Joe,
lets assume we want to obtain full space farfield FR valid from ~10Hz to 20kHz. In that case I miss one step in your procedure below (and I agree with all written below, just one step is missing). And it is adding diffraction response to NF response before merging, see following Vituix Merger video, time 0:50s:
YouTube
Diffraction response can be created in Vituix Diffraction tool, or excel sheet done by Jeff Bagby.
Of course, you can omit that step, but resulting FR is NOT full space response. It is just some hybrid response, part of it valid for NF, rest of it for FF.
This merging method is just one of the ways to obtain full space farfield response valid from low frequencies. If one is able to achieve ~20ms reflection free time window, or measures outdoor, or uses GPM for merging, than there is no need for the step described above.
lets assume we want to obtain full space farfield FR valid from ~10Hz to 20kHz. In that case I miss one step in your procedure below (and I agree with all written below, just one step is missing). And it is adding diffraction response to NF response before merging, see following Vituix Merger video, time 0:50s:
YouTube
Diffraction response can be created in Vituix Diffraction tool, or excel sheet done by Jeff Bagby.
Of course, you can omit that step, but resulting FR is NOT full space response. It is just some hybrid response, part of it valid for NF, rest of it for FF.
This merging method is just one of the ways to obtain full space farfield response valid from low frequencies. If one is able to achieve ~20ms reflection free time window, or measures outdoor, or uses GPM for merging, than there is no need for the step described above.
Oh boy, how much typing would I need to do? But quickly, as much as possible, correct a few things...
So many issues are being conflated and confused here. Like the 150Hz bump being now a vertical pipe like resonance when I have already stated that it disappears when moving the microphone moderately away and does not show up in RTA measurements near the listening position. This bump has nothing to do with diffraction loss or a vertical resonance inside the box, it is a matter of the microphone technique used.
Note the vertical length of the box and the distributed drivers (four) along the length, does it not make sense that a vertical pipe resonance is virtually impossible, it can't happen.
THAT WAS PART OF THE ORIGINAL PLAN TO AVOID!
So its a bit much to be confronted by this as an issue that was thought about ten years ago. Indeed, vertical line sources don't have this problem and that influenced me.
Also: If that was true, you would also have a wriggle at 150Hz in the impedance plot, but there is NONE!
Next:
Omnis don't cause proximity effect???
Yes, they DO!
And, they DON'T!
But we are comparing apples with oranges.
Am I saying the linked wiki page must be wrong when it says that in recording a voice the omni mic avoids (largely) the proximity effect? No, the article is essentially correct, but it is talking about recording voices.
But that is not what we are talking about here, where we are trying to emulate/simulate an infinite flat baffle (a la IEC) is completely different and using an omni mic will definitely produce a proximity effect when this measurement is made. We are trying to get a similar effect to an IEC baffle.
For example, make a large IEC baffle, then cut a hole in it and get somebody to sing through the hole and put ANY mic up to within 1/4" of the baffle and it will produce a proxikity effect. Even omnis will do it.
In fact the measurement relies on it, but it has a likely side-effect that we need to be aware of. Indeed omnis are the best, but ALL mics will exhibit this behaviour.
There are a lot of misunderstandings about the use of omnis, but in fact they are best for measurements for various reasons, they are potentially technically the best, like B&K, DPA and Earthworks measurement mics are omnis. John Atkinson for Stereophile only (as far as I can see) uses omnis. I only use omnis. Omnis also have greater LF extension, and less phase shift and impulse response (time domain).
So many issues are being conflated and confused here. Like the 150Hz bump being now a vertical pipe like resonance when I have already stated that it disappears when moving the microphone moderately away and does not show up in RTA measurements near the listening position. This bump has nothing to do with diffraction loss or a vertical resonance inside the box, it is a matter of the microphone technique used.
Note the vertical length of the box and the distributed drivers (four) along the length, does it not make sense that a vertical pipe resonance is virtually impossible, it can't happen.
THAT WAS PART OF THE ORIGINAL PLAN TO AVOID!
So its a bit much to be confronted by this as an issue that was thought about ten years ago. Indeed, vertical line sources don't have this problem and that influenced me.
Also: If that was true, you would also have a wriggle at 150Hz in the impedance plot, but there is NONE!
Next:
Omnis don't cause proximity effect???
Yes, they DO!
And, they DON'T!
But we are comparing apples with oranges.
Am I saying the linked wiki page must be wrong when it says that in recording a voice the omni mic avoids (largely) the proximity effect? No, the article is essentially correct, but it is talking about recording voices.
But that is not what we are talking about here, where we are trying to emulate/simulate an infinite flat baffle (a la IEC) is completely different and using an omni mic will definitely produce a proximity effect when this measurement is made. We are trying to get a similar effect to an IEC baffle.
For example, make a large IEC baffle, then cut a hole in it and get somebody to sing through the hole and put ANY mic up to within 1/4" of the baffle and it will produce a proxikity effect. Even omnis will do it.
In fact the measurement relies on it, but it has a likely side-effect that we need to be aware of. Indeed omnis are the best, but ALL mics will exhibit this behaviour.
There are a lot of misunderstandings about the use of omnis, but in fact they are best for measurements for various reasons, they are potentially technically the best, like B&K, DPA and Earthworks measurement mics are omnis. John Atkinson for Stereophile only (as far as I can see) uses omnis. I only use omnis. Omnis also have greater LF extension, and less phase shift and impulse response (time domain).
Last edited:
For those who want a clearer understanding, this is a necessary read about how to test speakers and capture data. It conforms with the work done by D. B. Keele.
Buy:
Testing Loudspeakers Book
Testing complete loudspeaker systems, which is a related and yet different topic, watch the YouTube video presentation by Stereophile's John Atkinson. Even if everything is not understood (takes a nerd like me), you will still get a lot out of it, and better understand the scope of the topic. Be sure also to take in the Q&A at the end of it, some interesting bits there two:
YOUTUBE VIDEO LINK
As Allen B asked about what I use, will come back a bit later with that list.
Buy:
Testing Loudspeakers Book
Testing complete loudspeaker systems, which is a related and yet different topic, watch the YouTube video presentation by Stereophile's John Atkinson. Even if everything is not understood (takes a nerd like me), you will still get a lot out of it, and better understand the scope of the topic. Be sure also to take in the Q&A at the end of it, some interesting bits there two:
YOUTUBE VIDEO LINK
As Allen B asked about what I use, will come back a bit later with that list.
Indeed
for you guys who keep asking back and forth about 150Hz, BSC, then goes to basic theory of measurement,
I just opened your profile and I don't find you posted any your own design yet, which is supported by proper theory and design. and most importantly followed by so many members . while this thread started from 2007 and that moment you might still learn basic of Loudspeaker Design, or maybe had not.
are you guys competent of all those fields that you are asking? or just by reading from internet?
I assume there more than 50 Elsinore have been built or maybe >100. build threads have been posted not only in this forum but others too, i have never read any dissapointment yet
Salute to you Joe to respond flawlessly for all this time, for a free design you have shown great value here.
There are people who ask to seek the truth, while many who just keep asking for nothing
The RTA measurement is not anechoic, so it mainly shows the room effects, especially at low frequencies. I am not asking about the sound of the room where the measurement was taken, but about the nature of the 150 Hz peak seen in the nearfield measurement (which is not influenced by the room). If the peak exists in the nearfield and you want to say that it does not exist in the far field, you need an anechoic measurement as I described it in my previous post.
As long as there is energy injected into the box at the frequency of the longitudinal mode, it will resonate. It does not matter much where the resonance is excited. Using multiple woofers on top of each other does not cancel the 150 Hz resonance in any way. Remember that the 150 Hz wavelength is about 2.3 m, which is much more than the spacing of the woofers in the Elsinores.
It seems we are using the "proximity effect" term in different ways.
For me, "proximity effect" is what is described on the Wikipedia page I linked earlier. It does not happen with omni mics.
Your description / example with the baffle is about going from the 2pi nearfield to the 4pi farfield. This transition happens when you go from "immediately in front of the baffle" to "far away from the baffle", and it does not depend on the type of microphone simply because it happens no matter if there is a microphone or not. The nearfield / farfield transition is related to the baffle step.
For those who want a clearer understanding, this is a necessary read about how to test speakers and capture data. It conforms with the work done by D. B. Keele.
Yup, that's a good book. I have read it many times since it came out, I built my audio analysis software based on it, and I still consult it frequently.
I am not sure to whom this is addressed. If you are talking about me, judge for yourself:
GitHub - mbrennwa/mataa
https://audioroot.net/wp-content/uploads/2014/08/MATAA_aX.pdf
GitHub - mbrennwa/osmcdoc: Documentation for the Open Source Monkey Coffin
Open Source Monkey Box
Don't get me wrong. I am not at all saying that there is something wrong with the Elsinores. I do think it is a good design.
However, I believe the SPL curve shown on the Elsinore website is misleading, and the explanations given for it don't seem right to me.
It seems that any confusion can be resolved simply this way: I use my experience to build the box and alignment (with option to go back and change it) and then do all the measurements, both electrical (impedance/current phase) and acoustic in situ.
That means "in box."
I don't have to calculate the baffle step, I just have to capture it via farfield and nearfield measurements. It is that simple. No simulation software/tools based on the width of the baffle and all that. I just don't trust them, too many variables. But from an academic viewpoint I applaud that they are available and that they can be used to exploit the issue. But my method is mistake proof.
I use the industry stand Clio FW01 and Clio 11 Software. For many years it was either the MLSSA or Clio that was used by professionals, still is, but other options now are available. I had a chance to talk on the phone with Roger Sanders (after he left Martin Logan) a few years ago. Guess what he uses? Clio FW01. Many will know that I have had a friendship with speaker designer Brad Serhan (Orpheus, Brigadier, ABC TV monitors, Axis) and guess what he uses? Clio FW01 as well. Mike Lenehan of Lenehan Audio still uses Clio 4 with DOS interface.
Clearly the microphone is also important and it must be an omni and I believe that for this purpose, this mic is as good as it gets (and Audiomatica, the maker of Clio, top recommendation):
More info from Earthworks HERE. And more HERE ALSO
My other goto is Clio's own MIC01, but it is not fully balanced:
I have a custom made utility amplifier capable of around 20 Watts with linear Vernier controls on levels. It is all calibrated using the Vifa XT25 Tweeter, because on axis and 2.83V it always looks like this and knowing it is 91dB-SPL I can get very accurate acoustic data. Levels have to be spot on!
MY CALIBRATION DRIVER - VIFA XT25 TWEETER
Of course there are heaps of cables, stands and so on, all various different lengths.
I have a rather large open area at the front of the house, so all acoustic measurements are taken outdoors, must pick a good day and NO wind. Also, the measured DUT (device under test) is raised well off the ground and a multilayered kind of very thick mat on the ground between DUT and microphone to control the main reflection that I have. This helps to get the most out of the measurment, we want to window the longest possible time and get resolution low down and it is HERE that we can capture the diffraction/baffle step before merging with nearfield measurement.
The whole process has to be done with precision and diligence, as we want the data into Clio's Software to be very accurate, double and triple checked.
Because I have captured the relative phase between drivers (don't even breathe to move the mic between 'takes'), I can process the files in Clio and the next step is to bring them into SoundEasy (another industry standard and it is Australian) and create complex files for all the drivers and now "build" the crossover in SoundEasy.
If you done a good job and no mistakes, SoundEasy is brilliant to "design" your crossover, and when you believe you have worked it out, then pysically build the crossover and get the speakers up and running. Use Clio's many tools to check that the design works as it should. It is very satisfying to see it come together. My favourite Clio tool to check the final resuls is Pink Noise and 1/6th Octave in room response. Of course I know my room very well and I know what to expect. I move the mic around and get a full sense of what it is doing and you can grasp what the spatial result/performance is like.
The final point I want to reiterate is that I have top notch industry standard tools. But there are now many low cost alternatives, but using a sound card rather than the Clio? That doesn't strike me as professional like Clio or even AP.
So there you are. Very brief indeed, even if this is a long post.
That means "in box."
I don't have to calculate the baffle step, I just have to capture it via farfield and nearfield measurements. It is that simple. No simulation software/tools based on the width of the baffle and all that. I just don't trust them, too many variables. But from an academic viewpoint I applaud that they are available and that they can be used to exploit the issue. But my method is mistake proof.
I use the industry stand Clio FW01 and Clio 11 Software. For many years it was either the MLSSA or Clio that was used by professionals, still is, but other options now are available. I had a chance to talk on the phone with Roger Sanders (after he left Martin Logan) a few years ago. Guess what he uses? Clio FW01. Many will know that I have had a friendship with speaker designer Brad Serhan (Orpheus, Brigadier, ABC TV monitors, Axis) and guess what he uses? Clio FW01 as well. Mike Lenehan of Lenehan Audio still uses Clio 4 with DOS interface.
Clearly the microphone is also important and it must be an omni and I believe that for this purpose, this mic is as good as it gets (and Audiomatica, the maker of Clio, top recommendation):
More info from Earthworks HERE. And more HERE ALSO
My other goto is Clio's own MIC01, but it is not fully balanced:
I have a custom made utility amplifier capable of around 20 Watts with linear Vernier controls on levels. It is all calibrated using the Vifa XT25 Tweeter, because on axis and 2.83V it always looks like this and knowing it is 91dB-SPL I can get very accurate acoustic data. Levels have to be spot on!
MY CALIBRATION DRIVER - VIFA XT25 TWEETER
Of course there are heaps of cables, stands and so on, all various different lengths.
I have a rather large open area at the front of the house, so all acoustic measurements are taken outdoors, must pick a good day and NO wind. Also, the measured DUT (device under test) is raised well off the ground and a multilayered kind of very thick mat on the ground between DUT and microphone to control the main reflection that I have. This helps to get the most out of the measurment, we want to window the longest possible time and get resolution low down and it is HERE that we can capture the diffraction/baffle step before merging with nearfield measurement.
The whole process has to be done with precision and diligence, as we want the data into Clio's Software to be very accurate, double and triple checked.
Because I have captured the relative phase between drivers (don't even breathe to move the mic between 'takes'), I can process the files in Clio and the next step is to bring them into SoundEasy (another industry standard and it is Australian) and create complex files for all the drivers and now "build" the crossover in SoundEasy.
If you done a good job and no mistakes, SoundEasy is brilliant to "design" your crossover, and when you believe you have worked it out, then pysically build the crossover and get the speakers up and running. Use Clio's many tools to check that the design works as it should. It is very satisfying to see it come together. My favourite Clio tool to check the final resuls is Pink Noise and 1/6th Octave in room response. Of course I know my room very well and I know what to expect. I move the mic around and get a full sense of what it is doing and you can grasp what the spatial result/performance is like.
The final point I want to reiterate is that I have top notch industry standard tools. But there are now many low cost alternatives, but using a sound card rather than the Clio? That doesn't strike me as professional like Clio or even AP.
So there you are. Very brief indeed, even if this is a long post.
Last edited:
I couldn't agree more. A good measurement is better than simulated theory.
However, a good measurement of the baffle step can be difficult. I know I am repeating myself, but I can't change the physics: to capture the baffle step in the far field, you need anechoic data that extends down to about 100 Hz or so. Analysing the nature of the 150 Hz peak in the farfield requires an FFT resolution of about 50 Hz, which corresponds to a 20 ms long anechoic impulse response. Let's be honest that getting such long anechoic measurements is not trivial. If you can do it, great. If you can't, an estimate based on some theoretical simulation is better than no data at all.
But as Martin King showed early on in the development of his TL model showed that the position of the driver does effect the magnitude of the resonances. In a TL a driver is often offset from the closed end just enuff that it almost completely kills the 1st unwanted harmonic. One does also get a reduction in the desired fundamental, but because the 1st harmonic is no longer an issue one can use less damping (fill) and get the low bass lost back.
In a tall line like Joe’s there will be a longitudanal resonance, and the box is really an ML-TL, but how the placement of the 4 drivers, and the stuffing (i’ve not paid attention to whether Joe stuffs these (like a TL) or lines them (like a BR)) affect the 1/4 wave resonance could be modeled, but Joe has had lots of time & development to work the LF tuning out with real world measures.
It is physcically possible to rearrange the drivers so that one gets less issues with centre-to-centre spacing, and optimize the functioning of the TL aspect to get more bass. The XO might need some tweaking.
dave
Also a comment on Joe’s book: It is fairly thorough, but does date itself somewhat (at least the edition i have), but i always take anything he does with a gran of salt since he showed that even he doesn’t always pay attention his own work, having made a clear mistake that voided the LF measures of his prototype Thor and subsequently the box needewd a redesign to produce real bass below 100 Hz.
dave
dave
I agree, but with most speaker you don't need 100Hz, unless the speaker has a well over 1 metre wide baffle. But your point is well taken, this is why I use 2 metres for my main acoustic measurements, you can get reliable data down to a lower frequency and with the Elsinore width of 280mm. That way I had no problem seeing the diffraction loss with lowering frequency.
But later, with in-room RTA type measurements, you will be able to confirm if you got it right, and generally I do get it right, but I think it also comes down to experience of the person doing the merging of FF and NF in software that comes with the Clio FW01. There is an equation that Joe D'Appolito has published, it may have come from Keele, I am not sure, that lets you calculate the attenuation of the NF to match the FF. I have to modify that by 6dB because my measurements are at 2 metres. I use the equation as a guide, but when measuring the port at the rear of the Elsinores, the degree of attenuation for that NF measurements is a pure judgment call. But that has no effect on the 150Hz bump shown as it is well under 100Hz.
Others here note that Dave is referring to Joe D'Appolitos book - I didn't write any published book. OK?
Agreed. But then again using my own judgment system (it's called experiece), I do that anyway, a bit of grain is good, not just in bread.
Also, his type of crossovers are not my cup of tea. I have modified speakers that he designed for Taiwan's Usher. Not your cup of tea either, I know. But hello, I am not on any kind of war path preaching the purity of science, the white lab syndrome. I have been in the room when scientists didn't agree with each other, and that is OK with me. Both Allen and I had what might be called a 'metaphysical' sides to our personalities, and yet a love of science, especially experimental science. But science is the happy pursuit of knowledge and what we can do with it, not as a set of absolutes. I was told by a scientist I respect that the only thing that matters were results. I go for results every time rather than dogma.
Better get off my soap box now, or I will get into truble?
PS: And guys, guess what system Joe D'Appolito uses? *CLIO!*
In the book he does mention this and it looks like Clio 4 (has a whole chapter), a still working DOS program, limit is 20KKz whereas Clio 11, which I use, can capture up to 96KHz, but 48KHz is more than enough unless special circumstances. But the book is now old, circa 1998.
PPS: Yes, Joe's NF calculation of attenuation is based on Keele's work, just confirmed it.
With a baffle width of 28 cm, the Elsinores will exhibit a baffle step that extends to well below 100 Hz. See Online Baffle Step Calculation to get an idea of the shape of the baffle step.
Unfortunately, it does not work this way. It's actually the opposite. If you move the microphone further away, the anechoic part of the impulse response gets shorter. Consider this back-of-the envelope calculation (see attached drawing):
- (A) Microphone at 1 m from baffle: The direct sound arrives at the microphone after a "flight time" of 2.9 ms (1 m flight time). The distance travelled by the floor echo is 2.69 m (Pythagoras is your friend), which corresponds to a flight time of 7.8 ms. The anechoic data between first arrival of the direct sound and the arrival of the floor echo is therefore 4.9 ms long, which corresponds to a lower cut-off of 204 Hz (this is also the FFT resolution).
- (B) Microphone at 2 m from baffle: The direct sound arrives at the microphone after a "flight time" of 5.8 ms, whereas the floor echo arrives at 9.3 ms (same math as above, with different numbers). The anechoic part of the impulse response is therefore only 3.5 ms long, which corresponds to a lower cut-off (and FFT resolution) of 286 Hz only.
Depends on what you mean by "it". The in-room RTA measurement mostly shows how the speaker interacts with the room. It will not show the baffle step curve, and it will not show what is going on with the 150 Hz peak in the farfield. You need an anechoic measurement for this.
Joe, I guess what I am trying to say is that you'd need meaningful measurements to back your claims about the low-frequency bahaviour of the Elsinores. You need a 20 ms long anechoic impulse response recorded in the farfield. You could get such data in a (very) large room or outdoors. This might be a bit of work, but it would clear up the "expert judgement" fog and you'd make sure to avoid the issues that Joe d'Appolito had with the Thor loudspeakers. If you think this is too much work (I would!), it would be better to say that you don't have the measurements -- no harm done!
Attachments
I’m now one of many who have/are being tempted to build these speakers
I’m in QLD, Aus - can anyone recommend any good CNC fabricators on the east coast to get the panels cut and routed?
The wood working aspect is going to be my main hurdle i expect....
Elsinore cabinet shape is very basic and any woodworking workshop must be able to cut it for you. no need for fancy cnc
Hi Joe R.,
Thank you for the detailed explanation of your measurement process and equipment. CLIO looks nice and seems accepted industry standard.
For DIYers, something like a Focusrite Solo and a third party calibrated mic from Cross Spectrum Labs and REW (free) will be able to do pretty much similar measurements for about $250. If one gets a calibrated USB mic like the UMIK-1, you don’t even need a Focusrite and save $100. Couple that with XSim tools from Bwaslo or PCD software from Jeff Bagby (both free), we can all design accurate crossovers (if we have the experience and knowledge).
Here are links for everyone to get started.
Third party calibrated mic:
Cross·Spectrum - Calibrated Dayton Audio EMM-6 Microphones for Sale
Focusrite Solo audio interface w phantom 48V for mic:
https://www.amazon.com/dp/B07QR6Z1JB/ref=cm_sw_r_cp_api_i_qblcEb7PNFGYM
REW software:
REW - Room EQ Wizard Room Acoustics Software
XSim software:
XSim free crossover designer
Jeff Bagby’s PCD and FRD Blender software:
Jeff Bagby's Software Page
Thank you for the detailed explanation of your measurement process and equipment. CLIO looks nice and seems accepted industry standard.
For DIYers, something like a Focusrite Solo and a third party calibrated mic from Cross Spectrum Labs and REW (free) will be able to do pretty much similar measurements for about $250. If one gets a calibrated USB mic like the UMIK-1, you don’t even need a Focusrite and save $100. Couple that with XSim tools from Bwaslo or PCD software from Jeff Bagby (both free), we can all design accurate crossovers (if we have the experience and knowledge).
Here are links for everyone to get started.
Third party calibrated mic:
Cross·Spectrum - Calibrated Dayton Audio EMM-6 Microphones for Sale
Focusrite Solo audio interface w phantom 48V for mic:
https://www.amazon.com/dp/B07QR6Z1JB/ref=cm_sw_r_cp_api_i_qblcEb7PNFGYM
REW software:
REW - Room EQ Wizard Room Acoustics Software
XSim software:
XSim free crossover designer
Jeff Bagby’s PCD and FRD Blender software:
Jeff Bagby's Software Page
Last edited: