I can not design a passive xo, diy with dsp-xo is my way. When selecting drivers, one must know how they will be used - dsp/passive, 2-/multiway, desired directivity profile, waveguide? and max spl requirement come to my mind first.
About the highest response range, membrane material, diameter and frontplate construction have significant role. Dispersion above 8kHz is important in giving the sparkle and timbre. If the driver has high directivity there, sound character changes a lot when sitting off-axis and also room (power) response at long distance drops too easily.
Basically the designer should decide if (s)he is aiming for minimal, moderate or high directivity, and starting from which frequency range? Abrupt changes in directivity are always bad, they make the speaker very tricky or impossible to make sounding right.
I have noticed that we have sort of trend for large diameter (32-35mm) Be or diamond tweeters now. The SEAS mentionded and BlieSMas for example - and expensive! They managed to eliminate or push resonance very high and have smooth directivity. Excellent choice for 2-way speakers and 3-ways with low xo!
With 3-4-way speakers that have max 5" mid, one has more choices even with 3/4" tweeters that have minimal directivity up to 16khz. Xo typicaly around 3kHz. Take for example textile dome SB21SDC It has mild on-axis rise above 10kHz to compensate increasing directivity, this is excellent in my eyes. High xo gives some problems with vertical off-axis response though.
With 4-5" waveguide the game changes. Large diameter creates a cavity resonance around 10kHz that rings like a bell. Throat shape and fit become very tricky to avoid this even with 1" drivers. SEAS DXT has minimal and shallow waveguide, thus avoids these problems at the expense of having only moderate directivity.
Baffle dimensions and driver placement make wiggles typically in 2-6kHz range, a wide waveguide helps a lot there.
About the highest response range, membrane material, diameter and frontplate construction have significant role. Dispersion above 8kHz is important in giving the sparkle and timbre. If the driver has high directivity there, sound character changes a lot when sitting off-axis and also room (power) response at long distance drops too easily.
Basically the designer should decide if (s)he is aiming for minimal, moderate or high directivity, and starting from which frequency range? Abrupt changes in directivity are always bad, they make the speaker very tricky or impossible to make sounding right.
I have noticed that we have sort of trend for large diameter (32-35mm) Be or diamond tweeters now. The SEAS mentionded and BlieSMas for example - and expensive! They managed to eliminate or push resonance very high and have smooth directivity. Excellent choice for 2-way speakers and 3-ways with low xo!
With 3-4-way speakers that have max 5" mid, one has more choices even with 3/4" tweeters that have minimal directivity up to 16khz. Xo typicaly around 3kHz. Take for example textile dome SB21SDC It has mild on-axis rise above 10kHz to compensate increasing directivity, this is excellent in my eyes. High xo gives some problems with vertical off-axis response though.
With 4-5" waveguide the game changes. Large diameter creates a cavity resonance around 10kHz that rings like a bell. Throat shape and fit become very tricky to avoid this even with 1" drivers. SEAS DXT has minimal and shallow waveguide, thus avoids these problems at the expense of having only moderate directivity.
Baffle dimensions and driver placement make wiggles typically in 2-6kHz range, a wide waveguide helps a lot there.
Last edited:
Your first tweeter is HiVi TN 25, quite a decent product for $17.90 and the second one is Seas T29B001 Beryllium, a somewhat horn loaded dome tweeter, and as a result of this kind of loading, there is a roll-off tendency towards HF.
George the attached snip from the Morel CAT378 datasheet (a semi horn loaded soft dome tweeter) demonstrates just how easily the rising response from a horn loading can be counteracted. It would seem in fact that they designed the tweeter so that it could be easily used with 1st order electrical crossover (I haven't checked but suspect 2nd order acoustic rolloff).
The second and third attachments are measurements of one of my morel DMS37 tweeters, the predecessor to the cat378. The first is horizontal polars taken in 10 deg increments from on axis through 120deg (after 120deg I started to get problems due non symmetry of the testing space).
Third is distortion plot. The highish numbers are due to the way REW is calculating. If you change the graph to only show from 1Khz up the figures are much lower. In any case above 2Khz I feel that the distortion is quite respectable. Drive level was 2.83V measured at 1M.
Tony.
The second and third attachments are measurements of one of my morel DMS37 tweeters, the predecessor to the cat378. The first is horizontal polars taken in 10 deg increments from on axis through 120deg (after 120deg I started to get problems due non symmetry of the testing space).
Third is distortion plot. The highish numbers are due to the way REW is calculating. If you change the graph to only show from 1Khz up the figures are much lower. In any case above 2Khz I feel that the distortion is quite respectable. Drive level was 2.83V measured at 1M.
Tony.
Attachments
Also of note is the extra -6dB applied to the Seas curves. All Seas curves have this. The rolloff likely won't be as dramatic in a box.
Later,
Wolf
Later,
Wolf
Yes, seas have information on the actual measurement setup of their woofers (datasheet for tweeters mentions .6 X .8M baffle but not the actual positioning of the tweeter) DATASHEETS
Tony.
Tony.
This is just for my own education and clarification - are you talking flat-FR at 1m, at 2m, or at the listening position - which of course is variable?
If at the listening position is this in a single spot or an average over the listening window, or over a larger/smaller region?
Is this independent on room treatments and size etc. i.e. Is this anechoic (measured outside on a ladder for instance), or in room in situ?
I ask because this is incredibly important to what you are saying. Room gain, speaker placement, room reflections and listening distance will obviously completely change spectral tilts and voicing and therefore the sound of a speaker.
I optimize for on-axis response, so they are toed-in to face me.
I measure at 1m, gated in room, and sit at about 10'. You will always have the optimal result at the listening position. Everywhere else is less than optimal.
Speakers are about 3' from front and side walls.
I really only have carpet on the floor, and 10' ceilings.
There basically is no rear wall, as it opens through a double-door archway.
However, besides being on-axis when listening, what you stated really has no bearing on my impressions of the sound. If they are done as I suggested against, they will still sound wrong to me. How do I know? I've done a lot of experimentation. I find even at a longer distance that the speaker will still hold it's composure and tonal balance.
This also equates to other places I've heard my speakers like conference rooms, friend's houses, etc. My impressions have not changed regardless the environment.
You can still tell how the speaker presents sound tonally.
Later,
Wolf
I measure at 1m, gated in room, and sit at about 10'. You will always have the optimal result at the listening position. Everywhere else is less than optimal.
Speakers are about 3' from front and side walls.
I really only have carpet on the floor, and 10' ceilings.
There basically is no rear wall, as it opens through a double-door archway.
However, besides being on-axis when listening, what you stated really has no bearing on my impressions of the sound. If they are done as I suggested against, they will still sound wrong to me. How do I know? I've done a lot of experimentation. I find even at a longer distance that the speaker will still hold it's composure and tonal balance.
This also equates to other places I've heard my speakers like conference rooms, friend's houses, etc. My impressions have not changed regardless the environment.
You can still tell how the speaker presents sound tonally.
Later,
Wolf
Sorry, a little late returning to the party here... but I wanted to answer your question, at least how I see it.
So, one problem with the classic 2-way speaker with a 6"-7" woofer and 1" dome tweeter crossed over at 3kHz is the directivity. As Juhazi mentioned, smooth directivity tends to sound best. In this 2-way speaker, there are two places that the loudspeaker is becoming directive - beaming if you prefer that term. One is just below the crossover point, where the woofer is starting to beam. The other is above 10kHz, where the tweeter starts becoming directive.
The SB Acoustics SB21SDC tweeter that Juhazi posted the responses for in POST 21 is an example of a classic tweeter with the dome sitting proud of a flat faceplate. The response is relatively flat and similar in SPL on and off axis response up to some frequency, usually just under 10kHz but sometimes lower, and the the off axis responses start to fall off while the on axis response remains relatively flat. The responses are diverging. When you design the speaker with flat on axis response, the loss of off axis response causes a peak in the directivity index, which can also be described as a dip in the power response. This non-smooth directivity has been listening tested as less overall balanced of a sound.
Some tweeters now include a small waveguide. The Seas tweeter you mentioned is one of these. The waveguide causes a downwards trending response above some breakpoint with the on and off axis responses tracking each other (appear more parallel to each other than diverging). By employing some simple correction like a low Q crossover, or a shunted capacitor, you can pretty easily re-flatted the response. What you get is a more constant directivity at the top end, and the speaker will sound similar on and off axis. The "parallel" responses at the top end of the Seas plot is what I was referring to in my "can you spot it?" comment.
The first speaker I heard with a tweeter using a small waveguide was MarkK's ER18DXT back in about 2006 or so I think, at a DIY meet. I had seen the spec sheet for the DXT tweeter and thought "why the hell would anyone want a tweeter with such a rolled off response"?! It has a hump like the Seas you mentioned AND a huge breakup peak at about 27kHz. But the speaker sounded really good, and I was intrigued, and after that I learned more and more about directivity and off axis responses, and how they influence the sound of a loudspeaker in a room.
Over time I got more and more interested in controlled directivity. Some loudspeaker use large horns to do this, but I have found that kind of speaker to sound very dull as soon as you move off axis - they are too directive at higher frequencies and not at all directive at low frequencies. Eventually I found myself building open baffle speakers, as they can have very flat direcitivity and sound about the same at any angle on or off axis (when designed correctly). This is one reason why I listed off-axis response as a key metric to keep in mind when building your speaker. It can influence everything from driver selection, number of drivers, crossover points, and so on.
Once you decide on the general class of speaker you want to build (e.g. passive crossover 3-way, or whatever you choose) then you can start to look for drivers having suitable characteristics for the project. That's a good time to look for measurements of a driver's distortion performance - it will also strongly influence where you choose to cross over a particular driver. How low you can cross over a tweeter is often determined in this way.
Thanks that is helpful. So it is the 1m gated in room measurement you refer to as flat?
That helps. I have mainly DSP controlled speakers so am always playing around with measurements and tonal balance.
I don't think it's advisable to take response measurements any further away than that, too many room induced unevennesses (what a great word!)
Yes- flat at that measurement. Of course, it's usually only good to about 500Hz or so.
Also of note- I optimize for in-phase at the xover point. This is important for acoustic alignment. Order is dictated by the drivers' best application. I guess there are always criteria that designate details that are rather important. You really can't just chalk it up to one parameter like FR.
Later,
Wolf
Also of note- I optimize for in-phase at the xover point. This is important for acoustic alignment. Order is dictated by the drivers' best application. I guess there are always criteria that designate details that are rather important. You really can't just chalk it up to one parameter like FR.
Later,
Wolf
Absolutely great comments esp final paragraph. And makes the wannabee-engineer posts that follow sound thoroughly naive and uninformed. Some people mistake acoustic measurements for perceptual measurements.
Pity your insightful remarks sparked no meaningful support for 30 posts.
I am agnostic about what is the best FR curve tendency for drivers fresh from the factory. Also, FR alone doesn't mean much without considering polar response and your room.
As Ugg10 says, FR, etc. just doesn't matter much because systems need to be tuned after installation and you are sitting in your chair and auditioning your modal music choice.
B.
Footnote: a speaker manufacturer needs to deliver a box that has expected characteristics. But that's a silly model for DIY builders who are able to configure their HiFi to their environment by having scope for adjustments all along the reproduction chain and available for fine-tuning to their better liking any time.
Last edited:
You are likely just EQing some problems in your system with the "smiley face". The gently downward trend you mention is often used because the usual design sounds harsh. These are just bandaids employed to try and make some inherent deficiencies of the loudspeaker "less bad".
Wow, hard to believe that this stuff resonates with you. I mean why should anyone use a scientific approach at all? Just wiggle some dials on your DSP until it sounds right in your cave, I guess. All this engineering design is just fake news after all, right?
The question is, a flat response speaker at the anechoic room will be a desirable speaker for the end user or not.
If someone perfectly tunes his speaker for his taste at his listening position by ears, and he brings back it to anechoic room, I guess most probably the measured FR would not be flat at all.
If someone perfectly tunes his speaker for his taste at his listening position by ears, and he brings back it to anechoic room, I guess most probably the measured FR would not be flat at all.
Thank you for your help. Guidance here and further research shows me that some high end tweeters like the Seas in my example are horn loaded which naturally creates that response hump. Once you tame the system response and flatten it out you get better directional performance, better distortion specs, etc. who knew?
George
George
I kind of agree with this, but we may have to consider one thing. Since many manufactures have been voicing their speakers with the "smiley face", many mastering engineers have been master the music to fit those smiley speakers. We can hear that from many masters that sound too harsh for flat speakers. 😛
Still better not to EQ the system anything other than flat, just use tone controls to taste, systems like the Cello Palette were designed in order to "remix" recordings
I recently found that I prefer old horn with obvious hump to the modern waveguide which FR is much smoother. My head kept telling me this waveguide is better, but my heart not. I was very confused, but I decided to follow my instinct. I realized that my time is so limited to listen to the unlimited amount of music that is on Tidal. Don't think, feel. 🙂
Picture added.
Attachments
Last edited:
My personal feelings are this:
DSP controlled speakers are slightly different to the norm. The level of control is such, that speaker placement, listening position and room modes can to some level be accounted for and controlled within the EQ and crossovers.
I too usually start with a 1m on axis measurement as my baseline certainly above 500hz. But for the woofer and sub-woofers, I individually EQ each driver according to multiple listening window measurements in situ. There are obviously pitfalls to this process and EQ has to be carefully applied - such as cut modal peaks but do not attempt to boost dips etc. etc. but this helps to address baffle step and room modes.
I also believe after experimentation that polar response and therefore direct vs indirect energy needs to be carefully accounted for too - listening distance and placement of speakers of course affects this too along with room treatment/furnishings.
None of this is easily achievable without DSP control of course.
I also feel the use of Listening Window Response, Spinorama and Sound Power as pioneered and advocated by Dr. Floyd Toole is one of the best evidence based scientific approaches to speaker measurement and voicing and is in large what I try to do.
If you are unfamiliar with Spinoramas and their use the link below is quite basic but gives quite a nice general background on it all.
Loudspeaker Myths: Anechoic Chambers , the NRC and Flat Frequency Response | Audioholics
DSP controlled speakers are slightly different to the norm. The level of control is such, that speaker placement, listening position and room modes can to some level be accounted for and controlled within the EQ and crossovers.
I too usually start with a 1m on axis measurement as my baseline certainly above 500hz. But for the woofer and sub-woofers, I individually EQ each driver according to multiple listening window measurements in situ. There are obviously pitfalls to this process and EQ has to be carefully applied - such as cut modal peaks but do not attempt to boost dips etc. etc. but this helps to address baffle step and room modes.
I also believe after experimentation that polar response and therefore direct vs indirect energy needs to be carefully accounted for too - listening distance and placement of speakers of course affects this too along with room treatment/furnishings.
None of this is easily achievable without DSP control of course.
I also feel the use of Listening Window Response, Spinorama and Sound Power as pioneered and advocated by Dr. Floyd Toole is one of the best evidence based scientific approaches to speaker measurement and voicing and is in large what I try to do.
If you are unfamiliar with Spinoramas and their use the link below is quite basic but gives quite a nice general background on it all.
Loudspeaker Myths: Anechoic Chambers , the NRC and Flat Frequency Response | Audioholics