Well your mention of a 94dB mid tweets confused me then !
In this case only the mid needs to be 94dB to be BSC'd
back to 88dB, the tweeter can be a normal 88dB to 90dB.
What your really saying is for two XLS per side the main
speakers would need to be MTM 2way with BSC or TMM
2.5 ways (BSC'd by definition).
And for one XLS per side a BSC'd MT is all that is needed.
This is presuming passive crossovers of course,
sreten.
In this case only the mid needs to be 94dB to be BSC'd
back to 88dB, the tweeter can be a normal 88dB to 90dB.
What your really saying is for two XLS per side the main
speakers would need to be MTM 2way with BSC or TMM
2.5 ways (BSC'd by definition).
And for one XLS per side a BSC'd MT is all that is needed.
This is presuming passive crossovers of course,
sreten.
Yes that is what I was saying, he wants to use two XLS per speaker, I dont think he wanted to use an MTM or similar. I was pointing out you dont want to have the bass more sensative then the rest with regards to massive waste of power. Unless he were to use one of the focal mids which are super sensative and cross it at its lower limit, two XLS would most probably be more sensative then the mid/tweet assembly.
Yes you are right the tweeter doesnt need to be 94dB sens a normal standard tweet will be fine.
Philip it looks as if you may be better off considering one XLS per cabinet with a matchin 12"XLS passive radiator which will end up being cheaper then two XLS per cab. Two XLS in a ported enclosure will get tricky as the port length will be huge, that graph I produced in the other post didnt factor in the additional port diameter required for two XLS. This would again increase port length making it extreamely hard to work with.
I dont know how much you were thinking of spending on the rest of the system but I doubt it was as much as the expensive focal mid drivers as you were concerned with funding.
It would help considerably if you would tell us what you had in mind for the rest of the system. Yes I realise I have said this before.
Yes you are right the tweeter doesnt need to be 94dB sens a normal standard tweet will be fine.
Philip it looks as if you may be better off considering one XLS per cabinet with a matchin 12"XLS passive radiator which will end up being cheaper then two XLS per cab. Two XLS in a ported enclosure will get tricky as the port length will be huge, that graph I produced in the other post didnt factor in the additional port diameter required for two XLS. This would again increase port length making it extreamely hard to work with.
I dont know how much you were thinking of spending on the rest of the system but I doubt it was as much as the expensive focal mid drivers as you were concerned with funding.
It would help considerably if you would tell us what you had in mind for the rest of the system. Yes I realise I have said this before.
If he parallels two XLS drivers, then he needs an amp capable of driving a 2 or 2.5 ohm load.
Unless he has such an amp, he would be better off hooking the two XLS drivers up in series and then going 8 ohm for the rest of the system. This would bring the woofer part of the system down to 88 dB sensitivity.
Unless he has such an amp, he would be better off hooking the two XLS drivers up in series and then going 8 ohm for the rest of the system. This would bring the woofer part of the system down to 88 dB sensitivity.
I was going to mention something about the strugle an amp will have of driving parallel XLS but I assumed philip would already know about this and have looked at the drivers impedance. However this may not be the case and is indeed worth pointing out. I thought the XLS were 8ohm nominal as per peerless so I went and looked at the site to see, they do indeed swing about a bit, however the range in which they are used is mostly under 8ohms (not including 20-30hz booms as per HT) so using them in series is not a bad idea. A passive xover is going to be used on these too so having a higher overall impedance is not a bad thing because it gives more leeway when regarding what effects the xover could have on the impedance.
I still think a single XLS is a better bet partnered with a passive radiator. Most people build one box, XLS and passive radiator for sub duty and dont have problems filling the room. I dont know how big philips room is but four XLS is going to fill one damned huge large space and probably require a biggish one to sound right. To me it just doesnt make sense to use four XLS when two will work well, unless you need to fill a massive room.
If you were going to use passive radiators to port the two XLS then I wouldnt be as bashing of the idea, but ports are going to be hard!
Taking 15-20m/s to be the limit for the speed of air in a port to minimise chuffing you need a port roughly 13cm in diameter this needs to be 134cm long to tune it correctly in the aforementioned box in philips other thread of 54 litres.
I still think a single XLS is a better bet partnered with a passive radiator. Most people build one box, XLS and passive radiator for sub duty and dont have problems filling the room. I dont know how big philips room is but four XLS is going to fill one damned huge large space and probably require a biggish one to sound right. To me it just doesnt make sense to use four XLS when two will work well, unless you need to fill a massive room.
If you were going to use passive radiators to port the two XLS then I wouldnt be as bashing of the idea, but ports are going to be hard!
Taking 15-20m/s to be the limit for the speed of air in a port to minimise chuffing you need a port roughly 13cm in diameter this needs to be 134cm long to tune it correctly in the aforementioned box in philips other thread of 54 litres.
Attachments
This is an interesting discussion. Something I've been thinking about lately is the relation between system Q, vented vs. ported, driver quality and room response. For instance, I recently finished a North Creek kit using SS 18W/8545 and I've found that the bass, although not capable of real HT levels, has the perfect balance of tightness vs. punchiness for my tastes and works great for both movies and music. The interesting thing is that it sounds much different than my (now sold) shiva 85L ported sub. The shiva seemed not as tight but at the same time it didn't have the same punch for movies, although it could obviously go much louder. Eventually I would like to build a new sub with the same characteristics as my main speakers, but I can't figure out what factors really control the sound I'm hearing. I thought the 'punchiness' might be related to my room effected response but my speakers measure close to flat from 30 to 90 hz in room. Since both my old sub and my speakers were/are ported I'm not sure that's it either. Is it as simple as system Q (my sub was .7) or does the driver itself play a major factor in this? I've also thought about using a tumult in the system but I'm worried that it might actually be 'too clean' if you know what I mean. I'm not sure though if that ultimately has an affect on punchiness or not. Maybe this will just come down to buying a woofer and trying out two or three different boxes/alignments and see what matches the closest. Anyone have any comments?
George
George
Wright said:
Classic driver design for sealed or reflex box usage indicates a
Q of 0.7 should be used sealed, and a reflex version would be
~ twice the box volume of the sealed version.
Personally I prefer overdamped reflex alignments to classically
flat. In the above case detune the port by ~ x 0.7, which
gives a response more suited to room gain, the white line
is ~ 6dB octave. Sealed volume for Q = 0.7 = 11.5L.
(Q of 23L sealed box = 0.55, 0.6 is a good maximum for reflex)
I presume the major influence of a subwoofer driver is distortion.
sreten.Attachments
Sensitivity changes?
Hey guys, thanks for all the posts and the interests. As I say, driver selection is still up in the air a bit, but what you guys were saying about sensitivity is really interesting. My mid and tweeter would be in the 90's, but it looks like the woofers I might use would be in the high 80's. I was trying to follow the sensitivity discussion but got a bit lost. What were you guys saying about how sensitivity changes with two drivers? With Baffle Step Correction (BSC?)? What are these two phenomena? Thanks,
Philip
Hey guys, thanks for all the posts and the interests. As I say, driver selection is still up in the air a bit, but what you guys were saying about sensitivity is really interesting. My mid and tweeter would be in the 90's, but it looks like the woofers I might use would be in the high 80's. I was trying to follow the sensitivity discussion but got a bit lost. What were you guys saying about how sensitivity changes with two drivers? With Baffle Step Correction (BSC?)? What are these two phenomena? Thanks,
Philip
I think the driver is the most important part of percieved sound. I started building speakers 33 years ago, in the beginning without that much knowledge, good teachers thou. Learnt a few things on the way, mainly related to horns and tqwt but also a lot regarding drivers, have built own, both permanent and fieldcoil magnets. I think that as long as the enclosure design is not totally wrong, nothing has more influence on the sound than the driver itself. The Gale 401 for instance sounds just like an AR MST
but with even more definition in the bottom end, uses the same AR 8" bass/mid but two pcs. Most Brittish Spendor/Kef clones with bextrene bass units sound exactly the same, extended but slow and maybe even dull. In the Altec "voice of the theatre" horn there´s a huge difference between original drivers and modern replacements/replicas (actually the fieldcoil original imo is best of all).
OK here goes with the BSC phenomenon. Look back to copy from previous post!
Basically a loudspeaker radiates every sound it produces in the high frequencies forwards from the drive units out to the listener. But at lower frequencies the sound starts to radiate round the back of the enclosure. If you have listened to a loudspeaker from behind you may notice that you dont get as much high frequencies the lows, this is because of this.
What happens is above certain frequencies the speaker radiates 2pi and at others 4pi, when this happens is determined by the width of the front baffle. As the wavelength of high frequencies is small in comparison to the baffle width they all fly foward. But as the wavelength goes beyond the baffle width the sound starts to "Creep" and bend round the sides of the cabinet. This means you get some sound going direct to the listener and some dissapearing round the back of the speaker.
http://www.seas.no/excel_line/excel/e0015.pdf
look at this data sheet from seas, it highlites BS brilliantly.
As you can see here the frequency response starts to decrease in SPL at about 800hz, which is odd because this is a bass driver and can easily operated below this keeping up the SLP. The driver can operated lower then this before the roll off in the bass, look further along at about 150hz, you'll notice that the response evens out again, then rolls off as you would expect at about 100hz. This roll off between 150 and 800hz is due to baffle step.
Bare in mind that theoretically BS gives a slow roll off to the tune of 6dB from start to finish. In the real world however you get some gain from the room so the full 6 isnt normally required, usually 3dB will suffice.
Sreten then added..
Close but no cigar !
You just needed to add why the Seas plots show this effect whilst
other plots don't. Seas tested drivers in real boxes in free space
(as you know) (I used the past tense because they seem to have
stopped doing it) whilst other manafacturers use a large flat baffle.
(and that with 3db of BSC = forced close to wall mounting,
and 6dB of BSC = free space mounting away from walls)
Sooooo taking all that into acount. If you intend to use your speakers in free space then 6dB will be required. So your 90dB tweet mid will be lowered by 6dB to 84dB. Because you would cross over low to the bass driver BS would have already taken place, so its passband would be 6dB lower then the sensativity quoted by peerless. This means the XLS will play at 82dB, this would be two dB lower then the mid/tweet so not much more padding would be required to get this flat.
However if you were to use two XLS they would play 6dB louder so would be 88dB instead of 82, this is four dB louder then the mid and tweet. Padding bass is not a good idea, dont even think about doing it massive waste of power.
I hope thats clearer.
Basically a loudspeaker radiates every sound it produces in the high frequencies forwards from the drive units out to the listener. But at lower frequencies the sound starts to radiate round the back of the enclosure. If you have listened to a loudspeaker from behind you may notice that you dont get as much high frequencies the lows, this is because of this.
What happens is above certain frequencies the speaker radiates 2pi and at others 4pi, when this happens is determined by the width of the front baffle. As the wavelength of high frequencies is small in comparison to the baffle width they all fly foward. But as the wavelength goes beyond the baffle width the sound starts to "Creep" and bend round the sides of the cabinet. This means you get some sound going direct to the listener and some dissapearing round the back of the speaker.
http://www.seas.no/excel_line/excel/e0015.pdf
look at this data sheet from seas, it highlites BS brilliantly.
As you can see here the frequency response starts to decrease in SPL at about 800hz, which is odd because this is a bass driver and can easily operated below this keeping up the SLP. The driver can operated lower then this before the roll off in the bass, look further along at about 150hz, you'll notice that the response evens out again, then rolls off as you would expect at about 100hz. This roll off between 150 and 800hz is due to baffle step.
Bare in mind that theoretically BS gives a slow roll off to the tune of 6dB from start to finish. In the real world however you get some gain from the room so the full 6 isnt normally required, usually 3dB will suffice.
Sreten then added..
Close but no cigar !
You just needed to add why the Seas plots show this effect whilst
other plots don't. Seas tested drivers in real boxes in free space
(as you know) (I used the past tense because they seem to have
stopped doing it) whilst other manafacturers use a large flat baffle.
(and that with 3db of BSC = forced close to wall mounting,
and 6dB of BSC = free space mounting away from walls)
Sooooo taking all that into acount. If you intend to use your speakers in free space then 6dB will be required. So your 90dB tweet mid will be lowered by 6dB to 84dB. Because you would cross over low to the bass driver BS would have already taken place, so its passband would be 6dB lower then the sensativity quoted by peerless. This means the XLS will play at 82dB, this would be two dB lower then the mid/tweet so not much more padding would be required to get this flat.
However if you were to use two XLS they would play 6dB louder so would be 88dB instead of 82, this is four dB louder then the mid and tweet. Padding bass is not a good idea, dont even think about doing it massive waste of power.
I hope thats clearer.
Re: Sensitivity changes?
Baffle step compensation you can search and find out about.
Matching of you units capabalities in only a major issue if you
are planning a passive crossover bass to mid, if its a active
then its basically a non-issue except in determining the power
ratings required in the amplifiers for the bass and mid sections.
sreten.
philip said:
Baffle step compensation you can search and find out about.
Matching of you units capabalities in only a major issue if you
are planning a passive crossover bass to mid, if its a active
then its basically a non-issue except in determining the power
ratings required in the amplifiers for the bass and mid sections.
sreten.
Re: Sensitivity changes?
The way most speakers work is, they are driven by a certain voltage and then they take as much current as they need at any given frequency. Current requirements vary frequency to frequency-voltage requirements do not.
Take a 4 ohm driver that individually plays at 90 dB at 2.8 volts, (not one watt).
Now take an identical driver and hook it up in series with the original. The series pair will still play at 90 dB at 2.8 volts.
Now disconnect the second woofer, and re-hook it up in parallel this time. The parallel pair will play at 96 dB at 2.8 volts input.
Parallel hookup yields 6 dB increased sensitivity to voltage. It increases the amp's current needs, but that is another story.
philip said:
The way most speakers work is, they are driven by a certain voltage and then they take as much current as they need at any given frequency. Current requirements vary frequency to frequency-voltage requirements do not.
Take a 4 ohm driver that individually plays at 90 dB at 2.8 volts, (not one watt).
Now take an identical driver and hook it up in series with the original. The series pair will still play at 90 dB at 2.8 volts.
Now disconnect the second woofer, and re-hook it up in parallel this time. The parallel pair will play at 96 dB at 2.8 volts input.
Parallel hookup yields 6 dB increased sensitivity to voltage. It increases the amp's current needs, but that is another story.
Headroom
Wright,
In terms of perceived "punch", it is to my way of thinking not entirely about quantity of low bass, such as the Shiva design, or even the Q, though I agree with the notion that slightly overdamped bass alignments generally sound better, display less overhang, etc.... It is about dynamics and headroom.
It is about how the high energy of a kick drum, for example, is delivered in a convincing manner. Not an easy task for any system. Best guess for a kick drum is 100 - 120 Hz for the fundamental. Right at the sub crossover or just above.
To my mind, two things need to occur for punch to really be felt, and the full spectrum heard. You must reproduce the sound with very little compression, and you must be able to reproduce the higher freq portion of the drum strike spectrum without screwing up the phase, if going through a crossover.
Take the usual 88 dB midbass cruising along to produce 85 dB at the listening position on, say, 15 watts. Suddenly, a piano chord is hammered or a drum kicked, and you need to reproduce a 15 dB peak.
It is very doubtful the usual 150 watt amp will deliver it, or the 88dB driver could handle it if it did (double the power for every additional 3dB of sound level x 5 = 480 watts). So, while the transient may be short in duration, compression or distortion usually result. Use 95 dB efficient drivers and you only need 120 watts for 101 dB peak in the same example, which is quite possible. You get punch.
All of the best systems I have heard have headroom, and don't sound loud when they are. If the distortion isn't there to give it away, you just get big sound with stunning clarity.
Incidentally, a DVD can deliver 6 dB/bit at 24 bit, or 144 dB in dynamics. Naturally, 85 dB speakers will just not reproduce an orchestra or full scale rock with anything close to realistic dynamics unless you can deliver huge power to them and they can take it. Start with high efficiency drivers. Less heat, less compression, lower distortion (with good quality), and the amps come cheaper as a result.
Tim
Wright,
In terms of perceived "punch", it is to my way of thinking not entirely about quantity of low bass, such as the Shiva design, or even the Q, though I agree with the notion that slightly overdamped bass alignments generally sound better, display less overhang, etc.... It is about dynamics and headroom.
It is about how the high energy of a kick drum, for example, is delivered in a convincing manner. Not an easy task for any system. Best guess for a kick drum is 100 - 120 Hz for the fundamental. Right at the sub crossover or just above.
To my mind, two things need to occur for punch to really be felt, and the full spectrum heard. You must reproduce the sound with very little compression, and you must be able to reproduce the higher freq portion of the drum strike spectrum without screwing up the phase, if going through a crossover.
Take the usual 88 dB midbass cruising along to produce 85 dB at the listening position on, say, 15 watts. Suddenly, a piano chord is hammered or a drum kicked, and you need to reproduce a 15 dB peak.
It is very doubtful the usual 150 watt amp will deliver it, or the 88dB driver could handle it if it did (double the power for every additional 3dB of sound level x 5 = 480 watts). So, while the transient may be short in duration, compression or distortion usually result. Use 95 dB efficient drivers and you only need 120 watts for 101 dB peak in the same example, which is quite possible. You get punch.
All of the best systems I have heard have headroom, and don't sound loud when they are. If the distortion isn't there to give it away, you just get big sound with stunning clarity.
Incidentally, a DVD can deliver 6 dB/bit at 24 bit, or 144 dB in dynamics. Naturally, 85 dB speakers will just not reproduce an orchestra or full scale rock with anything close to realistic dynamics unless you can deliver huge power to them and they can take it. Start with high efficiency drivers. Less heat, less compression, lower distortion (with good quality), and the amps come cheaper as a result.
Tim
Right , but
Kelticwizard,
You are right that you get no gain or 6 dB, but for a different reasons.
Two drivers in series:
Gain is 3 dB for double the surface area, while electrical loss is 3 dB for higher impedance/lower voltage. Net is zero.
Two in parallel:
The surface area again doubles for 3 db increase; the electrical impedance is halved, voltage goes up, gain is 3 dB also.
Net is 6dB
Tim
Kelticwizard,
You are right that you get no gain or 6 dB, but for a different reasons.
Two drivers in series:
Gain is 3 dB for double the surface area, while electrical loss is 3 dB for higher impedance/lower voltage. Net is zero.
Two in parallel:
The surface area again doubles for 3 db increase; the electrical impedance is halved, voltage goes up, gain is 3 dB also.
Net is 6dB
Tim
Thanks for the input Sreten, Ingvar and Tim; this is interesting stuff. I never realized that efficiency was that critical in a system but what you are saying makes sense. So in terms of gaining punch with a subwoofer, you would look for a high efficiency, low distortion driver with a good sized amp in a slightly overdamped alignment, and the ability to vary the phase to suit the location? I assume the driver should also be capable of sufficiently high output as well. Since I'm not familiar with any high efficiency-high output subwoofer drivers, would a lower efficiency driver still provide punch if driven with sufficient current? The Tumult could handle 1600 watts but the impression I've gotten from other posts is that a sealed tumult with high power would sound very clean and have tons of headroom, but still not have enough 'punch' for some. It is however entirely possible that my impressions are incorrect. Do you have any specific drivers you could recommend that are higher efficiency and still provide high output? I'm thinking I would need a 15" or larger.
George
Edit: I don't want to steal this thead so if this is getting too OT I can start a new thread. Hopefully this is still relevant to the topic at hand.
George
Edit: I don't want to steal this thead so if this is getting too OT I can start a new thread. Hopefully this is still relevant to the topic at hand.
Tim:
I wanted to keep things simple for Philip, so I did not go into efficiency versus sensitivity.
Hook the pair up in either series or parallel, and your efficiency doubles. However, since the way amp/speakers work are that the voltage only follows the input signal, and that the current drawn varies with the varying needs of the loudspeaker at different frequencies, I decided to limit the view to the ability of the speaker produce SPL at a given voltage.
At the same voltage level, when the speakers are hooked in series they will produce the same SPL as if one is hooked up individually. However, they will draw only half the current, and since power equals Volts X Amps, they end up producing the same SPL while drawing only half the power. Hence, their efficiency is doubled-or goes up 3 dB, whichever way you want to state it.
When hooked up in parallel, they produce 6 dB more SPL than one individually, but are drawing twice the power. Hence, their efficiency goes up 3 dB.
So if you hook up a pair, the efficiency is doubled or goes up 3 dB, whether in series or parallel. However, the voltage level must change for this 3 dB to become apparent.
So for the sake of simplicity, I just gave the apparent gains for series and parallel hookups at the same voltage level the single drivers were being fed.
I wanted to keep things simple for Philip, so I did not go into efficiency versus sensitivity.
Hook the pair up in either series or parallel, and your efficiency doubles. However, since the way amp/speakers work are that the voltage only follows the input signal, and that the current drawn varies with the varying needs of the loudspeaker at different frequencies, I decided to limit the view to the ability of the speaker produce SPL at a given voltage.
At the same voltage level, when the speakers are hooked in series they will produce the same SPL as if one is hooked up individually. However, they will draw only half the current, and since power equals Volts X Amps, they end up producing the same SPL while drawing only half the power. Hence, their efficiency is doubled-or goes up 3 dB, whichever way you want to state it.
When hooked up in parallel, they produce 6 dB more SPL than one individually, but are drawing twice the power. Hence, their efficiency goes up 3 dB.
So if you hook up a pair, the efficiency is doubled or goes up 3 dB, whether in series or parallel. However, the voltage level must change for this 3 dB to become apparent.
So for the sake of simplicity, I just gave the apparent gains for series and parallel hookups at the same voltage level the single drivers were being fed.
Re: Headroom
Hi Tim,
just being somewhat pedantic here, though I do agree dynamic
range issues are fundamental to "realistic" reproduction.
For a bass drum if you want the sound of the kicker on bass drum
skin to sound as though the drummer is giving it some with his
right foot you need loads of clean SPL capabability around 30Hz,
for that "woompf" to come through and integrate well with the
rest of the kicker on skin sound.
I used a sub in my system for a bit, other problems aside it did
allow "realisitic" reproduction of lightly played acoustic / jazz
bass drums, but this was very dependent on the sound being
unmolested on the recording, e.g.
"The Trinity Sessions" by the Cowboy Junkies.
sreten.
Tim Moorman said:
Hi Tim,
just being somewhat pedantic here, though I do agree dynamic
range issues are fundamental to "realistic" reproduction.
For a bass drum if you want the sound of the kicker on bass drum
skin to sound as though the drummer is giving it some with his
right foot you need loads of clean SPL capabability around 30Hz,
for that "woompf" to come through and integrate well with the
rest of the kicker on skin sound.
I used a sub in my system for a bit, other problems aside it did
allow "realisitic" reproduction of lightly played acoustic / jazz
bass drums, but this was very dependent on the sound being
unmolested on the recording, e.g.
"The Trinity Sessions" by the Cowboy Junkies.
sreten.About BSC, why not just implement this in an active filter, and use a pot, so you can adjust the frequency and amount?
Someone posted earlier that Bessel alignments have a Q of 0.577... that is not entirely correct, as 0.577 (actually 1/sqrt(3)) is only one of the Q values that satisfy the criterion for Bessel alignment.
For example, according to Rane, a Q of 0.403 or 0.308 will also satisfy these requirements.
An interesting point is that 0.308 will match very well to many rooms. You will note that the Peerless XLS simulation mentioned earlier has a Q of 0.34... That is based on an observation that 0.34 is generally the highest value (in Danish listening rooms, anyway) that will not sound too warm.
So, for the lowest woofer, you might want to go for a high-pass Q of about 0.308, and a low-pass Q of about 0.577..
Addendum: A sealed box with a Q of 0.308 might be unwieldly large, so you might want to build it to a Q of 0.577, and use the Linkwitz transform to bring it down to 0.308
Someone posted earlier that Bessel alignments have a Q of 0.577... that is not entirely correct, as 0.577 (actually 1/sqrt(3)) is only one of the Q values that satisfy the criterion for Bessel alignment.
For example, according to Rane, a Q of 0.403 or 0.308 will also satisfy these requirements.
An interesting point is that 0.308 will match very well to many rooms. You will note that the Peerless XLS simulation mentioned earlier has a Q of 0.34... That is based on an observation that 0.34 is generally the highest value (in Danish listening rooms, anyway) that will not sound too warm.
So, for the lowest woofer, you might want to go for a high-pass Q of about 0.308, and a low-pass Q of about 0.577..
Addendum: A sealed box with a Q of 0.308 might be unwieldly large, so you might want to build it to a Q of 0.577, and use the Linkwitz transform to bring it down to 0.308