 years. Look for TI TPA3255 chip amps, Hypex nCore and Purifi EVAL1. There still will be hiss even with those amps - u r too close to ur speakers :)
1. I don't understand how it is inevitable if I go off of what NTK seems to be implying. Unless passives pull nearly no power or the noise curve explodes as I move to the left past 20mW, if I'm interpreting it correctly I can't fathom how that would generate audible hiss at any distance audible to the ear even in a quiet studio.

4. Ah yes. The LS50 Meta falls to 3.7ohms but that is close to 4ohms.

5. That's interesting though probably arcane and hard for me to understand the underlying mechanics of.

6. I used a ground loop isolator and the noise is gone, so I'm assuming it was a ground loop. Or something ground loop isolator also filters out? Perhaps ground loop isolators can filter out EMI?

9. Well, a subwoofer I can place away from me. It's just bookshelves that have to be close to me so sub-bass is not an issue in this context. So I guess a speaker that measures flat sloping down with the in-room-response will actually be overly bassy if listened to too closely?

10. It sort of seemed like $500 amp that can power this comfortably with decent thd+n was relatively new, so thanks for that. As for hiss audibility though I'd refer back to point 1. I don't understand how based on Amir's measurements the hiss will be audible if it's like -40 at 0.01mW or something. If room is 15db the noise has to be somewhere around 15db for me to hear it, no?

With Neumann KH80 I am already having a hard time hearing the noise at 19in away with amp set to minimum power when my room is at its quietest. Sure, there are quieter controlled rooms out there but surely there are systems that run laps around a speaker I can get for $400 open box.
 

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,364
Likes
2,723
Location
US East
Thank you for the reply!
1. That 1W figure assumes the speaker draws 1W when idle, right? The two speakers I've ever measured the power draw of (Neumann KH80/Rokit 2 g6) drew 8.3-14.3w of power idle. Would a passive speaker like LS50 Meta really draw almost no power (<0.25w or less) when nothing is playing? If so why do I see such (relatively) high power draw from the wall from my active speakers?

However this question seems academic because drawing more power in this case pulling 8-20w *decreases* THD+N. In fact it's so low THD+N for a chunk of it at 8-20w is *lower* than Noise at 1w? On top of that, if worst case is true and a passive speaker can actually pull almost no electricity while idle, with the way the noise curve is sloping up towards upper left, it'll never reach anywhere close to audibility when nothing is playing, even with ear pressed to tweeter in a quiet studio environment it seems.

2. 65db average + 10db peaks + 5db EQ + 5db for moving around the room = 85db. You mentioned bass frequencies with an A weighted db meter. Wouldn't that be irrelevant if I the content played for testing loudness covers all the frequencies? If the mids and highs play at up to 75db, wouldn't bass only play a up to 75db, assuming my content was balanced and didn't have bass that was far louder than anything else?

9. It doesn't seem that way to me. Speakers can always be flatter, or have better and better directivity unless they're already perfect omnidirectionally. Or, they can be cheaper. If I look at the speaker preference rating, of the ones scoring higher than 7.8 or so under $1000 per speaker, there aren't *that* many choices which are confirmed to have no audible hiss nearfield (or are passive) and look good. If more and more speakers could just put more effort into doing well, we'd have more options and more enclosure styles to choose from.

In terms of speaker amps, the one I'd like to LS50 Meta are $500. I wish we could lower that price down to $250 or improve that THD+N figures to be closer to headphone amps. (The latter is more improvement for improvement's sake now I have your response on just how inaudible the noise from these nice amps are.) I'm assuming then, all the active speakers with hiss I can hear nearfield with amp turned all the way down just have incredibly garbage THD+N figures, 50db+ worse than the Buckeye? That seems unfathomable.

-
During the time you wrote your reply I added a question, and what was question 9 is now question 10.

1. I don't understand how it is inevitable if I go off of what NTK seems to be implying. Unless passives pull nearly no power or the noise curve explodes as I move to the left past 20mW, if I'm interpreting it correctly I can't fathom how that would generate audible hiss at any distance audible to the ear even in a quiet studio.

4. Ah yes. The LS50 Meta falls to 3.7ohms but that is close to 4ohms.

5. That's interesting though probably arcane and hard for me to understand the underlying mechanics of.

6. I used a ground loop isolator and the noise is gone, so I'm assuming it was a ground loop. Or something ground loop isolator also filters out? Perhaps ground loop isolators can filter out EMI?

9. Well, a subwoofer I can place away from me. It's just bookshelves that have to be close to me so sub-bass is not an issue in this context. So I guess a speaker that measures flat sloping down with the in-room-response will actually be overly bassy if listened to too closely?

10. It sort of seemed like $500 amp that can power this comfortably with decent thd+n was relatively new, so thanks for that. As for hiss audibility though I'd refer back to point 1. I don't understand how based on Amir's measurements the hiss will be audible if it's like -40 at 0.01mW or something. If room is 15db the noise has to be somewhere around 15db for me to hear it, no?

With Neumann KH80 I am already having a hard time hearing the noise at 19in away with amp set to minimum power when my room is at its quietest. Sure, there are quieter controlled rooms out there but surely there are systems that run laps around a speaker I can get for $400 open box.

1. The 1 W figure the amplifier output at the speaker terminals. The definition of THD+N is the total output (overall of all frequencies) of the amplifier, divided by the output at the test signal (1 kHz sine wave in this case, I believe), minus 1. It is the ratio of the energy of all the unwanted junk (i.e. distortions and noise) to the energy of the wanted 1 kHz signal.

The reason THD+N drops with output power, when THD+N is dominated by noise at the lower power range, is because noise largely stays the same regardless of output power. When the energy of the wanted output signal goes up, and the energy of the unwanted junk remains the same, the THD+N ratio goes down.

I'd differentiate between the speaker amplifier and the speaker drivers. Look at an active speaker as an amplifier connected to speaker drivers. Speaker drivers do not consume electrical power when not producing sound, nor produce noise when idle. Not all of the power the amplifier pulls from the wall translates into speaker output, since none can be 100% efficient. They all continuously consume a certain amount of power when idle, since some portions of the circuitry always consume power when they are turned on. Some amplifiers consume a lot more power when idle than others. It heavy depends on the design and design choices the engineers made.

2. Typically the power spectrum of music slopes down with frequency. For examples, see this thread:

The blue curve is A-weighting. You can see that the weights of the bass frequencies are very low, and the SPL measured do not include much of it. Since a large part of the energy in music is in bass (especially for non-classical music), the A-weighted SPL under-estimates the total sound power and therefore the required amplifier power. Also, because of the equal-loudness contour, our sensitivity to bass lowers quickly with SPL. For example, a note at 77 dB SPL 50 Hz sounds only as loud as 40 dB SPL 1 kHz, whereas at 89 dB SPL 50 Hz (12 dB increase), it sounds as loud as at 60 dB SPL 1 kHz, an 20 dB (phon) increase. Thus, if we go by hearing, we tend to underestimate the SPL of bass at the lower SPL levels.
(Figure from Wikipedia)

1024px-Acoustic_weighting_curves_%281%29.svg.png


9. Active speaker manufacturers are taking hiss more seriously now. Kali is revamping their product line to reduce hiss.

For passive speakers with 90'ish or less sensitivity, I don't think amp hiss is an issue. The Aiyima A07, which THD+N is -80 dB at 1 W 4 ohms, will give 7 dB SPL noise at 1 m for a speaker with 90 dB sensitivity. Not a practical hiss problem at all.

10. The THD+N figure is a ratio. It is NOT the noise power. It is the ratio of the noise power to the test signal output power. It means the noise power is 40 dB lower than the signal power of 0.01 mW, and -40 dB = 10^-4 = 0.0001X. Therefore the noise power is 1e-6 mW = 1 nW = 1e-9 W.
 
OP
BoredErica

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
1. The 1 W figure the amplifier output at the speaker terminals. The definition of THD+N is the total output (overall of all frequencies) of the amplifier, divided by the output at the test signal (1 kHz sine wave in this case, I believe), minus 1. It is the ratio of the energy of all the unwanted junk (i.e. distortions and noise) to the energy of the wanted 1 kHz signal.

The reason THD+N drops with output power, when THD+N is dominated by noise at the lower power range, is because noise largely stays the same regardless of output power. When the energy of the wanted output signal goes up, and the energy of the unwanted junk remains the same, the THD+N ratio goes down.

2. The blue curve is A-weighting. You can see that the weights of the bass frequencies are very low, and the SPL measured do not include much of it. Since a large part of the energy in music is in bass (especially for non-classical music), the A-weighted SPL under-estimates the total sound power and therefore the required amplifier power. Also, because of the equal-loudness contour, our sensitivity to bass lowers quickly with SPL. For example, a note at 77 dB SPL 50 Hz sounds only as loud as 40 dB SPL 1 kHz, whereas at 89 dB SPL 50 Hz (12 dB increase), it sounds as loud as at 60 dB SPL 1 kHz, an 20 dB (phon) increase. Thus, if we go by hearing, we tend to underestimate the SPL of bass at the lower SPL levels.

9. Active speaker manufacturers are taking hiss more seriously now. Kali is revamping their product line to reduce hiss.

For passive speakers with 90'ish or less sensitivity, I don't think amp hiss is an issue. The Aiyima A07, which THD+N is -80 dB at 1 W 4 ohms, will give 7 dB SPL noise at 1 m for a speaker with 90 dB sensitivity. Not a practical hiss problem at all.

10. The THD+N figure is a ratio. It is NOT the noise power. It is the ratio of the noise power to the test signal output power. It means the noise power is 40 dB lower than the signal power of 0.01 mW, and -40 dB = 10^-4 = 0.0001X. Therefore the noise power is 1e-6 mW = 1 nW = 1e-9 W.
1 & 10. I think I understand your examples and they helped me a lot in trying to understand all of this. The only question I have left is, do sounds always have to be louder than the background noise for a person to detect them? If background noise is 30db and there is 30db of noise coming out of the speaker arriving to my sitting position, will I notice noise at all?

2. But the extra "10db peaks" takes care of this, right? Or do I do 65db average + 10db peaks + 5db EQ + 5db for moving around the room +5db bass compensation = 90db?

9. I wonder if there is actually any scientific evidence that smaller speaker systems are not as good as larger speaker systems for a person sitting at the sweet spot listening at low volumes (with a sub), assuming the speakers perform at the same level. I was thinking maybe they could scale some speakers up but now I think about it I'm not sure if it helps at all.

People sometimes say larger speakers sound 'larger' but maybe that can be explained by lack of bass due to lack of sub, volume, or placebo?

--

11. Do you agree with what Bravomail said about how using Amir's in-room and early-reflections charts to listen to a speaker nearfield will cause bass to be louder than shown on those graphs? In the latest Genelec 8361A video review, Amir said that while the reflections and in-room graphs are for far-field, it's "close enough" for nearfield. It makes me wonder, because the LS50 Metas have too little bass to be flat but perhaps nearfield will boost those frequencies right up. I guess at the end of the day all of this is moot with a robust enough measuring/EQ/correction system.

12.
index.php

Isn't this graph of the Benchmark AHB2 suggesting it can alter the frequency response of whatever is played, reducing 2khz+ frequencies by as much as 0.5db? I thought that's not transparent?
 
Last edited:

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,364
Likes
2,723
Location
US East
1 & 10. I think I understand your examples and they helped me a lot in trying to understand all of this. The only question I have left is, do sounds always have to be louder than the background noise for a person to detect them? If background noise is 30db and there is 30db of noise coming out of the speaker arriving to my sitting position, will I notice noise at all?

2. But the extra "10db peaks" takes care of this, right? Or do I do 65db average + 10db peaks + 5db EQ + 5db for moving around the room +5db bass compensation = 90db?

9. I wonder if there is actually any scientific evidence that smaller speaker systems are not as good as larger speaker systems for a person sitting at the sweet spot listening at low volumes (with a sub), assuming the speakers perform at the same level. I was thinking maybe they could scale some speakers up but now I think about it I'm not sure if it helps at all.

People sometimes say larger speakers sound 'larger' but maybe that can be explained by lack of bass due to lack of sub, volume, or placebo?

--

11. Do you agree with what Bravomail said about how using Amir's in-room and early-reflections charts to listen to a speaker nearfield will cause bass to be louder than shown on those graphs? In the latest Genelec 8361A video review, Amir said that while the reflections and in-room graphs are for far-field, it's "close enough" for nearfield. It makes me wonder, because the LS50 Metas have too little bass to be flat but perhaps nearfield will boost those frequencies right up. I guess at the end of the day all of this is moot with a robust enough measuring/EQ/correction system.

12.
index.php

Isn't this graph of the Benchmark AHB2 suggesting it can alter the frequency response of whatever is played, reducing 2khz+ frequencies by as much as 0.5db? I thought that's not transparent?
1 & 10. The highly simplified view is the combination of both the room noise and noise coming out of the speakers will increase the total SPL by 3 dB, which is definitely noticeable.

Auditory masking is a rather complicated subject. I am not at all knowledgeable in the area. It depends on the frequency composition and level of the masking signal, and the frequency of the tone being masked. According to the post below, I think, a person may be able to sense tones 10 to 15 dB below a broadband masking noise.

You may also want to check out this video.

2. My personal desire for amplifier power is ~100X (= 20 dB) the power at my average listening volume. If you want to account for occasionally increasing the listening distance by 3X, a 5 dB padding is reasonable. That matches your +25 dB figure. Also, how the amplifier behaves when clipping can be important too.

Note that if the amp clips cleanly and recovers quickly with no drama, you may not notice that it clipped. A clean hard clipping of 6 dB (relative to the overall highest peak level of the music sample), which means the undistorted peak voltage is twice the clipping voltage (which will produce 22.6% THD at the large clips) isn't too objectionable. Therefore how well an "underpowered" amp clips is important, and we could be just fine with such an amp. Too bad that this information is rarely available.
clipping.PNG


9. I personally don't think so. I believe well integrated sub/sat systems can give large floor-standers a good run for their money. The "sound stage" is determined by the dispersion pattern of the speakers and the room.

11. Bass is heavily room, speaker placement, and listening position dependent. You'll most likely need EQ to get good bass anyway, so the "raw" bass response flatness is, IMHO, not important. What is important is sufficient capacity after EQ that will satisfy the listener(s).

I am not too sure about nearfield bass boost. My guess is that you'll need to get really close to the speakers to experience any acoustical nearfield effects. And for smaller speakers, their nearfield won't extend as far as larger speakers.

12. The droop is only 0.3 dB to 20 kHz. It is minor enough and high enough in frequency to be inaudible even in rapid switching AB tests, and is totally negligible relative to speakers. Not a concern IMHO.
 
OP
BoredErica

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
1 & 10. The highly simplified view is the combination of both the room noise and noise coming out of the speakers will increase the total SPL by 3 dB, which is definitely noticeable.

Auditory masking is a rather complicated subject. I am not at all knowledgeable in the area. It depends on the frequency composition and level of the masking signal, and the frequency of the tone being masked. According to the post below, I think, a person may be able to sense tones 10 to 15 dB below a broadband masking noise.

12. The droop is only 0.3 dB to 20 kHz. It is minor enough and high enough in frequency to be inaudible even in rapid switching AB tests, and is totally negligible relative to speakers. Not a concern IMHO.
1 & 10:
Aiyima A07, which THD+N is -80 dB at 1 W 4 ohms, will give 7 dB SPL noise at 1 m for a speaker with 90 dB sensitivity
I took your original example and calculated it and got minimum of 10db of noise. I heard the calc is not entirely accurate though.

Then I changed it up: Assume -90db or 10^(-9) x 1w of power for noise *Buckeye amp). 87db sensitivity. 1.5ft away, in a corner. 12.8db of noise. +15db, that gets me up to 27.8db of background noise and any higher I could theoretically hear noise? 30db would be a quiet room in a house, 20db more like a quiet recording studio right? (I'm not sure, I tried googling to no avail.) So it's conceivable one could hear noise even with the Buckeye?


11. Yeah, I guess EQ will sort all the problems out. There are limits to EQ though, right? Directivity issues, which should be more of a problem for mid and far field listening, which are caused by reflections off the room? If the bounce back produces sound that's not just the on axis FR but minus a few db? I remember rarely in reviews Amir would note specifically not to EQ some problems with FR. In one video he talked about phase shifts and how when it rose very sharply EQing won't help.

I should just read material and figure it out, but my friend and I look and there's a 2 hour video on how room acoustics work and it's making me die on the inside.

12. I guess 0.5db is really inaudible and that's why that's the threshold Amir has set up. I went to one of those sites to try to see if I can hear a difference and I couldn't really. So I guess it's fine.
 

mdsimon2

Addicted to Fun and Learning
Forum Donor
Joined
Oct 20, 2020
Messages
797
Likes
944
Location
Detroit, MI
So far the noise discussion has ignored a very important point, any residual noise from your DAC will be multiplied by your amplifier. @bravomail basically made this point but gain staging is very important. For example even if you have an amp that is low noise if it is high gain you can end up with hiss.

Michael
 
Last edited:

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,364
Likes
2,723
Location
US East
1 & 10:

I took your original example and calculated it and got minimum of 10db of noise. I heard the calc is not entirely accurate though.

Then I changed it up: Assume -90db or 10^(-9) x 1w of power for noise *Buckeye amp). 87db sensitivity. 1.5ft away, in a corner. 12.8db of noise. +15db, that gets me up to 27.8db of background noise and any higher I could theoretically hear noise? 30db would be a quiet room in a house, 20db more like a quiet recording studio right? (I'm not sure, I tried googling to no avail.) So it's conceivable one could hear noise even with the Buckeye?


11. Yeah, I guess EQ will sort all the problems out. There are limits to EQ though, right? Directivity issues, which should be more of a problem for mid and far field listening, which are caused by reflections off the room? If the bounce back produces sound that's not just the on axis FR but minus a few db? I remember rarely in reviews Amir would note specifically not to EQ some problems with FR. In one video he talked about phase shifts and how when it rose very sharply EQing won't help.

I should just read material and figure it out, but my friend and I look and there's a 2 hour video on how room acoustics work and it's making me die on the inside.

12. I guess 0.5db is really inaudible and that's why that's the threshold Amir has set up. I went to one of those sites to try to see if I can hear a difference and I couldn't really. So I guess it's fine.
1 & 10. My noise number of 7 dB for the Aiyima was for a single speaker. Two speakers will increase it by 3 dB to 10 dB SPL and match your number.

For your Buckeye estimates, you used a listening distance of 1.5 ft, with both speakers in a corner. This, IMHO, is very unrealistic. If the speakers and listener are in a equilateral triangle, meaning that the speakers will be 1.5 ft apart, the room is going to be not more than ~3 ft wide. (Note: The SPL calculator you use says "in a corner" is within 18 - 24 inches. This applies only for the typical listening distances of > 6-8 ft, not 1.5 ft.)

The audibility of a tone 15 dB below the noise level is the very best (or worst, depending on the perspective) case scenario. With room noise on top of amplifier noise, I believe the masking is much more effective. Again, how a signal is masked is highly dependent on the magnitudes and frequency compositions of the masker and the signal.

In the highly simplistic case where both (room noise and amplifier noise) are white noise, they will just be "additive" (compute using sum of power). If the amplifier noise is 12.8 dB and the room noise is 30 dB, the combined SPL will be 30.08 dB, which means the amplifier noise is undetectable.

Unfortunately neither room noise or amplifier noise is white noise, so reality is more complicated. But still, IMHO, amplifier self noise (i.e. hiss) from Hypex (or even the Aiyima A07) is not something that worries me, especially with speakers with < 90 dBSPL/2.83V efficiencies.

11. It is true that EQ won't solve directivity problems. But at low frequencies (< 150-200 Hz), most speakers are omnidirectional, so directivity is not an issue, but room mode is. Room EQ is almost mandatory for good bass. Full range EQ is only advisable if you have the speaker's anechoic on-axis or listening window FR.
 
OP
BoredErica

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
1 & 10. My noise number of 7 dB for the Aiyima was for a single speaker. Two speakers will increase it by 3 dB to 10 dB SPL and match your number.

For your Buckeye estimates, you used a listening distance of 1.5 ft, with both speakers in a corner. This, IMHO, is very unrealistic. If the speakers and listener are in a equilateral triangle, meaning that the speakers will be 1.5 ft apart, the room is going to be not more than ~3 ft wide. (Note: The SPL calculator you use says "in a corner" is within 18 - 24 inches. This applies only for the typical listening distances of > 6-8 ft, not 1.5 ft.)

The audibility of a tone 15 dB below the noise level is the very best (or worst, depending on the perspective) case scenario. With room noise on top of amplifier noise, I believe the masking is much more effective. Again, how a signal is masked is highly dependent on the magnitudes and frequency compositions of the masker and the signal.

In the highly simplistic case where both (room noise and amplifier noise) are white noise, they will just be "additive" (compute using sum of power). If the amplifier noise is 12.8 dB and the room noise is 30 dB, the combined SPL will be 30.08 dB, which means the amplifier noise is undetectable.

Unfortunately neither room noise or amplifier noise is white noise, so reality is more complicated. But still, IMHO, amplifier self noise (i.e. hiss) from Hypex (or even the Aiyima A07) is not something that worries me, especially with speakers with < 90 dBSPL/2.83V efficiencies.

11. It is true that EQ won't solve directivity problems. But at low frequencies (< 150-200 Hz), most speakers are omnidirectional, so directivity is not an issue, but room mode is. Room EQ is almost mandatory for good bass. Full range EQ is only advisable if you have the speaker's anechoic on-axis or listening window FR.
1 & 10.
1636937020458.png

My current sitting position forces me to be ~19-20in away from the speakers. I can't maintain an equilateral triangle because it's not physically possible. It's not up to me right now and due to home circumstances I don't have total liberty in rearranging the room. That might change in the future. It also happens the right speaker is pretty close to the corner of the room.

11. I think the only thing left for now about be what bad directivity looks like. My assumptions that bad directivity for fixing with EQ looks like reflection FR curves that don't mirror the same structure as the on axis FR curve except higher or lower down (quieter or louder). Instead they have a FR that is doing its own thing.

It's correct to assume though that nearfield, reflections don't play as big of an effect on actual in room FR right?

So far the noise discussion has ignored a very important point, any residual noise from your DAC will be multiplied by your amplifier. @bravomail basically made this point but gain staging is very important. For example even if you have an amp that is low noise if it is high gain you can end up with hiss.

Michael
Doesn't seem like the Buckeye amps even have a gain toggle... If it did though I would be on the lower gain side because I just don't need that much power.
 

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,364
Likes
2,723
Location
US East
11. Yes, for nearfield, or more correctly, direct field listening (as opposed to reverberant field listening) poor speaker directivity is less of a problem. One mitigation strategy is to absorb the side wall reflections as completely as possible. You may also want to place the speaker as far from the table surface as possible to reduce the table top reflections.

All 3 Hypex NCXXXMP amplifier modules (Buckeye currently offers 2 of them) have the same 25.5 dB gain.
 

mdsimon2

Addicted to Fun and Learning
Forum Donor
Joined
Oct 20, 2020
Messages
797
Likes
944
Location
Detroit, MI
11. Yes, for nearfield, or more correctly, direct field listening (as opposed to reverberant field listening) poor speaker directivity is less of a problem. One mitigation strategy is to absorb the side wall reflections as completely as possible. You may also want to place the speaker as far from the table surface as possible to reduce the table top reflections.

All 3 Hypex NCXXXMP amplifier modules (Buckeye currently offers 2 of them) have the same 25.5 dB gain.
Exactly. And even if you use a SOTA DAC (120 dB DR at 4 V) the noise contribution from the DAC will be more than the amp itself after 25.5 dB of gain.

Michael
 
OP
BoredErica

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
Exactly. And even if you use a SOTA DAC (120 dB DR at 4 V) the noise contribution from the DAC will be more than the amp itself after 25.5 dB of gain.
How does 120db sinad dac become bigger noise contributor vs the amp after 25.5db of gain? The worst case for Buckeye is -90db of noise at 1w and I tried to rearrange the numbers around for the dac and none of the numbers made sense.

Buckeye: (10^(-9))*1w=1e-9w of noise
Dac: ((10^(-12))*1)*(10^(2.55)) = 3.55e-10

The dac I'm thinking of buying has 112.8db SINAD @ 4v (Motu UltraLite Mk5). I need 1-4 balanced inputs (L, R, subwoofer, subwoofer 2 in the future perhaps) that has balanced out to the amp, ability to manage crossover, and for anything sub $1000 it seemed like Motu had the best SINAD.

OTOH for the Motu it seems the SINAD is being dragged down by harmonic distortions while the noise is actually way lower.
index.php
-140db of noise at 4v

Well, since Buckeye has no volume knob and I'm supposed to adjust volume via either a dedicated pre-amp or a multi-featured dac, I guess I'm supposed to be pushing a touch screen to change the volume instead of rotating a volume knob? Oh god can the Motu even do that? This is getting more and more complicated.
11. Yes, for nearfield, or more correctly, direct field listening (as opposed to reverberant field listening) poor speaker directivity is less of a problem. One mitigation strategy is to absorb the side wall reflections as completely as possible. You may also want to place the speaker as far from the table surface as possible to reduce the table top reflections.

All 3 Hypex NCXXXMP amplifier modules (Buckeye currently offers 2 of them) have the same 25.5 dB gain.
If I'm understanding correctly, that would mean moving speaker as high up and away from the wall as possible. This would mean moving the speaker even closer to my ears than 19 inches though. LS50 Metas are supposed to be more 'point source' due to its coaxial drivers, but still. Speaker/listening position is now very far from an equilateral triangle.
 
Last edited:

mdsimon2

Addicted to Fun and Learning
Forum Donor
Joined
Oct 20, 2020
Messages
797
Likes
944
Location
Detroit, MI
Let me start off by saying I have an Ultralite Mk5 and multiple NC252MPs and I love them, so personally I think you have good taste in gear. The Mk5 in particular is an incredible deal and nothing comes close for the price. I also think you are unlikely to have noise issues with most conventional speakers.

Mk5 dynamic range at 4 V is about 115 dB per Amir's measurements. The -140 dB noise floor you are seeing on the FFT is not the real noise floor, it is artificially lowered by FFT gain. This means that residual noise from the Mk5 is 4 V x 10^(-115/20) = 7.11 uV, this will always be present at the Mk5 output regardless of volume position.

Now multiply that noise by the 25.5 dB amplifier gain, 7.11 uV x 10^(25.5/20) = 134 uV which is just over double the amplifier residual noise of 63 uV as previously calculated by @NTK .

Michael
 

mdsimon2

Addicted to Fun and Learning
Forum Donor
Joined
Oct 20, 2020
Messages
797
Likes
944
Location
Detroit, MI
You may find the information in the thread linked below interesting, I did some measurements with a various DACs (including the Mk5) and a NC252MP amplifier as well as nearfield listening.


This presents a somewhat worst case scenario (other than super high efficiency drivers) as it is a relatively sensitive tweeter connected directly to an amplifier with no padding. I hadn't actually checked this before because I was skeptical of the measurements in an absolute sense but based on the calculations we just did I would expect combined noise from the DAC and amplifier (assuming uncorrelated noise summation) to be around 150 uV which is right about what I measured.

Michael
 
OP
BoredErica

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
Let me start off by saying I have an Ultralite Mk5 and multiple NC252MPs and I love them, so personally I think you have good taste in gear. The Mk5 in particular is an incredible deal and nothing comes close for the price. I also think you are unlikely to have noise issues with most conventional speakers.

Mk5 dynamic range at 4 V is about 115 dB per Amir's measurements. The -140 dB noise floor you are seeing on the FFT is not the real noise floor, it is artificially lowered by FFT gain. This means that residual noise from the Mk5 is 4 V x 10^(-115/20) = 7.11 uV, this will always be present at the Mk5 output regardless of volume position.

Now multiply that noise by the 25.5 dB amplifier gain, 7.11 uV x 10^(25.5/20) = 134 uV which is just over double the amplifier residual noise of 63 uV as previously calculated by @NTK .

Michael
Thanks, but you'd have to compliment my friend. :p We've been going back and forth looking at ideas on what to buy for our rooms. Motu and the Buckeye amp was his idea. I find audio very complicated, and it's only going to get more complicated whenever I have to do room EQ.

I calculated noise at 1.5ft away for a speaker of 85 db sensitivity (like LS50 Meta/R3) to be 10.8db with 2 speakers near a corner. (1e-9w of power.) If I take that and multiply by x2.12 I get 2.12e-9w of power, or 14.1db. That is still a ways off from 20db in a very quiet room or 30db in a regular room.

So the higher the gain on the amp, the more important dac noise performance is because it's multiplying more of the dac's noise?

Thanks for everyone's patience. For detailed questions on which piece of gear to buy I should probably make a new thread and provide new background info. I've been very annoyed by my old Rokit's noise and I never want to hear noise from my speakers ever again. I have a KH80 for testing and frankly the biggest upgrade is the lack of noise, moreso than FR changes.
 
Last edited:

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,364
Likes
2,723
Location
US East
So the higher the gain on the amp, the more important dac noise performance is because it's multiplying more of the dac's noise?
The rule of thumb is to put a much of the gain as possible to the front of the chain, and therefore as little gain as possible in the speaker amp (just enough to reach your desired listening level).

You can solve the problem by placing an analog volume control (can be a passive "pre-amp" or one of the headphone/pre amps) between the speaker amp and source. The combination effectively gives you adjustable gain for the speaker amp, so you don't have to run it at full gain all the time.

However, I don't think this is necessary with your proposed setup.
 
OP
BoredErica

BoredErica

Senior Member
Joined
Jan 15, 2019
Messages
429
Likes
500
Location
USA
The rule of thumb is to put a much of the gain as possible to the front of the chain, and therefore as little gain as possible in the speaker amp (just enough to reach your desired listening level).

You can solve the problem by placing an analog volume control (can be a passive "pre-amp" or one of the headphone/pre amps) between the speaker amp and source. The combination effectively gives you adjustable gain for the speaker amp, so you don't have to run it at full gain all the time.

However, I don't think this is necessary with your proposed setup.
Based on the calculations put forth in this thread, I don't think it's necessary either. By the time I want to have such high standards and I don't want to hear noise with ear pressed against the tweeter in my sound-isolated room, I'd be buying a $3000 Benchmark amp anyways.

Only tricky thing is adjusting volume on the fly.
 
Last edited:

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
1,364
Likes
2,723
Location
US East
You can still use digital volume control. Just set the analog volume so that full scale digital volume is the loudest you want to listen to.
 
Top Bottom