Head-Fi.org › Forums › Equipment Forums › Cables, Power, Tweaks, Speakers, Accessories (DBT-Free Forum) › I did a scientific test on a power conditioner--IT WORKS!!!
New Posts  All Forums:Forum Nav:

I did a scientific test on a power conditioner--IT WORKS!!! - Page 3

post #31 of 67
Quote:
Originally Posted by JohnFerrier
I don't know what is in the Furnman unit, but the ones I have use a 20 pound torroidal isolation transformer and rather large polypropylene capacitors.
Well if it is 20 pounds of doing nothing, than it can continue doing nothing while shipped to be tested side by side to the Furman . Now *that* would be an interesting test.
post #32 of 67
Quote:
Originally Posted by Ferbose
Do I need to offer a simpler analogy? There are two pieces of white paper, but only one has some faint stain marks. When the room is very dark, you can't see any stain mark and you declare both papers are blank. With better lighting you start to see the difference, and you know which paper is cleaner.
Turning on the light is analogous to turning up the volume beyond 110 db.

See ya
Steve
post #33 of 67
Quote:
Originally Posted by bigshot
107 db is about the volume of a car horn at close range. That means that you would have to have your music turned up pretty doggone loud to even get the noise floor to raise above zero. For all intents and purposes, a 30% improvement on 107 (or 104, depending on how you calculated it) is not discernable at normal listening levels.

See ya
Steve
That was in reference to a instrumental amplifier and not audio, and he already put a disclaimer saying that you couldn't really directly correlate it. In fact I'd suspect the 30% figure to be just a guestimate since you do not really know if it is a purely linear 30% improvement across the board in all conditions, etc.

Again it depends on where the noise is. If you are talking about 104 vs 107, no one cares as it is both ear bleeding inaudible situations (or if negative reference to zero still inaudible). But within a reasonable loudness, 3db is just on the border of audibility. So if there is 3db of *hash* or grain sitting on top of your signal, it is easily audible to those with keen ears. Of course there are online hearing tests out there that debate on the definition of easily audible since apparently some people can't even recognize clearly out of tune notes.

Also both aural senses and visual perception rely significantly on contrast. So maybe a 3db difference won't be noticable if both out of the audible range, but a 3db difference is audible if in the audible range. But no one I know can distinguish ultra-violet from infra-red either...

I'm not saying everyone's system automatically must have a layer of hash or noise that will be removed with a power conditioner. I nipped my problem in the bud by removing a ground loop. Plus I never get a chance to do any listening except the dead of night where my line conditions are pretty optimal anyhow. And I'd also have to agree that better power supplies in audio equipment should come first (i.e. a crappy DC switching power supply on a clean main is going to have a worse output than a very good regulated DC power supply on a noisy main).
post #34 of 67
Ferbose, it was interesting that you are involved (some how) with the measurement of signals from neurons (and posted something about it). I'm interested in neuroscience and try to read about it when I can. Learn all you can.
post #35 of 67
Thread Starter 
Quote:
Originally Posted by Tim D
It depends...in my experience the noise can just be part of the noise floor (i.e. sits at the bottom), but it can also sit on top of your signal. Obviously the noise that sits on top of the signal is more harmful than a slight 3db rise in noise floor.
The noise of the amp in the absence of signal can be called signal-independnet noise or self-noise. It is not going to go away when you have signal input. Suppose self-noise is 1 mV. When a 1 V signal comes in the output is going to be 1 V (+-) 1 mV, assuming voltage gain is 1. Every part of the noise floor is going to be superimposed on the signal.

It is also possible that amplifier has signal-dependent noise. If the 1V signal induces a noise of 5 mV in the amp. Then you are going to get 1 V (+-) 6 mV output. In other words, noise is never going away, but they just pile up higher and deeper.
post #36 of 67
Quote:
Originally Posted by Tim D
That was in reference to a instrumental amplifier and not audio, and he already put a disclaimer saying that you couldn't really directly correlate it. In fact I'd suspect the 30% figure to be just a guestimate since you do not really know if it is a purely linear 30% improvement across the board in all conditions, etc.
If that's the case then his measurements aren't measuring the right things. 3db of noise floor at 107db is great. A steady 3db of noise at all volume levels is a lousy amp, isn't it?

See ya
Steve
post #37 of 67
There are two measurements that explain what I am talking about. Noise Floor measurements and S/N Ratio's. One is noise without signal and one is noise with signal. I am just saying poor noise floor is the lesser of two evils compared to poor S/N Ratio. If you had an amp that had a -90 db S/N Ratio and a -120 db Noise floor, and an amp that had a -96 db S/N Ratio and -96 db noise floor...I'm telling you right now the -96 db S/N ratio amp will be the clear winner in sound quality (well assuming the noise components in the first amp isn't easy on the ear harmonic distortion that is ). That also is something to think about...amplitude or loudness is great and all, but the *type* of distortion is a huge factor in audibility. Noise that sits only on the floor (hence noise floor) and ignores signal for the most part is mostly of annoyance only to super sensitive IEM's...noise that sits on top (signal noise) is of annoyance for any headphone. Crappy power or ground noise can sound like hash and very grating. 2nd harmonic tube saturation would be the opposite.
post #38 of 67
Thread Starter 
Bigshot and John Ferrier, I think you misunderstand the audibility of noise itself vs. noise+signal.

Bigshot, you are right that hi-fi system does not go over 110 dB SPL (sound pressure level), which could cause permanent hearing damage. Now think about CD's 96 dB dynamic range. The last bit would be -96 dBFS (-96 db from full signal), meaning 14 dB SPL. A quiet room has noise level of 20-30 dB SPL, so you can't hear a -96 dBFS signal played through CD, at 14 dB SPL. But the truncation noise in the last bit of CD is definitely audible when the signal is -20 or -40 dBFS (normal music). This has been proven again and again by mastering engineers. That is wahy every digital mixer today has dithering. [Bob Katz's Mastering Audio has an excellent discussion on dithering]

Signal can make noise below hearing threshold audible. When signal is big and noise is small, you hear subtle changes to the signal. It also works the other way around: noise can bring out undetectable signal. If there is a signal too low to be heard, and you add a lot of audible noise to it, you can start to hear the signal. With big noise and small signal, you hear that the noise sounds funny and actually carries some signal.

For example, you cannot hear the difference between a 5 dB and 8 dB SPL signal. You hear just a quiet silence. However, you can hear the difference between 70 dB and 73 dB SPL--in fact 73 dB is twice as loud as 70 dB. Of course 3 dB difference in loudness makes a big difference. For loudness detection the ear can hear 0.1-0.5 dB difference. For noise detection, the ear is even more sensitive. It is thought that audio equipment should have >130 or >140 dB S/N to be considered noiseless (and hence 24 bit audio with 144 dB dynamic range).

Again a visual analogy. Suppose the reflection of moonlight is signal, and the shadow of a tree is noise. On most nights, you can't see the shadow created by the moon, because noise is below visible threshold and the signal is barely visible. When the sun comes out, the signal (relfected sunlight from the ground) is much stronger, and the noise (tree shadow) becomes so apparant.
post #39 of 67
My sources may be out of date, but I believe the ear can hear differences of around 1db in good conditions. .1-.5db is a bit optimistic...that is like saying everyone can tell channel mismatches in volume pots or sources outright and confidently which is not true since a lot of equipment aren't even matched this well. 10 db is the 'perceived' doubling of loudness, 3db is the doubling of actual intensity. Course what the human ear can detect is really just a vague guestimate. I know a lot of people would have a very difficult time detecting a 3db bass boost if the bass is cleanly reproduced in both settings. Course throw in a distorted bass boost and it doesn't matter what level the boost is, you can probably hear the difference. However our ability to sense bass is not as keen as midrange...see there are so many variables at play.

But yes I would wager we can and do hear aggravating distortion in extremely minute amounts. A clean signal vs a minutely louder clean signal no. A clean signal vs a minutely dirtier signal yes.
post #40 of 67
Actually, the signal makes noise more difficult to hear. Can you hear someone whisper in your ear better when it's quiet, or at a rock concert?
post #41 of 67
Quote:
Originally Posted by Tim D
But yes I would wager we can and do hear aggravating distortion in extremely minute amounts. A clean signal vs a minutely louder clean signal no. A clean signal vs a minutely dirtier signal yes.
I'll cast my vote for this. I can hear intermodulate distortion of particularly high midrange sounds because my hearing, for some reason, is exceptionally sensitive to them. It sounds like a sharp, painful slice against the background of "good" treble. I've actually had occasion to test my difference threshhold for 3500Hz (a resonance point on one of my speaker systems' titanium tweeter), and even at +.5dB it's moderately ringing and painful. I have to turn it to -3dB at ~3500Hz if I don't want my ears to feel like they're being stabbed after a few hours of listening to anything with a lot of frequency presence in that region.

However, in other speaker systems (particularly ones with silk- or paper-dome tweeters which lack that resonance), I can listen at the same average volume without equalization even though the drivers measure close to one another in linear frequency response; it's sort of a "sins of addition" versus "sins of subtraction" issue, I suppose. Subtraction, in the higher registers, sounds better to me.
post #42 of 67
Thread Starter 
Quote:
Originally Posted by bigshot
If that's the case then his measurements aren't measuring the right things. 3db of noise floor at 107db is great. A steady 3db of noise at all volume levels is a lousy amp, isn't it?

See ya
Steve
No, you misunderstand dB. dB is a relative measurement, not absolute. dB = 20 * log(V1/V2). If V1 is 10 times larger than V2, then the difference is 20 dB. dB is in logarithmic scale. For power, dB = 10 * log(P1/P2). 2V is 6 dB louder than 1V, but 2W is 3 dB louder than 1W. This is because P=IV=V^2/R. All audiophiles need to know this basic math about dB, seriously. When we say SPL=x dB, it means it is x dB louder than a reference we call 0 dB SPL. You can look up the actual sound pressure corresponding to 0 dB SPL if you like. Quiet rooms are generally 20-30 dB SPL. A super-luxury car is about 40 dB SPL when driving slowly. So 0 dB SPL is a very, very soft sound not normally audible.

If full signal causes the amp to output 10 V voltage swing, then -107 dB noise is 45 microVolts. Here the output is 10 V (+-) 45 microVolts (S/N = 107 dB). When signal is not as loud, say 1 V, the output is 1V (+-) 45 microVolts (S/N=87 dB). When signal is soft, say 0.01 V, the output is 0.01V (+-) 45 microVolts (S/N= 47 dB). When the signal is barely audible, say 1 miliVolt, the output is 1 miliVolt (+-) 45 microVolt (S/N= 27 dB!!!). Do you see how 107 dB S/N is not so great when you enter the softer passages of music? The ambience of that soft solo passage in a choral piece sung in a church is now destroyed by noise!

When the signal is zero or say something tiny like 100 microvolts, the output is 100 (+-) 45 microvolts (S/N=7 dB). This volatge is too small to move the speaker coils to produce an audible sound, and hence you think the noise is inaudible. But it is audible when there is actually music playing.
post #43 of 67
Quote:
Originally Posted by JohnFerrier
Actually, the signal makes noise more difficult to hear. Can you hear someone whisper in your ear better when it's quiet, or at a rock concert?
Using a Rolling Stones concert at a concrete baseball stadium as an example. Crowd chatter = noise floor. The louder the PA the better the S/N ratio i.e. signal over the noise floor. Horrible reverberations and pockets made from concrete seating and overhang would be noise inline with the signal. You won't hear it unless there is a signal or music playing. Finally turning up the sound so loud and selling enough beer that you no longer care about noise floor or S/N ratio is usually the solution of choice .

A power conditioner could be like putting up acoustic treatment in a crappy venue. But if you were at a well designed acoustic venue, you wouldn't need it. You can never assume that power conditioners won't benefit other people's setup...that is like saying that all live shows will only be held at well designed venues...I wish.
post #44 of 67
Quote:
Originally Posted by Ferbose
No, you misunderstand dB. dB is a relative measurement, not absolute. dB = 20 * log (V1/V2). If V1 is 10 times larger than V2, then the difference is 20 dB. dB is in logarithmic scale.

If full signal causes the amp to output 10 V voltage swing, then -107 dB noise is 45 microVolts. Here the output is 10 V (+-) 45 microVolts (S/N = 107 dB). When signal is not as loud, say 1 V, the output is 1V (+-) 45 microVolts (S/N=87 dB). When signal is soft, say 0.01 V, the output is 0.01V (+-) 45 microVolts (S/N= 47 dB). When the signal is barely audible, say 1 miliVolt, the output is 1 miliVolt (+-) 45 microVolt (S/N= 27 dB!!!). Do you see how 107 dB S/N is not so great when you enter the softer passages of music? The ambience of that soft solo passage in a choral piece sung in a church is now destroyed by noise!

When the signal is zero or say something tiny like 100 microvolts, the output is 100 (+-) 45 microvolts (S/N=7 dB). This volatge is too small to move the speaker coils to produce an audible sound, and hence you think the noise is inaudible. But it is audible when there is actually music playing.
I can follow your numbers. You will more easily hear the louder signal. Noise is not easier to hear when competing with another signal. If you wish to believe so, then that is fine. What you write makes sense otherwise.
post #45 of 67
When you want to hear something interesting on the television, do you tell all your friends to start talking? Or do you tell them to be quiet?
New Posts  All Forums:Forum Nav:
Head-Fi.org › Forums › Equipment Forums › Cables, Power, Tweaks, Speakers, Accessories (DBT-Free Forum) › I did a scientific test on a power conditioner--IT WORKS!!!