Hi and thanks a lot indeed !
So it is just for mid-high freq testing. Better than nothing.
A little off-topic instead i do not understand what means in the impulse response
Is not this graph a section of the waterfall at a specific frequency ?
http://cdn.stereophile.com/images/archivesart/Mlpfig5.jpg
Thanks again, gino
So it is just for mid-high freq testing. Better than nothing.
A little off-topic instead i do not understand what means in the impulse response
Is not this graph a section of the waterfall at a specific frequency ?
http://cdn.stereophile.com/images/archivesart/Mlpfig5.jpg
An externally hosted image should be here but it was not working when we last tested it.
Thanks again, gino
No. I really recommend reading. This information is available on the web, and would give you a clear explanation.
Last edited:
If you don't have an anechoic chamber, the sound you register on a mic contains the direct sound and also the reflected sound (from walls, floor, ceiling and objects). You need to throw away all the reflections, and since the sound travels at a finite speed, the reflected sound will arrive at the mic location after the direct sound, so you apply a gate so the measure (in your example, 5ms). Doing so you lose the info of the lower frequencies, for 5ms anything under 200Hz is lost.
Ralf
Ralf
This is actually the same information as the waterfall plot. A mathematical operation, the Fast Fourier Transform transforms the view from one to the other.
In response to some (much) earlier posts; of course an inductor in series with a driver does change the response, but it will not cause ringing. The crossover ideally will pass a square wave, but not to each speaker. The SUM of the crossover sections ideally will sum back to whatever the input was; but in practice, since the xo is used for phase and frequency correction as well as simple signal splitting, it ain't gonna happen, nor should it.
Mic placement. Yes, reflections from the walls, floor, and mic stand will introduce errors into the frequency response. If the mic is closer to the Unit Under Test, the speaker output will be louder compared to the reflections than if the mic is further away. I routinely used 18" (1/2 meter) for the mic spacing from the UUT. If there is more than one driver, it is important that the mic height is exactly the height of the middle of the driver array, so that the distance the sound waves travel from each driver to the mic. is the same as much as possible. Put the UUT in the center of as large a clear space as possible. (Not near walls if possible.) Some sound absorption on the walls (old shaggy wool rugs for example) will help a surprising amount.
The impulse response has its uses. The impulse spike will be delayed by the driver (less so) and the distance to the mic. It's worthwhile to have the delay from the two drivers the same, for a more nearly phase coherent crossover design. The sound actually originates at about the magnet gap, and then propagates through the come to the air; almost always to time-align the drivers, the tweeter will need to be set back to align its magnet gap with the mid-ranges'. This will lead to a step in the cabinet, which can lead to some unfortunate diffraction issues, so some absorption and a smooth curve are helpful.
You don't really need the math to understand the FFT. You DO need to understand what it's telling you, and what to do about it.
Again, I've not done this in a long time, so I don't know if there are easier sources than Dickason, but he is really excellent.
My ideal speaker does only direct emission. Focused in the sweet spot. Then i like horns a lot.
Woofers are problematic for this. But i would try to eliminate any boundary reflections.
We hear an obstacle when it reflects sounds.
In my ideal room bats would crash against walls ... hehehe.
Tests that can be related to the dynamic behaviour for me are of the utmost importance really.
Regarding alignment this is my ideal concept for a two ways driver
I love the impulse response like also the waterfall that i understand better.
I would love to see waterfalls in the drivers datasheets. Just to be sure that the problem is not the driver.
If math is needed i am out completely. It is my biggest regrets. Math is so perfect that is beautiful.
Thanks again for the advice. I am mostly interested in lab testing.
I know that measurements are not fashionable ... but they must tell us something in the end.
Presently i am trying to understand woofers behaviour ... and which are the best test to get an idea about their quality.
Kind regards, gino
Curmudgeon View Post said:
Hi!
That's not what he(?) said. He said that a one component filter will not ring. The IR will always be affected by any signal processing (time or spectrum).
False claim. No difference between active or passive in that regard. Only a 1st order text book filter will pass a square without distortion in the time domain, no matter if active or passive.
Digital solutions is nice becasue they are cheap and very flexible and powerful. Normally a DSP for this can do EQ and x-over + delays if necessary.
In the end very good systems can be built with any of - and mixes of - passive/active/digital processing.
Hi ! thank you a lot for the further explanation.
Maybe i am wrong but among L, C and R i think that L can have the worse effect on IR .. am i right ?
If this is true less L in series with drivers the better for IR ?
Is this important ? i know that some designers use only 1st order x-over.
Maybe they have a very strong point here ?
Loudspeaker designer John Dunlavy: By the Numbers... Page 2 | Stereophile.com
I can take this as a rule ? for me a speaker with a bad IR cannot be ok at all.
I see. There is no a best approach in the end.
I will go on reading. Hoping to catch the most important rules.
Thanks again.
Regards, gino
Last edited:
I have long suspected that one of the reasons for the success of first order networks is that the gradual roll-off requires the use of very well-behaved drivers to be workable.
I've also had success with the multi-slope drivers, popularized by Thiele; first order at the xo frequency, then a second pole about an octave out, etc. It's less demanding of the driver quality, but still retains much of the virtues of the first order xo.
I've also had success with the multi-slope drivers, popularized by Thiele; first order at the xo frequency, then a second pole about an octave out, etc. It's less demanding of the driver quality, but still retains much of the virtues of the first order xo.
Hi and thanks a lot for the helpful advice.
Actually i am noticing that some very high quality drivers like Scanspeak for instance show some irregularities in the freq respone and usually they are made more linear with some combinations of L, R and C in parallel with the driver.
After your explanation i would be quite sold on 1st order x-over.
The dynamic performance are very important for me (i love the SWT for instance).
And my first experiment could be a x-over modification on a pair of JBL L26 Decade where the woofer has indeed a not completely flat midrange.
That would be an interesting test bench.
I like already the sound of the original a lot.
This sounds complicated. I must keep everything very basic.
Thanks and regards, gino
- Status
- Not open for further replies.