Member
Joined 2009
It dawns on me now that "midrange" is a very general term and people have different ideas of what that encompasses. The size of the midrange will more or less be dictated by the design needs on the low end. The greatest benefit to this thread will probably be look up chart that shows how low and loud different size drivers can go and at what excursions. It would be the starting point for choosing a driver for a given range.
Member
Joined 2009
Thank you for running the tests. I did not think that small midranges suffer from dynamic compression above 300Hz at normal levels and having the numbers helps a lot.
It all depends on what you have in mind.
The JBL4341 uses a 10" for mids. Working bandwidth is 300-1250Hz.
There are many others that use smaller drivers.
Attachments
Member
Joined 2009
Simon,
Thank you for taking the time to contribute such an in-depth analysis.
I generally agree with everything you talk about but I get the feeling that you stretch some issues a bit in order to hit the 300-3000Hz range with an 8" driver. For example it's obvious from your directivity plots that an 8" driver mates perfectly with a wide dispersion tweeter at 1kHz and a narrow one at 2kHz. Yet you talk about crossing over an octave further, why?
I've heard the guideline to keep crossovers out of the 300-3000Hz region but you can't ignore so many successful designs that have a crossover right in the middle of that range. I get the idea that the audibility of a bad crossover is largely contributed by mismatched directivity. I understand that covering this range with a small driver will be limited by the target SPL level. I think an interesting design view to consider is using the sub and the tweeter to their maximum range and then have the mid be a filler to smooth out the transition.
Thank you for taking the time to contribute such an in-depth analysis.
I generally agree with everything you talk about but I get the feeling that you stretch some issues a bit in order to hit the 300-3000Hz range with an 8" driver. For example it's obvious from your directivity plots that an 8" driver mates perfectly with a wide dispersion tweeter at 1kHz and a narrow one at 2kHz. Yet you talk about crossing over an octave further, why?
I've heard the guideline to keep crossovers out of the 300-3000Hz region but you can't ignore so many successful designs that have a crossover right in the middle of that range. I get the idea that the audibility of a bad crossover is largely contributed by mismatched directivity. I understand that covering this range with a small driver will be limited by the target SPL level. I think an interesting design view to consider is using the sub and the tweeter to their maximum range and then have the mid be a filler to smooth out the transition.
The impedance is not a problem at all because you wire the drivers "series/parallel" such that the final impedance seen by the power amp is still 8 ohms (or 4 ohm if you choose that model). Also, the comb filter effect from more than one physically displaced drivers putting out the same thing is one thing (the more drivers, thoe more cancellations get filled in), but I'm at least as concerned about cancellations from room reflections. This is where the controlled directivity on the vertical axis becomes a huge plus with a vertical line array. And you want a narrow transducer so lower midrange imaging cues aren't blurred.
I appreciate you taking the time to run some numbers, but there's a few things other than just peak excursion you're overlooking, including:
Crest factor of music - not very high these days to be fair, but there was a time a couple of decades ago where the crest factor (average to peak levels) was as high as 15dB. (Try looking at the waveform envelope of Pink Floyd's The Wall
)That means for an average playback level of 90dB there would be brief transients peaking as high as 105dB. If the driver can't reproduce those high level transients, even for a fraction of a second, the dynamic range will be compressed on the peaks. The music will lose "attack", and its often not obvious until comparing to a speaker that isn't compressing.
Headroom - just because a driver can reach a certain SPL and stay within its Xmax doesn't mean that it can do so without dynamic range compression, especially in the midrange where excursion isn't generally the dominant limitation. If you want to reproduce a certain peak SPL with little or no compression you generally want a driver that can produce about 10dB more than your expected requirements.
Distortion - reaching a target SPL (just) with a single tone doesn't mean that it can do so with low distortion, particularly 2nd harmonic and IM distortion when multiple frequencies are present.
There is both AM IM distortion and FM IM distortion. AM distortion can be reduced with better motor designs and suspension linearity but FM distortion is fundamentally limited by excursion, as it's due to Doppler effect.
For a given excursion at low frequencies, FM distortion increases as the second frequency gets higher - so even a 1mm excursion at 300Hz introduces significant Doppler distortion at 3Khz. The only way to reduce it is use larger cone area, thus achieving the same SPL with lower excursion.
It's easy to overlook out of band sources of excursion too - if you have a 12dB/oct high pass (as often used) at 300Hz, excursion remains constant with frequency below cut-off until you reach the mechanical resonance of the driver, so if it takes 1mm excursion to produce 95dB SPL at 300Hz, the same drive level applied at 150Hz before the high pass filter will still produce 1mm of excursion, even though the response is 12dB down and below the usable range of the driver. (Excursion potential is being wasted by the inaudible part of the drivers response, as the woofer will be dominating there)
Strong bass signals even though significantly attenuated by the high pass filter can still cause significant excursions of the midrange driver unless you use at least an 18dB/oct slope. This all adds to the excursion requirements and makes intermodulation distortion worse than the signal requirements of the 300-3000Hz band would seem to indicate.
Other factors are that small drivers of the vicinity of 4" tend to be very inefficient. Your example is 86.5dB/2.83v/m, while an 8" full range driver can typically be in the 93-95dB/2.83v/1m range.
This means that for 95dB SPL you have either a small 4" driver dissipating 8 watts, or a large 8" driver with larger voice coil and magnet dissipating about 1.7 watts, but which is probably capable of more dissipation than the small driver.
Higher sensitivity from a bigger driver means less heating of the voice coil, less flux modulation, (one of the major sources of distortion in the midrange) and the lower excursion means less L(e) modulation, another large source of midrange distortion, as well as generally less 2nd harmonic distortion.
So yes, a small driver can produce "typical" listening levels at one frequency at a time, but play actual music on it consisting of many frequencies, excursion coming from out of band bass signals, voice coil heating, flux modulation and L(e) modulation, Doppler distortion, and high level transients, and a 4" driver won't cope in a large 3 way design. A good 8" driver in the same conditions would sound effortless.
Last edited:
well, in music midrange is more like 160-1600 Hz, just as bass is 16-160 Hz and highs 1600-16000 Hz
and here is midrange driver that - quite against laws of physics
Mms around 2g or so and Sd around 2000 or so
extremely high directivity though - so for fans of total direct sound only
The high directivity would be expected much like a large planar driver.
I wonder what your results would be if you tried a segmented crossover section with the rows of drivers,LIke what is done with a large segmented ESL driver?
This is a very interesting thread as I have been just recently doing some major studying on this subject lately.
I have found some documents that might help to put some light into this subject.
I will try to get them together later tonight as I have the PDF's, But, I lost the links as to were I found them.
In my situation my room is 10.5' X 18.5' and currently with no wall treatment due to the reflections Left and Right appear to be reversed when they are not.
Right now I am planning on a 4' ESL to target the midrange and up.
I am debating two width sizes, A 4" width will allow me to build two 8' long panels and a 5.5" width will allow me less waste of my mylar but will only allow me two 4' long panels with the structure material that I have on hand.
And the difference of the width may provide me a much narrower beam at 15khz and above than I care to have, Even though the seperation of the panels will be about 66" and I am 12' to 14' feet away.
I am in the process of drawing up a map of my room in order to find out where the sweet spot is going to be.
I am debating if it will be worth the while to build a 1" to 2" wide strip as a tweeter panel to combat this dispersion problem.
And go with the 5.5” width for a midrange.
When I first got my panels working I favored my little 3.75" wide one more than my 8.75" wide one as far as the beaming is concerned and now I know why!
The little panels handle the 300hz range nicely but at 200hz they would start max out.
When they got down to 200hz that seemed to be the limit, Due to driver clipping.
They would benefit from the wider 5.5" width, except for a much narrower
beam above 15khz. Hmmmmm...........
I have been looking for a direcivity calculator,But all I could find are some sample plots and I will post those later.
Anyway I will keep following this thread and if I can come op with any thing to help I will certainly post it.
Cheers ! jer
I'm a huge fan of having woofers in separate boxes so they can be independantly positioned in the room for best response at the listening position. I find this to be a substantial improvement in most cases. Because of that, when I go to choose a midrange driver, the question becomes "How low in frequency do I want the "satalite" speakers (everything non bass) to go?" I analyzed the talking voice frequency of me and a few friends and found that we typically talked with a fundamental frequency of about 111HZ. I therefore felt that the satalite speakers that create the sense of sound stage should go at least that far down. So I picked 100HZ. I figured four five inch drivers in a vertical line array on each side could handle 100HZ on up to where they become more directional than I wanted at around 1.4kHZ, where a tweeter that I really like was able to take it from there on up to 25kHZ. My finished system blows my mind when I put on good program material. Choosing a midrange driver without taking into account how it will contribute to the system as a whole seems foolish.
here is directivity simulation:
yes, especially one can try delay lines as in Quad ESL
Last edited:
Late but totally agree with the first long post, long but very pedagogical
And again totally agree with this follows, it means listen to music with attention
And again totally agree with this follows, it means listen to music with attention
I appreciate you taking the time to run
some numbers, but there's a few things other than just peak excursion you're overlooking, including:
Crest factor of music - not very high these days to be fair, but there was a time a couple of decades ago where the crest factor (average to peak levels) was as high as 15dB. (Try looking at the waveform envelope of Pink Floyd's The Wall
That means for an average playback level of 90dB there would be brief transients peaking as high as 105dB. If the driver can't reproduce those high level transients, even for a fraction of a second, the dynamic range will be compressed on the peaks. The music will lose "attack", and its often not obvious until comparing to a speaker that isn't compressing.
Headroom - just because a driver can reach a certain SPL and stay within its Xmax doesn't mean that it can do so without dynamic range compression, especially in the midrange where excursion isn't generally the dominant limitation. If you want to reproduce a certain peak SPL with little or no compression you generally want a driver that can produce about 10dB more than your expected requirements.
Distortion - reaching a target SPL (just) with a single tone doesn't mean that it can do so with low distortion, particularly 2nd harmonic and IM distortion when multiple frequencies are present.
There is both AM IM distortion and FM IM distortion. AM distortion can be reduced with better motor designs and suspension linearity but FM distortion is fundamentally limited by excursion, as it's due to Doppler effect.
For a given excursion at low frequencies, FM distortion increases as the second frequency gets higher - so even a 1mm excursion at 300Hz introduces significant Doppler distortion at 3Khz. The only way to reduce it is use larger cone area, thus achieving the same SPL with lower excursion.
It's easy to overlook out of band sources of excursion too - if you have a 12dB/oct high pass (as often used) at 300Hz, excursion remains constant with frequency below cut-off until you reach the mechanical resonance of the driver, so if it takes 1mm excursion to produce 95dB SPL at 300Hz, the same drive level applied at 150Hz before the high pass filter will still produce 1mm of excursion, even though the response is 12dB down and below the usable range of the driver. (Excursion potential is being wasted by the inaudible part of the drivers response, as the woofer will be dominating there)
Strong bass signals even though significantly attenuated by the high pass filter can still cause significant excursions of the midrange driver unless you use at least an 18dB/oct slope. This all adds to the excursion requirements and makes intermodulation distortion worse than the signal requirements of the 300-3000Hz band would seem to indicate.
Other factors are that small drivers of the vicinity of 4" tend to be very inefficient. Your example is 86.5dB/2.83v/m, while an 8" full range driver can typically be in the 93-95dB/2.83v/1m range.
This means that for 95dB SPL you have either a small 4" driver dissipating 8 watts, or a large 8" driver with larger voice coil and magnet dissipating about 1.7 watts, but which is probably capable of more dissipation than the small driver.
Higher sensitivity from a bigger driver means less heating of the voice coil, less flux modulation, (one of the major sources of distortion in the midrange) and the lower excursion means less L(e) modulation, another large source of midrange distortion, as well as generally less 2nd harmonic distortion.
So yes, a small driver can produce "typical" listening levels at one frequency at a time, but play actual music on it consisting of many frequencies, excursion coming from out of band bass signals, voice coil heating, flux modulation and L(e) modulation, Doppler distortion, and high level transients, and a 4" driver won't cope in a large 3 way design. A good 8" driver in the same conditions would sound effortless.
Very Cool,Graaf!!
What software is that you are using?
jer
the picture is courtesy of diyaudio user bjorno
Simon,
It follows from your addendum to my calculations we simply can rule any 2" dome plus also the somewhat larger 3" stuff such as ATC, VIFA DM 75 and even most well established typical 4" mids including hi End stuff like Thiel Ceratec etc.
I have a feeling here not everybody will agree with your + 15db= 105dB peaks. Furthermore I am not so sure short loud passages will cause easily audible distortion.
But let us see how others view this issue.
Eelco
It follows from your addendum to my calculations we simply can rule any 2" dome plus also the somewhat larger 3" stuff such as ATC, VIFA DM 75 and even most well established typical 4" mids including hi End stuff like Thiel Ceratec etc.
I have a feeling here not everybody will agree with your + 15db= 105dB peaks. Furthermore I am not so sure short loud passages will cause easily audible distortion.
But let us see how others view this issue.
Eelco
Member
Joined 2009
Let me first thank Simon(DBMandrake) for the exhaustive explanations. I think that I have a good understanding of the subject and yet your comments are very informative and an enjoyable read. In all seriousness you have a talent for teaching and you should probably look for an opportunity to write about audio concepts to a wider audience if you don't already.
-------------
I think this thread is inevitably going to hit the question of "how loud do you listen at home?". It's a personal question and everyone is entitled to their own opinion so I don't think it's worth looking for a definite answer. I will use myself as an example to make a point and not by any means to impose a target SPL level on other people. My girlfriend likes "listening" close to maximum and doesn't seem to hear when I yell at her that those levels can blow out the drivers.
I have an idea at what level the music from my system begins to sound distorted. I value clarity and my usual listening level is somewhere under that point. Interestingly, the living room starts sounding too live before I can detect any distortion from the speakers. I have no problem trading some dynamic range compression at high volume for working out other design issues because I can't hear that very detailed crescendo with my windows rattling.
-------------
I think this thread is inevitably going to hit the question of "how loud do you listen at home?". It's a personal question and everyone is entitled to their own opinion so I don't think it's worth looking for a definite answer. I will use myself as an example to make a point and not by any means to impose a target SPL level on other people. My girlfriend likes "listening" close to maximum and doesn't seem to hear when I yell at her that those levels can blow out the drivers.
I have an idea at what level the music from my system begins to sound distorted. I value clarity and my usual listening level is somewhere under that point. Interestingly, the living room starts sounding too live before I can detect any distortion from the speakers. I have no problem trading some dynamic range compression at high volume for working out other design issues because I can't hear that very detailed crescendo with my windows rattling.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.