marchel said:
Using a flat baffle does not cause such a large phase shift that necessitates the reversal of tweeter polarity. It only causes a bit of phase misalignment and as a result changes the vertical lobing pattern. If they must reverse the tweeter polarity, it's still because they used acoustic LR2 via the 1st order electrical filter.
It can be done 
First order highpass to tweeter + time-align driver acoustic center -> flat frequency response and nice pulse shape (low phase distortion).
Only the tweeter's acoustic response need to be first order, the mid/woofer response could be 2nd~4th order. The key is to apply excessive delay to the tweeter's acoustic center. This would compensate (partially) the group delay caused by woofer's lowpass response.
I used XT25, crossed at about 4500Hz. Works good for me.
Some more pics here
Seas 27TDFC or Peerless HDS rolls off too early. The 1st order electrical highpass combines with the acoustic roll off will result in an overdamped 2nd order acoustic highpass. B&W 800D is working this way.
First order highpass to tweeter + time-align driver acoustic center -> flat frequency response and nice pulse shape (low phase distortion).
Only the tweeter's acoustic response need to be first order, the mid/woofer response could be 2nd~4th order. The key is to apply excessive delay to the tweeter's acoustic center. This would compensate (partially) the group delay caused by woofer's lowpass response.
I used XT25, crossed at about 4500Hz. Works good for me.
Some more pics here
Seas 27TDFC or Peerless HDS rolls off too early. The 1st order electrical highpass combines with the acoustic roll off will result in an overdamped 2nd order acoustic highpass. B&W 800D is working this way.
Smooth polar pattern with maximum SPL towards the listener at all frequencies is the result to look for. Impulse response can be safely ignored. Polar pattern in the frequency overlap region depends strongly on crossover design rather than on the drivers.
First order crossovers are a true pain to get right because it's very hard to match the phase of the drivers across such a wide overlap region. Even 4th-order Linkwitz-Riley results in overlap across two octaves, and overlap is much wider for first order. Consider that two overlapped sound sources only radiate on-axis when they are perfectly phase matched, which ofen requires parametric equalization. A "small" phase mismatch like 45 degrees is enough to ruin polar pattern, and any minor narrow peaks and notches in driver frequency response will rotate phase locally far more than expected resulting in fuzzy polar patterns.
First order crossovers are a true pain to get right because it's very hard to match the phase of the drivers across such a wide overlap region. Even 4th-order Linkwitz-Riley results in overlap across two octaves, and overlap is much wider for first order. Consider that two overlapped sound sources only radiate on-axis when they are perfectly phase matched, which ofen requires parametric equalization. A "small" phase mismatch like 45 degrees is enough to ruin polar pattern, and any minor narrow peaks and notches in driver frequency response will rotate phase locally far more than expected resulting in fuzzy polar patterns.
banana said:It can be done
First order highpass to tweeter + time-align driver acoustic center -> flat frequency response and nice pulse shape (low phase distortion).
Only the tweeter's acoustic response need to be first order, the mid/woofer response could be 2nd~4th order. The key is to apply excessive delay to the tweeter's acoustic center. This would compensate (partially) the group delay caused by woofer's lowpass response.
I used XT25, crossed at about 4500Hz. Works good for me.
Your mixed-order crossover has an extremely downward vertical lobe, even more so than a pure BW xover, due to the woofer's exess phase delay. Your comparison of tweeter vs woofer axis measurements show this. If you always listen at the woofer axis or below, or if you use a stepped baffle, it'll be okay.
But I don't think your approach is generally recommendable.
I made a pair of wooden stand with inclined top, providing about 5deg of slope back to the speaker baffle. Even so, the tweeter setback is still not enough, optimal alignment is at the woofer level. The stand is taller than average, woofer level is 36" and is about the listening height with a low chair.Jay_WJ said:
There's a response dip located just a few degree above optimal axis. Similar case to Thiel, Spica and Meadowlark...
As for me, the only noticeable benefit from using low-order slopes is more natural driver integration due to their larger overlap around the Fc. This makes sense because a larger overlap avoids an abrupt change of system's nonlinear distortion profile across frequencies.
If I were you, I'd shoot for LR2 with Fc around 2.5 k to 3 kHz. IMO LR2 has a sufficiently wide overlap for natural driver integration. And you'll have more robust vertical off axis performance (and more robust horizontal off axis performance, too, due to the less effect of woofer's beaming). Why not try and compare it to your current setting?
If I were you, I'd shoot for LR2 with Fc around 2.5 k to 3 kHz. IMO LR2 has a sufficiently wide overlap for natural driver integration. And you'll have more robust vertical off axis performance (and more robust horizontal off axis performance, too, due to the less effect of woofer's beaming). Why not try and compare it to your current setting?
Jay,
That is interesting. Can you provide a bit more details? For example, should the tweeter be acoustical 4th order, and the woofer acoustical 3rd order, etc?
Also, what do you mean exactly by aliging the phase? Does it mean (1) at XO point, both drivers are at the same phase, e.g. both are at a 35 degree angle; or (2) the phase difference between the two drivers is kept at a constant within the XO region?
Regards,
Bill
HiFiNutNut said:
Right. Sometimes the woofer has a little shallow slope and other times the tweeter does.
For LR type crossovers, "aligning phase" should mean that both drivers are at the same phase. For other types (e.g., BW), it'll have different meaning (such as (2) you suggested).
But keep in mind that phase alingment in crossover design should depend on your design axis and what acoustic targets you intend to implement. For example, if you mostly listen at the tweeter axis and occasionally above it, you may want to place your in-phase axis slightly higher than the tweeter axis. In this case, matching your drivers' acoustic slopes to strict LR or BW targets will not work. You'll need to adjust the slopes so that their phase is aligned at your intended listening axis. For example, see what Zaph did for his L18/27TBFCG 2-way design.
Also note that people can have different perspectives. Ultimately what is important is the speaker's on-axis (or intended-axis) and power (or off-axis) responses and the resulting in-room response. So in practice, sticking with strict theoretical curves is not always necessary. What is important is that you know what's going on with your designed xover and can use your knowledge and skills flexibly to achieve your design goal.
I'm sure I'll do in the future, but not within a few years...my job will take most of my free time.
It'll be either my "ultimate" 3-way for music listening around the SS 15M driver or "low-cost yet extremely low-distortion" 3-way around the Dayton RS52A driver, both based on the Duelund all-pass 3-way xover.
It'll be either my "ultimate" 3-way for music listening around the SS 15M driver or "low-cost yet extremely low-distortion" 3-way around the Dayton RS52A driver, both based on the Duelund all-pass 3-way xover.
Jay_WJ said:
Trouble with the RS52 is it forces you to cross around 600-800. right in the middle of the ocal range, contrary to what I think you stated before.
That is why I am a firm believer that aiming to use drivers and an overall design that can cross at around 200-250 and again at 2400 or a bit above, is probably just about the optimum set-up.
Furthermore. If you choose the right drivers and use a 3-way series x-o, you are basically guaranteed to have approx 2nd order roll-offs with resultant nice phasing etc.
Then if you use an MTM for your top section, you get a nice symmetrical vertical lobe at the sitting position as well as giving the mids extra handing power to do their job properly.
Andy Graddon said:
I'm aware of this issue. But the answer depends on a person's perspective and how the design is implemented.
There was a discussion my friend had exactly about this issue. Take a look at this post:
http://www.htguide.com/forum/showpost.php4?p=362893&postcount=60
and some posts around it.
According to Wikipedia, "human voices are roughly in the range of 80 Hz to 1100 Hz"
Both the Fletcher-Munson and Robinson-Dadson curves show that the human ear is about the same sensitivity from 300 to 1000 hz with 1500hz being an unusual point of low sensitivity.
3200 hz apparently is a bad place to put a crossover. I prefer crossovers at 1600, 2200hz and 4Khz.
I have a love/hate relationship with the RS52. I started another thread on what I thought was metallic harshness of this driver. In the end I concluded that it could be controlled with either a notch or a steep crossover. Usher probably makes this for Dayton and has a soft dome version it appears.
Yes, you will probably crossover the RS52 at 600-800hz. But, what you gain is that you can cross over to a tweeter at 4kz or higher. 3 KHz is right around where auditory therapists say that intelligibility comes from. I think this is a good trade off.
http://en.wikipedia.org/wiki/Robinson-Dadson_curves
Both the Fletcher-Munson and Robinson-Dadson curves show that the human ear is about the same sensitivity from 300 to 1000 hz with 1500hz being an unusual point of low sensitivity.
3200 hz apparently is a bad place to put a crossover. I prefer crossovers at 1600, 2200hz and 4Khz.
I have a love/hate relationship with the RS52. I started another thread on what I thought was metallic harshness of this driver. In the end I concluded that it could be controlled with either a notch or a steep crossover. Usher probably makes this for Dayton and has a soft dome version it appears.
Yes, you will probably crossover the RS52 at 600-800hz. But, what you gain is that you can cross over to a tweeter at 4kz or higher. 3 KHz is right around where auditory therapists say that intelligibility comes from. I think this is a good trade off.
http://en.wikipedia.org/wiki/Robinson-Dadson_curves
Andy Graddon said:
That's not an excuse! I still read there occasionally even after I was banned there
There are a few excellent people there who deserve some of our attention, though I know a couple other people are really annoying.If you don't mind, here are some quotes:
"Just a quick question, Jon. I wonder if I have anything to learn by getting this question answered.
Sometime before I saw a post at diyAudio in which someone argued that he would never use the Dayton RS52 in any of his projects. The reason was that he would never put a crossover point within the critical midrange telephone band. What would you say against his claim? TIA. -Max"
"It's been a long tradition in many speakers designs to try to avoid specific parts of the speech range in implementing crossovers, because with many driver and filter combinations, the crossover region has significant perturbations in axial response, power response, group delay, and sometimes also contributes an off axis lobe.
Of course, a difficulty arises- where do you define this frequency range? Typically that would be 300 Hz to 3 kHz. Assuming the need for some effective driver overlap in the transition region, this would mean an effective driver response range of 150 Hz to 6 kHz. Hmmmm, not many of those around, are there? And by effective, I mean without cone breakup, and without significant linear distortion (energy storage, such as most midwoofers exhibit in the range above 1 kHz- see MarkK's and SL's shaped sine bursts for insight on that topic).
Did you read the section on crossover topology and design, here? This explains why this particular crossover type was used to optimize the performance of the RS52 and all the drivers together. Of special note is the second paragraph below the main heading:
Quote:
The most interesting aspect is that if properly executed the drivers of a three way are all in the same relative phase with each other at any frequency, with a net total phase rotation of 360 degrees from the low bass to the treble, similar to the phase rotation of a two way LR-4 crossover, but with lower peak group delay.
With appropriate driver spacing relative to crossover frequency, this produces the most seamless response I've ever heard in a three way system, not just on music, but on more demanding and immediately obvious signals like pink noise.
To my ears, the benefits are quite audible compared with running midwoofer drivers with their higher linear distortion (energy storage) in that frequency range, and the RS52 greatly eases the demands on the tweeter, too.
But hey, this is a free country, and if they guy posting on DIYAUDIO doesn't want to use RS52s, no problem with me. You can get similar midrange performance with a TB W4-1337S, but you'll need to use close to the same crossover frequency range at the low end- it just won't do 300-400 Hz clean.
In the case of this system, also, a small compact pre-built cabinet was desired, which ruled out using a cone midrange. The RS52 is the best performing dome mid I've tested by a wide margin, and I've used many in systems starting from the late 70's and the original Celstion dome mid.
Jon"
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.