<<Interesting comments, but I've been to two concerts today, and somehow ...I can't turn my brain on!>>
The high noise floor of the forum is useful to reset your brain between concerts
Noise Floor
- Thread starter Ambio
- Start date
The high noise floor of the forum is useful to reset your brain between concerts
Very interesting and insightful discussion. Thanks for kicking this off, Ambio!
Time to break out an analogy. So noise floor is the equivalent of the contrast level or "black level" of a television/display panel? Sounds like it (or looks like it!). So we are talking about the ability of an audio system to reproduce a signal, say 1KHz (especially because that is used as a reference point for SNR)
- without introducing additional noise to the signal
(or blackest of black in our TV analogy - equivalent of an OLED screen where the pixel is actually turned off)
- or with negligible addition of noise assuming the former is not possible. This is where things get relative. Since our ears detect sound on a relative basis and not an absolute basis, it is important for the correct sound to be relatively much louder than the incorrect sound (noise). Hence the S-N ratio. Again in our TV analogy, if the blacks are not pure blacks, the other colors have to be made much much brighter (higher contrast) for them to compensate the much higher noise floor (impure black).
Again, the thing to note is that merely increasing the brightness of the panel (volume or loudness for audio systems) is not enough as it will also equally increase the impurity of the black level (i.e. will raise the noise floor). So the more crucial aspect is that the ratio of signal (color quality) to the noise (black level) should be high. That is, the system should have a high contrast level.
And to clarify, the other type of impurity that we get concerned with is distortion, where the signal itself is incorrectly reproduced. Here, the analogy may break down but perhaps this is equivalent to a display panel showing incorrect colors (being incorrectly color calibrated)? And then there is incorrect timing (frames getting skipped?)
Apologies for my crude logic! So I too am puzzled by one aspect. I originally assumed that noise floor was about the presence of noise when no music is playing but the amp is on. But isn't that just a concern for analog audio? Digital audio in modern systems don't have this issue - as Ambio points out.
And does noise increase in volume too? Again, there seems to be contradictory answers here. If it doesn't then it is easy enough to suppress the noise by increasing the volume. If it does, then how does one "hear" it? I really don't hear any noise in my system, but fine, I am not an expert. But how do experts so confidently state in their reviews that they actually heard the noise floor go lower? Does this mean their expert ears typically hear a sea of background noise, and they can perceive this background noise become lower??
Or does this mean that they hear music itself getting distorted (say, the sound of a cymbal or piano note sound unnatural)? If so, why don't they just say that specifically - that piano F# key sounded more natural or real in a different system. That is how a musician would describe sound, for example. Why do they say instead that the "noise floor" got lowered? That is a strange way to describe something. No?
Time to break out an analogy. So noise floor is the equivalent of the contrast level or "black level" of a television/display panel? Sounds like it (or looks like it!). So we are talking about the ability of an audio system to reproduce a signal, say 1KHz (especially because that is used as a reference point for SNR)
- without introducing additional noise to the signal
(or blackest of black in our TV analogy - equivalent of an OLED screen where the pixel is actually turned off)
- or with negligible addition of noise assuming the former is not possible. This is where things get relative. Since our ears detect sound on a relative basis and not an absolute basis, it is important for the correct sound to be relatively much louder than the incorrect sound (noise). Hence the S-N ratio. Again in our TV analogy, if the blacks are not pure blacks, the other colors have to be made much much brighter (higher contrast) for them to compensate the much higher noise floor (impure black).
Again, the thing to note is that merely increasing the brightness of the panel (volume or loudness for audio systems) is not enough as it will also equally increase the impurity of the black level (i.e. will raise the noise floor). So the more crucial aspect is that the ratio of signal (color quality) to the noise (black level) should be high. That is, the system should have a high contrast level.
And to clarify, the other type of impurity that we get concerned with is distortion, where the signal itself is incorrectly reproduced. Here, the analogy may break down but perhaps this is equivalent to a display panel showing incorrect colors (being incorrectly color calibrated)? And then there is incorrect timing (frames getting skipped?)
Apologies for my crude logic! So I too am puzzled by one aspect. I originally assumed that noise floor was about the presence of noise when no music is playing but the amp is on. But isn't that just a concern for analog audio? Digital audio in modern systems don't have this issue - as Ambio points out.
And does noise increase in volume too? Again, there seems to be contradictory answers here. If it doesn't then it is easy enough to suppress the noise by increasing the volume. If it does, then how does one "hear" it? I really don't hear any noise in my system, but fine, I am not an expert. But how do experts so confidently state in their reviews that they actually heard the noise floor go lower? Does this mean their expert ears typically hear a sea of background noise, and they can perceive this background noise become lower??
Or does this mean that they hear music itself getting distorted (say, the sound of a cymbal or piano note sound unnatural)? If so, why don't they just say that specifically - that piano F# key sounded more natural or real in a different system. That is how a musician would describe sound, for example. Why do they say instead that the "noise floor" got lowered? That is a strange way to describe something. No?
Thad E Ginathom
Well-Known Member
I'm sceptical that these "experts" actually do hear the noise floor go lower. I suspect that, in line with your video analogies, what they hear is the contrast changing with different levels and maybe the greater definition of improved transients.
That makes a lot of sense. I still feel that they can use better adjectives though. For example, if they are hearing two piano keys in succession, how does a change in noise floor (or contrast level) affect the sound?
- Does it cause the two notes to get better separated (because the noise was clouding or blanketing the sound)?
- Does it cause the sound decay of one more to become smoother, and/or more prolonged?
- Does the sound itself (of a given note) sound purer or richer or more tonally accurate? Or more harmonic?
More than anything, two points really frustrate me or puzzle me:
1. Why don't reviews and reviewers use better adjectives? Why don't they talk about sound the way a musician would describe the sound, for example? Or at least describe it the way a normal person can understand it. By all means, use exotic descriptions - but please do that after you have done the simple but comprehensive description.
2. Why has no one attempted to define a baseline? I will even accept a baseline per reviewer (if not an industry standard) - such as stereophile and 6moons defining their own baselines. It could be as simple as beginning every review with a set of subjective tests. Play a recording of 7 keys on a piano for example, and use the same CD and recording for all your reviews. This could be test #1. Now define 9 other tests - that together cover the gamut of what an audio system is supposed to do. Then, if you want, get as subjective and floral in your adjectives as you wish.
Thad E Ginathom
Well-Known Member
Why don't we?
Like the guy who told the engineer, "I want it more... more... more... purple!"? Musicians are not necessarily the guys to set a useful standard. If we want a standard vocabulary, then we should probably look at audio engineering: ultimately, we'll be talking in numbers! That takes training, practice and, ultimately, its something we may not all be comfortable with anyway.
It's an entertainment industry. Within that, we may have reviews that we feel embody our personal taste, so are worth reading, and we may pick up some useful comparisons.
Thank you. Just doing my part to spice up the forum. BTW, I don recall mentioning NF being absence in digital audio. It maybe lesser but they are just there but probably way beyond then any other components.
Will get back to the rest of your post once I get pass post 17.
It actually depends which noise floor you are referring to. NF and SNR are not the same. SNR may be improved by increasing the volume if it involves the NF of the room. However, in the case of an amplifier or a preamp the SNR remains constant irrespective of the volume level.
This video explains it. Please let me know if you agree with the statement.
Last edited:
Ambio,
Nose floor is the measure of the amount of noise voltage which an electronic system can produce on its own without any audio signal being injected. It is the lowest most VOLTAGE which is generated by thermal phenomena out of electron collision occurring in the devices such as transistors or tubes when subjected to behave actively or biased in a circuit to perform desired set of features such as Amplification||Modulation. Usually in transistors the noise is measured in nV/?Hz and in the output of power amplifier in uV. The SNR thus can be given by 10Log[Vmax/Vnoise] in dB.
The SNR simple gives us a ratio of how max the signal can get amplified in terms of voltage with reference to the noise generation voltage which is inherent to every system. The if any signal voltage at output has amplitude lesser than the average amplitude of noise then it will get masked and lost and very difficult for the ears to differentiate from the original material, this is more prominent in Phono Preamps.
So the noise floor in otherwords is the lower possible voltage present at output of any amp or preamp or any regenerator.
And yes the U(t) in graph of Nikhil's post is a symbol for representing voltage as a function of time. Germans usually denote Voltage as U, whereas normally its denoted by V. Its technical nomenclature.
If you want to read more on types of noise phenomena occurring in electronics then search these words "Johnson Noise or thermal noise", "Shot Noise", "Flicker noise" & "Avalanche noise"
I hope this helps.
Kanwar
Nose floor is the measure of the amount of noise voltage which an electronic system can produce on its own without any audio signal being injected. It is the lowest most VOLTAGE which is generated by thermal phenomena out of electron collision occurring in the devices such as transistors or tubes when subjected to behave actively or biased in a circuit to perform desired set of features such as Amplification||Modulation. Usually in transistors the noise is measured in nV/?Hz and in the output of power amplifier in uV. The SNR thus can be given by 10Log[Vmax/Vnoise] in dB.
The SNR simple gives us a ratio of how max the signal can get amplified in terms of voltage with reference to the noise generation voltage which is inherent to every system. The if any signal voltage at output has amplitude lesser than the average amplitude of noise then it will get masked and lost and very difficult for the ears to differentiate from the original material, this is more prominent in Phono Preamps.
So the noise floor in otherwords is the lower possible voltage present at output of any amp or preamp or any regenerator.
And yes the U(t) in graph of Nikhil's post is a symbol for representing voltage as a function of time. Germans usually denote Voltage as U, whereas normally its denoted by V. Its technical nomenclature.
If you want to read more on types of noise phenomena occurring in electronics then search these words "Johnson Noise or thermal noise", "Shot Noise", "Flicker noise" & "Avalanche noise"
I hope this helps.
Kanwar
Last edited:
It depends, not always true.
Thank you Kanwar. This is now getting more and more technical. Too much to absorb in three days. I can smell something is burning inside my head. :lol::lol::lol:
If I may, nV/?Hz is used in Op amps or chips because the noise level is far too low for conventional units. All the noise measurements in resistor, transistor, chips are still far too low to affect the sound quality. The biggest problem with noise starts with the finished product and the design.
The final noise coming out the amplifiers output must be relatively small compared to the nominal output of 2.83V/1W. My understanding is they are usually in millivolt as uV is too small for any effect on the speakers.
Regarding the U(t), could it be representing amplitude? Voltage over time makes no sense for the sine graph shown there.
Ya it becomes a bit technical when you go deep
The johnson noise generation in resistor coupled with noise generation from semiconductors, all collectively reflect as noise floor at final output. Noise from resistors is not small when it comes to preamp or phono applications.
If you ever listen to 110dB sensitivity Pro-audio HF units, you will be surprised to see how easily a 150uV NF on the amplifier output gives you the HISS which is very annoying to your ears. Again it depends ,certainly no issues with 86dB dome tweeters.
But yes noise measurement if often done in uV not mV in case of good low noise amplifiers.You are missing a point here, there has to be a quantity for which the measurement of amplitude is being done and in that case[That graph] its Voltage U(t)and when we say amplitude, it means amplitude of either voltage or current varying in sine waveform as a periodic function of time [t] only. Time is there to make sure what is the max and minimum excitation of noise voltage at random in order to get an averaged value from set of variable reading over a period of time. You cannot represent a sinewave without the presence of time in X-axis factor.
Last edited:
No dispute there. I was referring why nV/?Hz. The noise level measurement there and in Prem's example varies. Could you please explain why different units were used. The rational behind it?
I wouldn't know as I have never measured SNR before but since you are familiar with the formula and pro amplifier. Could you give an example by using this Crown pro amplifier specs to show how loud the 150uV hiss going to be? (The SNR for this amp is 100dB.)
I should have worded it better. My point is the graphs were used to show how noise floor affects the sound. Noise floor would not alter the sine wave. If the Y axis were Voltage then could you explain the edginess that the NF produced as per the example graph? If it is V over time than it doesnt make sense.
Thank you.
The audio signal can eventually be considered a subset of the Johnson noise, since Johnson noise is present from DC to infinity. The difference here being the audio signal is not random like pure noise.
When two or more signals are present in the same circuit, they either add up or cancel each other, depending on their phase relationship. So the jagged little buggers at the bottom of the graph add/cancel with the pure sine wave in the example, producing the jagged sine wave. That shows up as distortion/alteration of the desired signal, however minute (THD figures are typically < a hundredth or even a thousandth of a percentage). Hope that clarifies your doubt.
Kindly elaborate in better way what kind of difference?
You are asking a wrong question !
SNR gives you the ratio of max to min range of headroom from noise floor to max unclipped voltage. It doesn't tells you how loud a noise voltage shall sound to your ears. The loudness depends on sensitivity of speakers and efficiency of your ears to be precise.
From that Crown amplifier, its 6K model quotes 103dB SNR A weighted @ 8ohms. Power output is 1200W. So voltage swing required for that much power into 8ohms is 100VRMS. Now take the 103dB and convert it to linear factor which comes out 141253.75 X. So the noise voltage of this amplifier in average would be 100V[vmax] Divided by 141253.75[linear attenuation factor], which comes around 700uV which is very much audible with 110dB sensitivity HF speakers.
Now if we take 150uV noise as reference average noise level for a certain amplifier then its SNR would be[Provided the max swing is 100Vrms into 8ohm as assumption] 116dB roughly.
Let it be in a simple way....
Try to imagine a MC phono cartridge output giving 500uV audio output and the phono preamp output noise floor is 150uV.......... your audio info hence gets masked by almost 1/3 noise floor level.
Got the point or not? That edginess is nothing but distortion of the audio waveform from the high presence of noise floor, so you are getting random of F1+F2 and F2-F1 in absolute values.
Last edited:
Thanks JLS001. I started this thread to understand how people could hear lower NF with change of , say , a cable. The graphs posted there were to explain NF and how they affect the sound. In the example above, ignoring the voltage and time which now it appears to be units taken out of an ocsiloscope, the only way the graphs is going to be jagged is when two identical frequency with similar pitch. What I am afraid, we are referring to something which may or may not be relevant to the topic. I am looking for practical explanation as to how the NF affect the pure tone as shown in the graph.
I do understand of various way that a our tone can be made to look like that but the relevance in this context of discussion eludes me. If you were to measure a pure tone from a vinyl then it may look closer to that but that not because of the NF of the equipment but due to distortion plus NF in the recording as cut into the vinyl grooves.
My understanding is for the 110dB speaker (depending on the nominal impedence) all you need is 2.83V for the ouput of 110dB SPL of reference 1KHz test signal at 1 meter distance. Since you mentioned that proamp produce 150uV of noise and it is very loud hiss, what will be the loudness level the noise playing simultaneously at 150uV? Or put it another way what would be the loudness of 150uV 1KHz signal?
I will get back to you on the first question as my laptop is not with me at the moment.
M
mpw
Guest
i remembered this very much..
lowering the noise floor to put it in simple terms is making the electronics quieter.
regards
mpw
Pun - noise floor is a term magazines use freely ... maybe it rhymes with dance floor :lol:.... maybe it somehow tinkles an "Aha !! " bell... somewhere in the deep audiophile mind..
lowering the noise floor to put it in simple terms is making the electronics quieter.
regards
mpw
Pun - noise floor is a term magazines use freely ... maybe it rhymes with dance floor :lol:.... maybe it somehow tinkles an "Aha !! " bell... somewhere in the deep audiophile mind..
Last edited by a moderator:
For a given pro-driver for HF with 110dB sensitivity, if we take 1W @ 8ohms @ 2.83V then 150uV becomes 25dB, which is way above threshold of hearing and easily recognizable if you sitting straight in front of the HF unit.
No where i have said it is very loud hiss, but if you are sitting in a normal silent listening room and enjoying a low level music, you can easily make out the hiss as an annoying factor with 110dB speakers during quite passages.
Follow HiFiMART on Instagram for offers, deals and FREE giveaways!
Similar threads
