Advice for Linkwitz Orion system with Nelson Pass-style pre-amp and amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
If you're going to build an active crossover, you're going to need isolation between stages, something you can tie a feedback loop around, something to offset insertion losses, and potentially add gain. This isn't all that difficult, conceptually. There have been numerous attempts at discrete opamps here, some more successful than others. You can swipe an existing design from another thread if you don't want to design your own. Better yet, steal the opamp out of Nelson's opamp paper and you're done.
Them what thinks chip opamps are the ticket can flip through a catalog and order in anything they want. I'd suggest something with a JFET input, though.
These days I'm going no gain and no feedback in my mid/upper active crossovers. Complementary JFET followers, self-biasing, used as buffers between filter sections. Works like a charm. You'll lose a smidgen of gain due to followers being less than unity gain and a bit to insertion loss, but it's about as pure and simple as you could ask for. It helps that I'm using first order crossovers for the mids and tweeters. You can cascade first order slopes to make a higher order slope, but it starts getting unwieldy; easier to go ahead and do a single discrete opamp and get it over with.

Grey
 
Member
Joined 2004
Paid Member
From the horse's mouth

I asked Siegfried Linkwitz at tonight's lecture, and he has no problem with anybody making any changes or redesigns to the Orion ASP (analog signal processor), as long as it isn't being referred to as 'Orion.' (Would it be stupid to call it diyAurion?)

He also pointed out that his Pluto system (with sub) comes quite close in performance to the Orion, although with slightly shallower soundstage depth (but wider 'sweet spot'). That makes me think Pluto might be the place to start when it comes to bang for the DIY buck.
 
I would be interested in finding out how the number of gain stages can be reduced. For one, the opamps in unity feedback can be simplified to, say, a follower stage. But that's not exactly reducing the number of gain stages. Perhaps Mr. Pass can shed some light here.

Take SL's functional block diagram of the Orions' ASP as a conceptual study. Things that you see being repeated are the all-pass filter, 24db/oct HP, and 24db/oct LP. From a circuit standpoint, how would you reduce the 2-opamp 24db/oct filter to just one gain stage and still achieve the same response. This would be an interesting exercise, and for discussion purposes, we can just draw the traditional cascaded 12db 2nd-order filter w/2 opamps (low-pass, high-pass doesn't really matter), and see how it gets morphed into a single gain stage. I seem to recall that 3rd order filters can be done with just one gain stage. Just not sure about 4-th order without getting into insertion loss and possibly other issues related to gain and phase.
 
I confess that just looking at the "Orion ASP Function block diagram" gives me a case of the heebie-jeebies. So many opamps--and chip opamps, at that--must sum to several hundred transistors...just in the crossover. All told, there's probably a cumulative hundred thousand dB of negative feedback in there. Yikes! That's the very antithesis of the minimalist school of thought.
Yes, I realize that he's got a reason for all the various bits and pieces, but still...I can't help but think that maybe it would be better to mechanically offset the mid/tweeter drivers than to add yet more opamps to an already overly busy circuit.
Just me being silly again. You guys go on with your project.

Grey
 
For Grey's benefit:
Nelson Pass said:
...The Orion is a sifficiently good and interesting loudspeaker that it deserves any improvements that can be made in the electronics. Linkwitz appears pretty agnostic about "audiophile" issues in electronics, but then his focus is the speakers. He isn't averse to using "better" active electronics. The opamps in the Orion circuits can easily be replaced with discrete versions, and there may be some opportunities to simplify the system overall - I'll take another look at it.
Nelson Pass said:
I spent some time reviewing the (possibly earlier version of) electronics for the Orions, and it is clear that you can replace
something like half the gain stages with discrete op amps and buffers and get the same nominal functions.
 
Paul,
I've seen what Nelson said, but I find Linkwitz's attitude strange--he goes to extraordinary lengths to attend to the most minute details in speakers--while compromising all the things he gains by using circuitry that sets his speakers back to the Stone Age. It just strikes me as odd. You would think that if he were that serious about the sound, he'd choose a better approach to all this eq/time domain/shelving/whatnot.

Grey
 
I find this conversation specially interesting, as I'm about to re-think the active filter of my dipole speakers, which are similar to Orion. But before that I have to measure again the drivers, so I have time to study what the alternative could be, alla Pass
 
The one and only
Joined 2001
Paid Member
This whole thing goes back to the project I did with Mobile
Fidelity back in the early 90's. The cutterhead drive system
had a massive amount of electronics in the signal path for
the purpose of gain, equalization, and feedback. The op amps
dated back to the 60's and early 70's, and there was an
isolated piece of electronics for every function, much like SL's
modular circuits. The advantage of this of course is that you can
play around with these in isolation from each other. The
disadvantage is that you end up running the signal through a
lot of op amps - in the case of the cutterhead system, maybe 10
or so.

What I did with Mofi was to integrate all these things around
a single op amp, the LT1028. To be sure, all the values of
every part were dependent on the others, and it was not a
trivial task, but I was able to accurately reproduce the response
curves of the original circuits.

We would look to do the same with the response of SL's Orion
circuitry with fewer discrete stages.

:cool:
 
Lacking a detailed schematic I can only make educated guesses about what he's doing in the circuit, but it seems that he's using electronics to do things that I would attend to in the physical design of the speaker. His background is in electronics; I'd lay odds that he's not particularly adept at woodworking and seeks to overcome that electronically.

Grey
 
Nelson,
Would it be possible to show us what you did with the Mofi 1-opamp circuit? It would help to generate some ideas here. I am not creative enough to figure out how one can combine the response of say an LR4 low-pass with an LR4 high-pass into a single opamp, or combining a notch filter with an all-pass. Hmm, maybe this in itself can be an article that you can post at your website? :rolleyes:

Thanks.
 
He's got a stage in there to roll it off at 140kHz? Okay, that one can go.
Jeez, he uses chip amps for his speakers and recommends anything from Doug Self to Rowland for customers. I guess if you put a signal in and one comes out, he's happy. Somehow, I had the impression that he was more performance oriented. It appears I was wrong. To be honest, the more I learn about his views the more I find myself wondering about the speakers.

Grey
 
Grollins, so I now know that you haven't heard the Orions yet. Me neither, but my speakers are similars to his, with cheaper mids, and I have to say that are excelent to my ears.

Let's focus on the electronics, as I'm sure that there are a lot of people listening to this thread, really excited with the possibility of improving an already really good sound. I would be really happy if I manage to change my filter to a better one. That would be great.


What would you choose instead of the gain module in the ASP?

The active filter of the JohnK dipoles has less active stages, and he's always trying to save on that (placing the poles in the right place, for instance)
 
Raka said:


Let's focus on the electronics...



Precisely. You'll note that I have not said one word about his choice of drivers or cabinet design.
Let me put it this way...if you heard people exclaiming about the steak at a particular restaurant you might become interested. As a red meat sort of fellow myself, I know I would. Now, let's say that before actually eating at this restaurant, you peeked into the kitchen and saw frozen chicken nuggets being microwaved, instant coffee instead of ground beans, and dried soup mix. These things are not necessarily inappropriate for a chain fast food restaurant, but this particular restaurant is being portrayed as an upper class dining establishment. Hmmm... Does it necessarily mean that the steak isn't up to snuff? No. Perhaps the restaurant has a particular cook who handles nothing but steaks; who goes to the local butcher every morning and hand selects each steak that he will be cooking that evening; who then spritzes a light dusting of herbs and spices over each steak so that it enjoys an entire afternoon absorbing flavor; who then grills each steak to perfection; who then bastes each side with a last minute addition of garlic butter before allowing it to be served.
Perhaps.
But I think you'll admit that it's not bloody likely, given the way the rest of the kitchen is run.
Now, as for my suggestion to dump the 140kHz "RF filter," there's a curious blindness that afflicts many audiophiles in that they look at each component in isolation. If they do consider interactions, it's usually along the lines of "will component X drive component Y?" But there are other things to ponder.
Consider the lowly filter. If someone tells you that there's a 140kHz low pass filter in a piece of equipment, your thought process is likely to be something along the lines of "140kHz is seven times the 20kHz upper limit of human hearing, so that's okay...in fact, that's more than okay...I'll never hear it."
I'll approach this from two directions:
--140kHz isn't 'seven times the upper limit of human hearing,' it's less than three octaves above the upper limit of human hearing. By comparison, compare 20Hz to 140Hz. Suddenly it doesn't seem quite as far away. Given that many instruments (e.g. brass, percussion, piano, synthesizer) have harmonics that go well above 100kHz, this begins to get a little scary. Can you hear 100kHz? No, not as a discrete tone, but it matters in terms of getting the waveform right down below the nominal 20kHz limit (which is pretty arbitrary, anyway).
--My second point--and here I come back to something I alluded to above--is that you're not listening to Linkwitz's crossover in isolation. You're listening to it with a source, possibly a phono stage, a line stage, the crossover itself, and an amplifier...all in series. Now suppose each of those components rolled off somewhere in the 100-150kHz region. In that sense, they're all filters. People don't look at what happens in the passband of a filter. They think that 140kHz is some sort of absolute. It's not. It's the -3dB point. But if you take a moment to look at a normalized filter response, you'll see that it has an effect for a couple of octaves below the nominal -3dB point. Okay, so what's a tenth of a dB between friends? Who cares? Consider what happens when you cascade each of the pieces listed above, each adding a tenth or two to the total. What happened at 20kHz? It's not at all uncommon to see response at 20kHz down by as much as a dB or even two (You'd be surprised how many circuits only make it to, say, 50 or 80kHz, instead of 100kHz; I was being generous when I said 100-150kHz.)
I can't speak for everyone, but I don't much like the idea of having my top end response lopped off before the signal even gets to the speaker.
I repeat--dump the 140kHz RF filter. Go away. Gone. Bye-bye. The easiest way to begin improving this crossover is to remove the unnecessary parts and the ones that harm the signal. Only then can you begin blending the rest into something more reasonable.
What would I choose instead of chip opamps? That's an easy one: Discrete. For that matter, I've already said so at least two or three times above. So now I've said it again.

Grey
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.