Hi folks!
I’m at a fork in my next 3-way build, and I’m hoping for some feedback. I’ve already chosen drivers that will play very nicely with each other in terms of crossover design and I’m working on my cabinets. While I was mulling things over late last night, I got to thinking… “What if I went with a DSP instead of passive crossovers?” What makes the question hard -- for me -- is that I have a very nice SET tube amp I absolutely love. The amp wasn’t cheap and I’m not in a position to buy 2 more in order to accommodate the additional 4 channels of an active DSP configuration. Right now, my budget for crossover components or DSP (to include DSP and two additional amplifiers) is ~$1000.
This brings me to my immediate question: do the advantages of DSP with cheap DSP/amplifiers outweigh the advantages of a single, very high quality amplifier in a traditional, passive system with very high quality crossover components?
I know there are some strong opinions about passive vs. active/DSP and it isn’t my intention to open that can of worms in the abstract. I’m inquiring specifically for my build constraints and resources, which is the path of greater acoustic efficacy?
If the consensus is “go with good passive crossovers, for what you have now” then that’s easy. If folks think DSP would be better… well, that’s where my knowledge base tapers off. I’ve built a couple DSP/active systems in the past, but they were not what I would consider serious audiophile endeavors (sorry, I know that’s a loaded term): stock GUI/software (e.g. SigmaStudio or miniDSP stock interface) and “decent” class D amps. So... in the case of DSP/active, I have myriad questions (but I’ll start with just a couple):
1. For a high quality build on a limited budget, what DSP? A miniDSP Flex Eight? Or something else?
2. For programming the DSP, should I use the basic GUI/software or tackle more complex FIR stuff? Or start with one and build up to the latter? Oh, and full disclosure, I don’t know squat about building/implementing FIR filters -- but I’m willing to learn!
Thanks in advance for the feedback and for sharing your knowledge. 🙂
I’m at a fork in my next 3-way build, and I’m hoping for some feedback. I’ve already chosen drivers that will play very nicely with each other in terms of crossover design and I’m working on my cabinets. While I was mulling things over late last night, I got to thinking… “What if I went with a DSP instead of passive crossovers?” What makes the question hard -- for me -- is that I have a very nice SET tube amp I absolutely love. The amp wasn’t cheap and I’m not in a position to buy 2 more in order to accommodate the additional 4 channels of an active DSP configuration. Right now, my budget for crossover components or DSP (to include DSP and two additional amplifiers) is ~$1000.
This brings me to my immediate question: do the advantages of DSP with cheap DSP/amplifiers outweigh the advantages of a single, very high quality amplifier in a traditional, passive system with very high quality crossover components?
I know there are some strong opinions about passive vs. active/DSP and it isn’t my intention to open that can of worms in the abstract. I’m inquiring specifically for my build constraints and resources, which is the path of greater acoustic efficacy?
If the consensus is “go with good passive crossovers, for what you have now” then that’s easy. If folks think DSP would be better… well, that’s where my knowledge base tapers off. I’ve built a couple DSP/active systems in the past, but they were not what I would consider serious audiophile endeavors (sorry, I know that’s a loaded term): stock GUI/software (e.g. SigmaStudio or miniDSP stock interface) and “decent” class D amps. So... in the case of DSP/active, I have myriad questions (but I’ll start with just a couple):
1. For a high quality build on a limited budget, what DSP? A miniDSP Flex Eight? Or something else?
2. For programming the DSP, should I use the basic GUI/software or tackle more complex FIR stuff? Or start with one and build up to the latter? Oh, and full disclosure, I don’t know squat about building/implementing FIR filters -- but I’m willing to learn!
Thanks in advance for the feedback and for sharing your knowledge. 🙂
Earlier this year I replaced the passive crossovers in a pair of Magnepan 3.7i speakers for use as playback monitors at a local recording studio with hypex plate amps with built in DSP. The improvement made possible by the flexibility of the DSP was huge. A pair of these is a bit more than $1000, but well worth it. The amplifiers are great. https://www.madisoundspeakerstore.com/speaker-amps/hypex-fusionamp-fa123/ The plate amp is great because you don't end up with a mess of boxes and wires. One can be slaved to the other and fed with a single coax SPDIF cable. These amps have every input you could want.
The miniDSP products have the advantage of offering the DIRAC room correction filtering (FIR filters) for just $200. The Hypex now has FIR filters available , but you have to come up with the filters.
FIR filters work well, but you will need some dedicated software or some advanced math skills and familiarity with FFT and inverse FFT to roll your own. Start out with a measurement microphone, some free software like Arta or REW ( Room EQ Wizard ) and roll some IIR filters (PEQ, Highpass, Lowpass, shelf filters) to get started.
The ability to easily implement and change crossover parameters and easily boost or cut small frequency ranges with the DSP make it much better than any passive crossover. I can easily try three different crossovers with the DSP and switch between them with a remote. You can compare 1st order, 4th order and 8th order crossover slopes or different crossover frequencies with a push of the button. Use a Linkwitz transform ( Asymmetric second order shelf filter) to extend the bass response .
The miniDSP products have the advantage of offering the DIRAC room correction filtering (FIR filters) for just $200. The Hypex now has FIR filters available , but you have to come up with the filters.
FIR filters work well, but you will need some dedicated software or some advanced math skills and familiarity with FFT and inverse FFT to roll your own. Start out with a measurement microphone, some free software like Arta or REW ( Room EQ Wizard ) and roll some IIR filters (PEQ, Highpass, Lowpass, shelf filters) to get started.
The ability to easily implement and change crossover parameters and easily boost or cut small frequency ranges with the DSP make it much better than any passive crossover. I can easily try three different crossovers with the DSP and switch between them with a remote. You can compare 1st order, 4th order and 8th order crossover slopes or different crossover frequencies with a push of the button. Use a Linkwitz transform ( Asymmetric second order shelf filter) to extend the bass response .
Last edited:
Solution is possible.
A simple minidsp 2x4 for 150euro and a big fullrange driver will do the job.
To lower loudspeaker cone induced distortion you can try out this tweak.
You find here also modified Fane Sovereign fullrange drivers as an example:
https://www.diyaudio.com/community/threads/how-to-make-a-loudspeaker-sandwich-cone.402917/
A simple minidsp 2x4 for 150euro and a big fullrange driver will do the job.
To lower loudspeaker cone induced distortion you can try out this tweak.
You find here also modified Fane Sovereign fullrange drivers as an example:
https://www.diyaudio.com/community/threads/how-to-make-a-loudspeaker-sandwich-cone.402917/
Thanks, Olsond3! I've heard Magnepans! Lovely, lovely sounds!! And that does open up an interesting option: start with one then build the other later down the road, as time/money permits! Then I can change as necessary or desired. Though I don't think I want to go with plate amps. I prefer the isolation of separate amplifiers which, as you mentioned, is a more cumbersome solution. But these speakers will likely be long-term and won't get moved around too often. Sadly, I was unable to bring my previous speakers with me to Japan due to their 380 lbs (each) cabinet weight. LOL
Thanks, Freedom666! But, as a I stated, I've already got my drivers (and full-range drivers are not the best fit for my preferences). And since I'm building a 3-way, I need 6 channels. So probably a 2x8 DSP.
Thanks, Freedom666! But, as a I stated, I've already got my drivers (and full-range drivers are not the best fit for my preferences). And since I'm building a 3-way, I need 6 channels. So probably a 2x8 DSP.
Combination of a tube amp and high efficiency fullrange driver is ideal.
Only the drawback in distortion should be tamed by good construction like Faraday copper rings and maybe loudspeaker cone tweaks.
But I understand the desire for a three way design at the expense of the tube amp not very well.
I would go then for passive crossover.
Only the drawback in distortion should be tamed by good construction like Faraday copper rings and maybe loudspeaker cone tweaks.
But I understand the desire for a three way design at the expense of the tube amp not very well.
I would go then for passive crossover.
I built 4-way speakers using two miniDSP 2x4hd boxes, one at each speaker and used optical cables to go out to each speaker. It worked pretty well. The remote even works pretty good, but the volume can get out of sync if you happen to point it so one box doesn't get the signal. The 8 channel miniDSP looks great. It's hard to choose which on to buy as there are several options for inputs. The FLEX products are great as they have the display that shows the volume setting etc. I always end up hauling my systems to club meetings and other peoples houses, so I've been going for more integrated and light weight solutions. The class D amps are looking better and better.
Oh, and Xsim 3D or Viston Boxsim both support designing crossovers with digital filters.
Oh, and Xsim 3D or Viston Boxsim both support designing crossovers with digital filters.
Take this solution.
Make a three way passive crossover only taking care of distortion and dispersion of the drivers.
Then correct frequency and phase with single minidsp with fir correction.
With that solution you come close to a good one way design which is phase linear also with iir filters.
And keep your good tube amp
Make a three way passive crossover only taking care of distortion and dispersion of the drivers.
Then correct frequency and phase with single minidsp with fir correction.
With that solution you come close to a good one way design which is phase linear also with iir filters.
And keep your good tube amp
Innovative! Does using the two DSPs potentially introduce any clock/timing issues? Or are you able to sync them somehow? Or does it not appear to be an issue?
The club meets sound cool! Sadly, all my more serious builds in the past followed a sort of Dunlavy cabinet philosophy, in terms of constrained layer damping, sealed, and just... stupid heavy. But I'm old enough now to appreciate compromise (and not trying to push 150 lbs panels across a table saw)! So I'll probably keep this set around 150 lbs, or less. At least, that's my goal. I'm also branching out more into TQTLs. Especially now that Martin King has really cracked the modeling -- and Hornresp means I don't have to be a rocket scientist to do the maths! LOL
I agree the FLEX Eight has some nice features. And better specs than the 2x4HD.
Slightly off topic, I just finished a 2-way desktop TQTL using a KABD-430 (Wondom J4, I think). It's cool for a budget, mid-fi solution but for serious listening I'm definitely wanting better. And to my credit, I was able to keep my computer speakers under 40 lbs (each)! LOL
The club meets sound cool! Sadly, all my more serious builds in the past followed a sort of Dunlavy cabinet philosophy, in terms of constrained layer damping, sealed, and just... stupid heavy. But I'm old enough now to appreciate compromise (and not trying to push 150 lbs panels across a table saw)! So I'll probably keep this set around 150 lbs, or less. At least, that's my goal. I'm also branching out more into TQTLs. Especially now that Martin King has really cracked the modeling -- and Hornresp means I don't have to be a rocket scientist to do the maths! LOL
I agree the FLEX Eight has some nice features. And better specs than the 2x4HD.
Slightly off topic, I just finished a 2-way desktop TQTL using a KABD-430 (Wondom J4, I think). It's cool for a budget, mid-fi solution but for serious listening I'm definitely wanting better. And to my credit, I was able to keep my computer speakers under 40 lbs (each)! LOL
LOL... Freedom, I was just contemplating something similar! Manage impedance with passive components inside the cabinet, and make the passive filters, external. Set the cabinet up for tri-wiring and connect the external crossovers that way. Then later if I want to DSP, I can just remove the external crossover go that route. Sure, it's more expensive, but I could build it up over time and it would be fun to A/B the different configurations!!
You can put the plate amps in a box outside the speaker. Currently there is just no other competitive solution to these plate amps within you budget constraints, and they are marvelous.
Thanks, sheeple! Not a bad idea.
Olsond3, the Hypex-FA123 looks pretty interesting. Anyone know what DSP chipset they use? I imagine it's something more robust than an ADAU1701. Especially if FIR filters are available.
Olsond3, the Hypex-FA123 looks pretty interesting. Anyone know what DSP chipset they use? I imagine it's something more robust than an ADAU1701. Especially if FIR filters are available.
The question of DSP vs Passive really depends on your drivers + their 'marriage' to enclosure & room.
If even one driver is intrinsically problematic, and bass extension is paramount then DSP always wins.
However, if you have a good room & truly wonderful drivers (plus the amazing budget of $1000)
then Passive has some nice advantages like keeping your system simple, maintaining your beloved amp. arrangement
and also the wonderful challenge & satisfaction that comes with the creation of truly excellent crossover network.
It is worth mentioning that many people are now going with the HYBRID option 🙂
If even one driver is intrinsically problematic, and bass extension is paramount then DSP always wins.
However, if you have a good room & truly wonderful drivers (plus the amazing budget of $1000)
then Passive has some nice advantages like keeping your system simple, maintaining your beloved amp. arrangement
and also the wonderful challenge & satisfaction that comes with the creation of truly excellent crossover network.
It is worth mentioning that many people are now going with the HYBRID option 🙂
The two DSPs worked great. I bought a 1 In, 3 Out powered optical splitter for maybe $20 on Amazon and just ran the cables out to each box.
I think modular in my builds and target about 25 lbs for each piece to keep my back happy. The current project is a two way of sorts with a 4" diameter spherical mid/tweet and a cube with a 10" woofer. The miniDSP works great for the crossover and compensates for the variations in the sphere response in the top end.
The spheres can use three channels of DSP if I want variable dispersion, omni, dipole or cardioid patterns. So with the woofer that's four channels of DSP per side. There's probably a FLEX 8 in my future. Hypex doesn't have a four channel plate amp.
That makes sense. I originally selected the drivers with the passive crossovers in mind, so they seem pretty amenable to that: Morel CAT408 (of the three drivers, I am least familiar with the actual sound of Morel), ScanSpeak 17PWJ00, and ScanSpeak 22W/8534G00. Running them through a quick sim in VituixCAD, I get very simple LR2s without any obvious need for anything crazy. Of course, theoretical sims don't mean much and I'll have to measure the drivers, in-hand, once they arrive. The ScanSpeak drivers I've heard over the years sounded well-behaved, overall. I'll be putting the 17PWJ00 in a 28 litre sealed box. I considered some kind of OB configuration, but my listening environment is my weakest link and I can't do much about it (It's a solid concrete building in Okinawa). I'm still poking around with the figures for the TQTL for the 22W/8534G00 in Hornresp. But very quick 'n dirty, Hornresp modeling (which obviously still needs work) shows adequate low frequency extension -- at least for my needs/environment.
Attachments
Oslond3, I love them! You have spheres! That's spectacular! What's the material? How did you make them? Either way, super cool.
Here's the desktop computer speakers I finished last week (with KABD-430 and the little plexiglass box I made for it). Still tweaking FR with the DSP and I'm obviously not done with the cabinets (paint, and I'll be sticking a .25 inch felt sheet into the front baffle). No one will ever accuse me of making pretty speakers. 😆 The SPL graphs aren't obscene but still need dialing in. Can you tell I live in a concrete box? LOL
Here's the desktop computer speakers I finished last week (with KABD-430 and the little plexiglass box I made for it). Still tweaking FR with the DSP and I'm obviously not done with the cabinets (paint, and I'll be sticking a .25 inch felt sheet into the front baffle). No one will ever accuse me of making pretty speakers. 😆 The SPL graphs aren't obscene but still need dialing in. Can you tell I live in a concrete box? LOL
Attachments
DSP's are really great for dialling in crossovers, quick and easy adjustments without having to deal with impedance affecting the response / amplifier. I've been using them for years on my projects to save having to do passives.
I'm hoping on my next project to try active analogue instead, as I want to hear analogue 'all the way' from my record deck to speakers.
If all the drivers you have chosen have good smooth raw responses I would give passives a go.
If they need a lot delay, notches for break up modes, EQ shaping beyond baffle step correction etc I would stick to DSP or active analogue.
Rob.
I'm hoping on my next project to try active analogue instead, as I want to hear analogue 'all the way' from my record deck to speakers.
If all the drivers you have chosen have good smooth raw responses I would give passives a go.
If they need a lot delay, notches for break up modes, EQ shaping beyond baffle step correction etc I would stick to DSP or active analogue.
Rob.
Having accumulated several years of experience with DSP x-overs (e.g. DEQX PDC, miniDSP) for me there's no way back to passive x-over speakers. That said, there are important side conditions: We need good AD conversion (if needed) and good DA conversion! Older and epecially cheaper miniDSP products did not meet these conditions. My current solution uses openDRC and miniSHARC with digital I/O (and clocking modifications), combined with external ADC and DACs. This solution is probably not suitable for everyone, so my recommendation is the Flex series, e.g. Flex eight for 3-way (with additional, external ADC if needed). Yes, additional power amps are a cost factor, but for me the advantages of active speakers outweigh the cost.
Greetings,
Winfried
Greetings,
Winfried
I'm in the DSP camp.
Active / amp per channel, optimise the amp for each driver pair.
Benefits are control, get the best X/O by design and then actual listening real time, no L-Pads if your drivers are off differing efficiency.
Proper time / phase alignment is relatively easy.
If you have deep bass, you have room mode control option.
Good DSPs are transparent.
I use 1 big silicon amp on TH subs. Chip amp on mid bass and 3 EL84 SET amps on the upper 3 channels.
Plus analogue in will work really well too. Your turntable still sounds like a turntable - believe it or not!
Active / amp per channel, optimise the amp for each driver pair.
Benefits are control, get the best X/O by design and then actual listening real time, no L-Pads if your drivers are off differing efficiency.
Proper time / phase alignment is relatively easy.
If you have deep bass, you have room mode control option.
Good DSPs are transparent.
I use 1 big silicon amp on TH subs. Chip amp on mid bass and 3 EL84 SET amps on the upper 3 channels.
Plus analogue in will work really well too. Your turntable still sounds like a turntable - believe it or not!
I ended up with a hybrid approach: passive crossover, DSP correction. That way, I save on amp channels but still have DSP control of the system response, which is particularly useful at LF.
IMO, above a few hundred Hz, it's possible to get excellent results with passive components. As you move down towards the room's modal region, EQ is required for good sound. Sometimes the EQ curves are simple and could be done with analogue circuitry, but complex curves become difficult to implement in analogue and DSP becomes the only way to go.
Chris
IMO, above a few hundred Hz, it's possible to get excellent results with passive components. As you move down towards the room's modal region, EQ is required for good sound. Sometimes the EQ curves are simple and could be done with analogue circuitry, but complex curves become difficult to implement in analogue and DSP becomes the only way to go.
Chris