ACD said:
Well that does change things, my approximate simulations
indicate 100W excursion limiting below 200Hz with the
higher sensitivity, though I only guessed at speaker Qts
and a sealed loading of ~ 4 litres.
Theorectically a L/R - 6dB at 200Hz should be fine,
though the amount of juice applied is an issue.
With a c/o frequency of 200Hz all baffle step correction
will be applied to the bass mid, so you need to factor
in a 6dB boost at 200Hz.
Most people shy away from a further c/o frequency, i.e.
a 4 way in the 300 to 3Khz range preferring a single
driver over this range.
If two more "special" PL18's are available I'd use them
for 0.5 way baffle step correction if you don't like a MTM
layout, you could of course go standard MTM 2 way.
🙂 sreten.
Remember the saying "Too many cooks spoil the soup" ?
This thread is taking a simple issue and making it so
complex when it doesn't have to be.
With a fully active setup and a flexible active crossover you
have control over the sound and you can adjust it to your liking
in real time. That is the benefit of a fully active system.
Connect whatever amplifiers you desire to the speakers,
set your active crossover to some generic default values
depening on the drivers, play some music at moderate
levels, then set your crossover output levels for each
driver in such as way that the sound starts to blend well
to your liking. Next, tweak the crossover points to your
liking. Reset levels again if needed.
Use an active crossover that has 18db - 24db slope for
the low frequencies and use 12db - 24db for the top end,
personal preference - you choose.
As far as the midrange driver being crossoved over low?
You can use a higher slope crossover and/or adjust the
levels of the midrange so the driver doesn't get overdriven,
you can play some music and 'feel' the cone excursion with
your hand to get an idea on how much abuse it's taking.
I wouldn't fear using excess power on the top end either.
I can connect midranges and tweeters to 600w amplifiers
on an active setup and not blow up the speakers because
the person controlling the volume knob is trained not to
overdrive the system.
/fun stuff/
This thread is taking a simple issue and making it so
complex when it doesn't have to be.
With a fully active setup and a flexible active crossover you
have control over the sound and you can adjust it to your liking
in real time. That is the benefit of a fully active system.
Connect whatever amplifiers you desire to the speakers,
set your active crossover to some generic default values
depening on the drivers, play some music at moderate
levels, then set your crossover output levels for each
driver in such as way that the sound starts to blend well
to your liking. Next, tweak the crossover points to your
liking. Reset levels again if needed.
Use an active crossover that has 18db - 24db slope for
the low frequencies and use 12db - 24db for the top end,
personal preference - you choose.
As far as the midrange driver being crossoved over low?
You can use a higher slope crossover and/or adjust the
levels of the midrange so the driver doesn't get overdriven,
you can play some music and 'feel' the cone excursion with
your hand to get an idea on how much abuse it's taking.
I wouldn't fear using excess power on the top end either.
I can connect midranges and tweeters to 600w amplifiers
on an active setup and not blow up the speakers because
the person controlling the volume knob is trained not to
overdrive the system.
/fun stuff/
ACD,
well I'll add to the soup anyway (3 posts since I started writing this.... arrgh)
If you add up drivers, you have the opportunity to chose "ideal" drivers for each frequency range. At the same time, you'll have a serious challenge to integrate them well.
For individual drivers, there is more than wattage that limits performance:
- linear excursion and radiating surface determine the maximum "undistorted" (haha) SPL limit of the driver regardless of how many watts the voice coil can take before evaporation. Siegfried Linkwitz has a calculator for that here: http://www.woodartistry.com/linkwitzlab/spl_max1.xls
Note how sensitivity and power don't even come into play.
- whether you can reach this maximum SPL with your amplifier, however, WILL depend on the amp's voltage and amperage capability. Here, Ohm's law is enough to calculate it.
The reason why a bi (tri, multi) amped system needs less combined power than a mono amped one is simple: power=voltage^2/R or R*current^2. The electrical power rises with the square of the usual "hard" limiting factors in amplifiers, namely voltage and current.
Power needs are determined by peak voltage or current before clipping, not by average levels. If you biamp (assuming for now into identical drivers and all frequencies), your peak voltage or current is now 1/2 of the monoamped peak demand. That means your peak power is the quare root of your previous peak power. It therefore falls to 1/4 of the mono amped peak power per bi amp amplifier.
Sanity check: one driver with voltage U gives power U^2/R. Two identical drivers with voltage U/2 each give power (U/2)^2/R each, but have 6 dB less SPL output each since they're at half the voltage. Since you have two of them, you get the 6 dB back, for again identical SPL as the single driver above. Your combined power input, however, was only
(U/2)^2/R+(U/2)^2/R = 2*(U/2)^2/R = U^2/2R
or half the power demand of the mono amped system.
You just saved 3 dB in power output ... for free 😉
Magic, not quite, but close.
Back to your question, the best compromise for undistorted sound could be a 3-way with a doubled up midrange that reaches down to the 100 Hz range. That would make integration easier than a real 4 way system, while preserving good undistorted SPL output capability in the midrange.
Generally speaking, the voice range is around 2-300 to 3000 Hz. The ear's greatest sensitivity is around 2000-4000 Hz. The region inportant for timbre perception is in the lower hundreds. So, whatever you do, you get problems in the 100 to 3000 Hz region:
- use a 4", and you get excursion limit problems at the bottom, or you cross it over at 400 or 500 and you dig into the region where timbre is critically perceived.
- use a 6.5" and you may reach 3000 on axis, but not off axis above say 1500 Hz, that will affect your in-room polar response and you will notice it.
- use a 8" and you must definitely cross over at say 1500 to 2000 Hz at the most, for dispersion reasons, which is close to the ear's highest sensitivity.
You could, however, use doubled up 4" to 6" drivers. You gain the excursion capability at the low end, and retain the good dispersion until you're over the voice range. With 4"ers, you even may use 4 per side without too much vertical lobing problems.
Some notes on the lower end: I recently realized how much the masking effect is real. I had a 4-5 dB broad room (?) resonance peak around 70-150 Hz, which I left unequalized for the longest time. Finally the boom started bothering me (and I am already using dipoles, so there is less boom to start with!). I took out that broad peak. Man, the midrange really cleaned up remarkably. And you would think that 4-5 dB peaks and dips are OK in-room. NOT! if they're broad.
Another tidbit: due to Fletcher Munson, whatever distortion your lower bass creates sounds a lot louder with regards to the fundamental, than it really is. Between say 80 Hz and 500 Hz the ear gains up to 20 dB in sensitivity. In other words, low level but high order distortions created in the bass will appear much louder than they really are, and they will definitely mask all the precision you thought you just gained in the midrange. Or the other way around, you wonder why your midrange is so muddy, when it's the bass's distortion that drowns it.
The secret to good sound may well lie in really clean bass. That means horns or dipoles 😉
well I'll add to the soup anyway (3 posts since I started writing this.... arrgh)
If you add up drivers, you have the opportunity to chose "ideal" drivers for each frequency range. At the same time, you'll have a serious challenge to integrate them well.
For individual drivers, there is more than wattage that limits performance:
- linear excursion and radiating surface determine the maximum "undistorted" (haha) SPL limit of the driver regardless of how many watts the voice coil can take before evaporation. Siegfried Linkwitz has a calculator for that here: http://www.woodartistry.com/linkwitzlab/spl_max1.xls
Note how sensitivity and power don't even come into play.
- whether you can reach this maximum SPL with your amplifier, however, WILL depend on the amp's voltage and amperage capability. Here, Ohm's law is enough to calculate it.
The reason why a bi (tri, multi) amped system needs less combined power than a mono amped one is simple: power=voltage^2/R or R*current^2. The electrical power rises with the square of the usual "hard" limiting factors in amplifiers, namely voltage and current.
Power needs are determined by peak voltage or current before clipping, not by average levels. If you biamp (assuming for now into identical drivers and all frequencies), your peak voltage or current is now 1/2 of the monoamped peak demand. That means your peak power is the quare root of your previous peak power. It therefore falls to 1/4 of the mono amped peak power per bi amp amplifier.
Sanity check: one driver with voltage U gives power U^2/R. Two identical drivers with voltage U/2 each give power (U/2)^2/R each, but have 6 dB less SPL output each since they're at half the voltage. Since you have two of them, you get the 6 dB back, for again identical SPL as the single driver above. Your combined power input, however, was only
(U/2)^2/R+(U/2)^2/R = 2*(U/2)^2/R = U^2/2R
or half the power demand of the mono amped system.
You just saved 3 dB in power output ... for free 😉
Magic, not quite, but close.
Back to your question, the best compromise for undistorted sound could be a 3-way with a doubled up midrange that reaches down to the 100 Hz range. That would make integration easier than a real 4 way system, while preserving good undistorted SPL output capability in the midrange.
Generally speaking, the voice range is around 2-300 to 3000 Hz. The ear's greatest sensitivity is around 2000-4000 Hz. The region inportant for timbre perception is in the lower hundreds. So, whatever you do, you get problems in the 100 to 3000 Hz region:
- use a 4", and you get excursion limit problems at the bottom, or you cross it over at 400 or 500 and you dig into the region where timbre is critically perceived.
- use a 6.5" and you may reach 3000 on axis, but not off axis above say 1500 Hz, that will affect your in-room polar response and you will notice it.
- use a 8" and you must definitely cross over at say 1500 to 2000 Hz at the most, for dispersion reasons, which is close to the ear's highest sensitivity.
You could, however, use doubled up 4" to 6" drivers. You gain the excursion capability at the low end, and retain the good dispersion until you're over the voice range. With 4"ers, you even may use 4 per side without too much vertical lobing problems.
Some notes on the lower end: I recently realized how much the masking effect is real. I had a 4-5 dB broad room (?) resonance peak around 70-150 Hz, which I left unequalized for the longest time. Finally the boom started bothering me (and I am already using dipoles, so there is less boom to start with!). I took out that broad peak. Man, the midrange really cleaned up remarkably. And you would think that 4-5 dB peaks and dips are OK in-room. NOT! if they're broad.
Another tidbit: due to Fletcher Munson, whatever distortion your lower bass creates sounds a lot louder with regards to the fundamental, than it really is. Between say 80 Hz and 500 Hz the ear gains up to 20 dB in sensitivity. In other words, low level but high order distortions created in the bass will appear much louder than they really are, and they will definitely mask all the precision you thought you just gained in the midrange. Or the other way around, you wonder why your midrange is so muddy, when it's the bass's distortion that drowns it.
The secret to good sound may well lie in really clean bass. That means horns or dipoles 😉
Konnichiwa,
I think the gains will be marginal. I'd say listen first. An X-Over point of 500Hz for the PL18 is a bit high (it is a fairly large driver) and there is little difference between (say) 300 & 200Hz X-Over.
I used to do this kind of tradeoff between SPL and lower X-Over on large Pro-Sound systems. Going up an octave from 125Hz to 250Hz did not really give that much more attaiable midrange level, maybe 3 - 4db (the limit BTW where not the Amp's), unless you are in such a situation where the last attainable db SPL counts I'd keep the X-Over ow and live with the fallout.
If you already have problems with midrange level then adding a 4th way is unlikely to fix it (it might improve integration between Subwoofer and Midrange by adding an actual woofer), you would beed more displacement (usually more cone surface) which means either MTM with the same driver (if you can get another Pair) or a different driver (which you do not want to consider).
If you don't have problems with levelk handling in the midrange don't loose sleep. ;-)
You can probably gain a good deal of distortion reduction and removal of compression by feeding the Midrange and Treble from a current output (as opposed to voltage output) amplifier. Then maybe some form feedback around the woofer to stop problems with distortion and compression there and you should be okay.
LF NFB can use a Linkwitz modded panasonic Mike capsula, then you can use an easy to build high power bridge amp to drive the heck out of your woofers. Current mode output on the Midrange and Tweeters are easily implemented.
Sayonara
ACD said:
I think the gains will be marginal. I'd say listen first. An X-Over point of 500Hz for the PL18 is a bit high (it is a fairly large driver) and there is little difference between (say) 300 & 200Hz X-Over.
I used to do this kind of tradeoff between SPL and lower X-Over on large Pro-Sound systems. Going up an octave from 125Hz to 250Hz did not really give that much more attaiable midrange level, maybe 3 - 4db (the limit BTW where not the Amp's), unless you are in such a situation where the last attainable db SPL counts I'd keep the X-Over ow and live with the fallout.
If you already have problems with midrange level then adding a 4th way is unlikely to fix it (it might improve integration between Subwoofer and Midrange by adding an actual woofer), you would beed more displacement (usually more cone surface) which means either MTM with the same driver (if you can get another Pair) or a different driver (which you do not want to consider).
If you don't have problems with levelk handling in the midrange don't loose sleep. ;-)
You can probably gain a good deal of distortion reduction and removal of compression by feeding the Midrange and Treble from a current output (as opposed to voltage output) amplifier. Then maybe some form feedback around the woofer to stop problems with distortion and compression there and you should be okay.
LF NFB can use a Linkwitz modded panasonic Mike capsula, then you can use an easy to build high power bridge amp to drive the heck out of your woofers. Current mode output on the Midrange and Tweeters are easily implemented.
Sayonara
More amp questions
The following graph is from a system 3 way system a friend of mine is making. The prototype filter was made using the DP226 digital filter, and the resulting analogue filter is implemented on my Active filter ONE board. (This was the project that I made the board for in the first place btw)
The amps driving the top end are high passed elsewhere (Only upper bass and treble on the graph) are both 250W (Similar types). The sub is driven by a 1 kW amp.
My question is now: could something be gained by using different amps for the tweeter and upper bass? Or is it ok to use similar type amps for this?
I don't have graphs of the acoustical output - sorry
Thanks
\Jens
The following graph is from a system 3 way system a friend of mine is making. The prototype filter was made using the DP226 digital filter, and the resulting analogue filter is implemented on my Active filter ONE board. (This was the project that I made the board for in the first place btw)
The amps driving the top end are high passed elsewhere (Only upper bass and treble on the graph) are both 250W (Similar types). The sub is driven by a 1 kW amp.
My question is now: could something be gained by using different amps for the tweeter and upper bass? Or is it ok to use similar type amps for this?
I don't have graphs of the acoustical output - sorry
Thanks
\Jens
Attachments
Re: More amp questions
the simple answer is yes and no, the latter not being optimum.
A reasonable balance might be 250w and 100w.
But to me they are excessive power levels, but does no harm.
🙂 sreten.
JensRasmussen said:
the simple answer is yes and no, the latter not being optimum.
A reasonable balance might be 250w and 100w.
But to me they are excessive power levels, but does no harm.
🙂 sreten.
In reply to my post on peak to peak voltage and passband sensitivties:
Is it really necessary in this thread to discuss the behaviour of what is effectively a single widerange driver in it's three frequency bands, full-space below resonance, full space above resonance and half space above resonance?
I design my active three ways so that they could produce any single tone within the limits of a digital source, no matter what frequency it is and ignore any considerations of music's power distribution. This gives incredible headroom, which is always beneficial to SQ😀
It's considered very poor netiquette not to include a point in a quote just so that you can get in a three syllable contradiction. 🙄Bill Fitzpatrick said:
Is it really necessary in this thread to discuss the behaviour of what is effectively a single widerange driver in it's three frequency bands, full-space below resonance, full space above resonance and half space above resonance?
I design my active three ways so that they could produce any single tone within the limits of a digital source, no matter what frequency it is and ignore any considerations of music's power distribution. This gives incredible headroom, which is always beneficial to SQ😀
Hmmm,
nothing wrong with overengineering, but the money/effort spent on amps may produce more significant results elsewhere.
A complete system design should consider the actual requirements and actual limitations of the components. The actual power requirements (acoustic power output) decline logarithmically with frequency at same SPL (pink noise vs. white noise) independently of type of music. The actual limitations are, on the top end, tweeter RMS power handling, and on the bottom end, excursion of the woofer.
So, if you have several amps covering different frequencies, the bottom end likely doesn't need all that much peak power to be driven to full excursion, and the top end likely can't handle the high RMS resulting from the same peak power. Where do you find a tweeter that handles 250W RMS?
Coherent engineering would mean, to tailor the various amps in such a way that all components reach their limits roughly at the same SPL level.
nothing wrong with overengineering, but the money/effort spent on amps may produce more significant results elsewhere.
A complete system design should consider the actual requirements and actual limitations of the components. The actual power requirements (acoustic power output) decline logarithmically with frequency at same SPL (pink noise vs. white noise) independently of type of music. The actual limitations are, on the top end, tweeter RMS power handling, and on the bottom end, excursion of the woofer.
So, if you have several amps covering different frequencies, the bottom end likely doesn't need all that much peak power to be driven to full excursion, and the top end likely can't handle the high RMS resulting from the same peak power. Where do you find a tweeter that handles 250W RMS?
Coherent engineering would mean, to tailor the various amps in such a way that all components reach their limits roughly at the same SPL level.
Or, quite possibly, you could find that the wholesale cost difference between a 50W and 500W amplifier module was negligible compared to the cost of the project as a whole. Indeed, one may find that from some manufacturers, perhaps one you find desirable to purchase from, the minimum amplifier module power rating is well in excess of what is needed in the real world. In that case it isn't a matter of bad engineering, but rather a symptom of economics of scale.
I also think that you can get away with less power for tweeter and midrange than you need for the woofers (and don't forget to take efficiency into account.
OTOH there are famous manufacturers of active speakers like Backes & Mueller who use the same type and wattage amp for each driver because they say it is needed for transients.....
Regards
Charles
OTOH there are famous manufacturers of active speakers like Backes & Mueller who use the same type and wattage amp for each driver because they say it is needed for transients.....
Regards
Charles
If you crib design elements from a very successful high end active, like the ATC SCM-300A, then the power is split 50W RMS to the tweeter, 100W RMS to the midrange and 200W a piece to the two 15" woofers. Crossovers are something like 300Hz and 4KHz IIRC, don't quote me on the last one, I could be wrong.
That is true of average acoustical power levels but it is not true of peak acoustical power levels. If we want a low distortion system, it must be able to play the loudest transient cleanly, whatever its frequency. With some types of music, the tweeter may be required to play sounds as loud as the loudest the woofer plays but the sounds are very brief (cymbal crashes, etc.) so they don't show up in the average power figures.
Bottom line, don't underengineer the tweeter or the midrange if you want a true hifi system.
Peak power vs. average power -- scroll down a bit for the tables.
Thanks for all your comments.
This is just thinking.... :
Power = heat
I have observed that after playing loud for some time that the big coils on the woofers get quite hot. Understandable as you put in several hundred watts....
However the tweeter coil doesn't get as hot as the woofer......
The woofer coil is app. 20 times bigger than the tweeter coil, and have bigger surface to dissapate the heat.
Question:
If the same amount of power is feed to both woofer and tweeter, wouldn't the tweeter coil get so hot, that it melts ????
This is just thinking.... :
Power = heat
I have observed that after playing loud for some time that the big coils on the woofers get quite hot. Understandable as you put in several hundred watts....
However the tweeter coil doesn't get as hot as the woofer......
The woofer coil is app. 20 times bigger than the tweeter coil, and have bigger surface to dissapate the heat.
Question:
If the same amount of power is feed to both woofer and tweeter, wouldn't the tweeter coil get so hot, that it melts ????
Again, it's average power vs. peak power. The average power to tweeters is quite low and that's what determines heat. But the peak power to the tweeter, for less than a second, can be quite high.
Catapult,
agree on peak vs. average, that's why you don't find 250 W RMS capable tweeters (as soon as I say this someone will find one 😉 ).
The small and poorly vented voice coil won't take it.
I can see the point of maintaining the same peak capability of doing 0dBFS for the tweeter. But this tweeter amp doesn't need to be capable of 250 W RMS continuous output. The woofer amp just might. So, you may dimension the PSU smaller for the tweeter amp for instance, while maintaing the necessary transient voltage and current capability as for the woofer to achieve equal SPL at 0dBFS. That can be achieved with a much smaller transformer on the PSU side.
Even so, you want odBFS to translate to equal *peak SPL*, not equal peak amp output. Most tweeters are quite efficient (92 dB / 2.83V is quite common) and they get a 6 dB boost from the baffle which the woofer doesn't, so the tweeter amp will need much less voltage gain. So we're at 98 dB vs, say, 88 dB / 2.83V for the woofer. The tweeter amp in this example needs 14 dB less in transient V and A capability than the woofer, that's a factor of 5, and it doesn't need it in constant RMS output. In this example, for a 88dB woofer and a 92 dB tweeter on baffle, a 50W tweeter amp will have same peak SPL output as a 250 W woofer amp. And while the 250 W woofer amp may need a 500 VA transformer for reliable 250 W RMS output, the tweeter amp may be happy with a 50 VA or less.
agree on peak vs. average, that's why you don't find 250 W RMS capable tweeters (as soon as I say this someone will find one 😉 ).
The small and poorly vented voice coil won't take it.
I can see the point of maintaining the same peak capability of doing 0dBFS for the tweeter. But this tweeter amp doesn't need to be capable of 250 W RMS continuous output. The woofer amp just might. So, you may dimension the PSU smaller for the tweeter amp for instance, while maintaing the necessary transient voltage and current capability as for the woofer to achieve equal SPL at 0dBFS. That can be achieved with a much smaller transformer on the PSU side.
Even so, you want odBFS to translate to equal *peak SPL*, not equal peak amp output. Most tweeters are quite efficient (92 dB / 2.83V is quite common) and they get a 6 dB boost from the baffle which the woofer doesn't, so the tweeter amp will need much less voltage gain. So we're at 98 dB vs, say, 88 dB / 2.83V for the woofer. The tweeter amp in this example needs 14 dB less in transient V and A capability than the woofer, that's a factor of 5, and it doesn't need it in constant RMS output. In this example, for a 88dB woofer and a 92 dB tweeter on baffle, a 50W tweeter amp will have same peak SPL output as a 250 W woofer amp. And while the 250 W woofer amp may need a 500 VA transformer for reliable 250 W RMS output, the tweeter amp may be happy with a 50 VA or less.
ACD said:
Yes and in double quick time, as its thermal time constant is much
shorter than the bass unit, its like comparing quick and slow blow fuses.
As pointed out above an ideal tweeter amp has high peak
capability and low average capability, such a beast does
not exist as far as I know, but can be approximated to an
extent by RC filtering of its power supply, though I wouldn't
take this too far.
In the days of analogue tape full peak levels in the treble simply
did not exist, - 10dB bieng a sensible approximation, of course
with digital 0dB peaks are possible, but I've no idea how often
this occurs if ever.
However an active speaker will separate out the components of
a wideband 0dB peak into each frequency range, and I still think
- 10dB is a reasonable approximation of max bass vs treble.
🙂 sreten.
sreten said:
Wouldn't this just be an amp with relatively high voltage but PS capacitors (and maybe XFormer) that are too small to maintain the voltage very long?
Why try to build an amp with deliberately restricted avearge output power capability at all ?
Simply connect it to the same (stable !) PSU as the "lower" drivers.The power distribution of the program material is responsible for generating the peak/average ratio !
Regards
Charles
Simply connect it to the same (stable !) PSU as the "lower" drivers.The power distribution of the program material is responsible for generating the peak/average ratio !
Regards
Charles
Surely with the kind of levels we are talking about, it makes more sense from a tweeter safety point of view to design a tweeter amp that never clips, rather than one that is too weedy?
A tweeter with it's sensitivity modified by the baffle step to a mere 95dBW would produce peaks of 111dB from a mere 50W peak. I think you'd fry the midrange and the bass before you fried the tweeter. You might as well use a 100W RMS design, just for safety's sake.
A tweeter with it's sensitivity modified by the baffle step to a mere 95dBW would produce peaks of 111dB from a mere 50W peak. I think you'd fry the midrange and the bass before you fried the tweeter. You might as well use a 100W RMS design, just for safety's sake.
- Status
- Not open for further replies.