I'd add selectable EQ (or bypass) on top of whatever crossover you have. I can make a static crossover (active or passive) to bring the best out of good recordings or make bad recordings tolerable. But a flat, tapered, BBC dip, high DI etc... etc... response won't always suit the recording I'm playing.
Sure you aren´t mixing tweeter and woofer up here or are you joking?
99% of the designs I´m looking at have the tweeter more sensitive cause otherwise how would you make a passive XO.
Am I missing something here?
DSP, FIR, passive, all useless if you don't know how to measure properly and understand what the measurements are showing...that should be the first thing to learn for a DIY'r
Just about every passive XO I see posted is premised on exactly comparable output with the drivers or faith in the manufacturer's specs. The XO takes no account of the difference between loudness on-axis and power output of drivers (reflecting directivity index and, in practice, room furnishings). So you can't predict ahead of testing just which driver needs trimming.
Sure, you can always stick a resistor in series with a woofer and get boomy sound. Not an audiophile way to enhance bass, eh.
Instead of a snarky and disparaging remark ("I think that question has been answered. You just don't seem to understand the answer.")... perhaps you could simply state in a few words just what "the answer" happens to be that I don't happen to get.
Sure, you can always stick a resistor in series with a woofer and get boomy sound. Not an audiophile way to enhance bass, eh.
Instead of a snarky and disparaging remark ("I think that question has been answered. You just don't seem to understand the answer.")... perhaps you could simply state in a few words just what "the answer" happens to be that I don't happen to get.
You're saying that with active crossovers and DSP available you can 'predict ahead of testing just which driver needs trimming'?
Come on. 😉 Active filters and DSP are very powerful and useful tools, and a perfectly valid approach to design and operation, but there is no need to imply alternatives are without value or validity, or insert spurious reasoning. What is rather unfortunate about most of these threads is the partisan attitude that frequently emerges. It's really not necessary.
That is not what was written. The insertion of series resistance to artificially raise effective driver Q is not unknown, and providing it is either accounted for in the enclosure alignment from the outset, or deliberately done to hit a desired alignment, there is no reason why you should get a 'boomy sound' unless that happens to be the objective. You seem to be announcing ignorance on the part of others without first considering they may know what they are doing, or properly reading what they said. In point of fact, all good loudspeakers should account for relevant series R in the enclosure alignment
Last edited:
Scottmoose -
"Partisan" is in the eye of the beholder.
You seem to have overlooked the importance of polar response and the room in greatly shaping the sound at your ears and yet is totally overlooked in the XO design.
Inconceivable to me that a speaker would be good in a plush furnished room and good in the hard-surfaced rooms all but universal today.... unless you could re-tune the speaker of the system. Scottmoose, can you tell us what design tool asks the user "How is your room furnished?" Or the DI of your drivers.
Back to something you may have forgotten from Audio 101: the series resistor greatly damages the damping factor of the amp. Whatever minor arm-chair conceptual influence Qes may have on the speaker box, the one-note undamped driver resonance is a terrible thing.
"Partisan" is in the eye of the beholder.
You seem to have overlooked the importance of polar response and the room in greatly shaping the sound at your ears and yet is totally overlooked in the XO design.
Inconceivable to me that a speaker would be good in a plush furnished room and good in the hard-surfaced rooms all but universal today.... unless you could re-tune the speaker of the system. Scottmoose, can you tell us what design tool asks the user "How is your room furnished?" Or the DI of your drivers.
Back to something you may have forgotten from Audio 101: the series resistor greatly damages the damping factor of the amp. Whatever minor arm-chair conceptual influence Qes may have on the speaker box, the one-note undamped driver resonance is a terrible thing.
Last edited:
It is indeed, and aggressive advocacy and selective commentary I personally find fits the bill rather well.
Which 'XO design'? Are you claiming that polar response is not considered in any crossover design except for those that use active filters and DSP? If so, that is nonsense I'm afraid.
Very few, albeit VituixCAD attempts to move in this direction as far as possible. The last time I checked though, any competent design uses measured data, however the crossover is implemented.
Wrong, I am forgetting nothing. As far as this is concerned we are simply dealing with a change in the effective Q of the relevant leg
Effective Qes', i.e. Qes with series resistance R in the circuit:
Qes' = (2pi * Fs * Mms * (Re + R)) / B*L^2.
Thus effective Qts', i.e. Qts with series R in circuit becomes
(Qms * Qes') / (Qms + Qes')
Thus, a driver with, for example, a baseline Qes of 0.333 and a Qms of 1.981 has a Qts of 0.285
The same driver with 2ohms of series resistance in circuit has an effective Qes' of 0.428, an unchanged Qms of 1.981 and an effective Qts of 0.352 [nearly].
Perhaps you could explain how an effective Qts' of 0.352 is underdamped, and, for example, the attached box alignment is likewise. None of this is new, surprising, news or anything else. It's been known and accounted for (and in some cases, exploited) since the 1940s in various ways. As noted, almost every loudspeaker will have some series R to contend with from wire loop & connection resistance / losses, + any passive components e.g. series inductors for passive designs, and these should be accounted for in a well-designed box / baffle alignment as a matter of course.
Attachments
Last edited:
>the one-note undamped driver resonance is a terrible thing.
Depends on where that one note is. I bet if it was at "20"... Ok, "30".
Then there's a debatable point that high amplifier damping factor - with its attendant "control" - necessarily gives the best sound. In some audio situations, yes - but not all.
Depends on where that one note is. I bet if it was at "20"... Ok, "30".
Then there's a debatable point that high amplifier damping factor - with its attendant "control" - necessarily gives the best sound. In some audio situations, yes - but not all.
Funny you should say that. I've thought that dipole woofers could benefit from the driver resonance bump (since they need all the low freq bump they can gather). But dipoles can "harvest" the natural resonance in ways that sealed boxes, etc. can not.
Ben,
I set off with (bad) passive crossovers, then moved to DSP, then back to passive crossovers, and then added some DSP.
You appear to be in the "moved to DSP" stage.
Here's how it happened:
- Badly-done passive crossovers don't work as one might hope. The way to go is to measure the actual drivers on the intended baffle, and use a good simulator to figure out the circuitry.
- A DSP-based system can be very good. Mine was tri-amped with a DCX2496, and it sounded excellent. However, I almost never used it. Too complicated. Too slow to get music playing, and no remote control to adjust volume etc. My laptop speakers got an order of magnitude more use.
- I moved back to passive crossovers when I realised that "easy to use" was high on my priority list. Why passive crossovers? Because I can use Any Commercial Amp (TM), to drive the speakers. Those amps come with remote controls, bluetooth (or streaming options), phono stages, etc etc.
- I then added a little DSP, in the form of a bit of <100Hz room correction, and small (low-Q) tweaks to the overall shape of the response. The DSP was running inside the primary source (laptop, permanently installed) under Equalizer APO. ie, I was still feeding stereo out to the amplifier.
- Finally, I moved to an amplifier with streaming options (it shows up as a speaker in Spotify and other apps) and DSP built-in. It also handles all of my video-based sources (games consoles of varying vintages) and a turntable.
The resulting sound is excellent. It's also very easy to use - if you stopped by, you'd be able to find the stereo on your phone and have music playing in seconds. You could also put a record on, retire to the sofa, and adjust the volume using the remote.
It's very difficult to get a multi-amped DSP-based system to be this easy to use. It's much easier to design & implement a decent passive crossover, so that's the route I took.
Chris
They can indeed. Albeit that is if you make the assumption that it's actually underdamped, which is hardly an invariable. As I asked above, perhaps you could indicate in which ways an effective Qts' of 0.352 and the alignment shown are underdamped. My manful best efforts have yet to locate the 'driver resonance bump' referred to, which would be a mite difficult with a Qt of 0.354, but perhaps that only applies in a new form of physics hitherto unknown to woman-born. 😉
Last edited:
I turn on one switch on a power-bar and everything is running in my tri-amped system, once I select the music which is played by something like J-River database. Or with a click or two, immediately switch to Internet access.
Is one switch too complicated? Didn't have a power bar?
B.
Footnotes
1. Actually some stuff like the ESL speaker bias and my laptop is always on
2. "like J-River" actually I use the fancy professional database "FileMaker" which helps me locate the music by being able to search on a dozen fields on 970 "cards" and also has a built-in Quicktime player tool. Hundreds of CDs, hundreds of records, dozens of tapes stored losslessly - and many of their visual wrappings and notes - occupy about 130 GB on an single SD card.
Last edited:
I agree DSP (or digital active filters / crossovers) are extremely useful for "speed" of change. Whether within the design of a system - or applying your DSP to a new system.
I've seen people use DSP "in lieu of measurements". I would wager if one refuses to invest in measurement, then DSP is the best (and really only practical) design by ear method.
The time and money trade off is a personal one. Beyond my skills are very steep passive crossovers. However, regardless of implementation, I am weary of those that believe there is no downside / con with a particular approach. Everything I've seen to date has some negative. So if 96dB+ crossover slopes are your bag, I'm sure they can be done in DSP.... the question is - what is the negative?
I've seen people use DSP "in lieu of measurements". I would wager if one refuses to invest in measurement, then DSP is the best (and really only practical) design by ear method.
The time and money trade off is a personal one. Beyond my skills are very steep passive crossovers. However, regardless of implementation, I am weary of those that believe there is no downside / con with a particular approach. Everything I've seen to date has some negative. So if 96dB+ crossover slopes are your bag, I'm sure they can be done in DSP.... the question is - what is the negative?
You make some good points there Dave. Now and then I also get the feeling that fewer people are using a crossover simulator, now that it has become so easy to get some kind of result. I'm not sure I understand what the measurements in the first post are meant to be demonstrating. Sometimes it takes a challenge to keep your skills up.
Dave Bullet: +1, lotsa good thoughts.
Maybe I am just an old lab researcher (which means I am skeptical by nature). You need to start with a design concept. My principal concepts are giving one speaker as much of the middle-ground as possible, esp if you are blessed with wide-range ESL panels. Also, if you have a great driver (like a ribbon), try to give it as many 8/aves as testing shows you is possible.
So you crank-up your DSP and see for each driver what band they can handle, looking at FR, distortion, and impulse profile. That approach will shine a very bright light on your design concept. In the case of the woofer or sub, how the room interacts with the low-range drivers can be taken into account automatically. Anybody have a sim that does that?
I hope explaining my personal approach will be interesting to AllenB.
BTW, since doing the testing in post #1, I've moved all my XOs to L-R48. I strongly recommend people try this. It really simplifies the system to have drivers overlap as little as possible. Specifically, you can address each driver and know that phase weirdness and other XO interactions will not show up to crap-up your work once you are happy with each driver alone. But Dave Bullet may be right that some downside might yet appear.
B.
Maybe I am just an old lab researcher (which means I am skeptical by nature). You need to start with a design concept. My principal concepts are giving one speaker as much of the middle-ground as possible, esp if you are blessed with wide-range ESL panels. Also, if you have a great driver (like a ribbon), try to give it as many 8/aves as testing shows you is possible.
So you crank-up your DSP and see for each driver what band they can handle, looking at FR, distortion, and impulse profile. That approach will shine a very bright light on your design concept. In the case of the woofer or sub, how the room interacts with the low-range drivers can be taken into account automatically. Anybody have a sim that does that?
I hope explaining my personal approach will be interesting to AllenB.
BTW, since doing the testing in post #1, I've moved all my XOs to L-R48. I strongly recommend people try this. It really simplifies the system to have drivers overlap as little as possible. Specifically, you can address each driver and know that phase weirdness and other XO interactions will not show up to crap-up your work once you are happy with each driver alone. But Dave Bullet may be right that some downside might yet appear.
B.
Last edited:
B,
An LR48 crossover has inherently more phase distortion than a LR24. Plus, the system polar response transition will be more abrupt at the crossover frequency.
That's just two possible issues with higher-rate crossovers. I hope you're not suggesting these are a panacea to solve myriad problems. As in all things in speaker design, numerous trade-offs exist.
Dave.
An LR48 crossover has inherently more phase distortion than a LR24. Plus, the system polar response transition will be more abrupt at the crossover frequency.
That's just two possible issues with higher-rate crossovers. I hope you're not suggesting these are a panacea to solve myriad problems. As in all things in speaker design, numerous trade-offs exist.
Dave.
Good to hear of possible shortcomings of sharp slopes. Until a few days ago, never occurred to me to explore anything like 48 - just not traditional.
But for phase you mention, doubtful it is heard on music, aside from direct effect on wave interaction from interacting drivers. And that interaction is greatly reduced with sharp slopes.
But for abrupt polar shift, that's listening with your eyes. When you are sitting in your chair with any slope XO, you might say, "Hmm, treble sounds different... could be polar response and my music room walls". But with a 48dB you could not say, "Hmmm, as we cross the XO band I note a shift in the music." That is not a sound parameter. Clear to your hypothetical eyeballs, I suppose, looking at a hypothetical chart if anybody ever made one like that.
B.
But for phase you mention, doubtful it is heard on music, aside from direct effect on wave interaction from interacting drivers. And that interaction is greatly reduced with sharp slopes.
But for abrupt polar shift, that's listening with your eyes. When you are sitting in your chair with any slope XO, you might say, "Hmm, treble sounds different... could be polar response and my music room walls". But with a 48dB you could not say, "Hmmm, as we cross the XO band I note a shift in the music." That is not a sound parameter. Clear to your hypothetical eyeballs, I suppose, looking at a hypothetical chart if anybody ever made one like that.
B.
Last edited:
None of this is hypothetical. Easily measurable and understood.
But I've got your shtick now. Hilarious. 🙂
Dave.
But I've got your shtick now. Hilarious. 🙂
Dave.
Well yes, this technique is a good option, as indeed it uncovered a problem. We might have different ideas on what it appears to be, but it's a way forward.