[xnor]

The graph seems to me to be as much use as a chocolate tea pot!

 
Mhhhmm yummie. I want one.

 

[krismusic]

Mhhhmm yummie. I want one.

 

[krismusic]

Can someone enlighten me? I cannot find an answer on the net. What does +or- dB actually mean?

 

[xnor]

Decibel is a logarithmic "helper unit" that describes a ratio of power or field quantities (such as sound pressure, voltages etc.).
Logarithmic means that even huge ratios can be described with relatively small decibel numbers.
 
Examples:
You put 1V into an amp and measure a max output of 5V:
 
5V / 1V = gain of 5x
 
20*log₁₀(5) = gain of +14 dB
 
 
 
You put 1V into an amp and the max output is 0.1V (a tenth):
 
0.1/1 = gain of 0.1x
 
20*log₁₀(0.1) = gain of -20 dB
 
 
As you can see, as soon as the gain factor is lower than 1x you get negative dB. For power you need to use 10*log₁₀(ratio).
 
 
Now if some specs say 20 Hz - 20 kHz +/- 3 dB this means that in the whole range the frequency response can vary by 3 dB up or down.
+/- 10 dB is perceived as twice/half as loud.
 
It gets more confusing since above numbers are completely relative, sometimes written as dBr. There are some common dB units that have a reference, such as dB SPL (reference is a sound pressure of 20 µPa), dBV (reference is 1 V) etc.

 

[krismusic]

That's a brilliant explanation xnor. Thank you very much indeed. I will now stare at it until I understand it. Couldn't be put much clearer. I'll get it! The - value was puzzling me and I now at least see what that represents.

 

[Steve Eddy]

Decibel is a logarithmic "helper unit" that describes a ratio of power or field quantities (such as sound pressure, voltages etc.).
Logarithmic means that even huge ratios can be described with relatively small decibel numbers.

Examples:
You put 1V into an amp and measure a max output of 5V:

5V / 1V = gain of 5x

20*log₁₀(5) = gain of +14 dB



You put 1V into an amp and the max output is 0.1V (a tenth):

0.1/1 = gain of 0.1x

20*log₁₀(0.1) = gain of -20 dB


As you can see, as soon as the gain factor is lower than 1x you get negative dB. For power you need to use 10*log₁₀(ratio).


Now if some specs say 20 Hz - 20 kHz +/- 3 dB this means that in the whole range the frequency response can vary by 3 dB up or down.
+/- 10 dB is perceived as twice/half as loud.

It gets more confusing since above numbers are completely relative, sometimes written as dBr. There are some common dB units that have a reference, such as dB SPL (reference is a sound pressure of 20 µPa), dBV (reference is 1 V) etc.

Uh... What was that middle part again?

se

 

[xnor]

Some completely random examples with voltage gain and how to convert it into dB.

 

[Steve Eddy]

Some completely random examples with voltage gain and how to convert it into dB.

No no, just being a smartass. It's a cliche line from Kevin Klein's character, Otto, in the film A Fish Called Wanda.

se

 

[kalston]

A great film

 

[miceblue]

You're zooming in on just 18kHz-22kHz. If you look at a conventional scale, from 0-22kHz, you'd probably see a natural rolloff from the bass to the treble as usual?

It's hard to tell with Audacity.

Maybe?


[rule]
Okay, so I'm trying to volume-match two cables from a source with a 1 kHz sine wave playing from the source (sine wave was generated with Audacity) at a set volume. How exactly do I do that?
Cable A's impedance measurement


Cable B's impedance measurement


Can Ohm's Law be applied here?
V = IR
If V is from the source, it should be constant, R is altered, so the current should be changing? The multimeter I was using couldn't measure current.

Anyway, I decided to measure the AC and DC voltages (I'm not sure what kind of signal is being measured so I just did both).
Cable A's AC voltage measurement


Cable B's AC voltage measurement


Cable A's DC voltage measurement


Cable B's DC voltage measurement


The DC measurements kept fluctuating, so I'm not sure if the displayed voltages are accurate.


So all in all, how am I supposed to volume-match cables? Their measured impedances are different, and one sounds like it plays music louder than the other. Were I to evaluate the "sound quality" of the cables, I would want to properly volume-match them, but as of right now, I have no idea how to do that.

 

[xnor]

If you know the headphones' and cable's impedance at 1 kHz you can calculate it like a voltage divider.
 
For example 20 ohm headphone, 5 ohm cable1, 3 ohm cable2, all at 1 kHz:
20/(20+5) = 0.8
20/(20+3) = 0.8696
 
20*log10(0.8696/0.8) = +0.72 dB for cable2
 
 
You could also measure the voltage across one headphone driver with the cable connected. (Gonna need a little self-built adapter for that so you have access with the probes.) For 1V output from the amp you should measure the same numbers as calculated above, assuming the amp has 0 ohm output impedance.

 

[miceblue]

If you know the headphones' and cable's impedance at 1 kHz you can calculate it like a voltage divider.

For example 20 ohm headphone, 5 ohm cable1, 3 ohm cable2, all at 1 kHz:
20/(20+5) = 0.8
20/(20+3) = 0.8696

20*log10(0.8696/0.8) = +0.72 dB for cable2

Oh, that was a lot simpler than what I was doing, hahaha. That being said, how do I adjust a source or amp's volume to such a precise number, if that's even possible. On Rockbox at least, I can change the volume by 1 dBFS, but even then, it's not that accurate.

[rule]
What's your guys' take on DSD? xnor I saw your posts from another website already. XD

The JDS Labs C5D uses a 32/384 DAC with delta-sigma DSD playback, but it's limited to 24/96.
http://www.ti.com/product/pcm5102a

 

[xnor]

Unless there is some extremely well-implemented noise shaping and filtering going on with both the ADC (recording) and DAC (reproduction), DSD is measurably inferior to even 24/96.
Some would even say worse then 24/48.

 

[Muinarc]

You reminded me of a question I had a while ago but never looked into. Back in the early days of CD I remember seeing (and owning) some equipment that had a "regulated 1-Bit DAC". I know DSD is 1-bit streaming or something to that effect.

How are the two different (I would assume quite different)? Also, were there any pros or major cons to the 1-bit DAC (or was it even a "thing" or just marketing)?

*edit* I did some more searching at it looks like it's mostly a marketing term. It's still an oversampling DAC I guess.

 

[krismusic]

Merry Christmas to all fellow skeptics. May 2014 bear scrutiny!