Grey, I agree that it is very likely that Linkwitz has a bias towards using fiters instead of other solutions, and that is what makes it an Orion IMHO. This project was started by a guy that likes his Orions but wants to go further.
You know well that I was making a point that every solution has its good and bad points including offset. In my limited experience, I would be tempted to approach the problem as you suggested, but I suspect the final speaker designed by me would be not as good as the current Orion, and would not be in the spirit of the Orion.
Clearly, as originally defined, the solution should be similar to one of your statements. The idea is to make everything as similar to the original Orion as possible except for the ASP, THEN we can answer the question which I feel is the "philosophical " one here: Will the sound of the speaker- already acknowleged by many as being quite good, be improved by:
1. discrete components
Or
2. better opamps
3. Strangely you didn't mention this one and I think it is the most powerful: Eliminating some opamps by artful manipulation.
I think 1 and 3 are of the most interest to people here and I think that only you seem to find this bothersome. There is also a practical matter here. The owner of Orions would find it very easy to compare the original ASP to the modified ASP, Whereas if the physical speaker panel is changed, then comparisons are going to be difficult, as is the possibility of most people here contributing to the design as it appears that many people here don't even own the speakers let alone have listened to them..
You know well that I was making a point that every solution has its good and bad points including offset. In my limited experience, I would be tempted to approach the problem as you suggested, but I suspect the final speaker designed by me would be not as good as the current Orion, and would not be in the spirit of the Orion.
Clearly, as originally defined, the solution should be similar to one of your statements. The idea is to make everything as similar to the original Orion as possible except for the ASP, THEN we can answer the question which I feel is the "philosophical " one here: Will the sound of the speaker- already acknowleged by many as being quite good, be improved by:
1. discrete components
Or
2. better opamps
3. Strangely you didn't mention this one and I think it is the most powerful: Eliminating some opamps by artful manipulation.
I think 1 and 3 are of the most interest to people here and I think that only you seem to find this bothersome. There is also a practical matter here. The owner of Orions would find it very easy to compare the original ASP to the modified ASP, Whereas if the physical speaker panel is changed, then comparisons are going to be difficult, as is the possibility of most people here contributing to the design as it appears that many people here don't even own the speakers let alone have listened to them..
Variac,
Given the large quantity of OpAmp involved, point number three above is where I would focus most of the efforts here. The M.O.J. Hawksford paper referenced before could be one viable alternative.
Also, using some passive filtering could be a viable solution as it should trade some S/N ratio for better distortion figures.
SL himself provides some guidelines for passive, line level filtering:
http://www.linkwitzlab.com/proto.htm
Regards
Giorgio
Given the large quantity of OpAmp involved, point number three above is where I would focus most of the efforts here. The M.O.J. Hawksford paper referenced before could be one viable alternative.
Also, using some passive filtering could be a viable solution as it should trade some S/N ratio for better distortion figures.
SL himself provides some guidelines for passive, line level filtering:
http://www.linkwitzlab.com/proto.htm
Regards
Giorgio
Grey, that is the exactly what you are doing, you judge the design without having heard it just because of the choices SL made with the physical design and the ASP.GRollins said:
The Orion(I have a pair at home) sounds great with the opamps IMHO, but if there could be improvements with the ASP I would like to try them.One easy way would be to replace some of the opamps(around the filter sections) with a couple of JFET buffers, it would not be that difficult to build an adapter with 2 buffers that could be put in the IC socket...
Isn't it important to remember that offsetting a driver and introducing a time delay really aren't equivalent? Like, at all? Since the Orion is a dipole speaker it's probably safe to assume off-axis response was one of SL's priorities in its design. Physically offsetting a driver changes the off-axis response in a way that electronic time-delay does not.
No?
No?
I like playing the name game.
First off, the Orion system I heard about two years ago is the best loudspeaker I have ever heard. I listened for several hours with many of my own reference tracks and other difficult passages, sometimes at loud enough volumes for the woofers to hit their stops. The delicacy of the plentiful bass alone is something which must be experienced first hand. These were driven by six Hafler DH-220 (or 200?) amps, but I still think that the lack of ultimate clarity (or liquidity or darkness) behind the treble came from the op-amp maze. (These were not Orion+ which add an additional rear tweeter.) I would still build the Orion system as-is (when I've saved up my pocket money) but would be much more motivated knowing there was a way to get around the op-amps by going discrete (and line-level passive where possible) and using as few transistors as possible. So I agree that the scope of this project should be only to simplify the ASP (and make it usable with balanced?) while leaving the drivers' physical arrangement alone.
Nobody liked the O'Brian/O'Brien idea? (I guffawed out loud!) The name "O'Brien" reminds me of the Seinfeld episode where George pretends to be O'Brien to get them a limo ride from the airport and instead gets everyone in a pickle. The impostor aspect applies for us here also, plus it's just too silly to pass up. Characters from mythology seem too arcane for use here, despite being very applicable. Especially Artemis which seems eerily appropriate as killer of Orion. And "Uranus" will be the perfect name when diyAudio launches a Pluto mission.
So I nominate O'Brien ASP.
First off, the Orion system I heard about two years ago is the best loudspeaker I have ever heard. I listened for several hours with many of my own reference tracks and other difficult passages, sometimes at loud enough volumes for the woofers to hit their stops. The delicacy of the plentiful bass alone is something which must be experienced first hand. These were driven by six Hafler DH-220 (or 200?) amps, but I still think that the lack of ultimate clarity (or liquidity or darkness) behind the treble came from the op-amp maze. (These were not Orion+ which add an additional rear tweeter.) I would still build the Orion system as-is (when I've saved up my pocket money) but would be much more motivated knowing there was a way to get around the op-amps by going discrete (and line-level passive where possible) and using as few transistors as possible. So I agree that the scope of this project should be only to simplify the ASP (and make it usable with balanced?) while leaving the drivers' physical arrangement alone.
Nobody liked the O'Brian/O'Brien idea? (I guffawed out loud!) The name "O'Brien" reminds me of the Seinfeld episode where George pretends to be O'Brien to get them a limo ride from the airport and instead gets everyone in a pickle. The impostor aspect applies for us here also, plus it's just too silly to pass up. Characters from mythology seem too arcane for use here, despite being very applicable. Especially Artemis which seems eerily appropriate as killer of Orion. And "Uranus" will be the perfect name when diyAudio launches a Pluto mission.
So I nominate O'Brien ASP.
Variac said:
Actually, I've already suggested same, but people acted as though I'd insulted their grandmother. I'll bow out and let you fellas debate what to change so that nothing is changed [sic]. After all the speaker is perfect, no?
Nicke,
Anything with, what was it? 22 opamps? speaks powerfully that the designer is, shall we say, besotted with one particular approach and is convinced of the strength of his position. You folks are in the paradoxical position of saying that the ASP needs changing, but that the Orion speaker system is the most marvelous thing on the face of the planet.
Uh, sorry, but I missed something. If it's that wonderful, you should leave it alone and be happy.
My impression is that some of you need to get out and hear more speakers. Get a little more experience with what can be done by other speaker designers--without opamps and an entire factory's worth of parts. There's an entire world out speakers out there. Go hear some electrostats: Martin Logan, Quad, or Sound Labs (some of the best soud I ever heard was through a pair of Sound Labs electrostatic speakers--simply gorgeous). Listen to some Magneplanars. Try out some horns. Give some of the better box speakers a shot. No. Really. Somewhere in there, you might discover that a parts farm may look impressive, but it's not the only solution to the various problems that come up during the design phase of building speakers. It may not even be (gasp!) the best answer.
For that matter, at least some of you might even get a kick out of the full-range speakers that Nelson (and others) are spending so much time talking about. Get rid of all the surplus electronics.
Grey
P.S.: Before anyone starts in with the inevitable complaint about the price tag of the speakers I mentioned above, please note that I did not say that you should buy them--I said you should listen to them. Get out. Live a little. Learn what is possible. Should you find, say, that the Martin Logans tickle your fancy, you'll find that it's not difficult to build an electrostat. If, by some chance, you like the Magneplanars, reflect on the fact that I built a planar driver of that sort in a single afternoon.
EDIT: If you're still searching for a name, might I suggest that you're looking at this the wrong way...call it the Cleopatra, who (according to some accounts) died from the bite of an ASP.
Geez indeed,Grey, you are the one creating the strawman that everyone here thinks that the speaker panel is perfect. I don't see that in the comments posted- and I personally don't think that's the case. Certainly the fact that Linkwitz designed the Audio Artistry loudspeakers, including the Beethoven model which was rated as an A, fullrange level loudspeaker by Stereophile I believe, implies that he can make a decent sounding loudspeaker. Granting that being blessed by them is not the ultimate criteria...
Instead, its as you said earlier- a philosophical choice that seemed to be made in the first post, in that ORIGINAL basic question is whether all those opamps and chipamps could be replaced or eliminated and would the speaker sound better? I think that probably everyone who posted here feels that it would.
I think that the problem was originally proposed with this angle, and that people feel that it would be a more pure experiment if some variables were left out. I agree that possibly the speaker could sound even better with a stepped baffle instead of some circuitry, but how are you planing to impliment that? Are you going to make the speaker your way? is anyone else? Instead some of us are proposing that a more controlled experiment is to just change SOME parameters and see how big an improvement (i have to say "if any " here to appear unbiased
there would be)
There is a practical aspect to this as I have mentioned. A current Orion owner could swap in the new design in minutes and compare the two . Another advantage is that the design could be done possibly more realistically in an online discussion format than a speaker baffle redesign..
On the other hand , I'm not too sure who is able to do this alchemy of artistically combining the components to use less opamps. It seems that the procedure is to try to combine functions, then measure to see if you achieved your goal. Most people here probably don't have the equipment for this..or the knowledge perhaps..
I'm not the promoter or organizer of this project, so any one can do anything that they= want and post it here. Grey, if you want to make a list of changes you would make to the speaker panel and the electronics, I don't think that you will be torn apart or anything, just do it - that would have taken a lot less verbage than complaining that everyone is jumping on you..
This discussion that we must decide on a philosophy came from you I believe..
Instead, its as you said earlier- a philosophical choice that seemed to be made in the first post, in that ORIGINAL basic question is whether all those opamps and chipamps could be replaced or eliminated and would the speaker sound better? I think that probably everyone who posted here feels that it would.
I think that the problem was originally proposed with this angle, and that people feel that it would be a more pure experiment if some variables were left out. I agree that possibly the speaker could sound even better with a stepped baffle instead of some circuitry, but how are you planing to impliment that? Are you going to make the speaker your way? is anyone else? Instead some of us are proposing that a more controlled experiment is to just change SOME parameters and see how big an improvement (i have to say "if any " here to appear unbiased
there would be) There is a practical aspect to this as I have mentioned. A current Orion owner could swap in the new design in minutes and compare the two . Another advantage is that the design could be done possibly more realistically in an online discussion format than a speaker baffle redesign..
On the other hand , I'm not too sure who is able to do this alchemy of artistically combining the components to use less opamps. It seems that the procedure is to try to combine functions, then measure to see if you achieved your goal. Most people here probably don't have the equipment for this..or the knowledge perhaps..
I'm not the promoter or organizer of this project, so any one can do anything that they= want and post it here. Grey, if you want to make a list of changes you would make to the speaker panel and the electronics, I don't think that you will be torn apart or anything, just do it - that would have taken a lot less verbage than complaining that everyone is jumping on you..
This discussion that we must decide on a philosophy came from you I believe..
Variac,
Indeed, you have touched on the nub of the thing. From where I sit, this appears to be turning into what I call a "herding cats" thread.
--Person A wants the Orion just the way it is now, but "better" in some unspecified way. Perhaps more discussion would clarify for this person the way to go, or perhaps they're just kinda "lurking" (is that the right word?), albeit with occasional posts.
--Person B thinks chip opamps are just peachy and wants to swap opamps around. (Clearly, I'm the wrong guy for that discussion.)
--Person C wants discrete opamps that will somehow fit into Linkwitz's existing circuit board. (I dunno about that...if the board I saw on his site is the same one they want to put discrete opamps on headers into, I don't think there's enough room.)
--Person D gets the idea that Person C's desire is destined for frustration and wants a completely discrete circuit that preserves all the original functional blocks of Linkwitz's concept. And he wants PCB artwork, too. In fact, every Person from here down is going to need PCB artwork to realize their circuit.
--Person E wants to combine all the original functions into half the number of functional blocks but still using opamps.
--Person F wants the same thing as Person E, but with discrete circuitry.
--Person G sees that there are, perhaps, some things that can be dropped, but wants to keep chip opamps because he has a box full of them.
--Person H wants what G is thinking of, but discrete.
--Person I has a table saw and...
(...and at least six or eight other permutations that I'm not going to take the time to list.)
Get the picture? This is turning into a herding cats thread. That's why I asked for someone to set some boundaries to the problem. As things stand now, Person A isn't really a problem, he just wants to talk about the speaker. Person B isn't really that much of a problem, either, at least not to me...I'm the wrong guy to ask about which opamp is 'hot' this week. But from Person C on, you begin tallying separate projects and the total is going to be something like eight or ten or more related-but-separate R&D projects, with all that implies about time, money, and commitment. No thanks.
And even if you were to set strict boundaries at this point the thread would wither, because the Persons whose desires were not being met would feel cheated. At which point the end result would satisfy perhaps two or three people...maybe.
I used to try to stick with such threads. These days, I tend to back away and let the cats alone.
No harm, no foul. You folks enjoy yourselves.
Grey
Indeed, you have touched on the nub of the thing. From where I sit, this appears to be turning into what I call a "herding cats" thread.
--Person A wants the Orion just the way it is now, but "better" in some unspecified way. Perhaps more discussion would clarify for this person the way to go, or perhaps they're just kinda "lurking" (is that the right word?), albeit with occasional posts.
--Person B thinks chip opamps are just peachy and wants to swap opamps around. (Clearly, I'm the wrong guy for that discussion.)
--Person C wants discrete opamps that will somehow fit into Linkwitz's existing circuit board. (I dunno about that...if the board I saw on his site is the same one they want to put discrete opamps on headers into, I don't think there's enough room.)
--Person D gets the idea that Person C's desire is destined for frustration and wants a completely discrete circuit that preserves all the original functional blocks of Linkwitz's concept. And he wants PCB artwork, too. In fact, every Person from here down is going to need PCB artwork to realize their circuit.
--Person E wants to combine all the original functions into half the number of functional blocks but still using opamps.
--Person F wants the same thing as Person E, but with discrete circuitry.
--Person G sees that there are, perhaps, some things that can be dropped, but wants to keep chip opamps because he has a box full of them.
--Person H wants what G is thinking of, but discrete.
--Person I has a table saw and...
(...and at least six or eight other permutations that I'm not going to take the time to list.)
Get the picture? This is turning into a herding cats thread. That's why I asked for someone to set some boundaries to the problem. As things stand now, Person A isn't really a problem, he just wants to talk about the speaker. Person B isn't really that much of a problem, either, at least not to me...I'm the wrong guy to ask about which opamp is 'hot' this week. But from Person C on, you begin tallying separate projects and the total is going to be something like eight or ten or more related-but-separate R&D projects, with all that implies about time, money, and commitment. No thanks.
And even if you were to set strict boundaries at this point the thread would wither, because the Persons whose desires were not being met would feel cheated. At which point the end result would satisfy perhaps two or three people...maybe.
I used to try to stick with such threads. These days, I tend to back away and let the cats alone.
No harm, no foul. You folks enjoy yourselves.
Grey
Variac said:
Yes, and maybe.
We have a pair of Orions (more specifically Wayne does) and
also a supply of SEAS drivers of all sorts. So we have the
materials at hand.
I have enough data to begin consolidating the circuits into a
more compact approximation of SL's circuitry, however it
chews up time, which is my most precious commodity. What I
think I'll do is to tackle it piece by piece.
The other temptation is simply to take the Orions and develop
a crossover set for it from scratch and see what I get.
Well, since you are the only one that seems to be tackling the issues means that you get to choose the issues that you tackle.
And the speed at which you do your tackling
Actually I'm in no hurry- I just think that this is interesting..
It is only my opinion, but I think it would be generally cool to tackle most (but not necessarily all) of the issues that Linkwitz did, just BECAUSE that's his approach and it helps answer the question as to how much better discrete opamps and amps would make it.. Leaving a few off is certainly kosher of course..since you are the tackler...
And the speed at which you do your tackling
Actually I'm in no hurry- I just think that this is interesting..
It is only my opinion, but I think it would be generally cool to tackle most (but not necessarily all) of the issues that Linkwitz did, just BECAUSE that's his approach and it helps answer the question as to how much better discrete opamps and amps would make it.. Leaving a few off is certainly kosher of course..since you are the tackler...
GRollins said:
About a year an 1/2 ago, having designed the NaO speaker system and having been asked many times how the NaO compared to the Orion I took it upon myself to build a pair of Orion speakers based on the info presented on SL's site; i.e, the Orion challenge. However, my approach was to look at SL's posted transfer functions and then set about designing active circuits which emulated the resulting acoustic transfer functions directly, rather than in the piece-wise approach SL used. It was a simple matter to eliminated a number of the opamp stages (6 per channel) by combining and optimizing specific active blocks.
Op-amps for inductors
Am I correct in surmising that an op-amp must use considerable feedback to simulate an inductor? If so, you must have enough feedback at the highest frequency that the op-amp is expected to be effective. How much feedback is required: 10dB margin, 40dB margin? At what frequency: 2x the filter freq, 10x freq., higher?
The buffers clearly require less open loop gain.
This is one necessary starting point to define the specs required when you begin substituting discrete for IC op-amps.
Jeremy
Am I correct in surmising that an op-amp must use considerable feedback to simulate an inductor? If so, you must have enough feedback at the highest frequency that the op-amp is expected to be effective. How much feedback is required: 10dB margin, 40dB margin? At what frequency: 2x the filter freq, 10x freq., higher?
The buffers clearly require less open loop gain.
This is one necessary starting point to define the specs required when you begin substituting discrete for IC op-amps.
Jeremy
Since the block diagram is posted on the web site, it seems to be fair game to try to emulate the general functionality of the Linkwitz ASP. However, in order to build an "Orion," one must purchase the construction plans, with which one presumably gets the crossover frequencies, etc. for the ASP circuit. As a result, this project should never get specific enough to infringe on that information. Any filter component schematics ought only to give equations on how the R's and C's affect the blocks' function and not give actual values for R and C, although suggested ranges may be appropriate (i.e. Cmax = 10 uF, Rmin = 1k, etc.)
Just my 2cents.
Jeremy
Just my 2cents.
Jeremy
Sorry to teast but as the designer of a completive system I don't think it is rightly my position to post the modifications I made. There is some history here and I don't want to come off as implying that my mods are superior to the original. However I will say that loosing 6 active stages certainly didn't hurt.
Over time I did make some changes to make the system balance to make it more suitable to my taste. I added some warmth in the lower midrange because I though vocals lacked body. But that is not specifically associated to the ASP topology.
At the time I build the system the ASP block diagram and transfer function response posted on SL's site (or the version I have from SL's site) gave sufficient information in terms of the gain of each stage and filter corner frequencies of the HP/LP filters. There is some ambiguity for delays but with these are pretty easy to gleam form response measurements.
I also ultimately purchased the plans set/boards.
The Orion is certainly an excellent speaker but from its inception I have felt it was overly complex in its active circuits.
Over time I did make some changes to make the system balance to make it more suitable to my taste. I added some warmth in the lower midrange because I though vocals lacked body. But that is not specifically associated to the ASP topology.
At the time I build the system the ASP block diagram and transfer function response posted on SL's site (or the version I have from SL's site) gave sufficient information in terms of the gain of each stage and filter corner frequencies of the HP/LP filters. There is some ambiguity for delays but with these are pretty easy to gleam form response measurements.
I also ultimately purchased the plans set/boards.
The Orion is certainly an excellent speaker but from its inception I have felt it was overly complex in its active circuits.
At the time, I was using Hafler T1600 amps. If we bottomed out the woofers, I must have had the Adcom arc welder on them.
Now they're Orion++. Even better!
The Hafler amps have had their weaknesses exposed by another set of amps on loan from a friend. The system is significantly more free of distortion on peaks, flatter (a bit of midrange bloat is gone), and has more clarity. These problems (such as they were, for the "best loudspeaker [you] ever heard") would appear to be the Halfer's fault, not the ASP's. Still, if the ASP could be simplified, I would be game to try out a simpler design.
To Grey: Extolling the virtues of the soundstage of the LS3/5A while saying the Orions do nothing new with soundstage is like calling the soundstage of an FM broadcast the gold standard, while ignoring live concert hall performances. As remarkable and curious and enjoyable as the 3/5A is (which I have listened to extensively BTW, along with countless other speakers and even, gasp, live music once in a while!) it's not in the same universe as the Orion, and not just in terms of soundstage. Perhaps you need to get out and listen to an Orion?
The RFI-rejection input stage is an analog filter on the input to the next (opamp-based) shelving high pass. So there's nothing to be gained by dumping it. SL also parallels the feedback resistors with small caps so there's no gain for RFI.
For the sake of clarity, there are 21 opamps implementing the 25 blocks shown on the ASP diagram linked to earlier (or optionally 22 opamps and 26 blocks with the optional bass notch filter around U10A. The builder can configure this to deal with one room mode. I turned mine into a switchable BBC dip filter for listening to DG recordings.) The detail-oriented will notice that four of those blocks do not have opamp designators.
Nelson: It's hard for me to see what can be replaced or combined. The input RFI filter is passive, as are the 50HP1, 2HP1, and the -7.9db gain. I believe the four opamps comprising the delays for mid and tweeter must be implemented with two opamps each because of the delay magnitude and phase. (SL discusses this in the Phoenix design pages on his site -- I don't recall exactly the rationale for the splitting the delay between two opamps). I suppose the two (or optionally three) notch filters which use opamps as simulated inductors could be replaced by real inductors. The two +/- 2.5db gain stages are trick circuits for usability: they make each tick on a linear one turn pot correspond to 0.5db of gain. The other blocks are either necessary buffers, shelving filters, or 2nd order HP/LP filters. But heck, if you can realize the same transfer function with passive components in the feedback loop of a single opamp per driver, I'd give it a try!
I suppose JohnK could have gotten rid of six active stages by dumping the two +/- 2.5db adjustments, compressing the two delays from four to two opamps, and doing the notch filters passively. Perhaps he just dumped the 5k notch on the mids since this is way out of band for the driver anyways (crossover is around 1.5k).
- Eric
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.