ttan98 said:
I use a third order electronic on the bass at 400 HZ and a first order line level on the compression driver mid horn at 350 that is summing acoustically around 500 HZ. Trial and error- lots of them
Magnetar said:
Doesn't this mean that something is seriously wrong? Either with the technique or the system. I would never accept a measurement where this was the case. And if it is, isn't moving the microphone around to find "the best response" kind of the wrong thing to do?
Early knows why microphone placement is ultra-critical with Magnetar's system, but he's not tellin'...
If you have to put you ear there every time, then that would be a problem.Patrick Bateman said:
😀
Patrick Bateman said:
Well I can't say for sure, but I do know that if you are doing the measurements right then the microphone position is pretty unimportant - several inches in any direction should not make much difference.
Now if these are steady state measurements in a reverberant room, then of course things will vary dramatically with position, but none of that is "real" and you should be doing spatial averaging. I would recommend my 20 year old paper on making in-room measurements and the problems that ensue. (AES paper on "Localized Sound Power")
Personally, I am not a big fan of any steady state measurement, especially in rooms. Its simply not that relavent.
Can I guess?
two midranges side-by-side.
five woofers in an "X" facing forward.
it's a seirous interference pattern generator.
details on one of his setups:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=110435
two midranges side-by-side.
five woofers in an "X" facing forward.
it's a seirous interference pattern generator.
details on one of his setups:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=110435
y8s said:
DING DING DING
Trying to get eight drivers to sum at twenty feet is a challenge... at any distance closer, it's impossible. That's why the measured response changes when you move the microphone one inch.
gedlee said:
Nope, it's a fact. You can change measurements with all speakers by moving a couple of inches. As you know it depends what, how and where you are measuring the speaker.
Patrick Bateman said:
He'd be lying if he didn't agree,
soongsc said:
LOL -
gedlee said:
It changes with LMS and other measurement methods too. You of course know that.
y8s said:
That sounded alright on axis, it has changed several times - first being the mids - The bass system however was better then 99.9999999999999 percent of the systems out there - LOL
Patrick Bateman said:
They sum fine at 10 feet, plus they don't have that soggy, low definition, low quality sound of a monopole like the most people use
Do you use omni directional bass too?.
gedlee said:
No, not always.
You must have an anechoic chamber and live in a perfect measurement world. Must be nice. The 'point' of me saying if you move the microphone the measurement changes stands in my world. An example would be for you to take a shot of the gated frequency response from a windowed impulse response of your horn at one foot away and move the microphone one inch at a time ten times. Then repeat at one meter. Then overlay the unsmoothed response plot on top of each other for the two measurements. You will then have not one but two charts proving whether or not I am wrong (in my world)
Magnetar, have you tried using a more symmetric baffle layout, tweeter in the center, midranges around it, then woofers around that, all in circles? Maybe, together with the right crossover points, this will minimize driver interaction, as it minimizes the distance between their centers. Kind of a synergy open baffle, like in Toms Synergy horn patent 🙂
MaVo said:
For the bass it doesn't matter but yes I have a board with several 5.25" drivers in a circle around a 10" conical waveguide. Didn't like it.
I don't have any major problems with driver interaction. The midrange and treble are single units now, The mid horn crosses in at 500 cycles and treble at 3000 cycles. The bass and low midrange is the array of tens and it is the only part that is dipole.
Magnetar said:
Had you been more polite I might have done just such a thing, but I'm not about to bend over backwards for someone who insults me.
At 1 foot away you are in the near field and there will be substantial differences in the measurements with small distances. But these differences are not propagated to the far field so why would we want to measure them? You can't get very useful data in the near field.
At about 10 feet, which to me is just about a minimum distance for a large system like the Summa, then 6 inches forward or backward will yield only about .1 dB change because of 1/r^2. Since the system is constant directivity there will be almost no change with angle in any direction, particularly near the central axis. Thus at about 10 feet, on-axis, any measurement within about the size of a basketball should not have a variation of more than about .1 dB. There will be larger random variations than this if, like me, you use a noise signal for the source, because of limited time averaging. But my point is that the response is not "ultra sensitive" to microphone location.
And no I don't have an anechoic chamber, I use my living room. But I do setup the measurement to get good data, or why bother? And if the measurement is "ultra sensitive" to mic placement then I still say that something is wrong and should be corrected.
gedlee said:
Thanks for your reply, The problem I have is trying to get the real data at 10 feet to correspond between measurements consistently. That is even with the microphone in place let alone moving it an inch.Can you explain what and how you use to measure the system. I use ARTA with a Behringer EC8000 and a M-Audio Fast Track Pro USB soundcard and mike amp. Now if I smooth the response the differences go away. Isn't that hiding the truth?
Arta Software
Magnetar said:
I use the same mic and I've used the same sound card so those aren't a problem.
I write all my own measurement software because I don't like anything out there, so I can't comment on specific application packages, but here is what you should look for. Hopefully you can get impulse responses from you measurement system. If you can't then it would be almost impossible to track down a problem. If the impulses are different from one measurement to the next then the system is actually changing. But a windowed impulse should be very stable against microphone location - its the noise and reflections that vary so much and even small mic location changes will alter the times and levels of these errors.
If the impulses look similar, (lets say identical, because if you can see a difference in the impulse response then this is a large difference in the frequency response) then the only hope is to do some averaging to remove the random errors that are not really part of what you are trying to measure.
I average several impulse responses, which are themselves calculated from many sets or records of data which are averaged - remember, I use noise, so averaging is a must not an option. I use almost ten seconds of data to get one impulse, which may have a duration of what a few ms. Thats a lot of averaging - remember that averaging in the time domain is not smoothing - they are completely different things. Averaging does not "smooth" a frequency domain aberation - in fact it accentuates it.
But I do also use some frequency domain smoothing as this, to me, only makes sense. Now we can argue extensively about what the bandwidth of the smoothing should be - I use either 1/6 or 1/3 - but I think that all would agree that some form of smoothing is essential.
If you do all this, then I cannot see how your measurement is going to be sensitive to microphone location.
I have taught many classes on acoustic measurements (in college and at profession seminars), and I am the chairman of ALMA's Education Committee. These are some of the most difficult measurements that one can make and courses on them is ALMA's #1 priority. It is my experince with measurements that leads me to say that people who believe that measuremenst don't correlate to what we hear are probably making bad measurements. I don't have this problem.
Hi Mike
"The problem I have is trying to get the real data at 10 feet to correspond between measurements consistently. That is even with the microphone in place let alone moving it an inch."
I use the same mic, Arta, and True RTA and I have the same issues, but I've found the problem with my system. I use an EMU-0404 USB sound card and there's a warm-up period with this card that can take as much as 15 minutes before it settles down. I don't trust anything I'm measuring now until I can get the same FR plots between Arta and True RTA and with the OBs I've been working with lately, I've found another trick. When the card is cold, it won't measure the dipole roll-off correctly. There's a point where suddenly, the expected roll-off simply appears in both measurement systems and from that point on, both of them become stable and consistent.
I've got a routine down now where I remove the EQ from the bass (4-5db boost averaged between 200hz down to 40hz) and run pink noise until the weird stuff below 200hz starts to settle down and show what is expected at these frequencies with a dipole. I've read that the EMU card does need a warm-up period, and I tend to believe that now. Also, I don't know how your card works, but mine has all its level controls on the unit itself and it's important for mine to have the rght amount of gain on the mic - too low, and measurements are inconsistent. Just a FWIW.
I haven't put too much stock in CSDs/BDs at this point because of obvious latency problems with an external card. I think the best I can get with mine is 7ms so I'm not too confident about what I'm measuring between output and input - I need to do some more research into that area.
Mavo,
"Magnetar, have you tried using a more symmetric baffle layout, tweeter in the center, midranges around it, then woofers around that, all in circles? Maybe, together with the right crossover points, this will minimize driver interaction, as it minimizes the distance between their centers. Kind of a synergy open baffle, like in Toms Synergy horn patent."
Mike's right - clusters suck. 😀
I've been playing with driver configurations on an OB for a coupla months now and I've had them in every possible setup that 4 direct radiators, 3 different horns, and 3 different compression drivers will alow and the tight cluster idea has, ummm, issues. 🙂 With this setup, it's very focused, but move your head 1 inch right or left and you fall into a horizontal null that no amount of crossover tweaking, or driver split roll-off will cure. Kinda nice when you're in the zone, but the zone is too narrow to be usable.
"The problem I have is trying to get the real data at 10 feet to correspond between measurements consistently. That is even with the microphone in place let alone moving it an inch."
I use the same mic, Arta, and True RTA and I have the same issues, but I've found the problem with my system. I use an EMU-0404 USB sound card and there's a warm-up period with this card that can take as much as 15 minutes before it settles down. I don't trust anything I'm measuring now until I can get the same FR plots between Arta and True RTA and with the OBs I've been working with lately, I've found another trick. When the card is cold, it won't measure the dipole roll-off correctly. There's a point where suddenly, the expected roll-off simply appears in both measurement systems and from that point on, both of them become stable and consistent.
I've got a routine down now where I remove the EQ from the bass (4-5db boost averaged between 200hz down to 40hz) and run pink noise until the weird stuff below 200hz starts to settle down and show what is expected at these frequencies with a dipole. I've read that the EMU card does need a warm-up period, and I tend to believe that now. Also, I don't know how your card works, but mine has all its level controls on the unit itself and it's important for mine to have the rght amount of gain on the mic - too low, and measurements are inconsistent. Just a FWIW.
I haven't put too much stock in CSDs/BDs at this point because of obvious latency problems with an external card. I think the best I can get with mine is 7ms so I'm not too confident about what I'm measuring between output and input - I need to do some more research into that area.
Mavo,
"Magnetar, have you tried using a more symmetric baffle layout, tweeter in the center, midranges around it, then woofers around that, all in circles? Maybe, together with the right crossover points, this will minimize driver interaction, as it minimizes the distance between their centers. Kind of a synergy open baffle, like in Toms Synergy horn patent."
Mike's right - clusters suck. 😀
I've been playing with driver configurations on an OB for a coupla months now and I've had them in every possible setup that 4 direct radiators, 3 different horns, and 3 different compression drivers will alow and the tight cluster idea has, ummm, issues. 🙂 With this setup, it's very focused, but move your head 1 inch right or left and you fall into a horizontal null that no amount of crossover tweaking, or driver split roll-off will cure. Kinda nice when you're in the zone, but the zone is too narrow to be usable.
An externally hosted image should be here but it was not working when we last tested it.
AJ said:
I'm sorry, but this just doesn't seem logical to me. What is it exactly that has to "warm-up"? I've used external sound cards before and have not seen this issue, and I certainly don't see how it could happen. These are not tube devices and any "warming up" would at most be a few .1 dB. Does the company acknowledge this problem?
The extrenal boxes usually have very little latency between channels, about 1/2 - 1 sample, so I don't see what the issue is with measuring input and output.
I'd also like to see this thread get back on topic and not diverted into an OB discussion. If we are going to elaborate on measurements then a new thread is warranted.
Well, I can't explain it either Earl, all I know it that's what happens. If there's a more rational explanation, I'm all ears.
Sorry for getting off-topic.
Sorry for getting off-topic.
Well thats why I bring it up - because I highly doubt that your hypothesis is correct and I think that you would be wise to track down these kinds of things as they could be symptomatic of other problems. Who's to say its correct after it warms up and not when it first starts?
What about the power amp - now those definately can have warm up issues. But again, any amp that has measureable differences after 15 minutes should not be relied on for measurements because something is wrong. But caps in the amp power supply or signal path warming up is a very common amplifier flaw.
The mic pre-amp could be suspect, or at least the phantom supply. I assume you are using a mic that needs power - any good mike would.
I'd be just very interested in discussing measurements as it is the basis for all we do (or at least it should be the basis!) and its so easy to get them worng. But I'd like for people to be justified in coming to this post to learn about waveguides. If one of you guys wants to start a tread on measuring loudspeakers I'd chime in. At some point I may even be talked into parting with my measurement software, which is, I believe, state-of-the-art. Its hard as hell to use, but infinately flexible and does some things that I've never seen other software do.
What about the power amp - now those definately can have warm up issues. But again, any amp that has measureable differences after 15 minutes should not be relied on for measurements because something is wrong. But caps in the amp power supply or signal path warming up is a very common amplifier flaw.
The mic pre-amp could be suspect, or at least the phantom supply. I assume you are using a mic that needs power - any good mike would.
I'd be just very interested in discussing measurements as it is the basis for all we do (or at least it should be the basis!) and its so easy to get them worng. But I'd like for people to be justified in coming to this post to learn about waveguides. If one of you guys wants to start a tread on measuring loudspeakers I'd chime in. At some point I may even be talked into parting with my measurement software, which is, I believe, state-of-the-art. Its hard as hell to use, but infinately flexible and does some things that I've never seen other software do.
Assuming you are using MLS test signals, it is necessary to determine the time difference between the direct sound and first reflection. When generating a frequency response from the impulse, the window width must not include the first reflection. Normally this takes care of most of the issues. But if you are still having problems, I suggest opening a thread to discuss it.Magnetar said: