(Quote)
And a 12 dB down point means 25% of the normal output, not 6%, on SPL scale. What is 6% is the input power needed to output 12 dB less than the reference SPL.
(End quote)
How on earth have you come to this conclusion? 12dB down is 25% of the voltage drive to be sure, but the current through the driver will also have dropped to 25%. -12dB is near enough 6% of the power. At this frequency the mid-range unit will be putting out 16 times more acoustic power than will the tweeter. Now of course if you put your head up against the speakers the output from the tweeter will be quite audible, but at normal listening distances the output from the tweeter will be significantly masked by the same (but un-attenuated) signal coming out of the midrange.
And a 12 dB down point means 25% of the normal output, not 6%, on SPL scale. What is 6% is the input power needed to output 12 dB less than the reference SPL.
(End quote)
How on earth have you come to this conclusion? 12dB down is 25% of the voltage drive to be sure, but the current through the driver will also have dropped to 25%. -12dB is near enough 6% of the power. At this frequency the mid-range unit will be putting out 16 times more acoustic power than will the tweeter. Now of course if you put your head up against the speakers the output from the tweeter will be quite audible, but at normal listening distances the output from the tweeter will be significantly masked by the same (but un-attenuated) signal coming out of the midrange.
You're confused between power conversion to SPL and comparison of SPLs. Check it out.
Assuming two drivers are in phase, an equal amount of their contribution (i.e., 50%) to summed SPL means 6 dB down of each driver. When a driver contributes 25% and the other 75%, the former is 12 dB down and the latter 2.5 dB down. The former is NOT completely masked by the latter.
Assuming two drivers are in phase, an equal amount of their contribution (i.e., 50%) to summed SPL means 6 dB down of each driver. When a driver contributes 25% and the other 75%, the former is 12 dB down and the latter 2.5 dB down. The former is NOT completely masked by the latter.
Its quite funny...I am now adjusting a notch and RC on my midrange(3way), and they seem to have a great deal of influence on each other...the funny part is that with every change on the mids notch and RC the BASS definition changes too...only explanation is that it is PHASE related (timing?)
Thanks for pointing out the problems in 1 order on the tweeter, but would these problems not be the same for the woofers and midrange, phase will alter the slopes.
My last system I made was an acoustic LR2 system, that is the safe way to go, It is also easy to check that the drivers are in phase. Electric It was 2 order on the woofer SS 18W4531 and 4 order on the tweeter SS 9700.
Well i guess B&W can make a 1 order system that makes the reviewers happy, then I guess we can do It too!!
My last system I made was an acoustic LR2 system, that is the safe way to go, It is also easy to check that the drivers are in phase. Electric It was 2 order on the woofer SS 18W4531 and 4 order on the tweeter SS 9700.
Well i guess B&W can make a 1 order system that makes the reviewers happy, then I guess we can do It too!!
Assuming first-order HP and LP filters, at the point that the tweeter is 12dB down, the mid-range will have very little attenuation.
Now I will agree that if the power level into the tweeter at this point, at the low-frequency, is enough to drive the tweeter into non-linearity, then the distortion will clearly be heard. But I don't think anyone is disputing this.
Can I put the following to Jay_WJ?
1. Consider a tweeter that is flat in amplitude and has minimum phase characteristics over a defined frequency band. If we now feed that from a HP filter that has a first-order characteristic, the measured SPL will also roll-off at 6dB per octave over this defined band. Earlier you seemed to deny this, but companies like Bruel & Kjaer have built very successful businesses on making equipment to check that speakers do exactly this.
2. If we now increase the defined frequency band to include the region where the tweeter itself is suffering from its natural roll-off, then this will cause the combined filter and tweeter SPL to have a region where the roll-off is greater than 6dB/8ve, and (more importantly) has excess phase shift compared to the desired response characteristic.
So far I can't imagine anyone would disagree.
Now for the contentious bit
It is my opinion (yes opinion only!) that provided this break point in the response from the desired 6dB/8ve to some higher rate happens at a point where the attenuation of the tweeter (due to the filter alone) is 12dB or more, then the effect is acoustically negligable at normal listening distances.
I'm happy to hear alternative views on this last point, and no, I don't claim to have golden ears, but I am paid to measure loudspeakers under proper lab conditions, which is why I'd rather trust the B&K kit than my (or anyone elses ears).
Now I will agree that if the power level into the tweeter at this point, at the low-frequency, is enough to drive the tweeter into non-linearity, then the distortion will clearly be heard. But I don't think anyone is disputing this.
Can I put the following to Jay_WJ?
1. Consider a tweeter that is flat in amplitude and has minimum phase characteristics over a defined frequency band. If we now feed that from a HP filter that has a first-order characteristic, the measured SPL will also roll-off at 6dB per octave over this defined band. Earlier you seemed to deny this, but companies like Bruel & Kjaer have built very successful businesses on making equipment to check that speakers do exactly this.
2. If we now increase the defined frequency band to include the region where the tweeter itself is suffering from its natural roll-off, then this will cause the combined filter and tweeter SPL to have a region where the roll-off is greater than 6dB/8ve, and (more importantly) has excess phase shift compared to the desired response characteristic.
So far I can't imagine anyone would disagree.
Now for the contentious bit
It is my opinion (yes opinion only!) that provided this break point in the response from the desired 6dB/8ve to some higher rate happens at a point where the attenuation of the tweeter (due to the filter alone) is 12dB or more, then the effect is acoustically negligable at normal listening distances.
I'm happy to hear alternative views on this last point, and no, I don't claim to have golden ears, but I am paid to measure loudspeakers under proper lab conditions, which is why I'd rather trust the B&K kit than my (or anyone elses ears).
Jay_WJ said:
But they won't be in-phase. The midrange will have about 15 degrees of phase lag, whereas the tweeter, at the point where the filter is 12dB down will have about 75 degrees of phase advance.
All my arguments have assumed that the drive units are impedance-compensated of course, so that they can be considered as resistive loads.
This is my last comment on this topic.
1st order series x-os on modern tweeters do not cause a problem at normal to very loud listening levels. So long as you cross a minimum of x3 above Fs.
The standard MTM's using series 1st order x-os sounds very good, and do not have any audible phase problems or tweeter overdrive issues.
So I guess its up to you whether you take the word of someone who has done some 50 speakers using this method, or someone who has done none.
Totally your call.
1st order series x-os on modern tweeters do not cause a problem at normal to very loud listening levels. So long as you cross a minimum of x3 above Fs.
The standard MTM's using series 1st order x-os sounds very good, and do not have any audible phase problems or tweeter overdrive issues.
So I guess its up to you whether you take the word of someone who has done some 50 speakers using this method, or someone who has done none.
Totally your call.
Ouroboros said:
If you assume BW1, yes, you're right since they're not in phase. I assumed LR2 in my reply. But even if BW1 is assumed, as you noted above, most 1" dome tweeters will severly distort at the 12 dB down point---with 3 kHz Fc, 12 dB down occurs at about 780 Hz. Only 12 dB down from the reference level is not sufficient to prevent the tweeter from distorting. And this distortion cannot be masked by phase canceling from the midwoofer since it's only from the tweeter.
In general, a true 1st order acoustical filter with Fc around 3 kHz is not possible with a 1" dome tweeter.
Can I put the following to Jay_WJ?
1. Consider a tweeter that is flat in amplitude and has minimum phase characteristics over a defined frequency band. If we now feed that from a HP filter that has a first-order characteristic, the measured SPL will also roll-off at 6dB per octave over this defined band. Earlier you seemed to deny this, but companies like Bruel & Kjaer have built very successful businesses on making equipment to check that speakers do exactly this.
2. If we now increase the defined frequency band to include the region where the tweeter itself is suffering from its natural roll-off, then this will cause the combined filter and tweeter SPL to have a region where the roll-off is greater than 6dB/8ve, and (more importantly) has excess phase shift compared to the desired response characteristic.
So far I can't imagine anyone would disagree.
Now for the contentious bit
It is my opinion (yes opinion only!) that provided this break point in the response from the desired 6dB/8ve to some higher rate happens at a point where the attenuation of the tweeter (due to the filter alone) is 12dB or more, then the effect is acoustically negligable at normal listening distances.
What I said in my earlier post is that if you apply 1st order electrical filter with Fc around 2.5 to 3 kHz to most 1" dome tweeters, you'll obtain something closer to 2nd order acoustical slope, not closer to 1st order slope. Most 1" dome tweeters roll off much earlier than 1 kHz. The RS28A, the XT25, or the SS AirCirc 6600 has a relatively extended response down to 1 kHz. And the SS D2904/9800 is an exception that has a flat FR up to 600-700 Hz.
To be regarded as a 1st order slope, alignment of the response to the target up to one octave below the Fc is not sufficient. Do not focus only on the SPL. In order for an acoustic curve to have the phase shift of BW1, the curve should mimic the BW1 target from much, much lower frequency. This is why we define the order of a filter slope as a slope at its tail, not one at Fc. BW1 has only about a 3 dB/oct slope around its Fc.
For example, if we apply an electrical first order filter to the XT25 with Fc at 2.5 kHz, we will have something closer to a 2nd order acoustic slope. See this design:
http://www.zaphaudio.com/ZD5.html
Andy Graddon said:This is my last comment on this topic.
1st order series x-os on modern tweeters do not cause a problem at normal to very loud listening levels. So long as you cross a minimum of x3 above Fs.
The standard MTM's using series 1st order x-os sounds very good, and do not have any audible phase problems or tweeter overdrive issues.
So I guess its up to you whether you take the word of someone who has done some 50 speakers using this method, or someone who has done none.
Totally your call.
Who said not? Andy, you still don't get my point yet. My point is that something people, including you I guess, call a 1st order xover is in fact something closer to a 2nd order xover, not closer to 1st order one. It's not about experience of building speakers---I wouldn't take your comment as offensive. It's about how we define the order of a filter slope. I just want to clarify a concept.
The "some 50 speakers" you've designed, if they used 1st order series electrical xovers, I bet the xovers are approximatley LR2, not BW1. That's my point.
Sorry, I had about 2 hours real sleep last night, no idea why, just a sh...y sleep. 
I do understand your point about the actual roll-off being more 2nd order. Although its not always the case.
Guess I was just a bit fed up with people saying can't cross tweeters under 10k, which is just plane rubbish, and that there were all sorts of other problems with using 1st order, when I know from experience that if these problems do exist, they are certainly not audible.
I do understand your point about the actual roll-off being more 2nd order. Although its not always the case.
Guess I was just a bit fed up with people saying can't cross tweeters under 10k, which is just plane rubbish, and that there were all sorts of other problems with using 1st order, when I know from experience that if these problems do exist, they are certainly not audible.
Andy Graddon said:
My experience was based on Scanspeak 8554 woofer, Dayton RS52, and Scanspeak 6600 used in a 3-way.
I also tried Scanspeak 8554, BG RD40 planar, and Scanspeak 6600 or Seas Excel Millenium.
And I tried Vifa tweeter and Vifa 5 1/2" mid woofer.
In all 3 cases, I (my girlfriend at the time and my buddy Jake) found that a 1st order crossover on the tweeter didn't sound good on my kind of hard rock music to me with my drivers and my crossover in my living room.
The sound for me of the 1st order crossover was VERY audibly different from higher order crossovers. All I can say is that 1st order doesn't work for me. That's just an alternate opinion. I don't like Rieslings or Bud Light either. They taste too sweet and too bland.
My interest in this thread started when I saw someone talking about B&W speakers which I respect. I haven't been able to duplicate their 1st order crossover and get the same sound quality.
In the case of the RS52 I can explain why 1st order couldn't work. The driver has a strong resonance at 13khz. I didn't have a specific notch filter.
In the case of a planar, the driver center-to-center spacing was either horizontal or 20" vertical.
But in the case of the Vifa, the center-to-center spacing was pretty close and mounted in slanted cabinets (Thiels CS 1.2). I actually could have lived with 2nd order on this speaker happily.
If there's a 1st order design using say Vifa drivers please point it out. I'll build it and take it to a DIY event.
Daveis said:
Madisound has a kit that uses the same drivers as the Meadowlark Swift that they call the Swifty.
http://www.madisound.com/catalog/ad...=1&inc_subcat=1&imageField.x=0&imageField.y=0
The Swifty crossover appears to be a 2nd order Butterworth from the pictures.
http://madisound.com/pdf/Swifty.pdf
Apparently, the original was 1st order. Madisound redesigned the crossover.
http://www.diyaudio.com/forums/showthread.php?threadid=96018
http://madisound.com/pdf/Swifty.pdf
Apparently, the original was 1st order. Madisound redesigned the crossover.
http://www.diyaudio.com/forums/showthread.php?threadid=96018
Andy recommended these designs a while back.
http://members.optusnet.com.au/~gra.../blackwood.html
http://undefinition.googlepages.com/diy-aethers
It just occurred to me that I've heard the Aethers at the Iowa DIY.
They won budget class. I think I remember thinking the imaging was good. I ranked them about in the middle of the pack of the 25 or so speakers I heard that day.
So I agree that 1st order crossovers can sound good, provided that they weren't designed by me.
http://members.optusnet.com.au/~gra.../blackwood.html
http://undefinition.googlepages.com/diy-aethers
It just occurred to me that I've heard the Aethers at the Iowa DIY.
They won budget class. I think I remember thinking the imaging was good. I ranked them about in the middle of the pack of the 25 or so speakers I heard that day.
So I agree that 1st order crossovers can sound good, provided that they weren't designed by me.
1st order x-os are all about driver choice as the very first thing. I helped Paul choose the mids and tweeter, he chose the bass unit because I couldn't suggest one.
It was the bass unit and its peak at 1k that meant Paul couldn't use the 3-way series x-o that I prefer. Not his fault, I have searched over there for bass drivers that are suitable for low order x-o wityhout correction circuits, and the only ones I could find over there would have pushed it way out of the budget class.
The Scan 8554 would not be suitable, and certainly the RS52 is not, so I'm not surprised you couldn't do a 1st order with them. Can't comment on the 6600 or BG ribbon.
Anyway, you could probably hear that there was no hint of tweeter issues in Paul's speaker. Also the way the drivers sound sort of "together" and the very tight imaging. This is one thing I have found with the 1st order x-os. I find the open back mids also provide a much wider sweet spot in a normal room.
You mention a Vifa 5.5" mid.. and a Vifa tweeter.. which ones particularly ??
It was the bass unit and its peak at 1k that meant Paul couldn't use the 3-way series x-o that I prefer. Not his fault, I have searched over there for bass drivers that are suitable for low order x-o wityhout correction circuits, and the only ones I could find over there would have pushed it way out of the budget class.
The Scan 8554 would not be suitable, and certainly the RS52 is not, so I'm not surprised you couldn't do a 1st order with them. Can't comment on the 6600 or BG ribbon.
Anyway, you could probably hear that there was no hint of tweeter issues in Paul's speaker. Also the way the drivers sound sort of "together" and the very tight imaging. This is one thing I have found with the 1st order x-os. I find the open back mids also provide a much wider sweet spot in a normal room.
You mention a Vifa 5.5" mid.. and a Vifa tweeter.. which ones particularly ??
Again, please don't see me nitpicking. I just want to clarify concepts here. When we talk about the order of a xover, we mean either the order of its acoustic rolloff or the order of its electrical topology.
Note that the latter only indicates the filter's physical/visual property. What is important is the filter's effect---its acoustic rolloffs. Sometimes a LR4 slope can be achieved by a 2nd order electrical filter and other times by a 3rd or a 4th order electrical filter, depending on the driver and the xover point.
This is WHy we normally use the order of the filter's acoustic rolloff, not its electrical order, when we say "1st, 2nd, 3rd, or 4th order xover." A filter's electrical order tells nothing about its acoustic property in real-world applications.
The above Undefinition's design uses first order eletrical filters for the midrange and the tweeter, but it's basically a LR2 design acoustically. I haven't looked at ALL of Andy's designs that use 1st order electrical series filters, but I can say that most of these designs should basically be LR2.
Note that the latter only indicates the filter's physical/visual property. What is important is the filter's effect---its acoustic rolloffs. Sometimes a LR4 slope can be achieved by a 2nd order electrical filter and other times by a 3rd or a 4th order electrical filter, depending on the driver and the xover point.
This is WHy we normally use the order of the filter's acoustic rolloff, not its electrical order, when we say "1st, 2nd, 3rd, or 4th order xover." A filter's electrical order tells nothing about its acoustic property in real-world applications.
The above Undefinition's design uses first order eletrical filters for the midrange and the tweeter, but it's basically a LR2 design acoustically. I haven't looked at ALL of Andy's designs that use 1st order electrical series filters, but I can say that most of these designs should basically be LR2.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.