I don't understand the ever need to get offensive and personal again?
Don't think it's very polite and respectful.
We are here just to help each other, and share ideas, nothing more and nothing less.
I literally only said something about a graph scale?
The reason why that even could be offensive totally beats me.
I would appreciate if the style of communication could be a bit more respectful, thank you.
Just stick to the subject, it's not that hard.
A lot of other people become tired of so many things, but have a decency to stay polite and respectful.
Don't think it's very polite and respectful.
We are here just to help each other, and share ideas, nothing more and nothing less.
I literally only said something about a graph scale?
The reason why that even could be offensive totally beats me.
I would appreciate if the style of communication could be a bit more respectful, thank you.
Just stick to the subject, it's not that hard.
A lot of other people become tired of so many things, but have a decency to stay polite and respectful.
Thank you for all the suggestions. The filter design I posted is very preliminary, and at this point it is just a feasibility exercise to ensure that a reasonable, plausible design is possible.
The goal of this first round of testing was (1) to assess cabinet resonances, and (2) to assess how much variation there is between measured data from the actual cabinet and measured data from the XPS foam board prototype.
The acoustic emissions from cabinet walls are very difficult to measure with accuracy. The driver must be playing in order to energize resonances in a representative way, and the driver acoustic emission will overwhelm any measurements of cabinet radiation. My approach is to not try to measure the cabinet wall acoustic emission with accuracy. Instead I measure it conservatively. If I can determine that the cabinet wall radiation is below some threshold level, it does not matter what the true level is. The cabinet signature is good enough.
Sound transmission through ¾ inch plywood or MDF is about 20 – 25 dB attenuated, depending on frequency. Structural resonances can greatly reduce the attenuation due to the panel acting as a diaphragm. So the threshold for resonance emissions is, in my mind, -20 dB below the radiation from the driver. Anything below this is most likely insignificant. I described this process in another project. https://www.diyaudio.com/community/threads/purifi-waveguide-project.394174/post-7349489
In this case, I played pink noise through the midrange, and then probed the cabinet with a mechanics stethoscope. Most areas sounded exactly like pink noise, which means no resonance. There was a spot on the rear wall and a spot on each side wall where the sound had a slight emphasis at about 450 Hz. Using sine wave signal, I was able to detect an elevated level over a broad band from 400 – 500 Hz, but no pronounced resonance. The impedance plot had a slight warble at 420 Hz. So this is an area to investigate further.
Making a near field scan of the mid driver at about 2 Vrms, the SPL was about 110 dB. At the same drive voltage, the side cabinet near field response was more than 25 dB down. Of course most of what I was picking up was not cabinet response, but driver response and room modes. But this did demonstrate to me that whatever side wall resonances might be present, they were at least 25 dB below the driver response. I got a similar result with the rear wall.
One more measurement worth looking at is the near field burst decay. Unlike a far field measurement, we can use a very long gate or no gate at all, and this give us good resolution down to low frequency. Here we see that the mid driver has a very clean decay in the usable range (50-4k), and there is no stored energy or resonance coming through the cone.
I repeated the same process with the woofer, and came to the same conclusion: The cabinet signature is below the threshold of significance.
The goal of this first round of testing was (1) to assess cabinet resonances, and (2) to assess how much variation there is between measured data from the actual cabinet and measured data from the XPS foam board prototype.
The acoustic emissions from cabinet walls are very difficult to measure with accuracy. The driver must be playing in order to energize resonances in a representative way, and the driver acoustic emission will overwhelm any measurements of cabinet radiation. My approach is to not try to measure the cabinet wall acoustic emission with accuracy. Instead I measure it conservatively. If I can determine that the cabinet wall radiation is below some threshold level, it does not matter what the true level is. The cabinet signature is good enough.
Sound transmission through ¾ inch plywood or MDF is about 20 – 25 dB attenuated, depending on frequency. Structural resonances can greatly reduce the attenuation due to the panel acting as a diaphragm. So the threshold for resonance emissions is, in my mind, -20 dB below the radiation from the driver. Anything below this is most likely insignificant. I described this process in another project. https://www.diyaudio.com/community/threads/purifi-waveguide-project.394174/post-7349489
In this case, I played pink noise through the midrange, and then probed the cabinet with a mechanics stethoscope. Most areas sounded exactly like pink noise, which means no resonance. There was a spot on the rear wall and a spot on each side wall where the sound had a slight emphasis at about 450 Hz. Using sine wave signal, I was able to detect an elevated level over a broad band from 400 – 500 Hz, but no pronounced resonance. The impedance plot had a slight warble at 420 Hz. So this is an area to investigate further.
Making a near field scan of the mid driver at about 2 Vrms, the SPL was about 110 dB. At the same drive voltage, the side cabinet near field response was more than 25 dB down. Of course most of what I was picking up was not cabinet response, but driver response and room modes. But this did demonstrate to me that whatever side wall resonances might be present, they were at least 25 dB below the driver response. I got a similar result with the rear wall.
One more measurement worth looking at is the near field burst decay. Unlike a far field measurement, we can use a very long gate or no gate at all, and this give us good resolution down to low frequency. Here we see that the mid driver has a very clean decay in the usable range (50-4k), and there is no stored energy or resonance coming through the cone.
I repeated the same process with the woofer, and came to the same conclusion: The cabinet signature is below the threshold of significance.
Serious question out of curiosity, but where and when did the idea rise up again that with a properly designed and build cabinet, cabinet resonances could be and issue?
I am rather confused that people start talking about this again (incl the ones that do reviews).
In the last 15-20 years I have seen countless of measurements investigating this.
In combination with just knowledge from the literate and books, the conclusion has always been the same.
Especially when you look at orders of magnitude and significance, it has never been a problem.
But the last couple of years, this subject seems to rise up again?
These kind of insights make me happy btw!
Thumbs up for the approach as well 👍
It is just due diligence. I was not expecting to find anything. But I sometimes discover something very significant while measuring a parameter that I already know the value of... and then I get surprised.
It is my opinion that with some good design rules, and with careful construction, it is possible to build a low-signature cabinet with a high degree of confidence. But it is very possible to build a cabinet with audible resonances. I have done so, and I had to trash them and start over. I hear commercial speakers from time to time that have an audible resonance problem in the upper bass / lower midrange.
As you said
Indeed, if the cabinet is properly designed, and properly built, structural resonances are insignificant. But the only true way to know if a cabinet has been properly designed and built is to test it for resonances...
Here is the SB26STAC tweeter response. I am showing both the response measured in this cabinet, and the response taken from the XPS foam board prototype (post 314). The prototype had straight non-tapered bevels, and I wanted to make sure that any conclusions I drew from that data are still valid. https://www.diyaudio.com/community/threads/compact-low-cost-active-3-way-speaker.402812/post-7471898
The two sets of tweeter data look very compatible. There are some small differences in diffraction, but not enough to affect my assumptions and conclusions. @PKAudio - hopefully this is what you are looking for...
Also the midrange SB15NBAC and the woofer RS270-4
j.
The two sets of tweeter data look very compatible. There are some small differences in diffraction, but not enough to affect my assumptions and conclusions. @PKAudio - hopefully this is what you are looking for...
Also the midrange SB15NBAC and the woofer RS270-4
j.
I have no beef in this but..
I dont know if English is your mothertongue? Its not mine either, but your posts generally dont come out polite or respectfull. -More like guns blazing, shooting at everyone and every detail.
I would urge you to re-read before posting, to check if something could be taken as offensive, could be omitted and if the post itself is really necessary.
-Otherwise I Thank you for all the projects and insight you share with our community.
Cheers!
A flat on-axis response is very nice for development work. I always start with this response before I do any voicing. It is very easy to compare the measured system performance to simulation, check that the simulation conforms to actual driver magnitude and delays. Once I am certain that the simulation is dialed in and fine-tuned, I can then confidently use the simulation for making small adjustments.
A flat on-axis response, to my ears, often sounds too bright in the lower treble, although with some classical recordings, it actually sounds quite correct.
During the voicing process, I usually focus more on the PIR response curve. But really, it is all a compromise, and each speaker's radiation pattern requires a slightly different compromise between on-axis, PIR and power response.
Here is the latest noodling...
And here is the passive filter with no DSP and no woofer
The passive crossover manages the transition from mid to tweeter. The shape of the DI curve from 500 Hz - up is controlled by the passive crossover. It also handles the notch filter on the midrange driver. All of the EQ, BSC, and woofer-mid crossover function is handled by the DSP.
j.
That's exactly how many other posts come across to me. (and not even a little bit)
It's not so difficult to just to stay decent and just stick to the subject, instead of getting all snarky and personal.
Even if some posts are already crossed many lines of respect to me and for many other people as well.
I already re-read my posts a billions times, because apparently we have to walk on eggshells here, since some people seem to be offended by quite literary anything.
So I simply can't predict if certain things will be taken offensive nor is it even my responsibility.
It's not my preferred way of communicating, I expect that people can have a conversation like adults.
Which on an international forum often means that sometimes people say certain things that might come across on a different way than you think it is.
Maybe ask for clarification first before judging? Something I very rarely see happening.
I am here to talk about audio and loudspeakers, provide feedback and receive feedback.
That is the only way to grow and get better.
Apologies Jim for putting noise in your topic again, but I was just responding to things regarding your measurements.
Nothing more, nothing less.
Last edited:
The reason I was asking, was basically what you said about sounding to bright.
Almost any speaker without waveguide with a woofer that is bigger than 4 inch, will have a missmatch in directivity, like we see in the measurements here.
In other words, we will have to much energy around 2-10kHz area.
Which can be perceived as being to bright.
Diffraction issues could even add more problems to this mix, typically around 800-3000Hz.
Although in this case they are a little smaller because of the baffle shap.
What I personally start doing is actually optimizing at 15-30 degrees from the beginning.
That way you already have a much nicer compromise between direct sound and off-axis sound from the beginning I think.
Self reflection is a difficult skill to master. In the spirit of being helpful I will point out two places where one of your previous posts could be misinterpreted and cause frustration.
As you correctly stated above it is good to question before unloading. Change it to "Why would you want to see it on a 30dB scale?" Leave out pixel peeping.
Stated bluntly as a fact when it is your opinion, Earl Geddes holds the opposite opinion that all normalized polar plots are useless. Who is right?
This is good because it explains your reasoning, unfortunately the damage can be done in the first line.
The differences might seem subtle but in my own experience they are the difference between communicating and winding people up.
There was never any damage to begin with?
Did I already mention to maybe clarify first before judging?
Sorry, but your entire response doesn't make any sense to me.
If you weren't pixel peeping, just simply explain WHY you aren't pixel peeping?
It's unfortunate that you didn't quote my actual entire argument about that as well, because you could read why a 30dB could be seen as pixel peeping.
There is absolutely nothing blunt about stating facts?
Other people might disagree with it, fine, let than just talk about the disagreements and use arguments.
In that journey we and other people might even learn a thing or two from it!
The difference is very obvious, that people get hung up about interpretation and wording instead of the subject.
But in any case, I NEVER got snarky and personal (which btw, I NEVER do) and you did.
Without even proper argumentation about the subject.
In many cultures, that's considered as extremely blunt and quite disrespectful.
Not only that, but you also automatically put your preferred way of communicating and ideas/culture about what's wrong/right/polite onto someone as well.
Truth probably lays somewhere in between.
There are many many many times I have to really bite my tongue because how incredibly rude and disrespectful certain people are and/or respond.
I could also get all personal about it, saying whatever I think about them.
Fact is that such a response isn't helpful for the thread nor the subject, plus it only creates additional noise.
Like we do now as well.
Last edited:
I would try to sacrify frequency response linearity of the passive filtering, for a better acoustic phase matching of tweeter to mid as a next step.
LOL, this is how misunderstandings happen. There was no snark or personal comments in my post just pointing out how your posts are being viewed. You did not see it the way I intended it.
This is actually a decent example to show why a normalized graph is so much more useful.
The entire section around 1.8kHz or so, shows how the directivity gets a lot lower all of a sudden.
Potentially a diffraction problem would be my first guess.
There is no way that this is so obvious in the left (non-normalized) graph.
Another good example would be when a woofer has a dip in the freq resp itself.
Which might give the impression that the directivity is getting higher, but when shown in a normalized response it would be clear it's actually not.
Meaning that you can just simply EQ the thing to make it straight for example.
Not only that, but directivity is a relative thing on itself.
Meaning it's all relative to the on-axis response.
(or any other axis you want to take as a reference if you like)
Last edited:
That's why I rather just stick to the subject, instead of calling people out on things.
I would rather have a passionate discussion about something involving the subject, than putting more noise into the conversation.
This is the funniest thing you have said yet
I would rather leave the passion out as well as the noise. I will do my best to to do that going forward.
Thank you for your thoughts. I believe the slight phase mis-match between the mid and tweeter helps keep the directivity index flat through the crossover region. However, I will try some sims with a closer match and see what they look like. Thanks !
Good phase matching helps too keep the overall speaker beam shape more independent of frequency in the vertical domain. Less reflections come from different directions with other intensity dependent on frequency, because the vertical beam shape is less "lobing around" and has a symmetrical pattern. When the phase is not matching, you hardly see it only in the power response views. I personally would sacrify for this a drop in the power response at the xover frequency, therfore i might aim then to 2,2-2,5 kHz XO, because the vertical drop here brings a bit foregiveness to the indirect sound as a nice tradeoff.