miniDSP and it's ilk allow rapid prototyping to come up with the cross-over that works best for the users requirements, that then allows an analog unit to be built right first time. Other than the (no longer available) Linkwitz boards there isn't much around off the shelf with any advanced features like all pass filters.
Plus with chip-Amps being so good, and sources all being high level soon the X-over will be the place where a DIY type can really do something rather than just stick an ebay purchase in a box!
Plus with chip-Amps being so good, and sources all being high level soon the X-over will be the place where a DIY type can really do something rather than just stick an ebay purchase in a box!
Bill,
I don't know about that. 🙂 I think for most DIY'ers it's a big step between developing a crossover with a miniDSP-like unit and realizing it with an analog counterpart.
The concept is sound, but a person still needs to know circuit design, how to create the optimum topology, notch filters, etc, etc.
I think most DIY'ers will probably just stop after the "development" phase and call it good. 🙂
For info: One of our members has produced some excellent three-way crossover boards that will allow a fairly high level of customization from a typical, generic active crossover design.
LX521 ASP Reworked
Cheers,
Dave.
I don't know about that. 🙂 I think for most DIY'ers it's a big step between developing a crossover with a miniDSP-like unit and realizing it with an analog counterpart.
The concept is sound, but a person still needs to know circuit design, how to create the optimum topology, notch filters, etc, etc.
I think most DIY'ers will probably just stop after the "development" phase and call it good. 🙂
For info: One of our members has produced some excellent three-way crossover boards that will allow a fairly high level of customization from a typical, generic active crossover design.
LX521 ASP Reworked
Cheers,
Dave.
Dave,
I might be wrong, I often am, but I hope I am right in some way. Whatever solution is chosen there is a steep learning curve, especially for those coming from pure electron herding into speakers. But that is the fun of it.
I have a couple of the buyout ASP boards when SL was selling them at $25 each. I will build them up once I know what I am doing with my speakers.
I might be wrong, I often am, but I hope I am right in some way. Whatever solution is chosen there is a steep learning curve, especially for those coming from pure electron herding into speakers. But that is the fun of it.
I have a couple of the buyout ASP boards when SL was selling them at $25 each. I will build them up once I know what I am doing with my speakers.
No, I would like to have one x-over frequency between tweeter (actually a compression driver) and mid/bass. 1.5 kHz is my target.
I consider the advantage to using a DCX2496 to be the ease of developing a prototype quickly and then implementing the requied EQ in the analogue domain.
Unless the miniDSP convertors and their associated analogue circuitry are exceptionally bad (don't know, never used one) I can see no reason why I would want to bother to replicate it analogue once it's dialled in.
Yeah and by that logic there is no reason for >99% of the stuff on this site. The fact that sensible is ignored here is half the fun.
I'm a proponent of passive XO. My current speakers never sounded nearly as good active (Minidsp Nanodigi), with any tried settings, as they did with even with the prototype passive xo (with the same DAC and amp).
Some things are completely overlooked in this thread so far...
We should remember that when all the drivers are in parallel (or in series) in passive crossover, they create electrically and acoustically "a new driver". Parallel drivers damp other parallel drivers, interfere with other drivers via acoustic parths (acoustic feedback, speaker acting as a microphone) and so on. Interactions are numerous. Passively crossovered speakers create a whole new circuit with the amplifier and act very differently as a system than when driven one by one. This "crosstalk" and interaction cannot be replicated with active crossover. Passive crossover also introduces current drive situations for the drivers (decouples the drivers electrically from the driving amplifier), which also cannot be replicated with active xo. And on a more basic level, also the load impedance and phase is very different to the driving amplifier. The amplifier-speaker interaction is different. It's getting late and I'm propably just repeating myself...
I believe all these "little" things, mainly the interactions part, is what makes passively xo'ed drivers work as one, cohesively, which is the main selling point of passive crossover's sound signature. They sound very natural and coherent in their tone, everything works in unison, and not like there is "three very different kind of drivers". Everything is smoother, more liquid, free-flowing, analog.
These two systems just cannot be equalled for the above mentioned reasons (and there is more, I just don't know them).
Another intertesting topic would be what quality of electrical crossover will surpass acoustical crossover? I'd say.... propably none, acoustical crossover might be the reference. It's the only kind of crossover that has the ability to reduce distortion that the driver generates. Hard to design and can only be used for woofers and midranges, not for tweeters which will always need an electrical HPF. Almost lost form of art nowadays in speaker design, cross where the fence is low? I highly suggest trying acoustical xo, let the purism take over you. 😀
Some things are completely overlooked in this thread so far...
We should remember that when all the drivers are in parallel (or in series) in passive crossover, they create electrically and acoustically "a new driver". Parallel drivers damp other parallel drivers, interfere with other drivers via acoustic parths (acoustic feedback, speaker acting as a microphone) and so on. Interactions are numerous. Passively crossovered speakers create a whole new circuit with the amplifier and act very differently as a system than when driven one by one. This "crosstalk" and interaction cannot be replicated with active crossover. Passive crossover also introduces current drive situations for the drivers (decouples the drivers electrically from the driving amplifier), which also cannot be replicated with active xo. And on a more basic level, also the load impedance and phase is very different to the driving amplifier. The amplifier-speaker interaction is different. It's getting late and I'm propably just repeating myself...
I believe all these "little" things, mainly the interactions part, is what makes passively xo'ed drivers work as one, cohesively, which is the main selling point of passive crossover's sound signature. They sound very natural and coherent in their tone, everything works in unison, and not like there is "three very different kind of drivers". Everything is smoother, more liquid, free-flowing, analog.
These two systems just cannot be equalled for the above mentioned reasons (and there is more, I just don't know them).
Another intertesting topic would be what quality of electrical crossover will surpass acoustical crossover? I'd say.... propably none, acoustical crossover might be the reference. It's the only kind of crossover that has the ability to reduce distortion that the driver generates. Hard to design and can only be used for woofers and midranges, not for tweeters which will always need an electrical HPF. Almost lost form of art nowadays in speaker design, cross where the fence is low? I highly suggest trying acoustical xo, let the purism take over you. 😀
Last edited:
It could also be that you 'like' the innacuracies that your passive X-over gives you. Without giving us any details on what you did hard to tell.
Nope! Negative. The Dsp on that Xfi Elite Pro card is *bypassed completely* in the "Audio Creation Mode" and signal is passed *bit perfect*. It is more productive here to be accurate. Bit perfect means bit perfect, NO DSP whatsoever is done in that card unless requested. One DA conversion ONLY and a very high quality one at that. Google "Creation Mode" and "Bit Perfect" please. Not being accurate is not helping, we all have the same goal here. FWIW, the DSP in some earlier card was god awful going through horrid and highly audible SRC. Not so in this case, though the SRC (if you choose to use it) is better. I avoid all SRC in mastering unless forced to use it, then I only use Weiss.
Last edited:
That is sort of how I use it, but then I redo it in JRiver digitally. My DCX unit could use some modification for direct out via caps, way too much in the signal path compared to a good interface, slightly audible imo, even using the AE digital in provided on input A.
Per above, the Creation Mode (bit perfect) is also available on the Soundblaster X7 series units I mentioned in that thread as well as most other recent Creative efforts. Jriver software crossover via convolution then through these inexpensive and remarkably clean cards is very easy to do
Last edited:
Partially agreed.
But I would go one step further: One would need to make use of more of the possibilities that a DSP offers than what is possible with a DCX or some other nicely priced DSP solutions in order to prototype refined active crossovers. At least addition and subtraction would be nice to have.
For this purpose I use Sigma Studio. This is a graphical development platform that is used to program a certain family of DSPs by Analog Devices.
But even then one has to be aware that the DSP crossover has to be transformed to an active analog one with care due to the "warping" of the frequency axis that is typical for discrete-time filters.
Regards
Charles
Much better unit: 2 in, 4 out, bulletproof, hi-fi quality (unlike the other digital processors that I see in this thread): used EV Dx38-- found ubiquitously. For instance: Used Electro-Voice DX38 Signal Processor | Guitar Center
Others that are recommended: used EV DC-One, Ashly Protea, etc. (all 24/48)
Better fidelity: any Xilica XP, XA, XD series, Yamaha SP2060 (all 24/96 internally)
List of units that I don't recommend (SNR or analog SQ issues)-- miniDSP, Behringer DCX2496, dbx driverack, etc.
List of analog active crossovers that I don't recommend (lack of EQ, delay, protection, etc.)-- all units.
Others that are recommended: used EV DC-One, Ashly Protea, etc. (all 24/48)
Better fidelity: any Xilica XP, XA, XD series, Yamaha SP2060 (all 24/96 internally)
List of units that I don't recommend (SNR or analog SQ issues)-- miniDSP, Behringer DCX2496, dbx driverack, etc.
List of analog active crossovers that I don't recommend (lack of EQ, delay, protection, etc.)-- all units.
The Behringer DCX units sound fairly nice if you use the digital in @ 96k They sound pretty dang good with a direct out mod. I tried it on two channels. Would not even go in analog to out analog with any of these units. Spitty sound for sure. But mine is from 01, they may have changed topology.
I am always amazed how simple conversation turns into black and white A vs B, while really there are so many shades of gray in between, and very often it is not A vs B but A + B or some aspect of both. Comparing passive crossover to analog or DSP based is such a wide, impossible conversation to have. Each one of these has what is better than others, as well as what is worst than others. The worst one is the wrongly implemented one.
I made them all, used them all, listened to them and feel pretty confidant what I learned in the process. I used passive ones, used analog active based on opamps, analog active based on jFet buffers, DSP based on DCX2496 that I heavily modified where only DSP chip stayed unmodified (see in DCX mod thread somewhere in mid 2000s) The last one is based on dual DACs and computer as an DSP with software.
My latest one is combination of passive and DSP (computer) based, see the last one here:
http://www.diyaudio.com/forums/newreply.php?do=newreply&noquote=1&p=4594198
or for analog jFet based one go at the very beginning of the thread.
In ideal world trip that signal has to travel from FLAC, WAV or MPx file to the speaker has to be the shortest one. Ideal speaker will have no crossover, but since that is not the case than the least amount of active devices or caps in that path is preferable. Talking about active devices single transistor or ideally jFet or tube is always better than any opAmp. OpAmps are great for many things, and they undobtly have their own space, but in my book, if possible not in the audio signal path. From that perspective passive crossover beats anything else, given that something has to be in front of them anyways. Adding additional electronic to the existing path will always apply audible tax. So, from that perspective there is the winner. But at the very moment you need to correct for a speaker imperfections, that task becomes much more complex with limited possibilities of what could be done.
Now welcome to the world of active crossovers. In analogue active, every higher order (steeper curve) exponential brings another set of active devices. Pretty soon there are so many devices and caps in the path of signal that there is only fog left that reaches the speaker. Obviously one could be sensitive here, and limit himself only to lets say 2nd order filter and minimize EQing but even that makes its stamp. If done with opAmps say good buy to anything that came before it, if done with jFets - much better, but still we added to the signal path caps and active devices. In commercial units it is certain opamps are used and who knows what caps.
For that reason passive analog crossovers are thing of past, welcome to DSP.
Think about DSP as your new DAC, preamp and volume control obviously in addition to a crossover. Here we could do much more, but with limited choice of what is offered on the market. But with lots of work great improvements could be done... a lots of work. Take DCX2496. Only when heavily modified it sounds very good.
I prefer software and computer for DSP as it offers much more what could be done in terms of speaker corrections and gives me choice to still have an ultimate DAC, clock, volume and analog circuitry driving amps. As an added benefit, there is nothing between amp and a speaker. In real world sound card take this place, but that is the worst one could do. The ideal would be stand alone DACs synchronized with source and clock. Here you could take no compromise approach but you have to do it all by yourself since on that level there is nothing commercially offered that I know. Mini DSP is one solution but still with limitations.
Running more than 2 way on active is complex process with the need of many amps. I like combining it with passive for 3 or 4 way systems.
I made them all, used them all, listened to them and feel pretty confidant what I learned in the process. I used passive ones, used analog active based on opamps, analog active based on jFet buffers, DSP based on DCX2496 that I heavily modified where only DSP chip stayed unmodified (see in DCX mod thread somewhere in mid 2000s) The last one is based on dual DACs and computer as an DSP with software.
My latest one is combination of passive and DSP (computer) based, see the last one here:
http://www.diyaudio.com/forums/newreply.php?do=newreply&noquote=1&p=4594198
or for analog jFet based one go at the very beginning of the thread.
In ideal world trip that signal has to travel from FLAC, WAV or MPx file to the speaker has to be the shortest one. Ideal speaker will have no crossover, but since that is not the case than the least amount of active devices or caps in that path is preferable. Talking about active devices single transistor or ideally jFet or tube is always better than any opAmp. OpAmps are great for many things, and they undobtly have their own space, but in my book, if possible not in the audio signal path. From that perspective passive crossover beats anything else, given that something has to be in front of them anyways. Adding additional electronic to the existing path will always apply audible tax. So, from that perspective there is the winner. But at the very moment you need to correct for a speaker imperfections, that task becomes much more complex with limited possibilities of what could be done.
Now welcome to the world of active crossovers. In analogue active, every higher order (steeper curve) exponential brings another set of active devices. Pretty soon there are so many devices and caps in the path of signal that there is only fog left that reaches the speaker. Obviously one could be sensitive here, and limit himself only to lets say 2nd order filter and minimize EQing but even that makes its stamp. If done with opAmps say good buy to anything that came before it, if done with jFets - much better, but still we added to the signal path caps and active devices. In commercial units it is certain opamps are used and who knows what caps.
For that reason passive analog crossovers are thing of past, welcome to DSP.
Think about DSP as your new DAC, preamp and volume control obviously in addition to a crossover. Here we could do much more, but with limited choice of what is offered on the market. But with lots of work great improvements could be done... a lots of work. Take DCX2496. Only when heavily modified it sounds very good.
I prefer software and computer for DSP as it offers much more what could be done in terms of speaker corrections and gives me choice to still have an ultimate DAC, clock, volume and analog circuitry driving amps. As an added benefit, there is nothing between amp and a speaker. In real world sound card take this place, but that is the worst one could do. The ideal would be stand alone DACs synchronized with source and clock. Here you could take no compromise approach but you have to do it all by yourself since on that level there is nothing commercially offered that I know. Mini DSP is one solution but still with limitations.
Running more than 2 way on active is complex process with the need of many amps. I like combining it with passive for 3 or 4 way systems.
One of the commenters here apparently takes issue with my recommendation to avoid miniDSP. Perhaps reading on these threads will give you a much better idea of why I don't recommend them: https://www.google.com/search?safe=....0....0...1c.1.64.hp..6.10.1782.0.8_BomlgAEEs
The portions quoted above are easy to agree with.
When there are implementations that go from digital source to power amplifier inputs with highest fidelity, with fully customizable signal processing algorithms and support for all current stereo and multichannel bit streams--to bypass the need for an AVP--economically--I'd like to know when that day occurs. It'll be a good one.
I suspect that the MPAA and RIAA will try to stop it legally, however...because "it's too easy"...
Chris
Last edited:
- Status
- Not open for further replies.