I didn't say that a step response test was invalid, and I don't think that I even implied that it might be invalid. I'm just observing that it isn't so readily available "in the wild", so to speak. It needs more careful interpretation than other available test signals and/or measurements.
And yes, critically-damped loudspeakers do have applications. Just that "tighter" bass is not one of them.
And yes, critically-damped loudspeakers do have applications. Just that "tighter" bass is not one of them.
If you listen to string quartets then anything that resembles a step or low-frequency square is unlikely to appear, but in many modern music such signals are present.
The whole rap/hip-hop genre lives off of hard-clipped kick drums, and squares are common in electronic music.
And then we have the famous cannon shots in Tchaikovsky' 1812 Overture.
Therefore we better make no assumptions what counts as "music" vs "test signals" (same goes for chirps, noises etc) ;-)
@KSTR, can you please provide an example waveform of a step-like signal in electronic music that has been extracted from the music signal? I am very curious to see one.
Also, I'd like to see an example of those hard-clipped kick drums that you refer to. If you can show some waveforms, that would be much appreciated.
I also understand the use of square-wave type signals in some genres of music. However, these are not really step signals, such as the ones that are being discussed here, are they?
Also, I'd like to see an example of those hard-clipped kick drums that you refer to. If you can show some waveforms, that would be much appreciated.
I also understand the use of square-wave type signals in some genres of music. However, these are not really step signals, such as the ones that are being discussed here, are they?
Last edited:
A sine wave is also a signal that is quite unusual and is highly unlikely to occur on music program material. But it's clearly useful; yet frequency response is not even close to the whole story.
Impulse response is response to an infinitely short "upside down T" pulse and it shows you very simply and plainly how much energy the speaker stores. Speakers should not store energy.
Most speakers butcher impulses.
Step response is an impulse response that has an infinite 6dB/octave low pass filter added. (The mathematical integral of an impulse function).
Step is popular in audiophilia because it's fairly easy for the eye to judge what the speaker is doing from a step graph and it's why we're discussing it here.
I think good phase and impulse response are among the things that separate the men from the boys in speaker design.
If you listen to a song with a wideband bass drum sound ie "Fear of a Blank Planet" by Porcupine Tree, on many subs you can literally hear the "thump" arrive on a later date than the "slap." If you have a sealed sub with good step response and you insert a DSP and add a 20Hz high pass filter, and switch the slope from 12 to 18 to 24 to 36 to 48dB per octave, you can hear the thump arrive later and later. It is plainly audible. Each 12dB/octave adds 1/40th of a second of delay (If I'm doing the math right in my head).
This is demonstrably audible at low frequencies and I have a strong suspicion that it's audible at HF too, just not "as such." I think good impulse response results in better imaging, more clarity, better resolution, better microdynamics.
Given that you can easily and inarguably measure step and impulse response, it's never made sense to me that we shouldn't try to optimize for it. In audio people argue about all kinds of things you can't measure. So why not optimize what you can measure?
I took great pains in the Bitches Brew and Live Edge Dipole [#1 at Parts Express speaker design competition 2023] designs to achieve linear phase and clean impulse response and in my personal opinion it paid off in spades.
Still, I think that using a real pulse as stimuli would be interesting (but hard wrt. SNR aspects). But think about it...:
# a sweep is a kind, gentle signal - a pulse is a jolt/explosion.
# Sweep to IR FFT transformation requires a linerar system or the math don't hold water. Speaker is anything but linear.
So what I'm getting at is that we perhaps make it to easy for the speaker when we do FR sweeps and bask in the glory of the nice looking calculated impulses respons while perhaps in reality, it looks like sheit.
Is my worry anchored in reality?
//
# a sweep is a kind, gentle signal - a pulse is a jolt/explosion.
# Sweep to IR FFT transformation requires a linerar system or the math don't hold water. Speaker is anything but linear.
So what I'm getting at is that we perhaps make it to easy for the speaker when we do FR sweeps and bask in the glory of the nice looking calculated impulses respons while perhaps in reality, it looks like sheit.
Is my worry anchored in reality?
//
You can transform the sweep measurement into a pulse with simple math. But your point is valid - which is that a sweep does not stress the system in the same way dynamically as a strong pulse does.
Which highlights the tradeoff between power and bandwidth.
Which highlights the tradeoff between power and bandwidth.
I'm not so sure about that. Below is an example of a low-frequency sinewave that has other music signal elements riding along with it.
In a (relatively) linear system, the frequency response, which includes the phase response, generally tells a great deal about the system's behaviour.
It's not that easy to make judgements from a step response as might be implied by the simple nature of the plot. Some performance parameters can show up relatively clearly, but low-frequency transient response is unlikely to be one of them, especially as far as judging bass "tautness" goes.
Last edited:
Random sample from https://soundcamp.org/tag/hip-hop-kick-drums
The original waveform certainly was two or three times the max amplitude.
There is a whole industry of outboard hardware as well as DAW plugins for the sole purpose of clipping drums to make them stand out more in the mix.
Square bass is everywhere, take any house/garage techno track....
Usually, linear enough unless you severely overdrive it.
Also, exactly how linear a speaker is, with real-world signal, can be examined by comparing the original response to an (arbitrary) input signal with a linear copy obtained by convolving the measured IR with the test signal. Can be quite revealing at times to listen to the residual, sometimes shockingly so once you slightly overdrive a speaker, notably ported ones. But usually the residual signal is small enough that there is zero visible difference between the real and the emulated response waveforms.
Everything we want to know about the linear response of system to steps, impulses, etc... is in the transfer function. This is the frequency response (including phase) in the frequency domain and the impulse response in the time domain. They are the same information just presented in a different way. A long time ago it was fairly common to measure an acoustic response with an impulse but the signal to noise was not good because loud impulses become non-linear. A "gentle" sweep is usually a better way to get an acoustic response which is why it has become common these days although it is not without it's cons. There are also other ways as well.
Good loudspeakers become audibly nonlinear at higher SPLs (bad speakers are always audibly nonlinear). How loud a speaker can play cleanly is an important part of a loudspeaker's specification that tends to receive little attention. K&H/Neumann include it as part of the spec of their speakers but few others do. With this information and a knowledge of the average listening level one knows how cleanly transients will be handled. Good quality music recordings tend to work with something like a 20 dB peak-to-average overhead which sets a practical maximum for a loudspeaker. Some strongly percussive sounds like internal combustion engines can exceed this by a fair margin but for commercially available music it is reasonable and indeed dropping it a bit to say 15 dB usually doesn't do too much harm.
I would like some explanation why an amplifier with very low damping factor, say 5 and one with damping factor say 8000 produces a very different step response in the speaker. Consider this, the output of the amplifier at zero output signal is not an open circuit it is a short circuit. My point is a system impulse response is the important factor, not only a component of it.
@perrymarshall, would you mind extracting a portion of the the waveform that relates to the "thump" and the "slap" that you are referring to, and provide a plot? I'm having trouble identifying what you are describing.
The bass spectrum on that track is certainly "wideband", that's for sure, as shown in the plot below.
Note the distinctly large peak at about 5Hz as well. Where did it come from?
Last edited:
I would say you are basically right.
But we need to take the following into account: When it comes to sloid-state amps we have often damping factors of much greater than 50. While a difference in damping factor between 5 and 50 might make a big difference in actual damping the differnece in DF between 50 and 5000 isn't that great anymore when it comes to the behaviour of the whole system.
It is even possible to build amplifiers with a negative outupt resistance in order to reduce Qts BTW !
Regrads
Charles
The reason for that behaviour is simply that the frequency response of the DF=5 amplifier is modulated by the impedance of the loudspeaker to which it is connected. There is a bit of EQ happening that is associated with the motional impedance resonance peaks in the low frequencies. This usually results in boosted bass response at those frequencies. Other peaks and dips also occur elsewhere due to the impedance swings of the loudspeaker.
See the example below.
Last edited:
Just listened to "Fear of a blank planet" IMO it contains nothing that I can even imagine have anything to do with Q being any number between 0 and 1. It is just plainly unimpressive and disappointing. I think what you listen too and enjoy or dislike can sway you in any direction. To me it is the same as having your car dented, you will do anything to remove the damage before anyone notices that you are driving an imperfect car.
Last edited:
In the days long gone, audiophiles inserted series resistance in line with their speakers to make it sound "better" Also placed dynamic resistance i.e. light bulbs in series with the speaker to make it soft clip or whatever. I just listened to Kodo Big Drum and it sounds pretty impressive, I don't even know what the Q of the speakers are and even if it is still even applicable after some 50 years (Rodgers LS3/5a) and I have never owned or felt the need to own a sub woofer. When I hear the body panels shaking in a passing car, I feel I want to puke.
Resistors inserted can change the FR and the way bass gets noticed. Some find it for the better.
I don't have a problem with it, as long as it's safe.
I don't have a problem with it, as long as it's safe.
You do realise that the prerequisite for this math to be valid is that it is bad on a perfectly linear system?
I advice you to check the theory!
To be really clear: I have no doubt of the mathematical relation you refer to - it's perfect. It's the potential pitfall when applying to the real world that I'm raising a warning flag for.... but please do prove me wrong in this "real world meeting math theory" conundrum...
edit; I se now that KSTR commented about this - he acknowledge my observation but says that it is linear enough to relay on if not overdriven... fine.
//
I tried to but was unable to confirm these observations. I took the original track in Audacity and then high-pass filtered it with a 24dB/octave filter with a −3dB cut-off frequency of 25Hz. I couldn't tell any difference between the filtered and unfiltered tracks using an instantaneous A/B comparison. The only thing that was plainly audible to me was that I couldn't hear any difference.
I then went and applied a 48dB/octave filter with a −3dB cut-off frequency of 20Hz to the original track. I still couldn't hear any difference.
I then went and changed my audiophile-grade headphones to another pair of audiophile-grade headphones. Still no difference, although there was a difference between headphones. Phew! I was worried that I needed to get my hearing checked. 🙂
Last edited: