Because the parts of signal from the room are reflections and pollute the original signal of the drivers. The room response introduces a non coherent response because it is a sum of reflections. It is not possible to make a precise crossover.
But in the low frequency below 200Hz, it is difficult to make good measurements, you should make them outdoors or in an anechoic room. It becomes very problematic when you make 3 ways or open baffle. You have the solution to do nearfield measurement, apply the diffraction rules and after reconstitute the overall signal. John k. proposes other techniques...
And you must integrate "room response" in the crossover when you compensate the floor reinforcement 😱 without integrate it in the data.
In my big OB, i integrated this compensation by ears. S. Linkwitz did it with a shelving high pass in the area 100-200Hz.
😉
Last edited:
Caught me generalizing a bit too much here. Above the modal frequency you want the direct sound to be flat and this must exclude the room. But in the rooms modal range it makes sense to include the room in the measurements. Basically the room and speakers are one thing at LF and a steady state measurement is the correct approach. Above the modal range the room-free direct sound is what needs to be measured and optimized.
Thanks, makes sense.
Is there something else going on above the modal point of a typical room? Do we discern those sounds and reflections differently? What I may be asking is: "Is there a certain frequency range above which we more easily detect what is the reflected and what is not?"
From my experience I would say yes, but have not strictly put it to the test.
Is there something else going on above the modal point of a typical room? Do we discern those sounds and reflections differently? What I may be asking is: "Is there a certain frequency range above which we more easily detect what is the reflected and what is not?"
From my experience I would say yes, but have not strictly put it to the test.
Well that is absolutely the case. The ear can basically only detect in the steady state at LF - the period of a wave is longer that the "integration time" (excuse the loose terminology, but its the way I think about it) and as such fast changes in the sound at LF cannot be detected. We don't "hear" reflections at LF anything like we do at HF. The physiology and nuerology of our hearing ,makes our discrimination a maximim at about 1-3 kHz. It falls on either side of that range, but in different ways and for different reasons.
That's the reason why I've always believed you should avoid placing a crossover in that range--which is very common place to put it. I've always been able to hear this hand-off. Now the strange thing to me is is that I just built a speaker with a crossover right in right in the heart of that range like I've tried so hard to avoid in the past. The cool part is that it's the least audible crossover I've heard. To be honest, I can't hear any ill effects of it. The speaker actually sounds perfectly coherent in a way FR enthusiast gush over, but better.😱 IMO matching the directivity at the hand-off is the key. I also have my woofer delayed 0.3 mSec and that may help--it makes the hand-off flat in amplitude. I bet the good Doctor could shed light on this situation.
Dan
Dan
Thats my opinion too. But its not so easy to do. You basically cannot do it when going from one piston to another.
It's not that easy with a WG either, but far more probable.😀 I guess that's just my experience. These things may be easier for others. I'm still not 100% smooth in the crossover region as it is a work in progress, but it still sounds more "coherent" than any FRer I've heard or built let alone 2-way or multi. I think that's at least as much a property of a fairly narrow and uniformly collapsing directivity (whatever you want to call it) and getting the crossover correct makes that possible.
As to keep this somewhat on topic, here's a graph with a relatively clean impulse response(though not beautiful):
An externally hosted image should be here but it was not working when we last tested it.
Now with a much longer gate. Note that these are not the exact same graphs, but taken in similar but not identical conditions. Notice how the off axis gets very noisy as more proximal boundary reflections enter in the measurements as does the graph at lower frequencies.
An externally hosted image should be here but it was not working when we last tested it.
There, safely back on topic,
Dan
Hi Elbert,
Just a quick note to say this is an excellent thread for us newbies who are only just getting into measuring loudspeakers. So much so, that I think there should be a link to this from the holmimpulse site (if anyone from the company is reading this).
It takes courage to admit you are a beginner and to put yourself up to be shot down by more knowledgable people.
It's good to see the thread keeping to the basics, as the user guide is not very instructive on this basis. Hopefully this will eventually form part of a 'holmimpulse for dummies' thread
I downloaded a copy late last year and have been working through many similar problems to yourself. Yes we've got a lot to learn, but holmimpulse is a great piece of software.
If you're ever in my neck of the woods I'll buy you a beer (or three)😀
Just a quick note to say this is an excellent thread for us newbies who are only just getting into measuring loudspeakers. So much so, that I think there should be a link to this from the holmimpulse site (if anyone from the company is reading this).
It takes courage to admit you are a beginner and to put yourself up to be shot down by more knowledgable people.
It's good to see the thread keeping to the basics, as the user guide is not very instructive on this basis. Hopefully this will eventually form part of a 'holmimpulse for dummies' thread
I downloaded a copy late last year and have been working through many similar problems to yourself. Yes we've got a lot to learn, but holmimpulse is a great piece of software.
If you're ever in my neck of the woods I'll buy you a beer (or three)😀
Fantastic thread Elbert 
In regards to measuring for the actual crossover design, what horizontal off-axis measurement angle will you use?
Would 10 degrees off axis be a reasonable estimate of a real world listening position ?
In regards to measuring for the actual crossover design, what horizontal off-axis measurement angle will you use?
Would 10 degrees off axis be a reasonable estimate of a real world listening position ?
Last edited:
Fantastic thread Elbert
In regards to measuring for the actual crossover design, what horizontal off-axis measurement angle will you use?
Would 10 degrees off axis be a reasonable estimate of a real world listening position ?
"Real world listening position"? What about room reflections?
What if the design is only for a specific room, like in-wall type stuff? If the goal is to build something for a specific response in a specific room then room reflections are not an issue since the end result will be flat accurate response in that room.
Even in active designs the room can be part of the design if its for that room. If the set of speakers are ever moved to another room then of course the measurements have to be redone.
Honestly, I do not see a whole lot of difference in the OPs gated measurements (warehouse vs room) so is this all just a splitting hairs topic?
In the end people are also going to optimize in room also so if its an active deisgn why not do it all up front with the measurements?
I 100% agree with gating but I have not see big differences when gating in a decent size room vs measurements in a parking lot.
Im doing an active crossover project right now myself and its a great thread to follow since Im also using HOLM to do my measurements. Its being done in a treated room with 4" thick (OC703) panels behind the speaker and on the floor infront of the speaker. Like you posted before the ceiling is harder to handle (its only 6 feet from the speaker to the ceiling, the speaker is on a stand 3 feet high).
I talk about the phase if the response is flat, the phase is not good, because you measure an incoherent response. The phase is not very accurate if you include the room.
😉
If the measurements are done correctly you should. Or did I misunderstand the statement? Do you mean both measurements being gated? Or neither?
An approach that I would take would be to measure (pseudo-) anechoically and then design for "no room" (i.e., exclude boundary effects). Next, implement room correction to equalise for room effects. However, this should be an additional stage, not done in the crossover proper. You don't want to have to redesign the crossover whenever you change locations.
You don't want to equalize for room effects above the modal region. And below the modal region you don;t want psuedo-anechoic data.
I envisage a two-step process. First, measurements that exclude room effects to design the crossover; then in-room measurements of the system in situ for room correction. I take your earlier points and the latter summary of the caveats of room correction.
No. If you make a loudspeaker with a flat anechoic response then the direct wave - the first arrival - will be flat. If you change it with EQ to "compensate" for the room, then you havejust messed up the nice flat system that you just designed.
Never EQ a system based on "room data" above the modal region if the speaker is know to have a flat response going in. And if it doesn't have a flat response, then why are you using it?
Earl, I have not disagreed with what you said. I have a loudspeaker design concept that has dial-in EQ to make what is essentially a full-range floorstander more closely compatible with different placement options. Effectively, to counter-equalise boundary reinforcement. And, yes, it works only in the lower frequencies. It will not be as optimal as is possible when DIYing ones own LS, but will have pre-selectable curves that will make it more widely useable, short of custom room EQ. And on room EQ, I said "I take your points", not to litter this fine thread with too much redundant information. In other words: let the reader be informed by other points that have been made above.
- Status
- Not open for further replies.