I’ve read somewhere long time ago about crossover point selection trick that one could make a steeper acoustical slope by applying an electrical slope combining with a natural roll-off of the driver. For example, a conventional dome tweeter could be considered its low-frequency loading as a closed box system since it has been sealed on the back. As a result, the tweeter would have a natural roll-off at 12dB/octave rate. Once we apply a 12dB/octave electrical filter to it, the final acoustical response should have going to be a 24dB/octave slope by implication.
First of all, is it true?
Then, by performing that, is there a risk for damaging the tweeters, because the selected crossover point would be located at quite low and close to the Fs of the tweeters?
Moreover, since the x-over point is near Fs, will it introduce a large distortion to the speakers?
Finally, would the principle still be valid for the dome midranges, as their constructions are identical, except for the size?
First of all, is it true?
Then, by performing that, is there a risk for damaging the tweeters, because the selected crossover point would be located at quite low and close to the Fs of the tweeters?
Moreover, since the x-over point is near Fs, will it introduce a large distortion to the speakers?
Finally, would the principle still be valid for the dome midranges, as their constructions are identical, except for the size?
First question yes tweeter slopes plus Xover slopes sum in terms of the roll on slope as you say.
You just need to get them to the manufacturers recommended Xover frequency.
Manufacturers often specify their tweeter Xover recommendations as say, 12db second order, at 2.5Khz which allows a basic comparison. between tweeters. Its a good place to start, and to keep in mind.
You can cheat on this if you don't play loud. But, If your wife or children don't know this fact they may unknowingly blow a tweeter.
If you want to adopt a lower than recommended Xover it is also worth looking at the distortion measurement of the tweeter with the designed Xover.
Impedance chance at FS will affect the crossover slope in that area, which means it can be beneficial to add an impedance equalisation network to allow the correct Xover performance. This should be viewable in Xover sims.
You just need to get them to the manufacturers recommended Xover frequency.
Manufacturers often specify their tweeter Xover recommendations as say, 12db second order, at 2.5Khz which allows a basic comparison. between tweeters. Its a good place to start, and to keep in mind.
You can cheat on this if you don't play loud. But, If your wife or children don't know this fact they may unknowingly blow a tweeter.
If you want to adopt a lower than recommended Xover it is also worth looking at the distortion measurement of the tweeter with the designed Xover.
Impedance chance at FS will affect the crossover slope in that area, which means it can be beneficial to add an impedance equalisation network to allow the correct Xover performance. This should be viewable in Xover sims.
Tweeters should never be pushed near their limit. Usually the crossover is set 1 or 2 octaves above the resonance.
Yes. It rolls off below resonance frequency (similar to a sealed woofer).
But tweeters are not made to be used down to the resonance frequency:
- They will easily get damaged by too much power and high excursion
- They will produce lots of tumbling movement, leading to high harmonic distorsion (they don't have a double suspension as most cone speakers do)
- The electric crossover will be badly influenced by the driver's resonance frequency impedance peak
The same applies to dome midranges - but it can be even worse.
You could try a small fullrange driver as tweeter. That might work!
Good data examples here that reinforce what everyone said:
http://www.zaphaudio.com/tweetermishmash/
Many tweeters Zaph tested have rising distortion with decreasing frequency significantly above Fs.
One example: "Vifa XT25 ($48) - Smoothest and most extended response curve in the group, and resulting CSD is excellent. Good tall order HD above 2kHx, but average 2nd order HD. Poor HD levels of all types below 2kHz, even considering the extended low end. It may have a 500 Hz Fs, but don't think about crossing it below 2kHz LR4 or 2.5kHz LR2"
http://www.zaphaudio.com/tweetermishmash/
Many tweeters Zaph tested have rising distortion with decreasing frequency significantly above Fs.
One example: "Vifa XT25 ($48) - Smoothest and most extended response curve in the group, and resulting CSD is excellent. Good tall order HD above 2kHx, but average 2nd order HD. Poor HD levels of all types below 2kHz, even considering the extended low end. It may have a 500 Hz Fs, but don't think about crossing it below 2kHz LR4 or 2.5kHz LR2"
In fairness you have to be a mite careful when reading HD plots as it depends on conditions. A tweeter with a lower Fs and / or flatter response to a low frequency will tend to have a rising LF distortion trend relative to another of equivalent size & with an equivalent quality motor. Assuming you're not actually trying to use that extended LF to a significant extent though, once HP filtered, the significance falls away.
The XT25 is an interesting unit & has to be treated with a degree of caution since in terms of Sd, it's actually a 3/4in tweeter, although it often gets thrown in with 1in types, which doesn't help it in that regard. That said -you can cross it successfully lower than John suggested, but realistically, only for systems intended for modest levels, and you're accepting there is some compromise in doing so. I wouldn't myself, as there are better choices, but it's been done. A comparison to the still highly-rated Seas Millennium is interesting: with the exception of HD2 (since the XT25 is essentially just a couple of concentric roll surrounds) HD is actually lower across the board, although the Millennium does have more excursion available. More to it than that (a lot more 😉 ) on both sides of course, but notable.
The XT25 is an interesting unit & has to be treated with a degree of caution since in terms of Sd, it's actually a 3/4in tweeter, although it often gets thrown in with 1in types, which doesn't help it in that regard. That said -you can cross it successfully lower than John suggested, but realistically, only for systems intended for modest levels, and you're accepting there is some compromise in doing so. I wouldn't myself, as there are better choices, but it's been done. A comparison to the still highly-rated Seas Millennium is interesting: with the exception of HD2 (since the XT25 is essentially just a couple of concentric roll surrounds) HD is actually lower across the board, although the Millennium does have more excursion available. More to it than that (a lot more 😉 ) on both sides of course, but notable.
Last edited:
Yep. It's not a 'trick' though: accounting for the driver's actual response is a natural part of filter design.
Yes.
It isn't necessarily quite low / close to Fs -depends on the design. For e.g., ferrofluid has fallen out of favour in recent years for various reasons, but automatically tended to put a damped LF response trend to most tweeters using it in the octave or > above Fs.
Damage is possible, depending on how much electrical power gets to it & what the excursion is. You need at least 2nd order slopes just to keep excursion constant, while LR4 in theory (or at least as far as excursion goes) provides the highest power-handling of any regular flat-summing topology. Depending on how hard you're pushing the tweeter, you may need to consider the electrical power handling on top of just the acoustic slope though, so a 3rd order or > electrical may be needed.
Per the above really: it's not necessarily all that close to Fs, and distortion depends on the tweeter itself. You're pushing harder than if you were crossing higher, but many can easily take it if used well -at least within the context of what they are. 2in vintage horn-loaded Altec compression drivers capable of 120+dB they ain't, so they'll never rival those in that regard -but they weren't designed for the same purpose.
Yes.
Last edited:
When a manufacturer states that a driver has a certain power handling it coms along with a x-over frequency and steepness. This is the limit - it doesn't matter where the driver falls off without a filter. It's already in the calculations so to say. So yes, exceeding these numbers puts the driver at risk.
//
//
I'd rephrase this a little. The final resposne curve of any driver is the combination of the driver's natural rolloff plus that of any crossovers or filters applied before it. I wrote about this here. See my pretty pics. 🙂
Well, it's the maximum power it can take under IEC or equivalent testing (usually 2.5KHz with a B2 high pass). That doesn't mean that's the lowest safe crossover frequency though, it's just the maximum short / long term electrical power it can take under those conditions.
Well, this narration described one way of making steeper slope but it only pointed to the “high-pass” section.
I wonder if the "low-pass" section should also be eligible.
However, on the low-pass side, there is a driver break-up/beaming occurrence. So, how do we deal with it?
By quality crossover design? 😉
Bad humour aside, yes, it is (or can be depending on whether it's relevant or required). And the same for the second. There's no one answer about how though, other than it depends on the design specifics: type of driver / drivers, on & off axis frequency response, distortion performance, target goals for average power handling, dynamic range at the listening position, power / polar response etc. But as has been noted, quality crossover design means you are working with the actual on-baffle driver responses, so assuming they have a natural low pass slope, or obvious resonant modes within a couple of octaves of the low pass (especially if you're running low order filters) then you wil need to account for either or both as relevant in the low pass filter design via whatever methods you prefer to use, are in-budget and achieve your goals, be it altering the filter frequency or order, notches of various kinds in the transition or even stopband to either linearise the FR or reduce passband distortion & so on & so forth.
Bad humour aside, yes, it is (or can be depending on whether it's relevant or required). And the same for the second. There's no one answer about how though, other than it depends on the design specifics: type of driver / drivers, on & off axis frequency response, distortion performance, target goals for average power handling, dynamic range at the listening position, power / polar response etc. But as has been noted, quality crossover design means you are working with the actual on-baffle driver responses, so assuming they have a natural low pass slope, or obvious resonant modes within a couple of octaves of the low pass (especially if you're running low order filters) then you wil need to account for either or both as relevant in the low pass filter design via whatever methods you prefer to use, are in-budget and achieve your goals, be it altering the filter frequency or order, notches of various kinds in the transition or even stopband to either linearise the FR or reduce passband distortion & so on & so forth.
Here is an example of the dome midrange driver: https://www.daytonaudio.com/images/resources/285-022--dayton-audio- rs52fn-8-2-in-reference-fabric-dome-midrange-specifications.pdf
Look at its frequency response graph, it’s clearly seen that it has a natural corner frequency at about 300 or 400 Hz. However, the Fs is indicated at 394.3 Hz according to the table.
Regarding the conversation above that we can combine the electrical slope and the natural roll-off of the driver to obtain the steeper acoustical slope, therefore, the natural corner frequency (F3) of this driver is located at about 400 Hz.
Would it cause any problems as the F3 is located very close to, or maybe lower than, the Fs?
Look at its frequency response graph, it’s clearly seen that it has a natural corner frequency at about 300 or 400 Hz. However, the Fs is indicated at 394.3 Hz according to the table.
Regarding the conversation above that we can combine the electrical slope and the natural roll-off of the driver to obtain the steeper acoustical slope, therefore, the natural corner frequency (F3) of this driver is located at about 400 Hz.
Would it cause any problems as the F3 is located very close to, or maybe lower than, the Fs?
Presscot,
It's clearly seen (by me..) the -12dB per octave rolloff begins above 400Hz, at 300Hz response is near -6dB down.
The driver has only 1mm Xmax, the excursion where harmonic distortion reaches ~10%.
If your don't mind that much distortion at a relatively low SPL, no problem 🙂
At 400Hz it would reach Xmax ~95dB SPL at 1meter using ~4 watts.
At 500Hz it would reach Xmax ~100dB using ~7 watts.
At 800Hz it would reach Xmax ~107dB using ~30 watts (half rated power).
At 1000Hz around 60watts (rated power) would be required to reach Xmax at 111dB SPL.
The usual practice of crossing around an octave above Fs makes sense if you want much output level, otherwise the "usable frequency range" suggestion of 500Hz is prudent, assuming 100dB@1m, (94dB at 2 meters..) with ~10% distortion is enough SPL for you.
Art
Attachments
So, if one’d like to cross any drivers where it could generate a steeper acoustical slope by means of combining driver’s natural roll-off and electrical slope (passive filters) as discussed on the original post, he won’t have to concern about the Fs location, correct?
Because the 500Hz point suggested in the example driver above is located fairly close to the impedance peak or Fs.
Because the 500Hz point suggested in the example driver above is located fairly close to the impedance peak or Fs.
Well, that happens then is that you utilise the driver in more or less the range where the big resonance is - so if thats OK with your - by all means go ahead.
A resonance in the pass band will mean the same decisions as for a bass driver; closed or vented, what system Q do I want, xmax, GD etc...
//
A resonance in the pass band will mean the same decisions as for a bass driver; closed or vented, what system Q do I want, xmax, GD etc...
//
If your goal is to generate a steeper acoustical slope rather than protect the driver from overexcursion and attendant distortion, the Fs is irrelevant.
500Hz is more than 1/3 octave above the example driver's Fs.
1/3 octave is an 8dB change in a 24dB/octave slope.
Raising the driver's crossover point by 1/3 octave also reduces it's excursion by 1/3 for the same SPL.
The Fs is located at 394.3 (~400) Hz. If one octave from 500Hz is 250Hz, the 1/3 octave is about 416.7 (~420) Hz. Am I correct?
However, in the FR graph, the -8dB is about 200Hz. So, what should it be? (200 vs. 420Hz)
I don't know, I just use the standards that have ben around since the early 1900s.
Standard ISO 1/3 octave bands from 16 Hz to 20 kHz are: 16, 20, 25, 31.5, 40, 50, 63, 80, 100, 125, 160, 200, 250, 315, 400, 500, 630, 800, 1k, 1.25k, 1.6k, 2k, 2.5k, 3.15k, 4k, 5k, 6.3k, 8k, 10k, 12.5k, 16k, and 20k.
The frequency response is what it is, 200Hz is around -12dB from the 420Hz response on the graph.
If you look at a 24dB per octave filter response, each lower 1/3 octave interval (24/3) is -8dB from the previous.
Oops, my math was off there..
Four times the displacement (excursion) is required for each halving of frequency to maintain the same SPL.
Art