Several years ago I bought a Meridian DSP-5000 (96/24) surround system (VERY inexpensive/cheap) (6 speakers -5 working) ,center and sub. I have only used one pair of the 5000's for stereo ( others are spares) and 95% of my listening is to LP's. Not only do I have too many Phono cartridges and pre-amps -but -maybe 8 or 9 ? ADC's - in 50 + years of untreated audio obsession these are hands down the best speakers I have ever owned . DSP crossovers just make sense to me. Having recently moved to a much bigger/better room - I will have to assemble the Linkweitz LX-mini kit that's been waiting for years because of not having the right room to use them in.
Active Crossovers, anyone dabbled?
- Thread starter Columbo
- Start date
Not sure why you say this. IME DSP works perfectly well with analogue inputs. In fact, the best LP playback I’ve heard is with a Sugarcube in the playback path. it uses DSP to process and de-noise/click LPs. At this point I would not listen to records without a Sugarcube.
Last edited:
irb
pfm Member
The trouble with this is that accurate electrical slopes and phase will not yield accurate acoustic slopes and phase unless the drive units (in the actual boxes/baffles) are ruler flat for a couple of octaves either side of the crossover point. Which they never are. So the theoretical superiority of active crossovers remains theoretical.
As James said, you need to measure and adjust, whether the crossover is active or passive. Also, analogue active crossovers generally offer little or no ability to tailor response and phase to a particular system.
As a practical example, I have a two-way diy speaker using a Scanspeak 10" crossing to a Bandor 2" at 600 Hz. I used to cross it actively. But in practice it never sounded quite right. Modelling it in software suggested that standard active electical filters just couldn't give quite the right acoustic slopes and phase. I ended up (after more modelling and measuring) with a mixed solution - asymmetric active electrical slopes and an inductor/resistor combo for passive baffle step correction. That was better. But I've now gone fully passive with it, and it sounds the best it ever has.
Maybe I was cack-handed with the active setup, or just lucky with the passive. But right now I feel it's less than optimal to put extra line-level circuitry into a system (or even dsp manipulation) when you can tailor acoustic response accurately with just a few well-chosen passive components. (Crossing/equalising at low frequencies is a different matter, mind you.)
Realysm_42
pfm Member
Any issues with this approach and coherence between drivers / overall sound?
My main system is DSP based and active and got to agree that DSP is the way to go, using generic LR crossovers will not end well. The generic ones do not take into account phase of each driver as an example and often the different drivers need different slopes to take into account the inbuilt roll offs of the drivers.
IMG_8411a by A H, on Flickr
IMG_8415a by A H, on Flickr
The system is based around a Minidsp Nanodigi (discontinued), this is powered by a Anker 6 port USB block, input is either toslink or coax digital, onto the coax I have added a XMOS208 usb card. Output is through 4 Khadas Toneboard DACs (ESS9038K2M) to give 8 channels and I have also added a USB power socket on the rear. The outputs are played through a Rotel RMB1066 6 channel power amp. The case is an old 10 series Rotel DVD that was been hacked and a new front plate added.
rotel.com
Speakers are home made again - Repurposed Akai SW125 boxes (approx 40L) sealed, SB WO24P-4 woofer, Model EM1308 Mid and ET448 Tweeters.
I am still fine tuning but finding that alignment of driver phase makes quite a difference to the apparent 3D sound and instrument separation even with almost identical on axis frequency plots. The benefit of DSP is that you can add any "house" frequency curve you like plus takeout the worst room low frequency nulls/peaks. I have used REW for measurement and VituixCAD for crossover modeling.
Untitled by A H, on Flickr
IMG_8411a by A H, on FlickrIMG_8415a by A H, on FlickrThe system is based around a Minidsp Nanodigi (discontinued), this is powered by a Anker 6 port USB block, input is either toslink or coax digital, onto the coax I have added a XMOS208 usb card. Output is through 4 Khadas Toneboard DACs (ESS9038K2M) to give 8 channels and I have also added a USB power socket on the rear. The outputs are played through a Rotel RMB1066 6 channel power amp. The case is an old 10 series Rotel DVD that was been hacked and a new front plate added.
RMB-1066
The RMB-1066 multi-channel power amplifier is an enormously flexible unit intended primarily for use in surround systems and delivers high power in a variety of configurations.
rotel.com
Speakers are home made again - Repurposed Akai SW125 boxes (approx 40L) sealed, SB WO24P-4 woofer, Model EM1308 Mid and ET448 Tweeters.
I am still fine tuning but finding that alignment of driver phase makes quite a difference to the apparent 3D sound and instrument separation even with almost identical on axis frequency plots. The benefit of DSP is that you can add any "house" frequency curve you like plus takeout the worst room low frequency nulls/peaks. I have used REW for measurement and VituixCAD for crossover modeling.
Untitled by A H, on Flickr
James
Lord of the Erg\o/s
It is possible to add subjugate stages (e.g. notch filter) to an active analogue filter circuit just as one would with a passive filter. You just need software to model the design. It is much easier with a DSP set up, but I argue you still need to factor in the measured native response of the driver complement.
FWIW, I did not need any of the extra stages in the active crossover I designed for my E-X. It was easy enough to model the C/R values in circuit to shape the electrical transfer so the target acoustic transfer is achieved.
Hoopsontoast
pfm Member
irb
pfm Member
That's the Scan 10" I'm using, in 55 litres sealed. Embarrassingly, due to house renovation, they're still in unfinished boxes, about three years after I built them, no stuffing or anything.
Currently crammed into corners, with our tv on top, as my listening room's been dismantled. There's too much boundary reinforcement, but needs must. (I really need to get this renovation to get finished, so I can get a room for the system again. My head's full of speaker projects, but I won't have time/space to do anything until the house is fixed.)With the little Bandors, I know I'm out on a limb not using them below 600 Hz, but I just don't like them so much when crossed low - a squashed, transistor radio sound to my ears. Even crossing them at 300 Hz there's still a hint of that, I find. Of course, crossing them high puts a lot more attention on the woofers, but I'm pretty happy with how they sound. I'm also planning to add a ribbon tweeter back in when I have the time to work on a crossover, but I can live with what I have if it doesn't improve things.
I have lots of other drivers waiting for a try out, including pairs of Bandors, Scanspeak 25W carbon/paper woofers, and some bigger stuff. But the sound I get from the 26W/Bandor combo just works, for me. (Especially with a rather dodgy looking FW F5 clone I picked up recently - I'm amazed how well it copes with the tough load.)
I've had plenty of good, well regarded speakers in the past, (Harbeths, Revels, Thiels, etc., several electrostatics) but I don't really miss any of them (apart from Dali Skylines, maybe). If my DIY efforts never produce anything better sounding than what I'm using, I'll still be OK.
irb
pfm Member
Yes, if you're designing your own analogue active crossover, and can model the circuit, that gives you a lot more flexibility. I suppose I was thinking of shop-bought units, like the Bryston I had for a while. But to go active you'll still need the extra circuitry to be transparent, and you'll still need more power amp channels, which isn't great if you like class A amps.
I've tried to tell myself that all amps with decent specs will sound much the same, but then I hear something by Nelson Pass and all that goes out the window.
As someone who listens to vinyl, I used a 3-way analogue active XO (Rod Elliott's DIY design) on my Maggies for 20 years. Then I added a pair of subs and because they were further away from my ears than the Maggies ... I needed to impose a delay on the Maggies.
So I swapped the Rod Elliot XOs for a miniDSP 10x10HD - which:
a. provided all the necessary XOs - including the additional sub/bass panel XO, and
b. enabled the necessary delays to be implemented.
I then upgraded from the 10x10HD to the (digital in / digital out) nanoDIGI - which required me to add an A2D converter for my phono stage. (My other 3 sources were already digital.)
I then added Roon to my setup - so I could take advantage of Roon's:
* convolution ... to add a 'FIR filter overlay' to the IIR filters delivered by the nanoDIGI. (This improved SQ significantly!
)* and its 'Live Radio' facility, to live-stream my vinyl through Roon.
So I have a digital-based system ... used for playing analogue music.
gingermrkettle
Deep vein trombonist
irb
pfm Member
You could indeed! Lots of people do say such things. But are they right?
Why should line level crossing be more predictable?
gingermrkettle
Deep vein trombonist
Because you are working into an impedance which does not have significantly varying impedance by frequency when compared to the reactive components?
You could indeed! Lots of people do say such things. But are they right?
Why should line level crossing be more predictable?
I think you're being disingenuous.
As I see it ... the main advantage of a line level XO is that the amp channels are then directly connected to their drivers. No passive components "in the way" to (negatively) affect the signal.
cooky1257
pfm Member
ATC make both.
atc.audio
Voicecoil temp will change a passive network's transfer function.
Active Amplification | ATC Loudspeakers
atc.audio
Last edited:
James
Lord of the Erg\o/s
Until DC output from the directly-connected amp blows the tweeter.
irb
pfm Member
Yes, the impedance you're working into is much less varied. But if you model and measure, you can take account of the varying impedance of loudspeaker drivers when you design a passive crossover. The result will be absolutely predictable, surely.
irb
pfm Member
I don't think it's fair to say that passive crossover components affect the signal 'negatively', but line level components don't. Yes there are some negatives, as there always are with any added component/circuitry, but that's true of line level too.
The principal effect of passive crossover components is positive - tailoring the frequency and phase of the drivers to the desired response. There will be some series resistance, but the idea that this is a huge drawback is rather overblown, I think. Active line level crossovers need reactive components too (capacitors), and they also need active devices, usually opamps. They have their own negatives.
@andyr Please don't accuse me of being disingenuous i.e. dishonest. I may have less than perfect knowledge, but my posts say what I think. I'm trying to contribute to a genuine discussion. Can't we do it without personal attacks? I'm getting really p***ed off with this stuff on PFM these days.
irb
pfm Member
Slightly off topic - I happened, quite by accident, to get a 5 minute demo of some speakers a few days ago, in a shop. Very decent, they sounded. It turned out they were Magico M7s. Yes, £450,000 worth. They use passive crossovers. Maybe someone should tell Magico about the overwhelming benefits of active crossovers. Because clearly, by using passives, they're being disingenuous.