DSP Active Crossover

@simonmoon wrote:

I’ve heard several systems with the mini DSP, and have been unimpressed.

I’ve also heard several very high end DSP systems, including the Kyron. And that system kills! Even Mr. analog, Michael Fremer loves it.

But my overall evaluation is, that it takes a very special DSP speaker system to best the best passive system.

It’s not as much the DSP unit as it is the implementation. My IIR-based DSP unit cost me just over just $1,000, and it’s an excellent piece of equipment for what it’s supposed to (make me) do, and one that also holds up perfectly well to Lake units (costing much more). That’s for nothing however if you don’t know how to turn those settings into proper, audible effect, of which there are different routes for that to be accomplished. I still prefer setting DSP-values manually by ear with the aid of measurements (and input from friends), and while a painstaking and lengthy process the results can be, and actually are extremely good.

Of a range of passively configured speaker setups I’ve heard that were converted to outboard active configuration - that is, bypassing their build-in passive crossover completely and replacing them with line-level DSP’s/electronic XO’s and more amps - each and every one of them eventually saw a substantial upgrade in sound quality over their passive iteration (and that was obvious fairly early on), to everyone listener involved. That’s all I need to know and a testament to the potential of active configuration, not least from the important basis of comparing the same speakers with different filter configurations.

Many are, on principle, against DSP due to speculated, negative effects of A/D-D/A conversion steps with analogue inputs only, but with no experience to really speak of that would actually single out this particular aspect as the detrimental factor. Why even make any assumptions as to what may or may not, technically, be the reasons for a speculated deficit?

One of the best setups I’ve heard, an outboard active one at, comprises the exact same DSP unit I’m using. For anybody wanting to tell me it’s an IIR-based filter and not a FIR ditto, while implying perhaps it’s the lesser solution, I can only stress the importance of seeing the forest for the trees in actually listening to a properly implemented active setup and let your ears decide. Should the FIR-filter hold the upper hand sonically, which theoretically it does not least in being able to generate linear phase response, that’s only an added bonus that will potentially distance active from passive even further.

@jaytor --

Kudos for venturing into outboard active configuration (it’s what I do myself). It will be interesting to learn of your findings (if you’re willing to share them) compared to your passively configured main speaker system once you’ve initiated the process and has gotten your head around it with actual impressions of the sonics to follow. Cool main setup, btw. - I’m sure it’s very capable.

My biggest reservations are 1) giving up my Denafrips Terminator+ DAC and nice-quality DIY preamp, and 2) using the DAC’s digital volume adjustments.

Your main priority is using the digital input on the dspNexus, I see, which - as a simplified system, with all that may entail - is what prompts above quoted concerns of yours. While on paper/in theory I understand worrying about introducing extra conversion processes, as an outset at least I wouldn’t be too concerned about them. What do you gain by avoiding them, while at the same time losing out on your Denafrips DAC and separate preamp?

Seems to me some effort of yours has been invested in sonically "shaping" the sound of your setup with these very components, so why not start with them remaining in the chain and instead use the analogue inputs of the dspNexus directly from your preamp? Yes, you’ll have an extra A/D conversion (the effect of which is blown out of proportion, if you ask me, if it’s even noticeable), but this way I’d argue you have better grounds for comparing the sound from the passive speaker config. and the one introduced with the active ditto on the basis of this alone.

There’s also the option of a software-based DSP preceding the DAC, which then necessitates a separate DAC channel for each of the amp dittos - like from the Okto Research dac8 PRO. This to some can bring into question whether a theoretical lack of overall signal symmetry of the DAC channels may have an effect on perceived sonics vs. only using 2 DAC channels that are then branched out on the output side of the digital crossover. I gather it’s in the splitting hairs department, but I have no experience to speak of here - not via my own setup, that is.

Regarding setting filter values good advice has been provided already. I’d experiment with steeper slopes than 4. order - myself I use 36dB/octave L-R throughout (BW slopestyle on the subs HP). Does the dspNexus use linear phase filters (FIR)? If not it’s not an issue, as I see it. In any case I gather you’ll be able to get even better integration between you sub and main towers fiddling with delay, although you may come about this by simply moving them back and forth to each other the way you've positioned them. With my speakers we used nearfield measurements of the MF/HF horns to set precise filter notches and make a light peak suppression. On top of that the real work was getting the delay right - not least between the woofer section of my main speakers and the tapped horn subs, where the "origin" of the front wave is somewhere inside the horn path, and not simply on the cone backside exposed at the mouth. With the tools at your disposal already I’d say you have a very good outset for interesting results.

@jaytor wrote:

Danville provided a sample 3-way crossover design which I have modified to meet my needs, although I’m sure I’ll make a lot more changes before finalizing my design. I’m starting out with the crossover points set at 180Hz and 1800Hz, using 8th order (48db/octave) filters. These are the approximate crossover points that my passive crossovers provide, although with much more shallow slopes.

The Audio Weaver software is very flexible and powerful, but requires that the user set all the filter parameters. In other words, you can’t just hook up a measurement mic and have the software automatically determine all the filter parameters to match a target curve. There is a lot more trial and error, but at the same time, you know exactly what processing is being done.

I have to make some speaker patch cables to connect between my speaker drivers and amps (bypassing the passive crossover), and then I can start testing on my main speakers. I’m hoping to do this over the weekend.

It will be interesting to learn of your findings when finally bypassing your main speakers’ passive crossovers for a fully active configuration. Not many do this on this forum, i.e.: switching their passively config. speakers to active ditto, nor those who start from scratch with speakers that came sans passive crossover to begin with. You don’t just buy speakers at hifi retailers without passive crossovers as an intended package for outboard active config. - except the rare likes of JBL M2’s, Sanders Sound and a very few others - indeed for that you’d typically have to convert your passive speakers into active, go the DIY route or pro segment. The latter two is what I did with DIY (shared) sub designs and pro cinema speakers intended for active config.

My main inspirational source for going active is a friend of mine who converted his passively config. S.P. Technology Revelation speakers to active ditto, and as it eventually turned out that switch was quite the, well, revelation. The passive S.P. Tech’s were power hungry beasts as few and developed on the Crown Studio Reference I amps, which are very powerful with a crazy high damping factor in the lower regions. However very few hifi amps can muster up that kind of speaker control and power delivery, and you’d wonder why the S.P. Tech’s didn’t come with the mandatory recommendation of being paired up with the likes of Ref I amps they were developed on. The bottleneck though proved not only to be about power requirement, but that bypassing the passively crossovers entirely - upon thorough implementation to active after a lengthy process - simply meant a massive upgrade in overall sound quality. Initially it was still very obvious that going active was the right thing to do - that was apparent from the very get-go.

Myself I’ve now used about two and a half years steadily upgrading my Xilica DSP-based, actively configured speaker setup in stages, lately only a few weeks ago. Detailed factory specs from EV manuals were initially involved to get a bearing on the crossover point (2-way main speakers, + subs) and where to place some of the filter notches; soon after near field measurements came to the aid for more precise adjustments, and what followed was a bit of restructuring acoustic room treatment, main speaker placement, some horn damping, amps experimentation (using similar amps top to bottom is a must), and eventually the lengthy process of making filter adjustments by ear mostly involving delay and subtle PEQ corrections. Gain structure, filter slopes and types (36dB/octave L-R here) and crossover points are usually found and settled on fairly early in the process, but lately we found some rather critical improvement in overall coherency with asymmetrical crossover points over the main speakers and delay reconfiguration incl. the subs as well.

Some may find it an impediment with the active speaker approach and setting filter values by yourself that it’s likely a never ending process, but once you settle on the rougher structure of your speakers, amps, source, cables and overall rudimentary implementation, what follows is the process of fine tuning all that into a progressively coherent whole in your acoustic environment in ways of specificity and accuracy that a passive speaker setup simply can’t equal. It’s not that much about swapping hardware any longer and being on that merry-go-round to find the perfect matches component-wise, but rather seeing the gains that can be made from a much better outset with a DSP acting as a digital crossover, and the many options that are offered here implementing the existing hardware components. To me at least that makes much better sense.