why can't we test our loudspeaker and crossover in rooms which will be used there
Since, we don't have the facilities of anehoic chamber to test diy speakers. I doubt the accuracy of Quasi-anechoic theory of measurement. But why we can't test in the rooms in which it is going to be placed. We know that frequency response of speakers in anechoic chamber will vary in real room environment. And with this we can really see so how this speakers is going to behave in that room.
Since, we don't have the facilities of anehoic chamber to test diy speakers. I doubt the accuracy of Quasi-anechoic theory of measurement. But why we can't test in the rooms in which it is going to be placed. We know that frequency response of speakers in anechoic chamber will vary in real room environment. And with this we can really see so how this speakers is going to behave in that room.
Last edited by a moderator:
Well you can but you have to decide what you want to measure. If you want to measure the speaker itself, then obviously you want to avoid anything that sends sounds to the test mike that does not come directly from the speaker, like room reflections. There are techniques to do that, except for say below a few 100 Hz.
If you want to measure how the speaker and the room together make sound, you can too.
So depends on what you want to measure.
Jan
what is the disadvantages for measuring speaker with direct sound and room reflection sound? After the measurement, i can use the equalizer to flatten the frequency response of sound coming from both speaker and reflection sound. After equalizing, i will have linear frequency response within the room itself which is good. Long delay Reverberation in the room issue, we can sort out reverberation by using acoustic panels.
If the sound at the listening position doesn't resemble the output of the speaker then something is less than ideal. Precisely what that is is the question. It may or may not be fixable at the crossover, and the optimum crossover might not be what the in room measurement suggests. The more you design the speaker for the room, or possibly treat the room, the less this should be a problem but knowing why means understanding what the differences are.
Suppose we take measurement of speaker within a room of which ideal position i will be always listening. Now with that frequency response graph, I can use active crossover for 2-way or 3-way design and use DSP to get maximum linear frequency response from 20-20000 hz. Then we need others procedure like Quasi-Anechoic or Ground Plane etc which will vary greatly with actual room listening position.
The very idea of the quasi anechoic measurements exists for a very long time and has been put to test only goodness knows how many times with success, so there is no objective foundation to doubt its usefulness. You have to get acquainted with it to appreciate it fully. Would be a good idea to get a book or two on the subject, like Dr.D'Appolito's "Testing loudspeakers" or Toole's "Sound reproduction - loudspeakers and rooms".
Best wishes.
@Jan: Congrats on a huge success with silent switcher! You deserve it.
If the response measures differently at the listening position, say it's too loud at one frequency when you measure a steady tone because there is a near wall reflection, you'll adjust and the direct sound will be more quiet, the reflected sound is louder than the reverberation which for some other frequencies will be the first significant reflection, it will also be much sooner, it will be at some near angle to the direct, it will be odd in its response in that there may be some comb filtering and other issues, plus the reverberant level at those adjusted frequencies will be altered eg. maybe they will be taken down too low.
This may or may not be fixed with a crossover. Sometimes it can only make it less noticeable but turning it down may not be a positive thing either. If this acoustic problem causes problems with reproduction, you will want an acoustic fix. Otherwise you need to plan for this when designing the speakers.
You can use gated measurements. They are most important above a few hundred Hz. They are not necessary below 100-200Hz, here you want the room response.
If you understand how the XO components effect the subjective audible response as you change them, then setting up the system initially with measurements anechoic or outdoors and THEN voicing them in your room would be the ideal way to proceed. Voicing is the purely subjective and creative aspect of speaker design and it's time consuming as you've gotta spend time listening critically which can be fatiguing and has lots of variables that include bio stats like heart rate, blood pressure, glucose levels and psychological factors like suggestive bias and your overall mood at the time of listening.
We can. Software like REW, Dirac, Audiolense, Denis' DRC, Acourate, HolmImpluse and others incorporate Frequency Dependent Windowing (FDW) in their analysis algorithm.
FDW means the analysis window is longer at low frequencies and becomes shorter as frequency increases to virtually analyzing the direct sound at high frequencies when measuring speakers in a room at the listening position. The idea is that FDW mose closely emulates what our ears hear in small room acoustics and is based on the research of JJ Johnton's Acoustic and Psychoacoustic Issues In Room Correction: AES PNW Meeting Report - Acoustic and Psychoacoustic Issues in Room Correction
At low frequencies we are hearing both the room and loudspeaker, whereas at higher frequencies we hear more of the direct sound of the loudspeaker and less of the room, as JJ points out. So using software like Acourate one can create digital XO's, linearize frequency response of the drivers, time align the drivers, provide FDW frequency correction at the listening position with an overall timing correction (i.e. excess phase correction).
Toole has authored an open access AES paper called, The Measurement and Calibration of Sound Reproducing Systems: AES E-Library The Measurement and Calibration of Sound Reproducing Systems See Figure 14 below:
This would be a good starting point for comparing ones measured in-room frequency response (using FDW) to a preferred target frequency response that is neutral to ones ears. I prefer the trained listeners target response in Tooles diagram above. This is the frequency response of my 2-way JBL 4722's at the listening position using Acourate and its capabilities listed above. Note the vertical scale:
There is also a target timing response (i.e. time alignment of drivers in a multi-way system), but is beyond what can be written in a forum.
If interested, I have captured this information, and a lot more like polar response, group delay, etc., plus step by step "how to" guides in an eBook in my sig below. By clicking on Look Inside one can check out the Table of Contents and read the first couple of chapters.
Hope that helps
Kind regards, Mitch
FDW means the analysis window is longer at low frequencies and becomes shorter as frequency increases to virtually analyzing the direct sound at high frequencies when measuring speakers in a room at the listening position. The idea is that FDW mose closely emulates what our ears hear in small room acoustics and is based on the research of JJ Johnton's Acoustic and Psychoacoustic Issues In Room Correction: AES PNW Meeting Report - Acoustic and Psychoacoustic Issues in Room Correction
At low frequencies we are hearing both the room and loudspeaker, whereas at higher frequencies we hear more of the direct sound of the loudspeaker and less of the room, as JJ points out. So using software like Acourate one can create digital XO's, linearize frequency response of the drivers, time align the drivers, provide FDW frequency correction at the listening position with an overall timing correction (i.e. excess phase correction).
Toole has authored an open access AES paper called, The Measurement and Calibration of Sound Reproducing Systems: AES E-Library The Measurement and Calibration of Sound Reproducing Systems See Figure 14 below:
This would be a good starting point for comparing ones measured in-room frequency response (using FDW) to a preferred target frequency response that is neutral to ones ears. I prefer the trained listeners target response in Tooles diagram above. This is the frequency response of my 2-way JBL 4722's at the listening position using Acourate and its capabilities listed above. Note the vertical scale:
There is also a target timing response (i.e. time alignment of drivers in a multi-way system), but is beyond what can be written in a forum.
If interested, I have captured this information, and a lot more like polar response, group delay, etc., plus step by step "how to" guides in an eBook in my sig below. By clicking on Look Inside one can check out the Table of Contents and read the first couple of chapters.
Hope that helps
Kind regards, Mitch
This IS how I recommend you work. Certainly worked for Allison, to a point. That is, design your speaker and crossover for a specific application, within reason. The LM-1 I designed are specific to a book case or desktop. Would sound crappy on a stand, but I gain significant efficiency.
Now, let's get into some of the hazards....
If you are doing a 2-way, gated far-field measurements is all you need. I use OmniMic which automatically blends my gated high frequency with un-gated low so it's relatively simple for me. The challenge for you is how well you can see through the ripples in the rest of the response, especially the baffle step. If you get used to looking at baffle step effects then you'll be better able to compensate for them in the crossover.
However this approach is not without negative side-effects. Below the gate threshold we can see the actual room gain plus baffle step (a very good thing which I think you are concerned about) but this is also co-mingled with comb filtering caused by room interactions, so it's not a purely positive choice. While yes, the far-field is what you hear, moving a few inches towards or away from the speaker will significantly change these readings. If you are going to create a 3-way with a crossover point at 200-400 Hz for instance this now gets pretty messy. Your crossover choices could be adversely impacted. It may sound good only exactly 1m away from the tweeter, or you may find yourself attempting to compensate for a local null that 5" further away does not exist.
I use a hybrid approach. I use the far-field as my macro-scope. It sets my guides for the levels and splice points of the near-field measurements for the woofer and port. Where to splice them, and how much to trust them.
Best,
Erik
Last edited:
Isn't the latest generation of DSP and particularly REW / DDRC, etc designed to do all of this?
Of course no matter what methodology is used, if a speaker system is "absolutely perfected" for a particular room and listening position, then even something as seemingly benign as adding / moving a piece of furniture, or leaving some doors open could affect that. Just how anal do we want to get with this?
As I've mentioned in a couple of other recent threads, my most recent big build included several approaches to crossover for a 2-way system. The most satisfying one was a parallel speaker level network, based on raw and gated quasi-anechoic ground plane measurements on axis, 22,5 and 45 dg.
Target the filter functions for desired results in the modelling and confirmation measurements in those same conditions - then deal with what the room throws at ya - by combinations of acoustic treatment, EQ / Room EQ, or acceptance.
Kinda like dealing with a major loss, I find the sooner you get to that last step, the better for all.
Of course no matter what methodology is used, if a speaker system is "absolutely perfected" for a particular room and listening position, then even something as seemingly benign as adding / moving a piece of furniture, or leaving some doors open could affect that. Just how anal do we want to get with this?
As I've mentioned in a couple of other recent threads, my most recent big build included several approaches to crossover for a 2-way system. The most satisfying one was a parallel speaker level network, based on raw and gated quasi-anechoic ground plane measurements on axis, 22,5 and 45 dg.
Target the filter functions for desired results in the modelling and confirmation measurements in those same conditions - then deal with what the room throws at ya - by combinations of acoustic treatment, EQ / Room EQ, or acceptance.
Kinda like dealing with a major loss, I find the sooner you get to that last step, the better for all.
Chrisb,
I don't think you can get to absolutely perfect.... but I think reasonable considerations should be considered while measuring. I think the quasi-anechoic method a little TOO optimistic, but overused for convenience. For instance, 2' from a wall. Up against the wall, in a bookcase or 5' from any walls. These are all reasonable, but imperfect, places to start.
In fact, I would only consider a pure quasi-anechoic method if I knew the speakers would be used with active equalization. That has significant benefits. For instance, baffle step compensation done actively can reduce wasted power, and allow for more placement freedom.
Best,
Erik
I don't think you can get to absolutely perfect.... but I think reasonable considerations should be considered while measuring. I think the quasi-anechoic method a little TOO optimistic, but overused for convenience. For instance, 2' from a wall. Up against the wall, in a bookcase or 5' from any walls. These are all reasonable, but imperfect, places to start.
In fact, I would only consider a pure quasi-anechoic method if I knew the speakers would be used with active equalization. That has significant benefits. For instance, baffle step compensation done actively can reduce wasted power, and allow for more placement freedom.
Best,
Erik
The only thing I did wrong was eq my system flat at the listening position
I think a lot is to do with the directivity of the higher frequencies. They have less room reflections than the lower range so should naturally measure lower (at the listening position you're getting the midbass + midbass reflections, whereas you're only really getting the direct sound from the tweeter. If you boost the treble to be as loud as the midbass with its reflections you will end up with a thin bass / harsh treble system)
edit: Just noticed that Mitchba's post above said this better than I did here..
Rob.
Last edited:
No disadvantage as such but you don't know how the speaker itself performs. In development you sometimes want to know that.
Jan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.