I was thinking of doing almost this exact thing using a Vifa P13WH-00 for the mid, a Vifa D27TG-35 (or something similar), and a Vifa P21WO-20 woofer (i think this is the one Vandersteen uses in the 2CE).
I changed my mind about the first order filters after reading this site though: http://www.silcom.com/~aludwig/Sysdes/Design_of_the_sound_system.htm#Crossover_design He seems to haave pupt a lot of work into his system, and he says that in the digitally filtered and crossed over music that he listened to that he couldn't tell the difference between a first order Butterworth and a 4th order Linkwitz-Riley. This, along with Linkwitz's site, made me think that a Linkwitz-Riley filter would be better. I finally decided to buy one of these instead of making my own though: http://www.partsexpress.com/pe/pshowdetl.cfm?&Partnumber=248-668&scqty=1 That way I can adjust things to my heart's content.
Recently though, I have decided that spending about $500 on speakers (not including the amp I would need to tri-amp them) is a bit much.
I changed my mind about the first order filters after reading this site though: http://www.silcom.com/~aludwig/Sysdes/Design_of_the_sound_system.htm#Crossover_design He seems to haave pupt a lot of work into his system, and he says that in the digitally filtered and crossed over music that he listened to that he couldn't tell the difference between a first order Butterworth and a 4th order Linkwitz-Riley. This, along with Linkwitz's site, made me think that a Linkwitz-Riley filter would be better. I finally decided to buy one of these instead of making my own though: http://www.partsexpress.com/pe/pshowdetl.cfm?&Partnumber=248-668&scqty=1 That way I can adjust things to my heart's content.
Recently though, I have decided that spending about $500 on speakers (not including the amp I would need to tri-amp them) is a bit much.
Davies, Andy, and others,
Guys, let's clarify the order of filter slopes here. I believe Davies meant the acoustic 1st order, not electrical 1st order. AFAIK, there's no 1" dome tweeter that can be used to implement a TRue 1st order acoustic slope at Fc of 3 kHz or below. If there's one, please let me know.
I believe the "1 order filters" in the thread title actually meant a 2nd order acoustic slope given by a 1st order electrical filter.
If you use a single series cap for most of 1" dome tweeters, you'll end up with something close to a 2nd order acoustic slope.
Guys, let's clarify the order of filter slopes here. I believe Davies meant the acoustic 1st order, not electrical 1st order. AFAIK, there's no 1" dome tweeter that can be used to implement a TRue 1st order acoustic slope at Fc of 3 kHz or below. If there's one, please let me know.
I believe the "1 order filters" in the thread title actually meant a 2nd order acoustic slope given by a 1st order electrical filter.
If you use a single series cap for most of 1" dome tweeters, you'll end up with something close to a 2nd order acoustic slope.
Hyldal said:
The MG14 looks like it would do a good job 1st order. I used 2 of the MG18's in "The Juke". nice drivers. What you need to decide is if you thing a single one of them is enough each side.
http://members.optusnet.com.au/~gradds55/ARGOS/TheJuke.html
Ouroboros said:
get 4 of them and wire in series in an MTM, that what I did with the MG18's (4 ohm) in "The Juke" (see link above)
Hyldal,
I'm 99% sure that the tweeter crossed at 3.8 kHz you heard did not have a true 1st order acoustic slope. Even at 3.8 kHz, there's no tweeter that can be used for a true 1st order slope. Here's why:
In the above are the Dayton RS28A-4 tweeter on infinite baffle and a BW1 target with Fc of 3.8 kHz. The RS28A-4 has good low end extension as a 1" dome tweeter (check Zaph's tweeter mishmash test page).
As you can see, if you align the tweeter's response to the target down to 1 kHz, the response below 1 kHz should have much faster rolloff. With BW1, a response below 1 kHz has big contribution. Note that at 1 kHz the SPL is only 12 dB below the reference (90.5 dB here). Also, note that the RS28's response used here was measured on infinite baffle. If you use it on a real finite baffle, you'll have even faster rolloff below 1 kHz due to baffle step loss.
Another consideration is, the fast rolloff below 1 kHz should result in a significant phase shift even above 1 kHz, from the target BW1 phase.
If you want to have as large an overlap as possible between drivers, LR2 is the way to go. Even LR2 is hard to implement. Not many drivers can satisfy its requirement.
For example, I believe Andy's "The Juke" above is an example of LR2 design. In fact, it should be something LIKE LR2, not real LR2, if the Fc is below 3 kHz. The 27TDFC has too fast a rolloff below 2 kHz. Not an ideal driver for an LR2 design with an Fc below 3 kHz, as Zaph mentioned in his ZD5 page:
"The Seas standards (like the 27TDFC and TBFCG) roll off a little too early for a LR2 crossover at the frequency I was shooting for."
I'm 99% sure that the tweeter crossed at 3.8 kHz you heard did not have a true 1st order acoustic slope. Even at 3.8 kHz, there's no tweeter that can be used for a true 1st order slope. Here's why:
An externally hosted image should be here but it was not working when we last tested it.
In the above are the Dayton RS28A-4 tweeter on infinite baffle and a BW1 target with Fc of 3.8 kHz. The RS28A-4 has good low end extension as a 1" dome tweeter (check Zaph's tweeter mishmash test page).
As you can see, if you align the tweeter's response to the target down to 1 kHz, the response below 1 kHz should have much faster rolloff. With BW1, a response below 1 kHz has big contribution. Note that at 1 kHz the SPL is only 12 dB below the reference (90.5 dB here). Also, note that the RS28's response used here was measured on infinite baffle. If you use it on a real finite baffle, you'll have even faster rolloff below 1 kHz due to baffle step loss.
Another consideration is, the fast rolloff below 1 kHz should result in a significant phase shift even above 1 kHz, from the target BW1 phase.
If you want to have as large an overlap as possible between drivers, LR2 is the way to go. Even LR2 is hard to implement. Not many drivers can satisfy its requirement.
For example, I believe Andy's "The Juke" above is an example of LR2 design. In fact, it should be something LIKE LR2, not real LR2, if the Fc is below 3 kHz. The 27TDFC has too fast a rolloff below 2 kHz. Not an ideal driver for an LR2 design with an Fc below 3 kHz, as Zaph mentioned in his ZD5 page:
"The Seas standards (like the 27TDFC and TBFCG) roll off a little too early for a LR2 crossover at the frequency I was shooting for."
I have 3 or 4 sims that people have done of the 3-way linked series.
The tweeter response in each case starts to roll-off at approx 6dB/octave half an octave above the top x-o point. It remains at approx 6dB until about an octave below the x-o point, where it transitions to a steeper slope , generally a tad less than 12dB
Because values in a series x-o can effect actual slope as well as x-o point (zeta) the slopes will generally not be exactly 6dB or 12dB, but could be a margin either way.
The tweeter response in each case starts to roll-off at approx 6dB/octave half an octave above the top x-o point. It remains at approx 6dB until about an octave below the x-o point, where it transitions to a steeper slope , generally a tad less than 12dB
Because values in a series x-o can effect actual slope as well as x-o point (zeta) the slopes will generally not be exactly 6dB or 12dB, but could be a margin either way.
Andy Graddon said:
Your description fits a LR2 target exactly. See below.
An externally hosted image should be here but it was not working when we last tested it.
The Fc of this LR2 target is 2.5 kHz. It is below 1 kHz that the slope becomes second order 12 dB/oct. Until then it gradually gets steeper. In the 2 k to 3 kHz range, it is approxomately only 6 dB/oct.
I didn't mean that your 27TDFC would sound bad. What I mean is that the 27TDFC is not an ideal choice for a strict LR2 crossover around 2.5 kHz Fc, due to its early rolloff. What will the effect be on subjective sound quality? I absolutely have no idea, because I haven't done the exactly same thing. If it satisfies you, no problem. That's eventually what we're interested in.
I do not force my response curve to a predetermined target, either. Consider what people do a LR4 design on a flat baffle in order to phase-align drivers. Usually asymmetric slopes do the trick. Sticking to predetermined LR4 slopes do not work here.
But for LR2, this trick does not work nicely, due to the required large overlap between drivers. That's why people use a slanted/stepped baffle or a delay network for time alignment of LR2.
In your LR2 implementation with the 27TDFC, there should be some misaligned phase due to the use of flat baffle and/or the 27TDFC's early rolloff. But the question is, is this necessarily bad? I do not think so. As long as you have a flat SPL sum at your listening axis and voice your speakers in your room, you'll have no problem.
The reason why some theoretically derived target rolloffs are useful is that when they are properly implemented, we can predict their various kinds of behavior such as phase tracking, vertical lobing, etc. With today's CAD available, however, we can always try many dfferent variants of them with good prediction and results.
But for LR2, this trick does not work nicely, due to the required large overlap between drivers. That's why people use a slanted/stepped baffle or a delay network for time alignment of LR2.
In your LR2 implementation with the 27TDFC, there should be some misaligned phase due to the use of flat baffle and/or the 27TDFC's early rolloff. But the question is, is this necessarily bad? I do not think so. As long as you have a flat SPL sum at your listening axis and voice your speakers in your room, you'll have no problem.
The reason why some theoretically derived target rolloffs are useful is that when they are properly implemented, we can predict their various kinds of behavior such as phase tracking, vertical lobing, etc. With today's CAD available, however, we can always try many dfferent variants of them with good prediction and results.
jerko said:
Yeah, the Xmax study in your first link sums up my experience. 6db electrical allows too much power at lower frequencies to reach the tweeter. It probably won't damage your tweeter to hit .3mm excursion, but it will raise distortion.
Also, higher order filters seem to negate the effects of acoustic center mis-alignment between drivers whose flanges are probably bigger than they needed to be.
A test I did a while back....
My test music consisted of Jimmy Buffet, Beatles, Tool, Sixpence None the Richer, Rush, Pink Floyd. I have a Dayton RS52 mid and Scanspeak 6600 tweeter. I experimented with 1st through 8th order filters digitally. I also varied the crossover point between mid and tweeter from 1500hz through 20khz in 1khz steps. I couldn't get any combination of 1st order filter with crossover point below 10 khz+ to not sound distressed on most test songs with the exception of Jimmy Buffet tracks. In particular, the Tool sounded bad with 1st order.
There was an allure to the 1st order. I wanted it to work because on the test tracks that sounded good, they sounded REALLY good.
The tracks that sounded best were vocals, not a lot of bass or drums. Maybe an accompanying acoustic guitar.
2nd, 3rd, and 4th order electrical all had things about them I liked.
As the order of crossover increased, the differences became less and less. 3rd and 4th sounded most similar.
And continuing to higher order slopes, the LR8 sounded similar to LR4, but with a pronounced electronic/processed kind of artificial quality.
The author you linked to noted that he had trouble hearing differences between crossover slopes when using a digital crossover. I've heard that same thing. The subjective difference I hear between 18db and 24db is pretty subtle. I speculate that the differences people hear may have more to do with number and tolerances of components in the crossover and less to do with the slope.
If the acoustic response of the LF tweeter roll-off (using a simple 1st-order HPF) stays 1st-order down to the point where the signal has dropped by 10dB, then I don't think you need to worry about the fact that it will have a steeper roll-off below that point (due to the inherent LF roll-off of the tweeter). After all, at 10dB down the tweeter is only putting out 10% of the power it is in its wanted frequency region.
The earlier poster who complained that in his example, the tweeter LF roll-off rate would become greater than 1st-order when the attenuation had reached 12dB, would only be getting 6% of the normal output from the tweeter at that point. The tweeter signal will be totally masked by the output of the midrange unit at that frequency.
The earlier poster who complained that in his example, the tweeter LF roll-off rate would become greater than 1st-order when the attenuation had reached 12dB, would only be getting 6% of the normal output from the tweeter at that point. The tweeter signal will be totally masked by the output of the midrange unit at that frequency.
Ouroboros said:
The point of the discussion was NOT whether the first order electrical filer would work or not. As I already said, it is possible to have good on-axis and power responses without using exact BW1 or LR2 acoustic rolloffs. The point I made was that an acoustic rolloff resulting from a 1st order electrical filter applied to a typical 1" dome tweeter is NOT a 1st order slope but almost always closer to 2nd order.
And your number description does not consider how sensitive human perception is. The contribution of a tweeter's output at a 12 dB down point is clearly audible. I know this from my experience of voicing my speakers. If you doubt it, test it yourself by feeding a test tone into your speaker at a frequency where the tweeter's output is 12 dB down, and listening close to drivers.
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.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.