|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
203.58.188.211
In Reply to: Re: If you are into convolution posted by Presto on November 20, 2006 at 14:53:54:
You are right about stereo vs surround VST effects. There are surrround VST effects available, but I am not sure the VST to Winamp bridge supports them.And in any case, the Winamp ASIO driver does not support surround (I think, because ASIO is inherently 2 channels). Maybe we can convince otachan to enhance his ASIO plugin to allow 5.1/7.1 to be directed to multiple ASIO outputs.
There's also a freeware VST stack that allows multiple VST plugins to be "stacked" as a single VST connector to the bridge. Search in www.kvr-audio.com for it. So you don't need a VST host if you have more than one VST plugin.
Incidentally, I'm bi-amping all three front channels, but I do it the analog way (just connect the amps in parallel). Fortunately, my amps allow this to be done safely (they have "out" connectors allowing me to daisy chain them).
Follow Ups:
Christine:I am going to try and run the KVR convolver VST plugin (and others if I can find them) ahead of the Allocator VST plugin using the Audio Teknika Console tonite.
Hey - are you bi-AMPING or bi-WIRING there? I think you're bi-wiring (that is, if you are daisy-chaining full range signals to multiple amps and each amps output is then "wired" to a separate leg of parallel passive crossover network). When your crossover networks are "active" you are bi-amping, and when "passive" you are bi-wiring. Or so I understand the lingo to go...
But now I'm being picky! ;)
Ie. my speakers have different connectors for different drivers, and each driver is driven by a SEPARATE power amp, in other words two amps per speakers.Bi-wiring is when you have run multiple cables from a single amp to multiple speaker drivers.
I think what you are trying to do is to apply filtering in the digital domain and route different parts of the frequency spectrum to different amps. This is a variant of bi-amping, but not as common as what I am doing. I prefer not to use your approach because believe it or not I do not like digital filtering and prefer the analog filter network built into the speaker. Also your approach does not work for analog signals, and I have a turntable plus I listen to SA-CD.
You might want to check my system details on the asylum :-) I am a bit of an anally retentive purist :-)
Christine.Right. You are indeed bi-amping. We're both bi-amping. You're doing passive (crossovers after the amps) bi-amping, I'm doing a variation of DSP active (crossovers before the amps) tri-amping. Active multi-amping can be done with analog filters that are passive, active/analog, active/digital, or DSP based.
We're both using "more than one amplifier" thus we're *multi-AMPING* and not *multi-WIRING*! :o)
As far as analog lovers go - I don't think the "mandate" of the Computer Audio Asylum is to convince all analog lovers to digitize their vinyl collections and get into DSP based crossover filters and room correction. The vast majority are concerned only with playback of sources that are already digital - and further to that, are accessible (precluding the use of SACD for example).
Then again, there are quite a few vinyl lovers who meticulously record their albums on the first play - be it an analog or digital recording. If record wear was not an issue, why would they go through all that trouble?
Hmmmmm....
There's actually some research on this in the 70s, and the general conclusion seems to be it's better to apply filtering post amplification (or "passive" as you call it) vs before amplification.The main advantage of bi-amplification is that multi-driver speakers present a complex impedance load on the amp. Amplifying each driver separately greatly simplifies the "load" being presented to the amp, and hence allow the amp to more precisely control each driver, achieving higher damping factors for the woofer and better preserving the delicate high frequencies of the tweeter.
If you accept the above explanation, then filtering post amplification simplifies the load further, thus even better results. However, filtering prior to the amp is not necessarily an improvement, since it's not the complexity of the input signal that challenges an amp, it's the complexity of the output load.
There is a counter argument that says filtering the amp used to drive the tweeter achieves better transparency. So I guess the optimal result is probably (using your terminology) "active" filtering the tweeter, and "passive" filtering the woofer. I don't know anyone who does that, except perhaps embedded in a powered monitor.
The reason I don't like digital filtering (and DSPs in general) is that in my experience they seem to add a veil or a "haze" to the sound. I first noticed this when I realised my system sounded much better without digital bass management than with (at the expense of slightly less deep bass). Then I noticed everytime I put a processing step between the music and my ears I always seem to lose some "life" from the music. So now I run everything unprocessed - even though as you know I have more than enough hardware to do lots of processing if I wanted to.
I even listen to Dolby Digital 2.0 soundtracks without applying Pro Logic II/IIx (for concerts and anything that's not surround encoded, they sound better that way). And I've given up on equalization, convolution, bass management, upsampling.
I know, it's heresy on this forum, but hey, at the end of the day it's the music that matters.
Hi Christine,I am confused by the following statements:
***"The main advantage of bi-amplification is that multi-driver speakers present a complex impedance load on the amp. Amplifying each driver separately greatly simplifies the "load" being presented to the amp, and hence allow the amp to more precisely control each driver, achieving higher damping factors for the woofer and better preserving the delicate high frequencies of the tweeter."***
***"If you accept the above explanation, then filtering post amplification simplifies the load further, thus even better results. However, filtering prior to the amp is not necessarily an improvement, since it's not the complexity of the input signal that challenges an amp, it's the complexity of the output load."***
I think I understand the first statement, but don't see the logic of how it leads to the conclusion in the second statment. I must not understand what "filtering post amplification" means, because the only way I know of doing that in a speaker would be via the speaker's passive crossover. From the first quoted statement, I would think that an individual driver would present a more benign load to an amplifier than an entire crossover network. How else does one filter post amplification if not via a passive crossover network?
CT "However, filtering prior to the amp is not necessarily an improvement, since it's not the complexity of the input signal that challenges an amp, it's the complexity of the output load."Presto > Correct. But when doing active filtering before amplification, you are reducing the effective bandwidth that each amp will see because you are reducing the bandwidth of the voltages for which it can will provide gain. When multi-amping with filters post-amp, each amp still sees the NET impedance of both the driver in the pass-band, the crossover in the stop-band and a combination of both in the crossover region. When filtering "pre-amp", there are only voltages present for the pass-band, which reduce proportionately through the crossover region(s) to zero voltage in the stop band. So, a big difference is that for active (pre-amp) filtering systems, the amps do NOT see crossover impedances in the stop band region(s), since there is simply no voltage AT those frequencies being amplified.
CT "The reason I don't like digital filtering (and DSPs in general) is that in my experience they seem to add a veil or a "haze" to the sound."
Presto > I think digital/active filters are much much more transparent than their op-amp based analog/active predecessors. Passive networks are not necessarily 100% transparent either. In fact, some of "deadest" sounding speakers are ones with elaborate passive networks that have every impedance compensation and EQ network under the sun and a component count like a laptop computer, and poor transparency and dynamics as a result. They measure ruler flat, but sound totally lifeless. The trick is to find DSP algorithms that do what "they are supposed to do" but also afford the listener the greatest possible transparency. Certain types of "digital mathmatics" seem to suck more life out of music than others, and even similar processes (like filtering or convolution) are not all created equal. Truth be told, I was about to "hang up my experimentation hat" with DSP filters and just stick to convolution/room correction stuff until I heard what the Thuneau Allocator can do. Al Jordon here at the computer audio asylum is getting similar results: the Allocator is the single most transparent DSP based crossover program we have encountered, and to top it all off, it's IIR based and performes "phase arbitration" for any filter within it's usable range.
There are also many cases where crossover frequencies for the low end are just not practical to do with passive components - this is why we see so so many crossover points in three way designs that are 300-500 hertz - right smack in the lower registers of the vocal range. Going active can often preclude the need for physically large and expensive passive components for lower crossover frequencies - meaning that the designer is less inclined to make a "design trade-off" due to prohibitive cost, size, or insertion losses associated with certain size inductors.
CT: "I even listen to Dolby Digital 2.0 soundtracks without applying Pro Logic II/IIx (for concerts and anything that's not surround encoded, they sound better that way)."
Presto > I thought I was the only one! ;)
CT: "And I've given up on equalization, convolution, bass management, upsampling."
Presto > I'm far too interested in that stuff to give up on it as a hobby. There are days I wish I had only a turntable and an integrated amp so I can put all of the technology away and just listen to music. But I am convinced that one of these days, all of the tweaking and experimenting will slowly turn into longer and longer listening sessions as DSP filters and room correction (AND my ability to implement them) get better and better. I'm really not a big fan of upsampling. I simply play all digital files back in their native resolution and bitdepth. I find that upsampling (on the PC anyways) is just a waste of computer resources that could be used for convolution or filtering! :D One thing people miss is that if you tax your system too much, you ARE getting degradation of SQ long before you are actually hearing dropouts or clicks and pops. So, for sure as you "Add DSP" into your signal chain you could be hearing lower SQ. But this does not mean there are additive "losses" (like how noise is additive in an analog signal chain). It just means one needs sufficient processing power to all of the DSP one wants to do.
CT "I know, it's heresy on this forum, but hey, at the end of the day it's the music that matters."
Presto > True - but in my mind, taking a digital master from a soundstage that was digital down to the mic mixers and pressing it into a vinyl disc that slowly wears out is just insanity. No less insane than digitizing a pristine analog recording anyways! For those of us who believe that digital can eventually come out of its infancy and sound really good, I think we're almost there - and some are there already. For those who believe it will never ever work, well, for them it will never ever work!
*** So, a big difference is that for active (pre-amp) filtering systems, the amps do NOT see crossover impedances in the stop band region(s), since there is simply no voltage AT those frequencies being amplified. ***Not sure I get your point here. Regardless of what the frequency content of the input is (and in particular whether it's bandwidth limited or not), what matters are the electrical characteristics of the load, which is different whether you are driving a driver directly vs through a filter. Depending on filter design, the load may be easier to drive (whether or not there are or are not any frequencies outside the passband) because the filter itself changes the impedance of the load.
*** I think digital/active filters are much much more transparent than their op-amp based analog/active predecessors. Passive networks are not necessarily 100% transparent either. ***
This is selective rationalization. You are effectively arguing that a really good DSP algorithm may be more transparent than a badly designed or overly-complicated passive network. Well, it may be, but a much better answer is: don't use a complicated passive network. If you have a dead sounding speaker, sell it and buy a better one!
*** but in my mind, taking a digital master from a soundstage that was digital down to the mic mixers and pressing it into a vinyl disc that slowly wears out is just insanity. No less insane than digitizing a pristine analog recording anyways! ***
Agree with you 100%. I also don't get vinylphiles raving about 180gm pressings of some digital recording. Why? I can understand people raving about original pressings of early recordings though - I'm sure the vinyl pressed from a pristine master tape will sound better than a digital remaster off a 30 year old tape.
*** But I am convinced that one of these days, all of the tweaking and experimenting will slowly turn into longer and longer listening sessions as DSP filters and room correction (AND my ability to implement them) get better and better. ***
For me, the Holy Grail is 64-bit processing - once that is common, I'll probably give it another try. But 32-bit just isn't good enough for me. Have a play with Sonar using it's 64-bit mixing engine and some 64-bit effects - you'll notice the difference switching 64-bit on and off. Chews up the CPU though.
*** For those of us who believe that digital can eventually come out of its infancy and sound really good, I think we're almost there ***
As far as I'm concerned, we're already there. I think digital (without processing) sounds really good. I'm not one of those "digital can never be as good as analog" people - analog has problems as well. The reason I still like listening to records is that I can buy them at $1 each from our local charity shop. A good LP has a certain crispness and clarity that's hard to match in digital (note: "hard" does not mean "impossible") but i don't like vinyl wear and tear either. And since I only buy second hand LPs, I see that a lot, and the difference between a worn out LP and a pristine one is about 3-4 dB of dynamic range - the worn out ones tend to shave off transients by around that, sometimes up to 6dB.
Christine:About multi-amping...
Yes, I do see how the crossover and the driver "add up" (vectorally) to provide a net impedance as seen by the amp. Let's look at three examples, all being different from one another.
1) Your speakers - each amp gets it's own unique leg of a parallel network, lets pick the woofer
2) A full range speaker - an amp gets the impedance of a speaker but no x-over
3) An amp in an active system driving a midrange (crossovers in preamp stage)In your case, the amp is seeing the net (vectoral) sum of the impedances of the crossover network and the driver. In the case of the fullrange, the amp is simply seeing the impedance of the driver. In the third case, the amp is also seeing the impedance of the midrange, but **only at those frequencies that are within the pass-band and crossover regions**. In the stop band, there is no "information" being passed, therefore there are no "voltages" being amplified at those frequencies, therefore there is no current drawn at those frequencies, therefore the amp does NOT see the impedance at those frequencies. A specific example would be an active bandpass filter so steep that it only allows one frequency to pass through, say 1000Hz. In this case, the amp would only see impedance at ONE frequency - 1000Hz. The impedance graph is now just a "dot" representing the vectoral sum of the reactance and the resistance at that particular frequency. The impedance at 100hz and 10,000 hz of a driver connected to such a filter would matter not one single bit, since no signal at those frequencies would ever be amplified in the first place.
So, to compare the amp in an active system to that of a passive system, in a passive system the amp or amps always see the sum of all parallel networks, both drivers and filters - even in the case of passive multiamping - where each amp gets its own unique filter and driver. This impedance "seen" encompases the entire audible bandwidth in either case. An amp in an active system only "sees" the impedance of the connected driver that are NOT in the stop-band (aka only the pass-band and crossover regions.)
I guess you could say that although a load has an "impedance" from DC to light, the only impedance that MATTERS to the amp are for those frequencies that it is called upon to reproduce. An amp (or any voltage source) only can "see" impedances at frequencies that it is driving voltage at. Impedances at other frequencies are there - they're just "not in use".
If what you are saying is true, then one could not use a ribbon tweeter with a near-zero impedance below its usable frequency range. The amp would "see" this almost-dead-short and go into overload immediately. This is not the case at all. You CAN connect this driver directly to an amp - so long as your ACTIVE crossover prevents your amp from ever GENEERATING a voltage that is sufficiently low enough in frequency to drive a current at those frequencies. No low frequency voltage, no low frequency current, no low impedance is seen by the amp, and nothing is overloaded. Try running that tweeter WITHOUT the active filter, and the amp will fry as fast as the tweeter will!!
*** Yes, I do see how the crossover and the driver "add up" (vectorally) to provide a net impedance as seen by the amp. ***This was the main point I ws trying to make, so I'm glad we both agree.
*** An amp in an active system only "sees" the impedance of the connected driver that are NOT in the stop-band (aka only the pass-band and crossover regions.) ***
I think you will find I have already aknowledged this earlier. To quote one of my earlier posts: "There is a counter argument that says filtering the amp used to drive the tweeter achieves better transparency."
I noticed you are mainly using the tweeter in your argument, and I seem to recall that research does suggest that filtering at the preamp primarily benefits the tweeter.
But a tweeter is very easy to amplify, typically only a few watts at the most. So the benefit of the approach that you suggest is minimal. Arguably filtering post amp is almost as good (but you do waste a lot of the power of the amp, admittedly).
Whereas it can be shown a woofer's impedance does change significantly with the addition of a filter post amp, and woofers require hundreds of watts to drive at high decibels (power required is inverse to frequency).
Anyway, I don't mean to say that there are no benefits to digital crossovers. And certainly you can potentially do digital crossovers, plus equalization, plus room correction more accurately than an analog only solution. I just don't like the "haze" that that imparts.
If you look at 32-bit floating point processing (which is what most software and DSPs do), it's not good enough to preserve 24-bit resolution. A 32-bit floating point number has 24-bit mantissa and 8-bit exponent. Numerical theory suggests you need approximately twice the resolution of your inputs when doing calculations in order to preserve the precision of the inputs. So 32-bit fp is barely adequate for 16-bit signals (the 24-bit mantissa has about 8 bit headroom for intermediate results), but clearly inadequate for 24-bit signals.
Ideally, you want to do all your calculations in 64 bit floating point (or simply use IEEE double precision which holds intermediate results to 80 bits), and then dither the final result to 24-bit fixed point before passing to the DAC.
I don't know any player or DSP that currently does this in real time. When that day comes, I'll be very keen to try it out. TI for example is starting to manufacture DSPs capable of 64-bit processing. And 64-bit VST and DirectX effects should hopefully be more prevalent when Vista is released. Sonar already has a 64-bit mixing engine, and I seem to recall Perfect Space is a 64-bit convolution reverb (but I could be wrong).
Christine:I was looking at Pristine Space - I had no idea it would be 64 bit capable - I will try and verify this. This would be phenomenal, since the Thuneau Allocator (according the website) now uses a 64 bit engine.
I got the SourceForge VST convolver working, but I think it's only 32 bit - I need to check this too.
What I really need these days is a really good prosumer or professional soundcard that has SOME ASIO mixing capabilities and goes beyound mapping only hardware I/O in ASIO. This will preclude the need to use ASIO4ALL and Virtual Audio Cable to stream data into the VST plugins via a kernel streaming output plugin from Foobar or Winamp.
The cards comes with a driver frontend called PatchMix.PatchMix allows dynamic assignment of ASIO inputs/outputs to hardware, and you can mix/route the signals, add effects etc. You can even do things like route an ASIO output to a Wave input etc.
The software works like a mixer so it should do what you are asking for. I can't remember if PatchMix is VST capable or not - would be cool if it was. Even if it doesn't, you can route an ASIO output to an ASIO input which an external VST host can then process and send back via another ASIO output back to PatchMix.
Christine:If Patchmix can handle VST plugins, this WOULD be awesome - eliminating the need for an external VST host.
I have looked long and hard at different cards - and the EMU 1820M is definately on my shortlist, along with others from Lynx, RME and Edirol.
Apparently, the 1820M uses some really good DACs in it as well. I can't remember if the Breakout box / dac module is PC bus powered or from an external source. The ability to power the DACs from a clean, regulated (or battery) source is a "must have" on my list.
ESI also makes some cards which use the "Directwire" patch software - hey didn't you mention this software some time ago? This is a neat "GUI" for those that want to "virtually connect" wires to create their PC's audio stream. Nothing wrong with an intuitive solution I guess...
Thanks again for the informative discussions.
No the 1820M is powered from the computer via a special cable (that patches into the computer's power supply directly), but it has an additional set of voltage regulators in the breakout box.The DACs are pretty good, Cirrus Logic 4398.
You might want to check out the 1616M PCI - which is the replacement for 1820M, which supports an external power source (I think).
The "directwire" facility on ESI cards is pretty primitive - it only allows routing, not processing.
Whereas Patchmix is a full mixer - you can do effects chains, multiple busses, mix and pan, etc.
Given that I don't do any processing (!), all this is wasted on me. But I realised from our discussion you may be very interested in it.
Another good thing is that the effects are run from the DSP on board the breakout box, so they don't take any CPU time :-)
Even if you don't want to use the on board effects, you can use an external processor, because all ASIO ins/outs are "virtual" on Patchmix and not tied to the hardware. Assume ASIO OUT 1/2 is the output from Foobar or Winamp. Within Patchmix, you can route this back out to ASIO IN 3/4, ASIO IN 5/6, ASIO IN 7/8. Now you can run three external programs that processes three separate stereo signals (for your digital crossover). For example, program 1 reads ASIO IN 3/4, processes, writes out on ASIO OUT 3/4. Ditto for 5/6 and 7/8
Now, back in Patchmix, you accept ASIO OUT 3/4 and route it the first physical DAC pair, accept ASIO OUT 5/6 and route it to the second DAC pair, etc.
You can set all this up so it's all connected and ready to go when you boot the PC.
COOOOOOOOOOOL!!! :D
Hey P,I think there is one possibility you missed. One can passively bi-amp using passive crossovers, but multiple amps.
If each amp has an in and an out, one could take a stereo amp and run into the "left" channel, and then out to the "right" channel. NOw BOTH outputs would have the same signal (say center), and you could use the stereo amp to drive the top an bottom of one speaker.
OH, and to complicate things, there are such things as line level PASSIVE crossovers too...
Dawnrazor:"I think there is one possibility you missed."
Actually I missed more than one. :o)
With multi-amping (2/3/4 way etc.), you can have passive filters after and/or before the amps, and "before" the amps there is also active/analog and active/digital, where active/digital could be in a PC or in an outboard DSP based unit.
The easier way to think about it is when you only have ONE AMP. No matter what you do then, you're bi-wiring / tri-wiring.
Multi-amping as Christine has done (passively on the ouput using parallel networks) is a great way to go. Each passive network gets it's own amp circuit back to the wall plug, thus greatly reducing (if not eliminating) the IM distortion that the all-too-common method of daisy-chaining multiple passive networks on one single amp can cause. It's also a great idea for people who NEED to have SET amplification, but also have speakers that employ parallel networks. If you're a 300B maniac, you can bi-amp or tri-amp your system, meaning you can get 16 or 24 WPC instead of just 8 watts - but you need to spend three times as much! :o)
"But honey, I need to get six 300B monoblocks. Christine is multi-amping, why can't I? C'mon... PLEEEEASE?!?"
Now we just need a solution for our friends who are stuck with series crossovers! ;)
Anyways, back to work. My ass is grass today and the boss in the lawnmower.
This post is made possible by the generous support of people like you and our sponsors: