Once again, thanks for bringing your knowledge to bear. But back to my question that you discussed in posts 632-635.
I'm still living in the age of "brass instruments" and in fact do want to look at impulse behaviour as seen with a real-life mic, not a virtual reality sim.
While I don't have access to any of those perfect impulses (or infinitely powerful amps to deliver them or cleaning tools to scape the woofer cone off the opposite wall), I do wonder what the mic pick-up will look like using a pretty-good impulse.
BTW, in addition to being puzzled about putting a wide-band signal into a band-limited DUT, also seems peculiar to use a DC-to-light signal which is all positive since that also isn't something audio systems do.
B.
If we say so then I'll just have to accept it, but obviously one shows change over time.
Yes I would have expected the silicone coating to upset the higher frequency perf. This was a cheap 15 inch poly cone with a surround that was a vinyl based material I believe. Basically a material that had a fair amount of hysteresis, I assume because the driver was designed to be used up to about 1khz . I however was using it below 300 in one system and as a sub in another. The material had a noticeably slow return to position when flexed. It was also very temperature sensitive.
The silicone coating gave the surround a much more spring like character as well as stiffening and raising res to about 40.
Basically I thought it was a test of two opposite extreems in suspension mechanical properties and their effect on bass perf. BUT I wonder how much if any it was this or how much it may have been simply the added control due to increased resonant point?
Years before this I was playing around modifying cones to build an open baffle woofer. I was trying to build something similar to the drivers used in the Carver dipole speaker. At first the motors were modified to increase Q. BUT what was really surprising was eventually I modified the suspension to a very high Q springy system. I cut out the spider completely and replaced it with two threads crossed at 90 deg through the VC former. These threads were tension ed and tuned on springs and a screw. This is where I noticed again a marked change in the sound of the driver. It suddenly had a "bouncy fast dynamic" quality in the bass. Sorry for the description but I just dont know any other way to describe it. The effect could be heard even at very low volumes hold driver close to ear.
How are you going to view it then? An O scope on the mic?
So put a mic in front of a speaker and whack it with the biggest charged capacitor you have and have a look, somehow.
But that's what a theoretical impulse is, DC to light, all frequency in zero time. If you want to use an "real impulse" that's what you would get. Why would you want to? If you are looking for a valid transducer measurement, using a signal that drives the DUT into nonlinearity makes no sense.
Sweeping a driver well beyond it's normal passband (in both directions) can yield information of value and constructing an impulse view from that data is not a useless VR sim.
What are you really driving at? Honestly you confuse me.
Resonances are much easier to see and to compare — the appearance of the Q of the ringing is the same across the frequency range — if the CSD uses periods on the time axis instead of sec. A well behaved driver will have equal spacing of the curves as the driver rings down.
dave
Actually that was what I was suggesting for SoundEasy to implement.
When drivers of same sensitivity and similar bandwidth are compared, the driver with a longer decay will usually be perceived as louder.
Large domes
Here's a comparison of a few large dome tweeters. The expensive seas one has what seems to be a much better csd plot, much tighter and uniform vs the others. This difference isn't apparent in the fr.
I have to say though the nonlinear distortion of the big seas dome doesn't look that stellar, not bad but not stellar either.
Here's a comparison of a few large dome tweeters. The expensive seas one has what seems to be a much better csd plot, much tighter and uniform vs the others. This difference isn't apparent in the fr.
I have to say though the nonlinear distortion of the big seas dome doesn't look that stellar, not bad but not stellar either.
You can see some of the larger effects in the FR plots, but some can decay faster, slower, and some are really hidden in the lower levels which do not show up in the FR unless you really scale in. These differences can cause different sonic qualities and are much easier identified in the CSD.
Hi Eelco, you are perfectly right, I started out with one sentence and then combined it with another so what it reads here indeed is nonsense. The 'perfectly flat' should have been deleted. But I didn't.
The point is: you can have a flat speaker FR-wise with troublesome energy storage issues and the other way around.
That's why we have all these different representations. CSD plots and FR plots can both be gated to show different things of interest. We have all this information at our disposal, so why not use it?
For those who think that CSD shows anything unique, consider this:
I have only the frequency response, but of course, I also need the phase. I take that response and do an inverse FFT on it to return the impulse response. Next I take this impulse response and calculate the CSD.
From this procedure you can see that the two things are completely related to each other. Some things are more obvious in one than the other but they both show exactly the same information.
I have only the frequency response, but of course, I also need the phase. I take that response and do an inverse FFT on it to return the impulse response. Next I take this impulse response and calculate the CSD.
From this procedure you can see that the two things are completely related to each other. Some things are more obvious in one than the other but they both show exactly the same information.
Earl, quite true for linear systems, and as you point out speakers are pretty linear for the purposes of discussion. However, humans are not nearly so linear, and to them looking at CSD brings different issues to mind as compared to FR graphs. I would vote for more use of CSD to help us more easily think about how well speakers perform and compare.
Last edited:
sidebar question.
as CSD measurements are only valid if done in an anechoic chamber what is going to make manufacturers lead the charge on providing this information?
and given that we don't listen in anechoic chambers how is it possible that this information can even help us determine performance in the "average" living room?
as CSD measurements are only valid if done in an anechoic chamber what is going to make manufacturers lead the charge on providing this information?
and given that we don't listen in anechoic chambers how is it possible that this information can even help us determine performance in the "average" living room?
As long as people recognize that CSD and frequency response are identically equal ways of looking at things, then I have no problem. It's when people believe that CSD shows them something that cannot be seen in the frequency or impulse response that I get concerned. If we are all on the same page that they are the exact same things - its just a preference issue - then everything is copacetic.
First, manufacturers don't even show a complete polar response measurement, which is more important than CSD IMO.
The anechoic response is the same thing as the direct response in the listening room, which is paramount to how we hear image. The image quality of a reproduction is established more by this direct response than the room response. The first few reflections can degrade this direct field image, but never enhance it. So in a room where the first few reflections have been controlled, the image quality will be entirely determined by the anechoic response of the speaker.
Don't loose your time too much, people seems blocked on concepts and numbers.
The real audio fidelity enthusiasts (as i immagine them) will try to build a loudspeaker able to convert an electrical signal to mecanical energy (soudwaves).
In their minds the only valuable parameter will be the ability of the soundwaves to mimic the elecrical signal.
Therfore the rendered measured signal (of music our sounds) should be compared to the electrical signal and then this comparaison will gives a ratio... a ratio of fidelity (if it is what people is looking for).
Never seen this number of my life
Last edited:
How about something like this? ESP SIM (Sound Impairment Monitor) one input connected to the amp and the other to a microphone, and simply measure the difference, why not? 