There is still some or even a lot of reading on digital if you want to start pushing the limits / doing custom FIR filters, etc.
First order/Time Phase-Coherent speakers discussions
"The game is done! I’ve won! I’ve won!"
I would like to use this thread to talk about this subject which I find rather fascinating and somewhat difficult to get my hands on. I went through a course in electromagnetism in college and I have to say this is even more confusing and you won’t find the answer in calculus, physics, Einstein relativity be damned it’s not in there either and definitely not in quantum physics. Listening to the "experts" from Vandersteens and Stereophile but ultimately it all came down to a missing link sort of argument ... something like this:
"Since if a speaker can produce a step response correctly, therefore it is time-phase coherent, and therefore it must be "good".
It’s like saying humans come from chimps since they share 90% genetic content with us, but we can’t find any missing links or evidence. FYI, we share a lot of gene with the corn plants as well. Another argument I’ve heard from John Atkinson that lacks any supporting evidence and he said that if everything else being equal, time-phase coherence tends to produce a more coherent and superior soundstage, but to the best of my knowledge, nobody has been able to produce some semblance of evidence since there is no way to compare apples to apples. Speaker "A" may have better soundstage simply because it’s a BETTER design, and the claim "time-phase coherent" is just a red herring. There’s no way one can say the "goodness" from "time-phase coherence" because you can’t compare apples to apples. Ultimately it’s a subjective quantification.
I’ve been doing some simulation and I will post some of my findings with graphs, plots, actual simulation runs so that we are discussing on subjective personal opinions. Some of my findings actually shows that intentionally making time-phase may result in inferior phase problem and NOT better! (will be discussed more in detail).
Having said all that, I am actually in favor of first order/time-phase coherent if POSSIBLE. I am not in favor of time-phase coherence just for the sake of it. It’s just that there are a lot of mis-information out there that hopefully this will clear those out. Well hopefully ...
Here my preliminary outline:
1. My "subjective" impression of what is "musicality" and how it’s related to first order filters.
2. Interpretation of step-response. I’ve read a lot of online writing with regard to the interpretations but I think a lot of them are wrong. A proper interpretation is presented with graphs and simulations.
3. A simulation of an 1st order and higher order filters with ideal drivers and why time-phase coherence is only possible with 1st order filter. This part will use ideal drivers. The next part will use real world drivers.
4. A simulation with actual drivers and how to design a 1st order/time phase coherent speaker. Discuss pros and cons. And why time-phase coherence may actually have phase issues.
5. Discuss real world examples of time-phase coherence with Thiel’s and Vandersteens speakers (and why I suspect they may not ultimately be time-phase coherent in the strictest sense).
6. I’ll think of something real to say here ... :-)
170 responses Add your response
@timlub +1 "With a active crossover, time/phase alignment becomes relatively easy." Agreed, and as someone who has those, if their frequency selection allows any choice from 20-20k which mine do, then it's a simple matter for me to just think up a combination of frequency selection vs slopes to play around with various drivers and compare the results on the fly from the lp. Does 1rst order have advantages? Sure. But they have disadvantages, too, like all the other slopes. I've experimented from 1rst to 5th order (LWR, Bttrwth) and can tell you easily that, for sound's sake, I would never start with a crossover design and try to pin it onto a pair of drivers. I'd have to start with the real-world drivers and come up with the right crossover. I'd certainly consider what the crossover might look like when choosing drivers, but that's only all the more reason to give the driver selection the most consideration from the start. But, 1rst order behavior advantages are not the only consideration to good overall design. But, with active digital crossovers, there's no need to spend all that time reading all the tea leaves (white papers, testimonials, reams of MLSSA charts, searching for who could be the most unimpeachable source of info on it, etc) in an attempt to get a handle on comparing them all from afar and then try to divine which might be best. With digital actives, you just dial it up and listen for yourself...a whole lot easier and faster that way. What I can tell from that is, yeah, 1rst order can be nice, but it's not the holy grail. The holy grail is a well orchestrated and executed speaker design overall...not just one or two aspects. |
you really have a lot of homework left.... you can't figure phasing without understanding the drivers being used and how slopes and parts will affect phasing.... it has not been mentioned, but remember there is acoustic phasing as you have been discussing, but there is also electrical phase. The rare time that I see that mentioned here on Agon is when discussing using tubed amplifiers, but individual drivers can vary phase when you change crossover points or slopes or even part grades when it comes to coils. Most of you have probably seen on occasion a tweeter or midrange is wired out of phase or reversed positive and negative. This is done on purpose when phasing gets too far out of whack to bring it back in line. On paper only, at 6db per octave, a driver goes 90 degrees out of phase, 12 db goes 180 degrees (which is when you reverse terminals), 18 db goes 270 degrees and 24 db come full around 360 degrees. In real practice most of the time, these numbers are not near accurate. So 6/6 slopes can get you close in phasing, but why did Joe Di'Apolitto use 18 db slopes in his mtm designs.... harder slopes often sound better, they roll off any bumps in a drivers frequency response and they give the same type (in reverse) phasing as 1st order slopes. I have seen a ton of discussion in this forum about why 1st order slopes, but it is a rare few that understand why and how the drivers combined with crossovers affect phase and even time alignment. |
With a digital active cross-over, time/phase alignment becomes relatively easy.OK, you have just changed the rule of the game. Sure with digital processing, you could, but I was clearly talking about pure analog xover filters. Digital processing has its own set of problems though which is just another can of worms that I am not sure I want to talk about in this thread. |
As someone has already done the work, I will simply post that from John Bau. By done the work, someone who has a product on the market and advertises this is what they do. It is not such an advantageous marketing thing these days. And that is only simple analog domain in the cross-over. With a digital active cross-over, time/phase alignment becomes relatively easy. John Bau of Spica can show you on his Tecron analyzer that the Bessel 4th-order low-pass filter he uses to cross his tweeter down to the woofer on his TC-50 and Angelus, when used with the particular drivers he chooses, is also phase coherent. |
You stated that "why time-phase coherence is only possible with a 1st order filter" ... I was filling in the blanks. 3. A simulation of an 1st order and higher order filters with ideal drivers and why time-phase coherence is only possible with 1st order filter. This part will use ideal drivers. The next part will use real world drivers. |
I think your conclusion w.r.t. filters may be a wrong conclusion based around the type of filters you were using. Different filter types have considerably different step/impulse response, independent of order.I actually I never said that. Or maybe you've read my mind but wrongly :-) I don't even know the meaning of that statement above you wrote there. Kind of like when the judge asked me "Did you see that girl". I was like "What girl are you talking about?" |
It’s like saying humans come from chimps since they share 90% genetic content with us, but we can’t find any missing links or evidence. Not to derail the thread, but I just can't let an ignorant comment like this go by without comment. We did not evolve from chimps, humans and chimps (and other modern apes) evolved from a common, ape like ancestor. The idea of a 'missing link' is archaic. There are many transitional species between ancient apes and modern humans. There are mountains of evidence: genetic, fossil, comparative anatomy, homology, and much more. Carry on... |
I think your conclusion w.r.t. filters may be a wrong conclusion based around the type of filters you were using. Different filter types have considerably different step/impulse response, independent of order. You also have to accept that nothing in nature has 0 mass and infinite force so this concept of a pure step response is theoretical. Add in some confusion from the likes of MQA and minimum phase filters in DACs which provide what could be considered an artificial output, i.e. they create something from the music that did not actually exist, and .... well draw your own conclusions. It is more a discussion at that point of euphonics ... what people like, not what is accurate. Just sending a sound through air softens an impulse response. If you are recording an instrument really close to it, then adding the above minimum phase filters, you could create a very artificial response akin to having your speaker right next to the instrument, not the typical 10-10’s of feet away. What is even "natural" at that point? |
Funny ... I really have no idea if there is such a thing as a "better" speakers. From personal experiences, designing first order speakers is really hard and it usually takes a long time vs. higher order speakers. This is not a subjective observation or myth lols, but a matter of fact since since I think I am capable enough to tell the flow of time ... by counting my gray hair :-) Usual things that are hard tend to be good. A beautiful hot blond is not going to say "yes" so easily. Ask me how I know. Oh well maybe I should stick with higher order speakers ... hahahahaha |
Hi Tim, I agree with what you said. Putting a cap on a tweeter for example does not mean a first order 6db/octave. The best I can hope for is 6db/octave in a limited frequency range. I think speakers from Thiel and Vandersteens that "claimed" to be first order, but it's an approximate, but I don't want to say too much here. And as you said, there are more than time-phase coherence. There are many ways to design a good pair of speakers. And I won't argue personal preferences. I will present an "objective" point of view: here is what first order/time phase coherence. Whether which one is better is not up to me. I took me a lot of works and a lot and lot more thinking to put together what I think is a comprehensive and meaningful discussion of first order / time phase coherence approach that I don't think I saw anywhere on the web. As I said, there are just too many misinformation out there. Some people ought to pay me lols. It costs me a good life! Also I am not really looking to argue with anyone here. I will present my case, and it's up to you. I suspect more will end up disagreeing with me than agreeing :-) |
@andy2 Not sure that there is a question here. I've run many simulations with 1st, 2nd, 3rd and 4th order slopes. I've run many comparisons with butterworth and linkwitz riley in 2nd and 4th order. Even putting a single cap on a part does not ensure that you'll end up with 6db per octave. And really isn't the closest thing that you can get to proper phasing coming all the way around with 4th order slopes? 1st order are normally at minimum 15 degrees out of phase. 4th order can be very close. I'd recommend physically time aligning your drivers then comparing your slopes. I've had surprisingly good results in soundstage and imaging running odd order butterworth (say 12/18) as well as linkwitz riley 4th order. Both of these solve issues with driver peaks also. again not sure that these comments are what you are after, but I hope they help, Tim |