Question on loudspeaker design, crossovers, and which driver gets the vocals

0 Members and 1 Guest are viewing this topic. Read 5490 times.

seasterl

  • Jr. Member
  • Posts: 10
I need a little help understanding this design concept, but according to the following website...

http://www.tnt-audio.com/topics/frequency_e.html

the range for the human voice is 87 Hz - 1174 Hz.  This covers the bass up to soprano.  I've checked other websites and they are in good agreement with this frequency range.

What puzzles me is that when you consider the crossover points (and their slopes), it seems that the above frequency range is not covered by the "midrange driver."  In fact, for a typical three-way speakers, this range is usually / mostly covered by the bass driver.  In four-way speakers, it's the mid-woofer.  However, when you play a sound that lies in this range (87 Hz - 1174 Hz), it seems to emanate from the "midrange" driver of the loudspeaker.  This is contradictory to what you'd expect given the speakers specification.  Why is this?  You can hear the sound coming from almost all of the drivers, including the bass driver, but it is much more prominent from the midrange.  Are resistors placed in the circuit to direct the signal to the midrange driver?

I've had some experience in the past using mini monitors and passive subs and a Bryston crossover, but the Bryston does not correct the phase and my system lacked coherence and the subs sounded disconnected.  (So much for a $2500 crossover, huh!)  However, I still like the idea of using a good electronic crossover to divide the frequencies prior to amplification, then use the amp that is, IMO, most appropriate for that range (i.e., SET on vocals and horns, etc., SS on bass) and then use a driver appropriate for that range.  In other words, I'd make sure the vocals are not directed to a bass driver and that low frequencies are not directed to a tweeter.  Are there loudspeakers out there that are setup so you can bi- and tri-amp like this?  If so, what type of crossover are folk using and how do you correct phase and achieve coherence?  Those of you who are doing this, do you plan to one day soon abandon your setup, sell all the gear, and go back to something simpler, like a single integrated amp and pair of full-range loudspeakers?

kfr01

First, the sound seems to come from the midrange driver because of the increasingly directional nature of higher frequencies.  A 100hz note is more directional than a 50hz note, etc.  Second, while the fundamental may be within a certain range, harmonics and silibances often extend to higher octaves.  These, sometimes as much as the fundamental, give a voice its signature.  Third, you'll find that vocals sung (as opposed to spoken) often fall on the higher end of the region you quoted. 


On the subject of active system design.  Try going with a speaker kit and build external crossovers.  Give the speaker a set of binding posts for each set of drivers.  This is what I've done with my Exodus 2641 3-way towers from DIYCable.com.

Frankly, I think if the crossover designer is competent, it won't matter if two speakers are playing the "midrange." Also, I think that starting with drivers that are meant to handle very wide frequency response regions without breakup and distortion in either direction is a very key factor in a highly coherent loudspeaker system.

Example:  the Extremis woofer from Adire.  http://www.adireaudio.com/Home/ExtremisMidwoofers.htm
It is a great mid-woofer for a 3-way system because it can reach low frequencies at high volumes without significant compression and distortion; all the while maintaining a relatively smooth and fast frequency response out to 5khz.  Thus, even if this driver -is- crossed over to a more traditional midrange driver in the key vocal region, it likely won't matter or be distinguishable from the midrange.

Many drivers from Usher, Seas, Scan-speak, etc. exhibit this same sort of versatility.

Try a DIY kit.  GR-Research, DIYcable.com, madisound.com, etc.

eico1

As noted in the chart you reference, those are fundamental frequencies listed. Voice intelligibility requires 300Hz to 3kHz, as basic analog telephone research and practice shows. It seems you can drop the fundamental frequency on things like voice and bass instruments without loosing the "voice" of the instrument, but need an octive or 2 above.

Note that most 2 way designs utilize this fact an place the crossover at 3kHz.

Not a speaker designer,

steve

andyr

I need a little help understanding this design concept, but according to the following website...

http://www.tnt-audio.com/topics/frequency_e.html

the range for the human voice is 87 Hz - 1174 Hz.  This covers the bass up to soprano.  I've checked other websites and they are in good agreement with this frequency range.

What puzzles me is that when you consider the crossover points (and their slopes), it seems that the above frequency range is not covered by the "midrange driver."  In fact, for a typical three-way speakers, this range is usually / mostly covered by the bass driver.


Great question, seaster.   :D

It seems to me the human voice is probably the most critical "instrument" to get right because we are familiar with it the most!   :)   But 80-1,200Hz is a difficult range for a single driver ... in 3-way speakers, many bass drivers cross over to mids at 120-350Hz (eg. "Orions" at 120Hz & Magnepan III/3.X Series at 200-300Hz) and mids can cross over to tweeters at 1,500-3,000Hz (eg. Orion at 1,500Hz, "NaO" at 2,500Hz & Magnepan IIIa at 2,600Hz).

A 2-way speaker might cross over at 2,500Hz but I suggest would not generally go down to much below 100Hz ... so they need a sub to make them "full-range" (turning them into a 3-way!   :D ).

Of course, even if a multi-driver speaker did have a driver covering, say, the 4 octaves from 80-1,300Hz (so the fundamental voice range was covered by it alone), harmonics would be dealt with by another driver ... or two!   :D  And the harmonics are very important ... that's what makes, say, a wooden flute distinguishable from a silver flute when both are playing the same note.


I've had some experience in the past using mini monitors and passive subs and a Bryston crossover, but the Bryston does not correct the phase and my system lacked coherence and the subs sounded disconnected.  (So much for a $2500 crossover, huh!)  However, I still like the idea of using a good electronic crossover to divide the frequencies prior to amplification, then use the amp that is, IMO, most appropriate for that range (i.e., SET on vocals and horns, etc., SS on bass) and then use a driver appropriate for that range.  In other words, I'd make sure the vocals are not directed to a bass driver and that low frequencies are not directed to a tweeter.  Are there loudspeakers out there that are setup so you can bi- and tri-amp like this?  If so, what type of crossover are folk using and how do you correct phase and achieve coherence?


It's easy enough to run a tri-amped active system - I've converted my Magnepan IIIAs to active 3-way operation (analogue crossovers), using the same slopes as the original stock passive crossover.  Like many, though, the 3-way Maggies have their mid-range driver connected out-of-phase ... so bang goes "phase coherence" with that crossover particular regime!!   :o

I suggest the only way you'll get a phase-coherent multi-driver setup is to use 1st order crossovers - so you can have all the drivers connected with the same polarity - which makes it even more difficult to select appropriate drivers (if one is aiming to have the frequency range 80-1,300Hz covered by just one driver).   :(

Or you buy a fancy digital crossover like the brilliant, Oz-designed DEQX - which allows you to play with phase, as well as slopes and cut-off frequencies.

BTW, you said your subs sounded disconnected ... was that because you were crossing over to them too high (ie. not 60-70Hz)?


Those of you who are doing this, do you plan to one day soon abandon your setup, sell all the gear, and go back to something simpler, like a single integrated amp and pair of full-range loudspeakers?

Absolutely not!!   :D

Regards,

Andy


JoshK

Interesting question and good discussion.   Basically most of the points here I'd agree with.   Consider this for pondering...the oboe is an instrument that does not produce the fundamental frequency only its harmonics but the ear hears the notes and fills in the missing fundamental. 

andyr

Interesting question and good discussion.   Basically most of the points here I'd agree with.   Consider this for pondering...the oboe is an instrument that does not produce the fundamental frequency only its harmonics but the ear hears the notes and fills in the missing fundamental. 
Hi Josh,

Is that so?   :o  I'm amazed!  Can you explain the science behind this ... shirley the hollow pipe of an oboe has a frequency at which it can be made to vibrate (by operating the keys to plug various holes) ... and changing keys changes the vibration frequency.  So how come it doesn't produce a fundamental?   :?

Regards,

Andy

JLM

  • Full Member
  • Posts: 10758
  • The elephant normally IS the room
This question hits on a couple of the reasons I use a single driver speaker.  Most "full" range drivers are better termed "extended" range, but they typically do cover 80 - 3,000 Hz.  No phase issues, great coherence, ideal imaging, and automatically provides the advantages (improved dynamics, increased bass response, better amp/driver synergy) of direct amp to driver design.

The best way IMO to use most "full" range drivers is to add a supertweeter (season to taste) and a powered sub to keep the crossovers out of this all important midrange.  Don't skimp on sound quality when shopping for mid-range or extended-range drivers.  I'm amazed at the quality of mid-range drivers in many expensive speakers.

seasterl

  • Jr. Member
  • Posts: 10
Thanks, guys, for all these great posts.  I'm taking it all in.  I realized the issue with the harmonics and distinguishable qualities about midrange voices, and didn't think about the need to extend the range up to about 3000 Hz.  It makes sense to me, and also why many designers will set their crossover point at 2500 Hz - 3000 Hz from midrange to tweeter.  The directional versus nondirectional issues I'm aware of, too. 

Regarding my (previous, now long gong) system when I used the passive subs, I had my crossover point at 70Hz, but forgot which slope I used to blend the response.  Problem:  after I got a good blend, the phase was off about 90 degrees and on some songs it was noticable, and on others it was not.  The subs and amp were so tight and responsive that it was very apparent.  A more relaxed setup and it may not have been.

Getting back to something JLM hit on, it seems to me that the best setup would be a 3-way loudspeaker where a single driver handled all of the midrange, with crossover points at, e.g., 70Hz and 3000Hz with appropriate slopes and phase shifts corrected for by physical offset of drivers.  If this same loudspeaker were also tri-amp'able (I'm not sure if this is a word!), then we could choose a low-powered SS amp for the sub driver, and good SET amp for the midrange, and perhaps an OTL (or whatever) amp for the tweeter.  Taking this one step further, if this same loudspeaker had no crossovers, we could use an active crossover and manually dial in the points and slopes and thus send a limited bandwidth to the amps so the amps don't have to work on the entire bandwidth, but only on a limited frequency range.  (I know that it's easier for an amp to do it's job if it only had to amplify a limited frequency range.)  This is what sounds like the best idea to me.

So what's wrong with this idea?  I guess most folk don't like the idea of tweaking.  The cost of extra amplifiers, interconnects, power cords, and speaker cables will be insurmountable for many listeners.  It will all but put half of the loudspeaker designers / manufacturers out of business.  What if,... we do all this, hook up our favorite source, sit back and listen, and the sound just isn't that good compared to a single integrated amp driving a full-range loudspeaker with built-in passive crossovers?  (You know, there's that issue with a single driver not being able to keep up with complicated musical passages.)  The manufacturers are smart and spent nearly 100% of their time perfecting their product so that they can be better than the alternative approach. 

Opinions?  Comments?


DSK

...it seems to me that the best setup would be a 3-way loudspeaker where a single driver handled all of the midrange, with crossover points at, e.g., 70Hz and 3000Hz with appropriate slopes and phase shifts corrected for by physical offset of drivers.  If this same loudspeaker were also tri-amp'able (I'm not sure if this is a word!), then we could choose a low-powered SS amp for the sub driver, and good SET amp for the midrange, and perhaps an OTL (or whatever) amp for the tweeter.  Taking this one step further, if this same loudspeaker had no crossovers, we could use an active crossover and manually dial in the points and slopes and thus send a limited bandwidth to the amps so the amps don't have to work on the entire bandwidth, but only on a limited frequency range...

That's exactly what a high end 2-way (NOT a mini monitor or small bookshelf) plus high end subwoofer will do. Subwoofer positioning, continuously variable phase dial on the sub, and decent measurement software/equipment will help optimise the phase matching and integration between sub and 2-way. If you are going to cross over at around 70hz, it may as well be to a sub rather than a woofer in the main speaker box. The ability to locate the sub separately can also help reduce room problems. It will also allow a narrower baffle on the main speaker, remove the bass vibrations from the main cabinets and give greater LF extension and weight. Done right it will give a greater sense of effortlessness, scale and power. A purpose designed sealed speaker will often provide a cleaner, smoother integration with the sub. You can use external active crossovers if you desire, or the sub's own LPF and place HPF's on the inputs to the main amp. Approx 50% of amplification is consumed by the range below 300hz.

Beware using different amp topologies for different frequency ranges. They can have different tonal and spatial behavior and cause a lack of coherence, especially when using speakers with steep slopes.

sunshine

  • Jr. Member
  • Posts: 7
May want to check out this website  http://www.bamberglab.com/
He advocates bi/tri-amped speakers with active crossovers.

Bob Reynolds

  • Full Member
  • Posts: 526
That's exactly what a high end 2-way (NOT a mini monitor or small bookshelf) plus high end subwoofer will do. Subwoofer positioning, continuously variable phase dial on the sub, and decent measurement software/equipment will help optimise the phase matching and integration between sub and 2-way. If you are going to cross over at around 70hz, it may as well be to a sub rather than a woofer in the main speaker box. The ability to locate the sub separately can also help reduce room problems. It will also allow a narrower baffle on the main speaker, remove the bass vibrations from the main cabinets and give greater LF extension and weight. Done right it will give a greater sense of effortlessness, scale and power. A purpose designed sealed speaker will often provide a cleaner, smoother integration with the sub. You can use external active crossovers if you desire, or the sub's own LPF and place HPF's on the inputs to the main amp. Approx 50% of amplification is consumed by the range below 300hz.

DSK nailed it. I'll add one other comment -- distortion. I've read several speaker reviews lately that publish distortion figures. Tweeters have distortion like good electronics, i.e., fractions of one percent. However, woofers when pushed to either higher SPLs or lower frequencies produce distortion in the range of 5% - 15% increasing very quickly. Using a sub with a high-pass filter on the main speakers solves that problem cleanly.

I'm using M&K reference monitors + M&K sub + M&K bass management controller. The setup using their test tones is very easy and it's very rare that I can detect the sub is in the room. I only wish I had gone this route years ago.

Duke

  • Industry Contributor
  • Posts: 1160
    • http://www.audiokinesis.com
If you start looking at where would be a good place to put a crossover, in my opinion the 3 kHz ballpark is far from ideal.  Human hearing is at its most sensitive between about 2 and 5 kHz, so anomalies in this region are more likely to be detectable.  I admire designers who can pull off a crossover in that region without audible consequences.

In my opinion the reason a lot of speakers have crossovers around 3 kHz has to do with what's available and easy to use.  A good midwoofer will go up that high, and a good dome tweeter will go down that low.

I used to be a dealer for a 6.5" three-way; that is, the speaker used a dedicated 2" dome midrange even though it could easily have been completed as a two-way.  The midrange dome covered 1 kHz to 3.5 kHz as I recall.  I felt that it outperformed its two-way competition on vocals.  The speaker has been sadly discontinued, but I'll mention it in case you ever see a used pair for sale.  It was the Cliffhanger Bulldog.

One other thing that might be of interest is that the ear's ability to localize a sound source vertically is very much frequency dependent, so a voice can appear to come from the high frequency driver when most of its energy is coming from the low frequency driver.  Below 500 Hz the ear is very poor at vertical localization of a sound source.  Above 1000 Hz the ear starts to get pretty good at judging the vertical height of a sound source.  Vertical imaging acuity peaks at about 4 kHz (right where several other sensitivities peak).   

Duke

bpape

  • Industry Participant
  • Posts: 4465
  • I am serious and don't call my Shirley
    • Sensible Sound Solutions
Good points.  Though the ear is better at vertical localization of sounds from below that of sounds from above.  This is why many people without perf screens in home theater setups prefer the center channel above the screen instead of below - it's much harder to localize

Bryan

mixsit

I wouldn't hesitate for a moment to recommend looking into the SP Tech line. His designs meet these issues head on. High on accuracy, time alignment, controlled dispersion, low distortion, ..
http://www.4sptech.com/version2/mrs.html

Cheers :wink:
Wayne Smith

jules

Duke,

I find this observation most thought provoking ...

Duke:
[/quote]I used to be a dealer for a 6.5" three-way; that is, the speaker used a dedicated 2" dome midrange even though it could easily have been completed as a two-way.  The midrange dome covered 1 kHz to 3.5 kHz as I recall.  I felt that it outperformed its two-way competition on vocals[/quote]

You are suggesting a 2" driver to cover the range from 1KHz to 3.5KHz for vocals. I think there might well be something in this [apart from great dispersion]. Still, it runs counter to pretty much any currently popular system you like to name where this role is covered by drivers around the 6.5 - 7" mark.

Given that our attention is usually drawn to the imposing size of woofers or maybe the cunning of a ribbon tweeter at the  other end, the simplicity of using a relatively cheap 2" dome to deliver the goods in the most important part of our hearing range suggests possibilities for three ways with simpler xovers and reasonable cost.

Interestingly, Aslan Acoustics use a couple of 2" domes to cover the 1kHz to 4.5KHz range. Hmmm ... interesting project somes to mind

Jules 

 

JLM

  • Full Member
  • Posts: 10758
  • The elephant normally IS the room
VMPS builds a standmounted 3-way that uses a single compact ribbon driver that runs 166 - 10,000 Hz.

The standmounted reference 3A speakers use extended range driver with cap protected tweeter.

Tri-amping gets complicated and very expensive.

andyr


Tri-amping gets complicated and very expensive.


As a 3-way Maggie owner, who has converted his Maggies to 3-way active, I'd have to say that the above statement is one of those oft-quoted, supposed "home truths" which have a skerrick of gold surrounded by a slagheap of dross!   :D

I guess the order of "complicated" might go something like this (from simplest through more complex):
*  integrated amp and single-driver speakers
*  integrated amp and passive, multi-driver speakers
*  preamp and active, multi-driver speakers (with in-built power amps - like ATC)
*  preamp, power amp and single driver speakers
*  preamp, power amp and passive, multi-driver speakers
*  preamp, stereo power amps and active, multi-driver speakers
*  preamp, monoblock power amps and active, multi-driver speakers
*  etc.

So, yes, there are more components to plug together in an active setup (except when you have ATCs) but is it really more "complicated"??   :o

And re. "expensive" ... yes, in a 3-way active setup you have to buy 3 stereo power amps but these could be (rated into 8 watts), say:
* 100wpc for the woofer
* 60wpc for the mid-range, and
* 25wpc for the tweeter.

The "rule of thumb" for comparing watts in an active vs. passive setup is that a single amp has to have twice the sum of the ratings of the two amps used in the active setup, to deliver equivalent power ... so for a 3-way active setup, you'd have to have a single stereo amp capable of delivering 370wpc for equivalent passive power to the above active system.

I suggest to get a "big" amp which sounds as good as the 3 smaller amps might, could in fact cost more than the 3 small, cheap amps.

Regards,

Andy