Columns Retired Columns & Blogs |
Loudspeakers Amplification | Digital Sources Analog Sources Featured | Accessories Music |
Columns Retired Columns & Blogs |
Loudspeakers Amplification Digital Sources | Analog Sources Accessories Featured | Music Columns Retired Columns | Show Reports | Features Latest News Community | Resources Subscriptions |
You should talk to teh Legacy designer, and then you would understand how top quality speakers come about and why, all your ideas are wrong, really. there is no such thing as too many capactiors, or having a crossover is bad, no single driver can make music, or anything like a live event, physically impossible....Bill d has a great explanation on how it can't be..He right, you are wriong, if you where right, everyone would have a midrange speaker doing teh full spectrum, we don't, we have full range systems using multiple drivers with crossovers.....how did you ever come up with thinking a mid range driver, ONE, is a full range system, no one thinking clearly would think that way..... And what kind of crappy amplifiers are you talking about that can't handle any load a speaker presents, you are buying some more inadequete stuff. If teh amp starts out at 3% distortion, how does that make for a more realistic reproduction? you are changing teh sound of anything going in, on top of too small a driver to move anything approaching a live concert of even 4 pieces, let alone a full orchestra, DUH....one midrange can't make a flute sound real off a recording. Remeber when Rdy Valli sang through his megaphone, didn't that sound so full and clean, no, it sounded constricted, lifeless, and like a sound effect...not natural voice
You don't belong on a discussion forum.
You speculate about Thiel speakers, my educated guess is that there are phase compensation networks, and perhaps notch filters in some of their designs however, none of these are evil as you describe them.
You claim that a single driver's input impedance is benign and I would disagree. A direct radiator driver in free air or in a sealed box is a very good approximation to a second order resonant system. Resonance is required to provide a flat response in the passband or what was described in the Rice and Kellogg paper as the mass control region:
http://mixonline.com/TECnology-Hall-of-Fame/1925-dynamic-loudspeaker/
The input impedance is a resistance plus inductance below the fundamental resonance (max Zin), and resistance plus capacitance above the fundamental resonance until the series resonance (min Zin) point is reached. The impedance becomes a resistance plus inductance above the series resonance point due to the voice coil inductance. Note that the fundamental resonance is what we call the motional impedance reflected into the primary. It is actually the mechanical parameters seen as their electrical analogies in the primary circuit. If you were to clamp the voice coil so that it could not move there would be no motional impedance, simply Rvc + Lvc. So, I would not agree that a single speaker is a benign load rather it is wildy reactive.
The driver motor is what we refer to as a reciprocal network in that current in the primary produces force on the secondary, but also velocity on the secondary produces voltage in the primary.
Most two way systems with a reasonable crossover follow this trend but with another parallel resonance in the midband, then the Rvc + Lvc is seen above this point. Conjugate load matching is implemented by placing a series resonant circuit in parallel with each parallel resonant point, and an R + C for Rvc + Lvc, such that they complement the curve and flatten the impedance. It is really the complex conjugate in that we reverse the sign of the complex part of the impedance so that the reactive terms cancel. What is left is a resistive load, hard to believe all those reactive components and you end up with a resistive load. Note that it takes 8 additional passive components to compensate the two impedance peaks and Rvc + Lvc. A resistive load is easier for the amp to drive, causes less heating in the output stage, less of a tendency to exceed the safe operating area of the output transistors, and less tendency to trip the output protection. HOWEVER, amplifier designers have known about difficult loads for a long time, remember the high current craze of the 1970s? Any good amp should be capable of driving 4 even 2 ohm loads, so this really is not an issue with a "good" amp. There are exceptions, if a speaker dips to 2 ohms, and is a reactive load, you might want to look for an amp that is rated for one ohm loads but this is certainly not the norm.
You question low efficiency and suggest this might be due to dissipation in all the passive crossover components; this is incorrect. Reactive components such as capacitors and inductors do not dissipate energy, rather they store and discharge it, ideally. Inductors will dissipate due to resistance in the copper wire, and capacitors due to lead and plate resistance but this is usually small to the point of being insignificant. Direct radiator speakers are inefficient because they do not get a good "grip" on the air that is being moved. The mechanical analogy would be that they do not have leverage to move the air. A good horn is a broadband impedance transformer; it provides acoustical leverage. Most direct radiator drivers for home use are .2 to .9% efficient, some direct radiator PA drivers get up to 3 or 4 % efficient. The wasted power goes into heating the voice coil. High performance speakers are a true challenge to design.
Let me also clarify your view concerning voltage sources. A true voltage source does not simply provide voltage, it provides whatever current is required to maintain the specified voltage. High current amplifiers are usually good voltage sources.
Conjugate load matching is nothing new, KEF used it somewhere around the 70s or 80s.
Pete B.
You seem annoyed by the 66% distortion figure for what was it the B&W 802 at 100W. That was measured by Don Keele in Audio I believe, who certainly knows how to perform these tests having worked for EV and JBL as I recall. Does this pass a sanity check? Going from memory the 802 is a vented system tuned to a nice low 25 Hz (Fb). Vented systems have the advantage of reducing the required cone excursion around the vent tuning frequency, however they do reach max excursion for a given SPL at about 1.2 to 1.4 times Fb. This is not to put down vented systems, they still have excursion advantages over sealed but we have to keep this max excursion point in mind. The simple answer to your question is that either the drivers were reaching beyond Xmax, or the 802 has an iron core saturation problem, or coupling to the midrange driver. These were both issues in the 801 as I recall. The simple answer is that you don't want to put 100W into an 802 at 41 Hz. B&W probably should have addressed this issue with better bass drivers or whatever was required.
Zaph's curves are taken at about 1 to 4 watts. No where near the upper end of the driver limits. All of the harmonics will increase with higher power.
Pete B.
Great stuff, based on reality and electrical science.
Pete, it looks like this discussion will very likely be between the two of us for awhile. Cyclebrain only posts on the weekends.
No, sir, That is not what I said. This is what I posted; "Given a single driver, the reactance of the load is fairly benign and easily predictable with frequency when compared to a multi-element passive crossover connected in parallel with the drivers (the most common connection type in consumer speakers)."
There is no claim to a totally benign characteristic of a single driver. Just that its actual behavior is more easily calculated (without even the presence of the driver being required) or actually measured than that of a reactive multi-component crossover feeding multiple reactive drivers. Not that the latter cannot be accomplished with a bit of information but that the real devices don't always live up to the end product of the calculation in large part because the designer either works in an on paper "ideal" situation where everything behaves perfectly or the design is made for worst case scenario and assumes some bad behavior along the way.
I'm not clear on where you're headed with this information. Of what value is a non-existent condition to this discussion?
Back EMF. Which in most cases in a multi-way system passes through the crossover (at least the ground plane of the crossover) on its path back to the amplifier and must be taken into consideration when discussing the distortion of the loudspeaker system. Correct?
And doesn't any loudspeaker system which must be driven by an amplifier (all of them) also depend on the amplifier's NFB and/or damping factor to control overall THD of the system? Not inherent distortions within the driver itself but overall THD of the loudspeaker system. There is a very important distinction to be made between inherent and overall distortion products as I see this issue.
Yes, it is all but impossible to think something like this; http://stereophile.com/floorloudspeakers/806legacy/index4.html, or this; http://stereophile.com/floorloudspeakers/205marten/index4.html, could be considered a "resistive" load. What you are telling me is common place in loudspeaker design is not matching up with what I observe in reality. A speaker with an electrical phase shift above 45 degrees while it's impedance is below four Ohms is not to my knowledge a resistive load for most amplifiers.
There are no two speaker systems I am aware of that have a similarly reactive character and therefore a similarly resistive character. How does your conjugate loading contend with the majority of systems on the market that represent "difficult to drive" loads? What has the designer done wrong to so drastically not result in a resistive load to the amplifier?
That is my argument. I am for a trend away from highly reactive loads being so commonly found in audiophile loudspeakers.
This is where you and I seem to be parting ways. I don't feel an amplifier is "good" only when it is capable of driving extremely low impedance loads particularly when that load is highly reactive. And, as I have stated, I am opposed to the trend toward highly reactive loudspeaker systems.
I have to say I find what you post to be typical of a designer who operates on paper and not in the day to day world of high end consumer audio. Please don't take that as an insult, it is just a fact I've noticed in my years in the hobby. I have long felt that speaker designers do not talk to amplifier designers and amplifier designers are quite often trying to build something that will drive what the speaker designer dreamed up without regard to what amplifier could drive their design. And, yes, I remember the high current, arc welding amps of not that long ago. They were a byproduct of speakers that should never have reasonably been put on the consumer market.
But defining a "good" amplifier as one capable of such behavior is not how I go about system building. One of the better budget amplfiiers of not that long ago was the original BK ST140, a product that received a Stereophile Recommended Component status. It was an amplifier that had far more virtues than vices and was capable of quite musical expression - as long as the load it was being asked to drive remained at a fairly high by the standards of the day six Ohms or higher without a highly reactive component. That amplifier was, to my way of thinking, a very good example of a "good" amplifier existing in a world not well suited to its talents.
Defining a "good" amplifier by your terms would eliminate virtually all tube based power amplifiers from contention. To believe that any tube amplifier not capable of driving a two Ohm load without incident is not looking at the larger picture IMO. Even with the standard fare of solid state amps, driving two Ohm loads under dynamic conditions will require a substantial power supply that only serves to drive up the cost and inefficiency of the total system.
Once again, I don't feel that's what I have stated. Yes, I do think more parts used to compensate for problems that need not exist is wrong headed, raises the system cost and lowers the overall efficiency of the system. When eight parts are required to compensate for a highly reactive load, I question the value of a highly reactive load.
The CS3.5 was designed with multiple notch filters. To some extent that was how many designers thought in the 1980-90's. My issue with overall system efficiency is why have we allowed the market to be driven by systems that require multiple notch filters and high current amplifiers?
I understand that everything is a trade off in audio but it seems we have been as blinkered about the "watts are cheap" concept as the US has been about its energy sources. I see very slow progression towards a better driver and instead a reliance on notch filters to build a more complex system. A better speaker system still seems to be represented by drivers which rely on multiple small filters to aid its overall performance.
I understand that concept. My point is, even with the use of fully horn loaded designs, which cannot be accommodated in most domestic situations, the best the audio industry can do is still a very low 10-12% system efficiency from the loudspeaker system. No different than in 1946. I am befuddled by why we have not moved beyond that point - and actually trended toward the opposite direction of lower efficiency - over the last half century plus. Low speaker system efficiency leads to power compression and higher overall distortion products, among other problems, when we ask those speakers to perform at semi-realistic levels for long enough to playback a full concert performance.
Lower efficiency in the loudspeaker results in higher cost for the "good" amplifier asked to drive such reactive loads. If we desire anything more than "good" in an amplifier, by your definition of "good", the costs rise expotentially.
Once more, I understand that. But, by your definition of "good", a voltage source with minimal current capacity would not qualify. That is exactly what I question as the route forced upon us by the current trend in audiophile loudspeakers.
To harken back to the overall theme of the thread, a higher efficiency in the loudspeaker system will result, from everything I know, in a lower overall distortion level in the system's performance. That other gains in system performance would be possible need not be restated.
We are still talking past one another. I am annoyed by the inclusion of such a figure in an advertisement for another product when less than full discloure and apparent obfuscation of facts is the order of the day. I dislike the way the ad was arranged.
Your reply does, however, raise the point of system efficiency once again and how distortion is to be measured and discussed. If the high THD results from a saturated inductor, the cause is poor parts selection. A revised parts list would rectify most of this problem I assume. If the problem is limited Xmax, we face another issue. If the THD measurement is merely the coupling of the midrange driver to the port's frequency range, then we face yet another problem. No one is denying the difficulty in designing a high quality system. I am merely stating that nothing we have available to us has reached beyond the level of introducing as many problems as we have solved.
That doesn't answer my question. Will the harmonics increase evenly with increased output? If so, that still puts the F5 at a level well below F2. I would consider the 1-4 watt to be of value since 1-4 watts represents a typical "average" listening level for most of us.
You asked to be educated and now you suggest that I misunderstood you. I took the time to read your post carefully, and would disagree that I misunderstood you so I'm going to drop out with this long winded back and forth.
It really doesn't bother me that my definition of a good amp is different from yours.
Pete B.
F4 and F5 are almost always lower than F2 or F3, you are stating the obvious, your implication earier was that there was little if any F4, F5 as I read you. It is actually possible to have a large reduction in even order harmonics through careful design, however this is beyond this discussion. Yes, Zaph's tests are useful but not representative of a system playing anywhere near a level close to concert levels. What is interesting is that most amps are quite clean right up to clipping, whereas most speakers are not.
I expect marketing literature to offer the best case for the product and worst case for the competition; this is how the game is played. I don't see it as a problem as long as data is not misrepresented.
Pete B.
Pete,
Thanks for all of the information and the descriptions. Fascinating stuff.
I appreciate your effort posting.
Yes, how is that a contradiction? If I ask a question and you supply an answer to another question, what am I to say?
And I took the time to read your's. If a simple disagreement is what has caused you to withdraw, this forum is going to be rough on you.
It wouldn't bother me either except you say you are responsible for the design of crossovers.
The idea of the forum is to exchange similar and differing opinions and information, Pete. If we all agreed all the time, there wouldn't be much to say. That's been proven.
Here's what I posted; "From all I know the major distortion product of most drivers will be harmonic in nature and low order in content with very little above third order." If what I post is obvious, I would also assume it's true. It was you who suggested high levels of F4/F5. I can't find them in the information you have provided. I asked if the distortion products were linear in their content as level rises but your reply hasn't supplied that information.
And most people can ignore short term clipping unless it is of the "hard" variety. A vc reaching the limits of its excursion would be similar in a speaker to hard clipping in a transistor amplifier I suppose. You certainly should know when it's happening. However, depending on the type of distortion product and its nature at onset - similar to the soft clipping of a tube amp - the total THD of a system might not be that noticeable. The B&W's seemed to survive their problems to become a recommended product by many reviewers and a reference for others.
That would seem to be the gist of this thread. How important is a total THD spec?
Once again we disagree.
Perhaps loudspeaker distortion is a bit like LP distortion. I posted this summary of an interview with Ken Kreisel (of M&K speakers) published in the February, 1996 issue of Audio over 10 years ago. I am PB2 at this forum:
http://www.diyaudio.com/forums/showthread.php?postid=1016793#post1016793
Perhaps it's not. I don't remember discussing inner groove distortion in a loudspeaker. What's your point?
I've never heard anyone suggest that low order harmonic distortion is bad. Yes, it would be ideal if there were no distortion but we haven't reached that point yet and aren't likely to in the near future. Since distortion is present in virtually every component I'll settle for a "pleasant" amount of distortion of a type that doesn't set my teeth on edge or make me leave the room.
Jan IS distortion of the ODD harmonic, not pleasant. Makes me leave the room. Jan is certainly not CLASS A...no wait, he might be a CLASS A, in the entire order as in the WHOLE thing. CLASS A WHOLE No crossover, no watts, no drivers, he's an audio UNICK....has nothing to use!!!!
upd, if there was another comment about 5" speakers in there, I'll never know. I've finally had enough of you and you are on my "ignore" list from here on out. Toodles, upd!
Buddha, I have this article. I will dig it up and post. It was a 2, 3, or 4 speaker comparison article. I think the Totem Mani-2 review was in the same issue. As I said, I'll dig it up. The Velodyne claimed unprecedented low distortion measurements. Listeners found it wanting. I believe there was another (Acoustic Energy?) speaker reviewed in the same issue.
Dup, you refer to "reality" and "electrical science." You have a working knowledge of neither. Shush.
The best way to evaluate speakers is to listen. Period. The huge variety of available electronics out there, today, makes the speaker a moving target, but you just have to do the best you can.
You can only measure what comes out of the speaker. And what comes out of the speaker is what went into it. Amplification, source components, and software. Of COURSE the room matters. But that is between you, your achin' back, your patience, and what you want to do with acoustic treatment options. None of this can be isolated within the boxes or panels. All you get, in the final test, is what you hear, and where you hear it. The problem with speaker distortion measurements is that they can't be isolated from the driving electronics and the room in which they are asked to perform.
Are you measuring the speaker, or the speaker driven by what went into it? The speaker has nothing to say unless you plug it into some music. When choosing a system, you must listen. With software that represents what you love most about home music.
This thread is interesting, but, once again, it has become a mere excuse for DUP to reveal his profound ignorance about all things audio.
If you are insecure about your own ability to make the right choices, you haven't listened enough. Some abstract number, in false percentages of distortion, won't help you. If you want to make technically sound decisions about speaker distortions, you will have to learn about how amplification and input source material interact with them. They do not "speak" unless they are plugged in. Distortion measurements take you backwards, not forward, into the electronics that are driving them.
Frequency sweeps, waterfall plots, impedance curves, and on/off axis plots can be useful, but only in terms of choosing optimal system components. Ultimately, you have to listen for yourselves.
DUP can't help you. All he can do is ridicule others' comments and shamelessly hawk Legacy models. You can either listen with your own ears, or you can listen with somebody else's. Or, you can read the non-musical text and graphs, and go a-guessing. Good luck, on the latter. I have already tried it, and it didn't work.
Just listen. Happy tunes.
Pete,
Do digital crossovers have less negative byproducts than analog crossovers?
For example, if I recall correctly, in a first-order analog crossover network,the phase difference between the low pass and high pass networks is 90 degrees at the crossover frequency. To obtain in-phase response, a second-order or higher order crossover can be used, but this adds cost and complexity and this changes the crossover slope, etc.
Does a digital first-order crossover exhibit phase issues? If so, can these be readily cancelled out in the digital domain?
As another example, is there phase shift in a digital EQ circuit?
I am trying to understand the degree to which the interplay between these factors is simply the nature of the beast, and how much is due to using analog v digital to accomplish the task.
Would we have less speaker distortion if we used digital crossovers?
That would be a reasonable area for many speakers to have problems. However, it would appear you are missing or not mentioning how things work both in music and in how we hear music. Many instruments have considerably higher output level at the second harmonic than at the fundamental. Many rooms would prop up and add to any problems in this area.
Fortunately, our ears do quite well at replacing missing information when we hear the correct clues from frequencies higher up the instrument's range. There isn't room in this forum to list all of the speakers music lover's have come to appreciate that would not perform well in a total THD measurement beneath 70Hz.
As you yourself state, if you desire room shaking bass down to beneath 20Hz, you'll know rather quickly these are not the type of speakers you should consider unless you plan to pair their superior virtues in other areas with a quality subwoofer.
I don't understand your terminology. How would a speaker be good at "hiding" distortion other than to render it inconsequential?
The test article you linked to indicated IM distortion in a driver (only a single driver, not a speaker system) is largely predictable by extending the plotted curve from F2 through F5. With F5 appearing as much as 100dB beneath F1 in most of those drivers the test results would indicate to me that IMD is not a major problem in most well designed drivers when they are not pushed to their limits. At average listening levels IMD doesn't appear to be much of a concern in modern speaker systems.
You are stating the obvious. What you are not stating are the numerous possible causes of this type of distortion and how they might affect the average listener.
I have no problem allowing you to believe what you wish to believe as long as you allow me to believe what I wish to believe. It's when someone begins to suggest there is only one way to approach this hobby that I have real problems.
Stereophile is a subjective review magazine - always has been. I hope it remains as is, a verbally informative channel to disseminate information about high quality audio products. If you would like to see more tests and less text, there are other sources for that sort of reviewing.
I always read the "Measurements" section of each review. There is generally valuable information to be found in each review test. However, each review test really begs for another review test. Without a full rigor of tests, other tests results are too often left in question.
Personally, I am far more interested in the subjective review content of a subjective review magazine. That's why I read Stereophile. If a reviewer states the perceived existence of a problem, then that sets off the warning LED's and I know I should pay particular attention to individual aspects of a product's performance if and when I have the oppportunity to give it an audition. The B&W mentioned earlier was never regarded as a rock and roll speaker. Therefore, a few words that say in effect, " The speaker doesn't carry the full weight of a rock concert when pushed to what would be considered realistic levels for such performances", would be sufficient warning to me that I might want to consider another option if loud rock played at full bore levels will be my go to musical choice.
If I do not listen to music that would show off the limitations of the speaker, I might still be interested in what else the speaker does well. To include more tests which will be meaningful to a specific group of readers will only cut down on the amount of room provided for what is the magazine's main purpose for existing - subjective reviews of the product reproducing music.
The B&W 801/802 won "recommended component" status in many a magazine and in many a music lover's home and also found its place in numerous high quality recording studios as a reference monitor. Those people who bought the speaker probably never heard the distortion component mentioned in the Legacy advertisement or it never bothered them for their specific use. The subjective term "Not a rock speaker" was probably sufficient advice to keep most headbangers away from the B&W. What total distortion measurements wouldn't have provided with these speakers is a clue to the many other charms of a true reference quality loudspeaker.
JGH made the point many years ago the only real usefulness of harmonic distortion tests is to reveal the spectrum breakdown of the nature of the distortion component. With no proof provided other than some drivers/speaker systems display low F5 and that IMD is not something to overly concern ourself with there doesn't appear to be much value in total distortion measurements. Static measurements using either single frequency test tones or frequency sweeps will reveal different aspects of performance. Which is most important to any one reader is something that would make such tests relevant to a specific group and more or less meaningless to another.
I see no need for loudspeaker THD measurements to be included in Stereophile. Give me more text about how the product does or does not involve me in the music's intent.
Elk, you should read some of Meridian's white papers on this topic.
I've seen them.
I treat all such publications as marketing - read with a very critical eye.
I feel that's a bit cynical when the author has proven themself to be of value to the audio community but that's your choice. How do you know when what you are reading isn't driven by the author's desire to prove their own point of view?
Earlier in this thread Pete posted this;
That was in response to my statement that the 22 element crossover in the dramatically low impedance Thiel CS3.5 was in part responsible for sucking "the life juice" out of an amplifier.
The Marsh article Pete linked to later in this thread can be opened from here; http://search.yahoo.com/bin/search?fr=yb...capacitor/marsh
Marsh states in this article;
Marsh then goes on to list the ways signal is lost to heat within capacitors, the single most common passive crossover device in use. I can think of no multiway speaker that does not employ at least one cap in the signal path.
As I read Marsh's article, I feel I am correct in stating signal loss is common in high component count crossover networks and even more so when the system impedance is low. I can only conclude that signal lost to heat results in a lowering of the overall system efficiency even when voltage sensitivity remains similar to another product with a lower component count. In other words, two speakers spec'd at the same voltage sensitivity will not necessarily play at the same level when fed "X" watts in.
I assume Pete won't reply to this matter since disagreement is not why Pete came here. I would, however, be happy to hear how others interepret the information included in Pete's linked article and what implications it might have to loudspeaker efficiency,system efficiency and loudspeaker distortion.
Nothing cynical about this at all. White papers are there to sell a product. I similarly analyze drug company epidemiological and efficacy studies with care.
Critical, careful reading is always appropriate when reading marketing materials.
You are of course free to accept all marketing publications at face value if you so choose.
How do you know when what you are reading isn't driven by the author's desire to prove their own point of view?
The same way all intelligent readers do.
Please put me on ignore. I will make a few more comments then drop this with you, if you let me. What you have done here is demonstrate your ignorance about capacitors and reading technical, really marketing literature. What you have to do to see your mistake is to look up the Q, or DF, or ESR for caps that are commonly used in crossovers and then calculate the heating under typical conditions. Your example is the Thiel which I believe has all polys so that should make your job easier. If you do this, you will learn that yes caps sometimes get hot (usually electrolytic types in highly demanding applications such as switcher power supplies) , but those do not under any real world situations. You would know this if you ever built even one crossover, or even simply outboarded one but again you demonstrate your ignorance. You read that "paper" desperately attempting to prove yourself. You are uneducated in my field, and you should give it up. I learned that caps do not normally get hot as a young child.
I notice that you've repeated that you think Zaph's tests are fairly representative of real world use. Let me ask have you ever burnt out a tweeter?
How many average watts do you think it takes?
Have you ever burnt out a typical 6-7" mini monitor woofer with a 1" voice coil, say 3-5 mm Xmax, aluminum former? How many watts do you think it takes?
What do you think of people who burn out speakers, they must be playing them quite loudly wouldn't you say so?
Clearly, this is the other end of the spectrum compared to Zaph's tests at about 1-4 Watts.
Pete B.
Wow, quite the block of posts.
As for the complaint of low speaker efficiency. Yes it is low. Is it because of the crossover? Does a bi or tri amp setup without a passive speaker crossover gain efficiency? I don't know, but I would guess not.
Unfortunatly system inefficiency being low is normal in all fields.
There is only so many BTU's in a gallon of fuel. Despite what we all have heard, there is no way to have a 100mpg carburator.
Any photocell will always be pretty inefficient relative to energy in vs. energy out.
The losses always seem to be big when there is a conversion in the type of energy. Electrical to mechanical. Mechanical to electrical. Thermal to mechanical. Chemical to mechanical.
Sorry.
As for the question about digital filters. While digital processing can do amazing things it also has its own problems.
But digital crossovers only exist in the low power side (amplifier input) and not on the high power side between the amplifier output and speaker input.
Welcome to the real world.
Thus the questions.
What are these problems? Are they similar to the problems with analog, such as phase shift?
Excellent point.
From what I've seen on this forum, if you grant that most of us are "intelligent", that would imply reading something that already agrees with our own opinions.
You may choose to read only those things that agree with your already chosen views. I take a much broader approach - which is what most intelligent, interested people do.
I answered your off-topic question only because you presented it multiple times.
Now do you have something to add to the topic itself? Any personal knowledge - not Googled - regarding digital filters?
Pete! I thought I was on ignore?
Hmmmmmmm ... so you weren't telling the truth about that either.
That was not my point nor was it Marsh's reason for using the words "heat producing losses". The point is the word "loss" not the word "hot". The word "hot" does not appear in either my post nor in Marsh's article to which you first linked. Heat is one way energy is converted to loss in any audio component even if the component stays cool to the touch. An inefficienct class A amplifier will have more energy consumed and then converted to heat than will a more efficient class AB amplifier. Now, what's your argument to that statement?
Yes, that's right, Pete, I was desparately attempting to prove myself. To who? You? Don't flatter yourself, Pete.
How did your insult go earlier? Oh, yes, ever hear of projection?
What I learned as a child was to read everything with comprehension. Something you obviously were not taught since you answer questions that have not been asked and link to articles that don't exist. Maybe you should give up this discussion forum stuff since you're not very good at it.
Here's my point, each component part represents loss, that loss is typically seen as a conversion to heat - not "hot" though as the component gets "hot" the losses increase. If one part has loss, more parts will equal more loss. If "hot" indicates more loss, even more "heat" will result in even more loss.
What is likely to make a component "hot"? How about a load that requires large amounts of voltage and current passing through it? If the driving amplifier is a voltage source, wouldn't such a load be one that begins as a low efficiency system which then dips to 2.5 Ohms with a high electrical phase angle at the same frequency? At that point, poly caps are just a bit more insurance the crossover will survive any abuse.
I did not claim the crossover was the major culprit in the low efficiency of the Thiel speaker. I said the high parts count in the crossover contributed to the overall inefficiency of the system, particularly as the system hits power compression which it is likely to do if the user wants to play at high SPL's. If each component has loss associated with it and there are more parts, there will be more loss and more compression and, therefore, lower efficiency of the total system. Less of what goes in will come out.
No, in all the years I've been working with audio components, I have never burned out a tweeter. Not one, not even when I was demonstrating speakers at high SPL's. It doesn't take a genius to hear amplifier clipping or driver distress. In my own systems I've always had sufficient power and system efficiency even when using a four watt amplifier or an 82dB LS3/5a to not be in danger of blowing a tweeter.
I'm at a loss however, as to where you are going with this.
Another trick, Pete?
I assume most of us know it is far more likely you will damage a tweeter when you have less power rather than more power and when the speaker's total efficiency is low rather than when it is high. That would argue for higher system efficiency.
Nope, haven't done this either. Like I said, it doesn't take a genius to know when to turn it down. It does take a bit of a fool to not realize they should turn it down especially when that voice coil starts banging away at its limitations.
I assume you have burned up drivers. Is that where you're headed with this, Pete?
No, not if the system efficiency is low or their preference is to abuse their equipment. I would say anyone who does this with consistency isn't the brightest bulb in the marquee as would most driver manufacturers when they stop honoring warranties.
I assume this is something you've done, eh, Pete?
How's that? If I take a rather pitifully designed table top system's amplifier, what, a couple of watts best, and use it to play "The 1812 Overture" through my 3/5a's at anything more than a whisper, I won't have 3/5a's when I'm done.
If you are of the opinion that all music must be played at 120dB to sound "good", and you believe the only "good" amplifier is one that can drive a two Ohm load, you and I are not on the same wavelength here and you should be responding more to upd.
I know most people listen at an average level that requires no more than 1-4 watts and that that requirement goes down when system efficiency goes up. How many watts does a 104dB speaker system require to play at average levels? How many at peak levels?
My point here has always been the total system efficiency should be going up which is not what's been happening over the last forty years, and when system efficiency rises, speaker THD will decrease. We are throwing away energy as it is converted to heat loss due to low system efficiency.
What are you talking about, Pete? Frying eggs on your speaker's crossover?
You seem awfully snippy, Elk. I didn't think the question was off topic. And so what if it was? This forum hardly stays on topic for long.
No, I've never dealt with digital filters, only read about them. Someone using a single driver speaker generally doesn't have much use for filters. And, at this point, reading about digital filters in consumer audio is somewhat like reading about hen's teeth. Obviously, active filters don't present the same problems as passive filters and digital filters supposedly do not present the same problems as analog filters. I do know the use of active filters of the analog or digital variety raises system efficiency but for many reasons not related to the many problems of passive filters. But, with a well designed SDFR, it hardly matters at all.
I do not think this is true. My understanding is that digital filters have significant issues, often analogous to analog filters, which is why my hope to get some solid information. Those that incorporating digital filters in their designs do indeed imply that digital solves all.
Bits is bits. If you believe that, upd has an CD player for you. Cheap.
Generally, digital filters are DSP equivalents to their analog counterparts. So with the same settings they have the same phase shift, ringing, and so forth. However, and understand I'm not a DSP expert, I believe digital filters can do some things that analog filters cannot. Or at least do things that would be very difficult or expensive to do as analog.
--Ethan
Digital filters should remain very constant and on spec over time and with varying conditions which would otherwise affect the typical analog filter. Voltage and current should not affect their performance as it does analog filters. Ideally a digitial filter is used in a system that remains in the digital domain from source to speaker driver. Conversion from analog to digital and D to A along with the "problems" of voltage to current conversion and back (as discussed in FK's review of the Krell Evolution 505 in this month's Stereophile - though some might view those remarks as manufacture propaganda) are also avoided. Of course, all the parts must talk to one another in a similar language and at the same rate or you find the problems most recording studios encounter.
Yes. There are many that are specifically made to replicate analog filters and processors. This makes sense in audio recording and mastering where the entire history sounds arose out of analog.
This is the part that I am wondering about. For example, as you know, various sample rate conversion programs employ various types of phase filters, etc. Yet they still have distortion artifacts. I am curious as to why this is.
Why is there any distortion at all? We have lots of processing power available, 64-bit floating data, etc. It's not an issue of accuracy or inability to handle the numbers.
Do digital filters, by their very nature, have some of the same artifacts as their analog counterparts? It appears that they do, even when operated solely in the digital domain. Why?
It's not so much that the goal is to imitate analog filters, but this is just how filters are implemented. But instead of using capacitors and inductors (or simulated inductors) to create the needed phase shifts and feedback loops, they use DSP code and shift registers. Here's the key: The basic algorithms are identical.
Yes! Again, it's because the algorithms are identical. A high-Q analog filter will ring due to its Q and "that's just what happens," and the same "just happens" occurs with a digital equivalent.
--Ethan
First of all I'm an analog guy. I have an excellent understanding of what is happening in one of our radar systems up to the point were I hand the digitized data to the processing guys. How they can take a string of sampled amplitude returns and convert them into a photographic quality image makes my eyes glaze over evey time I ask for an explination.
Digital processing can do anything that you can come up with a formula for. If you use the formula for a traditional capacitive/inductive first order, second order, etc, filter the results will be the same as its analog equivilent. But if one then adds a formula for phase shift cancellation equal to the phase introduced, then you have just done something that would be very difficult if not impossible to do in the analog world. All ideally done lossless too. But on the negative side, many of these formulas get info from other formulas or make incorrect assumptions.
Also for a vinyl/tube person the A/D to D/A part is unacceptable.
Let me just mention first that sample rate conversion between non-integer related rates is a very complex problem with no equivalent in the analog world. It is best to work at integer related rates when possible. I mention this so that perhaps you can pick another example.
Ethan is correct in that the first way we learn how to design digital filters is by building the digital equivalent of an analog filter where the theory is well developed. These would be the first homework problems in a DSP design course after the basics are covered. However, there are some very neat things about digital filters, it is easy to implement linear phase filters, and easy to have them be very steep. Delay for say time alignment is easy, parametric EQ is easy, and more.
Multiway speakers are a special case because the total output is the sum of two or more filter sections and this is a bit more complex than a simple highpass or lowpass. The sum of linear phase HP and LP filters is not necessarily linear phase. There are very few linear phase system solutions, not saying that this is critical just an observation.
You should also look into FIR and IIR digital filters which might answer some of your questions. Saturating arithmetic is also important when the amount of precision is limited in a processor. Some of the digital nasties that you are referring to might be the result of misuse, what happens when a digital filter is overdriven, does it clip?
It is a complex subject, but I agree today where speed and precision are inexpensive we can build some excellent systems.
You might like this thread, and in particular Eva's posts, I agree with much of what she says, see posts #11, 22, 30, 45 for example:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=127643
She overstates the issues with passive crossovers a bit, but covers the tradeoffs in the process. We can get by with passives just fine for most applications.
Are you aware that there are loudspeaker design tools where you can prototype the crossover in the digital domain, use a multichannel sound card to implement and evaluate the filters before prototyping the analog versions? Assuming you're going to passive analog.
Perhaps a new thread should be started for the discussion of digital crossovers, and loudspeaker processors. I just want to mention that when I speak of distortion I'm referring to non-linearities in the transfer function that result in harmonic, inter-mod, or any type of distortion that introduces spectra that were not there in the input signal. I do not consider non-flat amplitude or phase response "distortion". So when you speak of digital crossover filters correcting for "distortion" do you mean amplitude and phase or non-linear types? It turns out that if the non-linear characteristics of the driver are known, it is possible to "pre-distort" the signal so that the cascade of the two transfer functions is more linear. This is a very advanced topic and I've never seen it implemented in any real world system.
There was an old analog tape recorder that used analog pre-distortion to get more head room out of analog tape, this was the subject of an AES article.
Pete B.
Why? A/D/A is audibly transparent these days, so it seems unlikely you'd even notice that part. Especially if you're starting with vinyl and tubes where distortion and other artifacts are much higher to begin with. Even when someone prefers the sounds of tubes and vinyl, modern digital can capture that exactly. So it's a win-win for everyone.
--Ethan
Yes, it's a valid technique, and here's my take on that:
Pre-Distortion Techniques
--Ethan
Makes sense. The distortion spectra that I have seen (and have generated myself) from various SRCs have been conversions from 96kHz to 44.1kHz.
Yet wouldn't doing the conversion at the lowest common denominator and using lots of calculation bit depth resolve this issue? (Apparently not, or all the SRCs would be wonderful.)
This is what I would expect.
However there seems to be the need for compromises. For example, Wadia's CDPs giving up flat frequency response to 20kHz in their digital reconstruction filter in exchange for rolled off highs but better phase response.
Why is the tradeoff necessary?
Snippy? Jan, Elk never sounds "snippy." But there are others who occasionally do...
With all those tines, though, he certainly looks horny.
I am rather enjoying this thread. The woodies and the mere listeners are havin' at it. I never knew there could be such joy and strife in the mere capacitor.
Cyclebrain is an analog guy. So am I. But I am willing to wait and keep trying the new digital brainstorms -- wait and try, wait and try...
Hey. How long does this go on, until I croak? When doth the promises for fulfillment given in wired media get their fulfillment in a new dispensation that actually sounds better?
Last time I heard a live symphony concert, it was in analog and real air was carrying the message, from the composer to my heart and brain, via flesh-and-blood creatures sawin' wood, beatin' on drums and pipes, and blowin' brass.
I loved it. There wasn't a DAC in sight, and no sampling was allowed beyond the roped areas at the outskirts of the bar.
Still, being gullible and hopeful, I keep trekking on, to all the shops that sell the latest capacitor-less digitized modernities. And it keeps on sounding sampled, chopped, mutilated, and coming out the wrong end of the horn. There always seem to be at least a few precious notes missing in action.
I remain patient.
Really, Elk is the least snippy person who posts here. And he actually knows a diode from a dildo.
Wow. Is the brain all analog these days?
All this time I thought neurons were kinda digital in how they fired.
You don't have to convince me, I am not one that has a problem with digital conversion. I just know that there are those that will always be able to hear the inferiority of digital either real or imagined.
Just to be clear, I am "an analog guy" by profession, not by some unfounded belief that analog is good and digital is bad. I know that digital processing works, I just don't completely understand it.
Yes, but don't forget our a/d and d/a converters. Eyes, ears and vocals.
Most of us occupy this camp, digital remains mysterious - at least on the edges.
Pages