[MRC001]

Actually regarding volume 1 dB should be easily audible as 3 dB boost is a 50% gain in power at that frequency. I was playing around with Sinegen and sine wave frequency samples and pink noise sample. It was very hard on my ears so no wonder 1 dB difference didn't have an audible difference to me, as I didn't take any breaks to listen to actual music to let my ears lest. This is why equalization is hard work, finding prominent peaks might not be very difficult, but all the milder dips and peaks are quite difficult to hear for two reasons: the testing methods tire your ears and you start perceiving sound differently and not everyone has golden ears

Gain is expressed as voltage, where a 3 dB gain is an increase of 41% (ratio of 1.41:1), which would double the power at that frequency. That sounds like a big difference, but the ear perceives things exponentially, not linearly, in both frequency & amplitude. So it's not as big a difference as it seems numerically. A 3 dB gain or loss could be easy or hard to hear, depending on over how broad a frequency range it is applied, what that frequency range is, and whether the music you're listening to has much energy in that range.
 
For more on dB and ratios: http://www.mclements.net/Mike/mrc-blog/blog-140310.html
 
Like you, I don't like bright speakers or headphones. I don't like them dark either; I prefer a neutral presentation. With most audio equipment, too bright is a much more common flaw than too dark.

 
I agree, listening for differences one needs to keep in mind that the acuity of our perception varies from day to day, so do it when you've got fresh ears. Patience over time builds experience which builds acuity.

 

[InCreD1Ble]

  Gain is expressed as voltage, where a 3 dB gain is an increase of 41% (ratio of 1.41:1), which would double the power at that frequency. That sounds like a big difference, but the ear perceives things exponentially, not linearly, in both frequency & amplitude. So it's not as big a difference as it seems numerically. A 3 dB gain or loss could be easy or hard to hear, depending on over how broad a frequency range it is applied, what that frequency range is, and whether the music you're listening to has much energy in that range.
 
For more on dB and ratios: http://www.mclements.net/Mike/mrc-blog/blog-140310.html
 
Like you, I don't like bright speakers or headphones. I don't like them dark either; I prefer a neutral presentation. With most audio equipment, too bright is a much more common flaw than too dark.
 
I agree, listening for differences one needs to keep in mind that the acuity of our perception varies from day to day, so do it when you've got fresh ears. Patience over time builds experience which builds acuity.

+3 dB gain does increase the voltage by 41%, but that voltage is directly related to the power output, which in case of +3 dB boost is doubled. A 6 dB gain would double the voltage and quadruple the power. -3 dB gain would halve the power, reduce the voltage by ~29%, -6 dB gain would reduce the power 4 times, by reducing the voltage by 50%. Needless to say, in audio, there are other factors factors that make the results of these gains deviate from theoretical values, therefore you can't predict the exact volume increase.

 

[MRC001]

 

+3 dB gain does increase the voltage by 41%, but that voltage is directly related to the power output, which in case of +3 dB boost is doubled. A 6 dB gain would double the voltage and quadruple the power. -3 dB gain would halve the power, reduce the voltage by ~29%, -6 dB gain would reduce the power 4 times, by reducing the voltage by 50%. Needless to say, in audio, there are other factors factors that make the results of these gains deviate from theoretical values, therefore you can't predict the exact volume increase.

That is correct. My reply above was in response to your comment that +3 dB increases power by 50%. Mathematically speaking, that means increasing by a factor of 1.5, but I see now you meant it doubles it - a factor of 2.0.

 

[InCreD1Ble]

  That is correct. My reply above was in response to your comment that +3 dB increases power by 50%. Mathematically speaking, that means increasing by a factor of 1.5, but I see now you meant it doubles it - a factor of 2.0.

Oh yes, my bad! I was thinking of -3 dB which is a 50% decrease in power, +3 dB isn't the same. My bad

Unrelated question, why do sharp peaks and troughs in EQ's are better to be avoided?

 

[MRC001]

  Unrelated question, why do sharp peaks and troughs in EQ's are better to be avoided?

EQ with big enough amplitudes combined with sharp changes (steep slope in dB / octave) causes other kinds of distortion that can become audible. That distortion typically takes the form of pre and post echo or induced ripple or ringing on the waveform. This smears transients making them less crisp and natural sounding.
 
This is why I prefer parametric EQ over "graphic" EQ - it's far more transparent. A parametric EQ lets you set the center frequency, amplitude and slope. A graphic EQ doesn't let you set the slope, and it typically uses steep slopes so the bands don't overlap. Using several knobs on a graphic EQ to shape a gradual curve is NOT the same thing as a single wide parametric EQ!
 
In my own A/B/X testing, using purely digital parametric EQ applied in 24-bit with noise shaped dither, +/- 3 dB and 1 dB / octave is completely transparent audibly and even analyzing the wave on a computer shows no residual distortion. But +/- 9 dB and 18 dB / octave was easy possible to tell apart audibly (> 95% confidence) and you could see the induced ripple in an audio wave editor. Threshold of audibility is somewhere between these extremes.

 

[InCreD1Ble]

  EQ with big enough amplitudes combined with sharp changes (steep slope in dB / octave) causes other kinds of distortion that can become audible. That distortion typically takes the form of pre and post echo or induced ripple or ringing on the waveform. This smears transients making them less crisp and natural sounding.
 
This is why I prefer parametric EQ over "graphic" EQ - it's far more transparent. A parametric EQ lets you set the center frequency, amplitude and slope. A graphic EQ doesn't let you set the slope, and it typically uses steep slopes so the bands don't overlap. Using several knobs on a graphic EQ to shape a gradual curve is NOT the same thing as a single wide parametric EQ!
 
In my own A/B/X testing, using purely digital parametric EQ applied in 24-bit with noise shaped dither, +/- 3 dB and 1 dB / octave is completely transparent audibly and even analyzing the wave on a computer shows no residual distortion. But +/- 9 dB and 18 dB / octave was easy possible to tell apart audibly (> 95% confidence) and you could see the induced ripple in an audio wave editor. Threshold of audibility is somewhere between these extremes.

Ah I see, that makes sense. What causes the echoes though? The phase shifts? Also, don't parametric EQ's have echoes as well, for example it's quite prominent in linear phase EQ's.

 

[kazuya95]

It's easy to hear, I've double-blind tested and can hear the difference reliably down to 1 dB. The difference is subtle but not that hard to hear; it's in a very accessible frequency range that has a lot of energy from a wide variety of instruments. I actually prefer +2 dB as +3 is a just bit too much! To my taste, the LCD-2 has near perfect response. All it needs is a slight lift in the upper presence to compensate for its softness in this area.

The silver cable guy is someone else - I'm using the Audeze stock balanced cable driving from my Oppo HA-1.

With Silver cable you have thinner and dryer sound. More impact in low. More detail and separation.
If you like warm and thicker sound go with copper

 

[MRC001]

  Ah I see, that makes sense. What causes the echoes though? The phase shifts? Also, don't parametric EQ's have echoes as well, for example it's quite prominent in linear phase EQ's.

The echoes depend on the slope of the filter. The steeper the slope, the worse the echoes. Parametric EQ can be steep or gradual - you define the slope. It's transparent if you use gentle slopes.
 
Getting a shallow filter slope to minimize distortion is the primary reason for digital oversampling. At 44.1 kHz the Nyquist limit is 22k so the transition band of the filter is only 2k wide. A filter that steep may induce ripple or echoes. 2x oversampled, the transition band is now 24k wide - almost 10 times wider, so the filter has a much shallower slope - eliminating ripple, echoes and other distortion.

 

[Sound Eq]

  The echoes depend on the slope of the filter. The steeper the slope, the worse the echoes. Parametric EQ can be steep or gradual - you define the slope. It's transparent if you use gentle slopes.
 
Getting a shallow filter slope to minimize distortion is the primary reason for digital oversampling. At 44.1 kHz the Nyquist limit is 22k so the transition band of the filter is only 2k wide. A filter that steep may induce ripple or echoes. 2x oversampled, the transition band is now 24k wide - almost 10 times wider, so the filter has a much shallower slope - eliminating ripple, echoes and other distortion.

 
 
wish u were here 3 years ago when i had my audeze lcd to help me eq it
 
u have great eq knowledge

 

[InCreD1Ble]

  The echoes depend on the slope of the filter. The steeper the slope, the worse the echoes. Parametric EQ can be steep or gradual - you define the slope. It's transparent if you use gentle slopes.
 
Getting a shallow filter slope to minimize distortion is the primary reason for digital oversampling. At 44.1 kHz the Nyquist limit is 22k so the transition band of the filter is only 2k wide. A filter that steep may induce ripple or echoes. 2x oversampled, the transition band is now 24k wide - almost 10 times wider, so the filter has a much shallower slope - eliminating ripple, echoes and other distortion.

I understand that, but what is the CAUSE for digital filters? I know the origin of these ripples and passive and active physical filters but not the digital ones that are used in EQ's.

 

[MRC001]

Steep filters cause distortion whether or not they are analog or digital. Comparing the best engineered filters, digital vs. analog, digital filters can be more transparent because they get closer to the theoretical minimum distortion that signal theory says is possible. And they give you more control over what kinds of artifacts arise. So the general rule is, filter in digital domain when possible, and use the most shallow/gradual filter slope that will do the job.
 
I am not well-versed enough in signal theory to explain exactly why or how a steeper slope causes more distortion. But it seems intuitive when you consider that transient impulses contain a wide range of frequencies going into the supsersonic that all occur suddenly in perfect time sync when they occur in nature. We hear this as a light, crisp, clean natural snap. Now you apply a filter boosting or cutting only some of these frequencies. That spreads them in time differently from the rest - but how differently? A shallow filter treats similar frequencies more or less the same - that's the definition of shallow - so whatever happens, happens to all of them together. The steeper a filter is, the more differently it treats similar frequencies.

 

[InCreD1Ble]

  Steep filters cause distortion whether or not they are analog or digital. Comparing the best engineered filters, digital vs. analog, digital filters can be more transparent because they get closer to the theoretical minimum distortion that signal theory says is possible. And they give you more control over what kinds of artifacts arise. So the general rule is, filter in digital domain when possible, and use the most shallow/gradual filter slope that will do the job.
 
I am not well-versed enough in signal theory to explain exactly why or how a steeper slope causes more distortion. But it seems intuitive when you consider that transient impulses contain a wide range of frequencies going into the supsersonic that all occur suddenly in perfect time sync when they occur in nature. We hear this as a light, crisp, clean natural snap. Now you apply a filter boosting or cutting only some of these frequencies. That spreads them in time differently from the rest - but how differently? A shallow filter treats similar frequencies more or less the same - that's the definition of shallow - so whatever happens, happens to all of them together. The steeper a filter is, the more differently it treats similar frequencies.

 
That's a very interesting theory. There are a few questions that come to my mind however. I remember back in my signal theory lectures, learning about passive and active filters you could see how the ripples are directly derived from the capacitors and transistors used in those filter systems.

When EQ'ing however, you can choose various processing methods. For example, Electri-Q parametric equalizer has different processing modes: analogue, digital, linear phase, FIR, IIR, etc. Some of them mimic actual physical transistor models, some of them are a mix between digital and analogue, some of them are purely digital. In a purely digital mode where you don't follow any physical models and have a complete freedom in coding you can choose or come up with a completely different algorithm that doesn't incorporate any type of ringing. Why do the developers still choose to use such algorithms that have some sort of echoing when it can be completely avoided? Is it because a true physical equalizer will always have that "feature" so it's desired to have even in fully digital modes? Is it because they still apply same logic in digital modes like the physical filter systems do? Or is it something else?

 

[MRC001]

...In a purely digital mode where you don't follow any physical models and have a complete freedom in coding you can choose or come up with a completely different algorithm that doesn't incorporate any type of ringing. Why do the developers still choose to use such algorithms that have some sort of echoing when it can be completely avoided?

One never has complete freedom. While the digital domain frees you from the physical constraints of electrical components, there are purely mathematical constraints to waveform transformation; distortion can't be completely avoided. The best one can do is a purely digital transform whose side effects (distortions) are close to the mathematical limits. Essentially, it's mathematics that says distortion depends on the slope of the filter, among other things.

 

[InCreD1Ble]

  One never has complete freedom. While the digital domain frees you from the physical constraints of electrical components, there are purely mathematical constraints to waveform transformation; distortion can't be completely avoided. The best one can do is a purely digital transform whose side effects (distortions) are close to the mathematical limits. Essentially, it's mathematics that says distortion depends on the slope of the filter, among other things.

 
Actually you are right. Most digital equalizers utilize Fourier Transforms to manipulate the waveforms. It's the limitations of Fourier theories used that introduce distortions in digital equalizers, however I must say the digital processing modes are much more transparent and distortion-free than the analogue ones (the ones that model real physical filter systems).

 

[marcan]

Digital filters can be linear phase but will introduce pre-ringing which can make the sound artificial. In real life no transient can have an echo before it starts.
Analog filters can't be linear phase but won't introduce pre-ringing. A non linear phase digital eq can avoid pre-ringing. So an eq will always add distortions but you won't be able to hear it if you use it sparingly  Large band or shelves will have less distortions.