5th Element, I'm familiar with CSD and know the ridges you speak of.
If there's a ridge at 6 kHz and you adjust the equalizer to -24 dB SPL at 6 kHz then the ridge will disappear since all sonic content will disappear.
Aside from that, I don't see any relationship between SPL and TIME.
If the driver has decay, then the driver has decay.
If a church has decay due to echoic reflections them you call the Pope and say we need velvet curtains A-SAP.
If there's a ridge at 6 kHz and you adjust the equalizer to -24 dB SPL at 6 kHz then the ridge will disappear since all sonic content will disappear.
Aside from that, I don't see any relationship between SPL and TIME.
If the driver has decay, then the driver has decay.
If a church has decay due to echoic reflections them you call the Pope and say we need velvet curtains A-SAP.
That the -12 dB is set at 3000 Hz or in between 4k5
But we don't know where is it originated.
I have suspicions of how it is originated. But it moves too fast and I can't see it
Why are these two disagreeing?
Is there a difference when the notch is applied by a passive network attached to the speaker driver terminals and when the notch (active or passive) is applied to an amplifier and that amplifier effectively isolates the driver from the notch network?
Yes, electrical damping of the cone may be different.
Lower order cone/surround resonances may be succesfully treated in two independant steps, with active systems, i) try cure the source by finding a damping (termination) impedance applied by the amp that actually makes the resonance quicker in decay and lower in Q-factor (also minimizes off-axis effects), then treat the symptoms by ii) finding the electrical inverse of the frequency response -- looking off-axis, too -- at the region of interest and apply it (FIR filters in a DSP are a perfect choice), but be careful with boosted frequencies (don't flatten out notches, that is, you're just pumping more energy into an acoustic null).
Passive speaker level XO/EQ can do that, too, but there's much less degree of freedom.
Basic problem with nulling any narrow, high-Q mechanical resonance is that this is extremly sensitive to precision matching of the inverse filter, even if the mechanical parameters were linear and time-invariant. The slighest mismatch of frequency, Q and boost/cut level will result in a full length decay ridge, albeit at lower levels than the untreated resonance. Some metal or ceramic cone's nasty top end behaviour isn't likely to be fully cleaned up this way, and distortion products still undergo the full mechanical level boost from the resonances.
Last edited:
There isn't any specific disagreement, it's a common misconception I think that the CSD actually shows you something different than the frequency response. They are both different ways of displaying the same information.
If the frequency response is perfectly flat then the CSD will also show linear decay across the frequency spectrum. In other words if you've got a driver with awful ringing and a horrible ridge in the CSD, if you equalise it perfectly flat then the ridge will disappear.
Like I said before though, this wont stop the cone ringing in response to the non-linear distortion products and as KSTR says it is quite difficult to EQ them out completely. It's also worth pointing out that it's generally not worth doing as any driver that exhibits strong distortion amplification, due to cone resonance, should be crossed over before the peaks in the distortion actually occur.
If you've got resonances that don't create large peaks in the distortion and you want to run the driver full range etc, then you most certainly want to try and EQ them out otherwise they will colour the sound.
If the frequency response is perfectly flat then the CSD will also show linear decay across the frequency spectrum. In other words if you've got a driver with awful ringing and a horrible ridge in the CSD, if you equalise it perfectly flat then the ridge will disappear.
Like I said before though, this wont stop the cone ringing in response to the non-linear distortion products and as KSTR says it is quite difficult to EQ them out completely. It's also worth pointing out that it's generally not worth doing as any driver that exhibits strong distortion amplification, due to cone resonance, should be crossed over before the peaks in the distortion actually occur.
If you've got resonances that don't create large peaks in the distortion and you want to run the driver full range etc, then you most certainly want to try and EQ them out otherwise they will colour the sound.
From my experience: Most of the best sounding drivers I've heard, were made with paper cones of one sort or another. 2Nd to that is polypropylene. The draw back to paper is, care must be given in the driver design to avoid spiky response. The drawback to poly is, without proper design they can be sluggish and also have spiky response. I've only found metal drivers to be good at low frequencies. Metal in any other frequency band never has worked for me.
This is one of the more level posts I've read on DIYaudio in some time. It's good to see another critical thinker here who looks beyond the obvious and considers the factors and conditions.
Some metal drivers, such as Yamaha's beryllium domes were predominently second order harmonic, and carbon fiber weave drivers from Morel were third ordered. Some aluminum tweeters begin to break up at 20kHz, others closer to 30kHz. While berylium theoretically breaks up around 45khz, I also recall the berylium domes didn't exhibit any upper mode ringing, which was odd. It can vary greatly by design and application.
Last edited:
That's a good question and the truth is I don't know. One would expect the direct connection to the amplifier to perhaps offer a degree of improvement vs the passive notch filter, but it's probably very small. The passive notch filter itself doesn't do anything, as far as I am aware, to stop the cone ringing, whereas the amplifiers damping could offer some support.
Hi,
Just because three factors are intertwinially self-convoluted, does not mean a zebra is a horse.
Do you have some kind of link for your posit that FR and CSD are two noculars of the exact same thoroughbred?
Carbon Nanotube seems to be quite second order specific as well, at least in one piece of dataset I've seen.
I tend to wonder which metal has......
- The fastest impulse response / settling time
- The most accurate IR / ST
The Beryllium you mentioned is so expensive, that the profit-incentivized transducer designers usually prefer to refer to academic looking theories as to why Mylar or PET is in fact satisfactory, or simply avoid all theory and measurement altogether and focus on the marketing.
If all consumers understood concepts like THD, impulse response, break-up frequency, kJ/m3, Tesla, anechoic, conductivity, radiation, phase, space, tonal colour, ESTA, et cetera...... then they would buy accordingly and recalibrate the entire earth as we currently know it.
Last edited:
Thank you for your reply. Carbon nanotube and Carbocon were of great interest to me. The impulse response, step response and settling are largely influenced by the nature of the suspension and motor. The big advantages of berylium, diamond and other materials was their rigidity, low mass, and propagation velocities. For an equivalent dome and equal rigidity, berylium is much lighter, something to the order of twelve times lighter than a silk dome tweeter. This helps move breakup above the playback range. Depending on the geometry of the piston diaphragm, these properties can be put to great use.
I'm very curious to see what the future holds as materials and compositions enter driver design. I recall ATD was producing carbon nanotube sandwich woofer diaphrams with an outer layer of titanium film. Harmonic distortion was merely average, but its damping properties of cabinet reflections were interesting. They were still expensive at around $600 each for a 6" mid-woofer. I never got to try and measure its various performance attributes, as the distributor ceased to carry them just before I tried to buy one. At that point, all I had were third party tests.
I'm very curious to see what the future holds as materials and compositions enter driver design. I recall ATD was producing carbon nanotube sandwich woofer diaphrams with an outer layer of titanium film. Harmonic distortion was merely average, but its damping properties of cabinet reflections were interesting. They were still expensive at around $600 each for a 6" mid-woofer. I never got to try and measure its various performance attributes, as the distributor ceased to carry them just before I tried to buy one. At that point, all I had were third party tests.
Last edited:
Thank you.
Frequency response is a snapshot of time, it's just like a photo.
Impulse response is not a snapshot, it's a capture of time, like a video.
I will look at the thread you linked later in more detail and try to decipher what on earth they're thinking.
There are many whom would disagree, as the response, CSD, scope, impulse and step response each hold importance in demonstrating how a driver behaves. CSD is a useful means of presenting energy storage as frequency changes, as opposed to flipping through hundreds of separate plots. In contradiction to Zaph's advice, a driver may display a relatively flat response but excessive energy storage, even in narrow bands. I've seen this many times. Most interestingly, the tweeters I found sounded the cleanest turned out to have the quickest setting, none of which could have been disambiguated through frequency response measurements alone.Zaph Audio said:
CSD becomes unuseful only when the viewer is looking at manipulated results taken in someone's basement with excessive data reduction through gating and omitting data at lower levels. Some, and I would venture to say most CSD plots are gated to impede the capture of data after the event, include far too few data points, and omit data below -30dB. Imagine if one was to look at an image of a car, but the outside of that image was cropped. The missing data leaves much open to stipulation. It's as poor a practice as retailers and companies weighting and windowing their measurements.
Last edited:
Except that the frequency response isn't just a snapshot in time vs the CSD that displays vs time. When one analyses the impulse response and processes it into a frequency response you are using all the data within the impulse to do so.
The important part that I took away from that thread from the guys with the technical know how, is the mathematical side of things that govern the relationship between the impulse response, the frequency response and the CSD. Both the CSD and the FR are different ways of displaying the same data. You are not hiding data when you show the FR alone and although the CSD might make it easier to actually see certain things than the FR does this does not change the fact that the data is included within the FR.
It might be true that the necessary gating that Zaph uses makes the CSD pointless at lower frequencies, but that is not the point. I have never seen a CSD plot where one was unable to trace the ridges back to artefacts within the FR. Sure I've seen FR where nothing immediately jumps out at you, yet it does with the CSD, but then upon further inspection you can see why the ridges are there.
Both are just different ways of displaying linear distortion afaik though and linear distortion can always be attended to by applying the correct equalisation.
The important part that I took away from that thread from the guys with the technical know how, is the mathematical side of things that govern the relationship between the impulse response, the frequency response and the CSD. Both the CSD and the FR are different ways of displaying the same data. You are not hiding data when you show the FR alone and although the CSD might make it easier to actually see certain things than the FR does this does not change the fact that the data is included within the FR.
It might be true that the necessary gating that Zaph uses makes the CSD pointless at lower frequencies, but that is not the point. I have never seen a CSD plot where one was unable to trace the ridges back to artefacts within the FR. Sure I've seen FR where nothing immediately jumps out at you, yet it does with the CSD, but then upon further inspection you can see why the ridges are there.
Both are just different ways of displaying linear distortion afaik though and linear distortion can always be attended to by applying the correct equalisation.
How does one obtain a frequency response? By running an FFT on the impulse response. And how does one obtain a CSD? By running an FFT on the impulse response. FR and CSD are the same data, just displayed differently. The different displays may have utility, but they really are the same data.
BTW, if you are not measuring in a true anechoic chamber, you have to gate at the first reflection or you totally contaminate the CSD display. In general, DIY CSD's are worthless in the bass.
Bob
BTW, if you are not measuring in a true anechoic chamber, you have to gate at the first reflection or you totally contaminate the CSD display. In general, DIY CSD's are worthless in the bass.
Bob
IR = FR
IR = CSD
Plus
CSD = FR
However
FR != CSD
I.e., FR in solitary, without IR, can not in any way, shape or form calculate spectral decay.
Yes?
If they were the exact same data we could simply turn FR into CSD with accuracy, with the correct mathematics applied, inverse FFT or SPL tracing or whatever.
Yes?
Answering a post from that thread here. I don't want to post in a 2007 thread.
Here is one M.R.O.: Sony MH1C: Is a bad excuse better than none? [UPDATED]
In that link, this text "The bass response is pre-ringing"
Either way, linear and minimum phase are not the issue here.
The issue here is, to my amazement, if FR and spectral decay are the same data or not.
They are not the same data.
It's volume versus time.
FR is volume only. IR is volume and time.
Time includes the ADSR envelope.
See link http://krashan.ppa.pl/articles/u1synth/
See IR http://rinchoi.blogspot.com/2012/03/etymotic-research-er-4b.html
The ADSR affects sound quality as well, in my view.
Yes?
Answering a post from that thread here. I don't want to post in a 2007 thread.
Here is one M.R.O.: Sony MH1C: Is a bad excuse better than none? [UPDATED]
In that link, this text "The bass response is pre-ringing"
Either way, linear and minimum phase are not the issue here.
The issue here is, to my amazement, if FR and spectral decay are the same data or not.
They are not the same data.
It's volume versus time.
FR is volume only. IR is volume and time.
Time includes the ADSR envelope.
See link http://krashan.ppa.pl/articles/u1synth/
See IR http://rinchoi.blogspot.com/2012/03/etymotic-research-er-4b.html
The ADSR affects sound quality as well, in my view.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.