I have absolutely no idea what to make of this. There are so many fundamental problems with a passive Pluto alone, that the rest of the exercise seems wasted. Why he wouldn't just build up a proper Pluto and experiment with the correction software using a proven design is beyond me.
Now, now, boys...
Haven't many of us wondered, secretly, or even out-loud (on the forum) what might be done with digital cross-over/equalization?
It seems that the Pluto is a much easier basis for this type of experimentation than Orion.
This "experiment" seems to be one of letting the digital correction do (nearly) the whole job with the minimum of other electronics.
I do not believe such correction can adequately address distortion, so if it had been mine to do, I would have started with at least the basic active cross-over (sans equalization) and separate amps, THEN apply digital correction.
But it was not mine to do, so more power to the man who actually did it. The results do show nice improvement for the parameters he was trying to optimize. Of course, none of us (perhaps including mikecd) know the sonic comparison of this and the standard Pluto.
In any case, it would be appropriate for him to get Siegfried's name right. Even the cover of Audio Adventure had it wrong! While they can never correct that printed cover, perhaps you can update your page Mike?
I do still wonder (out-loud, apparently), what room-equalization and phase alignment might do to further refine the excellent performance of the Pluto (and Orion).
(wish I could find the spell check option on this new forum software... )
Sounds to me like MikeMcD is just playing with the names (Steve/Siegfried and Sharon/Charon), so I don't see why members here are taking umbrage on SL's behalf. And this project should also be viewed as the technical exercise it is and not as a criticism (or heaven forbid - improvement) of SL's design. Frankly I'm a bit disturbed at the tone and reaction to someone having obvious technical expertise offering a different approach to and variation on a similar goal.
FWIW, I will be applying Harman Kardon's Ezset/EQ room correction to my 5.1 PLUTO system, as well as using the processor's time alignment and bass management features. I'll try to post some impressions, but unfortunately I won't have anything very scientific as my measurement capabilities are limited.
My PLUTOs are finished!
The project is absolutely not a "stick in the eye" to Siegfried, for whom I have the utmost respect. I am happy to discuss the design and results with anyone who is interested. But, I am frankly surprised by the negative reaction it received on this forum, and I do not care to defend it against attack.
Davey: Please feel free to delete the thread if you believe it is inappropriate.
I am sure you had clear motivations, interests and goals in approaching your version of this project as you have. Hopefully they were satisfying for you.
Do you have standard Plutos against which to compare the result? I think most of us will not know what to make of your results because they are just too far from any basis we might have for comparison.
For instance, it would be instructive to see your full set of measurements from a standard Pluto. Then, at least some technical, if not sonic, comparison might be made.
Your measurements show the effects/performance of the DRC, and certainly there is impressive improvement, but it seems that there is so much more to correct than a standard Pluto would require. Speaker-level passive cross-over components are so far from ideal, I think really good ones cost more than another little amp channel.
Any motivation to push a little farther? Any plans to compare your DRC prowess on a basic (non-equalizing) active cross-over system? If so, I think that would be much more interesting and instructive to the Pluto community! It might result in an even higher level of sonic performance for you.
For instance, in the Cumulative Spectral Display plots, how would you account for the dramatic change in "stored energy" at 50 Hz? Certainly the effect of reducing initial over-driving is apparent, but what is the DRC doing to affect the subsequent time response? And does the Pluto system, with active cross-over (even without EQ?), show anything like this undamped behavior in the first place? Do you use acoustic fill in the pipe to dissipate the back wave? I also wonder about the reason for the level pots. I would have thought the DRC would manage that part of the "equation" easiest of all.
Without a standard Pluto for reference, I just cannot judge the quality (relative sonic merits) of the final result of so much reliance on digital corrections. And it would be nice to be able to learn a bit from your efforts about how effective this can be!
Thanks for you post.
Meanwhile, what is your aluminum extrusion and from where? If I could fit amps into a larger version, that would be a nice thermal solution for "standard" Pluto builds.
I do not have standard Plutos for comparison. I am not in any way criticizing standard Plutos - I simply chose to take a different approach, and I documented my results as a contribution to the field. I'm sure that standard Plutos sound outstanding, but I know that my version also sounds outstanding and I have an objective set of measurements to prove it.
There is probably a little more to correct in a passively crossed Pluto than in the standard version. But, this is small in comparison to the correction required to compensate for the impact of the listening environment.
Speaker manufacturers like to show their products have a flat frequency response. This is usually measured close-miked in an anechoic chamber. Put these units in a normal living room and measure the frequency response 12 feet away and you will see something very much like my uncorrected Plutos. Rarely are the step or phase responses published - they can't be made to look better in an anechoic chamber. I think when folks debate the merits of foil polypropylene vs metallized film capacitors in a crossover, they are dancing around the margins of sound quality.
I don't think we would see any improvement by digitally correcting the standard Pluto - not many areas in my measurements need a lot of improvement.
The impact of speaker energy storage is captured in the transfer function impulse response. The correction filter compensates for this by pushing the resulting impulse response towards the classic Dirac spike. Digital correction is also able to exercise some control over the reverberant sound field, thereby compensating for room reflections.
I used the acoustic fill specified by SL in his design notes. Yes, DRC would correct the speakers without the level pots, but this was a simple feature to add to the crossovers.
The enclosures are from Hammond.
But go easy on the capacitors there - I am courting Teflon Foil, at least in my mind. I think I love her, but her tastes might exceed my budget!
No question that the room is a BIG deal. I personally recognize that past a certain point of refinement, it is a major, or THE major issue in accurate reproduction. And no single speaker could be optimum for every listening environment.
Thanks for showing us your results. As you might have noticed in the thread above, Cameron intends some DRC with his stock Plutos. It will be interesting to hear his experience, before and after, though I don't think he will be able to match your measurements to facilitate comparison of technologies.
One more thing. Now you have a speaker that conforms to your microphone. What mic did you use? Any special calibration?
And how does it sound to you compared to other speakers you have had in the past?
I use the Behringer ECM8000 microphone without calibration. Again, it's not going to make much difference if it's off by 0.1 or 0.2 dB. This system sounds significantly better than other uncorrected speakers I have used. But, about the same as these speakers when corrected.
I noticed some excitement here over Sharon. I am not sure what the project is about except to apply DSP to a passive xo loudspeaker and use DSP for room correction. Seems perfectly plausible to me and has been done with varying success before. I do notice though an underlying belief that this approach is trivial and that the result would be superior over a speaker that is designed under free-field conditions and then placed in the room without room eq.
I suggest to take another look at the Orion+ and the resulting AES paper http://www.linkwitzlab.com/stereo%20reproduction.htm.
I am of the opinion these days that DSP room eq is very difficult, other than for a few low frequency modes. Certainly, placing a mic at the listening position and equalizing for flat response is not going to improve accuracy, though it will change what you hear. I am also of the opinion that the room is usually not the problem, but the polar response of the speakers is. By building something like the Pluto or Orion+ you get that one right. The overlooked and not fully understood parameter is our perceptual apparatus. You are likely to mess with the ASP between your ears when you apply DSP to the ear signals. So if you experiment with DSP try to assess what you are actually changing perceptually. Start with a known source like the Pluto and with adequate recordings.
Have fun. This could revolutionize how we build and set up speakers.
Thanks for your inputs!
You mention that just "placing a mic at the listening position and equalizing for flat response is not going to improve accuracy". I might understand what you mean. But Mike's results show substantial improvement in both stored energy response (note the reduction in 50 Hz decay) and in flat frequency response. Phase looks much improved as well. No comments on distortion, dynamics, or other terms of performance. Of course, this is applied to a passively crossed-over build, so there is more to "fix" than a Pluto.
So can you just say a little more about what you find are the more significant parameters of "accuracy" as pertains to perception? Is it primarily polar response in the room? If so, the Pluto hardware configuration seems a super platform on which to build!
Thank you, Siegfried, for your comments and insight. I want to be clear from the start that I am not advertising my project as superior to Pluto in any way. It is simply another, somewhat novel, approach that I believe produces good results. I enjoy seeing how far I can push non-audiophile components (not a criticism of Pluto in the least - but we must admit that it is just some inexpensive speakers, PVC pipe, and in my case, caps, coils, and resistors - a perfect platform for my thesis that high-priced audio gear is not necessary for extremely good sound reproduction).
I agree that DRC is non-trivial. I spent many hours tuning and testing before I obtained the results I published. But, it is more than equalizing for flat response. The audio is corrected in both the frequency and time domains, as can be seen in the impulse and step responses.
My objective is to have the sound in the recording reach my ears - no more and no less - just as it was recorded (good or bad). If I am listening to music recorded in a cathedral, I don't want it to sound like the cathedral is in my listening room; I want the room removed from the equation. DRC attempts to do this, and in my opinion, succeeds to nice degree. I attempt to demonstrate this with my objective measurements. If there are other measurements of accuracy, I would sincerely want to include them in my series of tests (I also test harmonic and intermodulation distortion, but these levels are so low (<1%) that DRC does not attempt to correct for them).
The field has progressed quite a bit since Fielder's 2001 paper, and most of the objections he raised have been addressed in the current state of the art (many of his objections were somewhat insubstantial). Specifically, modern digital correction employs: frequency dependent windowing (since the sensitivity of the room transfer function to the listening position is correlated to wavelength, listening position sensitivity increases with frequency, so correction is applied that is proportional to wavelength, and only about 20-50% of the time/frequency plane is corrected), separate treatment for minimum phase and excess phase, pre-echo truncation, ringing truncation, peak and dip limiting.
As a result, many of the audio artifacts and other anomalies created in early attempts at digital correction have been eliminated.
Mourjopoulos is one of the leading researchers in digital correction. For authoritative papers on room dereverberation and loudspeaker equalization, I would look at 2003-2005 publications by Mourjopoulos, in which he sites considerable success in subjective assessments of DRC.
Thanks, SL. I get your points.
A topic of general interest for me, but one I have no ability to pursue right now (or ever?), is the relative merits of correcting factors of major magnitude, like frequency response, at the potential expense of ones of minor magnitude, like filter ringing and sub 1% distortion.
In this case, it seems that the applied DRC did a great job of correcting major errors in the sound. But perhaps these corrections are of relatively lesser value if the ear adjusts and compensates for errors in those terms? It would seem the improvement must be well worthwhile unless there is a significant price in other attributes of the sound, like objectionable distortions. From something I read recently about well-executed analog filters (thanks Eric), we are apparently quite able to hear "coloration" in filter circuits due to inaccuracies at levels that most of us would likely have said were absolutely insignificant (well under 0.1 dB).
Our ears obviously do not treat all sounds equally. So it seems it is all about the balance perceived in the result. I must believe that improvement in a given parameter is always preferred IF there is truly no loss in another parameter. But once even an apparently "minor" penalty is involved at another parameter, it becomes much less likely that we can predict, technically, the desireability of the outcome. It would be interesting to have the opportunity to do good digital/analog comparisons someday. I think I would have to hear it.
Mike, it would be helpful to have a clear definition of what problem you are trying to solve with DSP. What is the cause for the problem and how is DSP the means to solve it?
For example, I would want to have the illusion in my living room of hearing the recording venue, whether it is a cathedral, jazz club or outdoor meadow. I do not want to hear my living room. An acoustic event has the dimensions of time, space, frequency and amplitude. They should be reproduced accurately.
Equalization, like leveling, is probably fundamentally the wrong approach because it throws away information.
As I have stated before, my objective is to recreate a recorded audio signal as accurately as possible. I am not trying to improve on the Pluto design and I do not deny that the Pluto is an outstanding speaker system. I do, however, find the Pluto enclosure design to be an attractive vehicle to use in my quest to prove that it is not necessary to spend a lot of money on high-end audio equipment to realize outstanding sound quality. I thought your followers would find it to be interesting.
I am afraid that you and I will have to agree to disagree on this subject. I completely accept your contention that the Pluto produces outstanding sound. However, you do not accept my similar contention about my system, even when supported by a hand-full of objective measurements of my results. Fortunately, there are numerous published academic papers supporting my contention that digital audio correction fundamentally works.
This thread has obviously backfired, and I apologize to you and the Pluto community for the trouble it has caused. I suggest that it be closed and forgotten. I will seek more receptive enthusiasts elsewhere.
Best of luck to you in the future,
You are missing the point.
I have no problem when someone takes the Pluto or Orion as a starting point for some other design. My whole website is an invitation to explore and to do just that. I am merely showing what I have arrived at and how everyone can duplicate that.
Please read http://www.linkwitzlab.com/#CAUTION:
Since a key piece of your strategy is to use DSP, and a firm belief in the validity of your measurement data to describe perceptual properties, I wanted to rattle your cage a little bit. You must not have noticed the room response measurements in frequency and time for Pluto and Orion: http://www.linkwitzlab.com/stereo%20reproduction.htm
I can assure you that I am perfectly happy with them and if you read further you might see why.
Would you want to apply DRC to these measurement data? Why and for what purpose? You did not clarify the quote from your earlier post. It would be fair that you explain your objective to other Pluto builders so that they can judge for themselves if there is something of validity and interest here.
Much has been written about DRC in technical journals. Maybe you understand the mathematics of DSP, but there is still a lot of research going on right now about their application to real rooms and loudspeakers. The problem is the intersection between DSP and the human cognitive faculties between the ears.
Do you want to make a contribution here with Sharon or are you merely tinkering? The implication was that you had come up with something. Is that so?
Lots to clarify. Just be assured that there are no ill feelings here.
Man, I wish I had more time to give to this post right now...
Mike, my comments here are just my opinion - nothing more. And they should be taken as broadly supportive, not challenging, at least not in any negative sense.
In your response to my question about the sound of the result, you said it was much better than uncorrected (no surprise), but not much different from other similarly DRC'd speakers (important). This certainly shows the method is capable of making "any speaker" sound approximately "the same". If the general direction of your work is just to show that anything can be made to sound the same, regardless of polar response, etc., then perhaps it is really more appropriate on DRC related boards. That is to say, if you are not contributing to "Pluto" experience and/or advise, if this is just another example of DRC with no particular relevance to Pluto, then perhaps this thread IS misplaced. If we cannot reasonably try out what you have done, it does not help Pluto users, or any potential "evolution" of its design, it is not strictly a "Pluto" post.
But I, for one, definitely found your post interesting and valuable. I would not have discovered your work it if it was not posted here. I AM interested in DRC (well, the whole potential of digital cross-over/EQ for active speakers). So I find your project and its results intriguing.
I am also glad to know a little more about progress being made on digital filter artifacts. Many of us have suspicions that substantial digital filtering will have unpleasant sonic artifacts. I believe that will eventually become false, and that reasonably available digital processing will surpass reasonable analog options, without noticeable artifacts. Whether that day is already here, would be exciting to know.
So, given that you have used a Pluto platform, I would be even more interested if you were able to do some other comparative work that informed us on the relative merits of the digital implementation vs. standard analog implementation. Alternatively, it might be similarly useful if you were able to provide enough support information for others to try out what you have done so they could provide feedback to other forum members. Please hold off on the reaction "How can I provide correction for anyone else's room - that is the whole point of the project". Conceptually, there are two parts to the correction: 1) idealizing the speaker system response, and 2) compensating the interactions of the room that perturb modify ideal reproduction as transmitted to the listening position. I know that your project does this all in one go, and that it might not be possible (or even desireable?) to separate the two parts. This might limit the value of the result to the average Pluto owner.
In any case, I will draw from your experience that a Pluto can be made to sound "correct" in a given room using current DRC. And I will further conclude that, if applied to a standard Pluto, the result might sound even better, since there would be relatively less for the DRC to modify. As justification for that hope, I noted that you say your work did not attempt to address deviations below about 1%. I know that SOME 1% differences are very significant when pursuing "outstanding sound quality". So if you can start with a minimally distorted, relatively smooth, flat speaker system, and THEN correct room response to within 1%, I imagine the result could be truly outstanding.
In summary, it would be clearly relevant to Pluto if you would do some comparative work that informed us of the relative merits of your approach versus the standard Pluto implemetation, OR if you could provide enough guidance for others to apply your method(s) to their own standard Plutos. You certainly do not owe us this, but any chance of either? Regardless, I respect and appreciate your work.
Very interesting to see this post, thanks a lot for the info.
Before 2000 I noticed from an exhibition a DSP was used to get near ideal speaker spec. I tried to write the the processor in Europe about the details - very sorry I forgot his name - and he was kindly replied that modern speaker drivers are very good that makes the DSP not that needed. At that time I did thought in another way that may be the technology was still in lab. Such technology is what I wanted for many years - using DSP to solve most problems in A/V.
In this Christmas days I went back to Siegfried's web site and noticed interesting updates, then ran into this post. What I feel is that, the progress of DRC makes it is near a real - I say near a real because I'm not aware of any real product, and not sure the details, not sure whether it makes use of new technologies in digital audio like MP3/AAC which I'm quite family with. BTW, digital filtering is able to achieve any precision we want, much more precisely then analog one whose performance is limited by component tolerance. Digital filtering is already used in most A/V receivers, and in all digital audio codec, but unfortunately, Jim's doubt is reasonable because most A/V receivers may use high precision algorithms.
Yet DRC will not solve all problems like Siegfried stated, it will solve system response problems, but not acoustic problems, you can't expect a 5" speaker sounds like 10", or Vifa sounds like Audax, or absorb material can be replaced by DSP. But it will get near ideal system transfer functions if designed properly, or if the measurement chain is ideal it can get a near non-distorted system much better than any analog filtering can do, and it will replace most analog filters just like CD replace tape.
To me, it is an new technology but not art, and innovative products like Pluto or other product SL made are sth like art. As an example, even today's computer can do most of things, we still prefer man made oil painting, that's the difference. Even it sounds better, people will still want to get SL signed product right?
DRC is simple in principle, in frequency domain a system pulse response is the transfer function, reverse the function then we get a constant or pure gain.
Get back to Sharon, Mike you should really use digital crossover and two bands amplifier.
Correct me if I'm wrong in any ways.
typo- "Jim's doubt is reasonable because most A/V receivers may NOT use high precision algorithms." we can see A/V receiver use 24/32 bit DSP, yet AAC decoder needs double precision (64bit) or even more (floating).
Who is online
Users browsing this forum: No registered users and 1 guest