DSP vs. active analog crossover vs. passive analog crossover. What is your take?

@mikelavigne wrote:

no simple answer to this question.

the crossover should compliment the media, and the degree of room acoustical work that has been done. dsp fixes room<->speaker integration issues. but it’s not the only way to do that. at lower price points and integrated systems dsp does result in a more listenable performance. but past a certain price point it starts to get in the way of signal path purity for 2 channel.

so no absolutes in this question.

it depends....

We may be addressing different aspects here. Digital Signal Processing can be many things, but in my use of the Xilica DSP unit it serves only one purpose: acting as a digital cross-over, and nothing else. No "room<->speaker integration," surround processing or other; 2 channels only (i.e.: 3 outputs pr. channel), with 4 outputs dedicated for the 2-way main speakers, and 2 outputs for the subs. It’s worth noticing that as a digital XO in my case there’s no passive ditto in the signal path, so each amp output channel sees its respective driver terminal directly.

Is this how your experience with DSP, acting actively as a digital XO solely, has also been formed - apart from being used as a room correction unit over already passively configured speakers?

i own the ultimate dsp processor, the Trinnov Altitude 16. i use it with my 9.6.3 surround sound system for my home theater. it uses ’object based’ dsp to create soundscapes that do make movies more real. horses for courses, and it’s the ideal solution for those type movie or concert recordings.

I looked up the Trinnov - an impressive beast of a multiple-feature, high quality apparatus. For home theater use in particular I gather it’s a godsend, and with digital inputs and a digital source one would avoid the A/D conversion, not that I find the added A/D conversion to make much of a difference sonically, if any.

for traditional 2 channel i prefer an analog crossover since i’m a big analog recording guy, as well as have no intention of ’double’ converting my digital files with dsp for my 2 channel listening. that would suck the life out of the music.

The A/D to D/A conversion necessitated with a DSP in your case with an analogue source - if the DSP were to act only as a digital XO sans any passive XO - to my mind would be the lesser evil compared to the influence of passive cross-overs on the output side of the amp, but again, that’s just me.

but i do have a separate dedicated 2 channel room. my 2 channel room is purpose built and highly tuned to avoid the need for any dsp. i’ve fixed the room, and retained the purity of my analog signal path.

With my former passive configured all-horn speakers I at one point had 3 processing stages involved: the passive cross-overs in the speakers as an analogue "processing" stage, the Xilica DSP to high-pass them, and JRiver Convolution hosting a software for room correction in both the amplitude and time domain. It was a capable setup overall, but I prefer my current fully active setup with only one processing stage: the Xilica DSP. Room acoustics have been optimized with both diffusion and light absorption, so no room correction.

my 2 channel room speakers have 2 towers per side; each tower is 7 foot tall and 750 pounds (3000 pounds total). one tower per side is passive, from 35hz and up, the other tower is active and powered for under 40hz, (on paper) -3db at 7hz and -6db at 3hz. so serious bass capability. the passive tower rolls off at the bottom and cannot be used as a stand alone speaker. the bass tower uses analog adjustments and gets it’s signal from the passive tower speaker terminal so it’s signal mimics the sound of the main amplifier. the crossover can be adjusted from 50hz down to 20hz for best room integration.

i get awesome performance, considering the years of work i’ve put into the room.

Indeed sounds like an awesome speaker setup - kudos. Have you tried running the whole system fully actively with no passive XO’s involved?

i think many situations can benefit from dsp, but at the tip top of the 2 channel music reproduction food chain it’s a penalty.

I differ here, certainly as outlined above using a DSP unit as a digital XO - sans any passive ditto - only.

I found going Active to be huge step up in Linn speakers/amps - incredible change in dynamics, clarity, spaciousness.  Unless a room is designed to acoustically match the speakers, properly applied DSP is requirement to achieve clean/natural/flat sound.

Great thread - thanks for posting it.

Below is my first hand experience.  DSP/passive/active crossovers (XO) doesn't have to be mutually exclusive.  For example, DSP can be an "and" not an "or" proposition when considering passive XO options.

My main stereo speakers have internal passive crossovers (XO) which I'm considering removing and housing externally to avoid the negative vibrational affects of being inside the speaker.  Ideally, I would bypass the passive XO for a digitally active implementation, but ripping out the XO usually voids warrantees and makes reselling difficult, so it's not a slam dunk decision for me.

With the above situation, it hasn't stopped me from using DSP for room correction and XO used for low-passing a pair of subs and high-passing the mains using a single stereo amp for the mains, or for crossover duties (and room correction) when i chose to bi-amp the mains (with subs).

Have you considered adding DSP to your speaker's passive XO to improve system quality as an interim step before going to a fully active XO?  If so, then read on . . .

Constraints/Pre-Requisits:

> DSP won't work for your record collection only on digital sources.  In theory you can run your record collection through an Analog-to-Digital converter, perform DSP, then convert back to analog but I'd expect this to rob the music of its analog charm.

> A multi-channel DAC is required as a foundational piece of the puzzle to allow speaker-by-speaker level control for XO and room correction purposes.  I use an exaSound 8-channel DAC and JRiver's DSP Studio to control things at a micro not macro level.  A two channel DAC will only implement DSP on a left or right macro channel basis and everything receiving say a left signal may "see" the DSP corrections, although the XO settings help funnel the corrections to the right speaker driver, but this won't work for phase alignment which needs micro level adjustments.  (I don't use AV receivers but many have built in room correction (Anthem ARC or other brands that OEM Dirac) so I am ignoring these for the purpose of my response.)

> An acoustical measurement tool (i.e. mic+software) is a requirement in my opinion in order to manage what you can measure.  REW or OmniMic are excellent ones, and I also use Audiolense XO for the creation of FIR correction filters (see "GOLD" quality level below).

3 Levels of Sound Quality

"Bronze" - passive XO with no DSP used

> this is what I call the raw signal that relies more on acoustical treatments to help rid the room of troubling modal peaks or nulls.  Measurements help determine best speaker or sub or listening position, time align speakers, or determine if acoustical treatments are doing what's needed or are in the best positions on walls or ceiling.  It doesn't do any freq response corrections and time alignment is all manually done, which requires a certain skill level.

"Silver" - passive XO with DSP used

> this improves upon the Bronze level by providing far more XO flexibility by providing nearly an infinite combination of XO frequencies and/or slopes that you don't get on a sub's plate amp, or an analog XO like the Bryston 10B that I used to use.  While the Bryston 10B XO placed between the preamp and amps sounded very good, I would say that the improvements it brought about lied mainly in the sub integration with a smoother freq response; it couldn't do anything in the time domain to improve things. But, it only have a handful of XO frequencies from which to choose and 3 slope options (6/12/18dB per octave).  When JRiver's DSP was used for room correction - used below 500Hz for cutting of peaks - it sounded quite satisfying.  But each parametric EQ filter affects phase so I minimize the number used which doesn't allow for the flattest freq response.  I wanted more flexibility for even better sound . . .

"Gold"  - passive with digital XO and DSP for freq and time domain correction

> Freq Domain: in a single stereo amp config, I use the speaker's natural bass roll-off to set the digital XO freq and slope to match it.  Same goes for when the mains are biamped - I create a high-pass filter to match the woofers and another high-pass filter to match the midrange's natural roll-off (the tweeters do not have their own speaker binding post so I can't do much to affect them). DSP is used to integrate the subs which are placed in different spots in the room and have slightly different XO frequencies and freq correction as they load the room differently from their unique spots.  The granularity of 1Hz increments and slopes within a digital XO makes for surgical precision on an individualized speaker basis.  For example, I use this granularity/flexibility to help ensure that nulls fall on frequencies that are not actual notes played by an instrument; middle C on a piano is 261Hz so I manipulate the XO freq such that the deepest part of the null will fall either side of 261Hz between the adjacent semitone notes of 246Hz and 277Hz. This digital flexibility helps reduce the number of notes affected by nulls which improves sound quality.

> Time Domain: I can manually set delay/phase and phase reversal easily but have recently upped my sound quality by using Audiolense XO to automatically create FIR correction filters to match the freq response to my chosen target curve and time aligning subs to each other and to the mains.  I use my ears to be the final judge of the timing domain - a good reggae song with strong bass downbeats and upbeats played by other instruments is a good way of ensuring the time alignment of the subs and mains (e.g. the downbeat played by the subs and the upbeat played by the mains shouldn't lag each other and should be correctly timed to the beats per minute of the song). Alternatively an impulse response (IR) using just say the left main and left sub could be used to align their timing so a single IR peak is seen.  The sound quality is drastically better than the Bronze or raw sound even in my well acoustically treated room and  better than the Silver but not by as huge a margin.  This was confirmed with the help of two audiophile friends' testing recently.  The Gold listening with FIR correction filters invoked is free of listening fatigue, smooth even sounding frequency response and the best time aligned bass I have been able to achieve.

Caveots - (1) as some have noted in this thread DSP can bring negative audible effects like phase issues or distortion.  Using a FIR correction filter will smoothen the freq response and handle phase issues very well but it comes at a "cost" of attenuating the overall volume to prevent digital distortion from occurring that's related to boosting nulls. This filter insertion loss (attenuation) reduces the IR amplitude which reduces some of the musical dynamics.  There are ways to minimize the filter insertion loss and ensure dynamic integrity which I implement as dynamics are important to me.  (2) your chosen target curve that helps guide what XO frequencies are chosen or which modal peaks to cut with parametric EQ has a huge impact on timbre integrity.  Boosting bass as most would do can make male voices overly chesty and thick if the boost is started at 300Hz but much less so if started at 100Hz, for example.  Experimentation is key.

Sorry this turned out to be such a long read but I wanted to share my experiences in hopes someone finds them helpful . . .  kevin

If you start with digital there is no way anyone could know if DSP is used or not. You would know if it was turned on/off but it is supposed to be doing something so I hope so. Can you hear an A/D and D/A conversion? With the right parts highly doubful but then some claim to hear the difference between different high end caps in crossovers so it is a moot point I think.

I would read what Bruno Putzey says about active speakers. He is designing to a price point but still next level in terms of what is possible. Impossible to replicate with passive.

Implementation will still trump basic technology. It will be like vinyl and digital. Digital is far more capable but a good mastering is more important than anything. A bad speaker will be bad no matter passive, active or DSP. However, similarly, it will allow very competent lower cost speakers as the tech progresses. Being mechanical good speakers will never be cheap of course.  

One thing is for sure, a small portion of the audiophile world will continue with their old tech, blissfully claiming the superiority and pureness of their old tech (using lots of incorrect concepts and poor understanding of how things work) while the rest of the world progresses without them :-)

Some here are talking about active crossovers with added DSP. I’m talking about Active performed within the DSP there are no passive filters. The signal goes in digital is processed within the digital domain  to DAC then sent directly to the amp then on to the transducer. Volume control can also be conducted within the digital domain. If an analog signal is sent it first goes through an ADC before DSP. The OP was asking about crossovers. There are passive and active crossovers both use analog components and then there is DSP crossovers which is an entirely different thing.