I think, at a high level, you can do a lot of manipulation with passives to achieve amazing results. But DSP offers greater convenience and immediate gratification. And the potential for FIR stuff gets crazy in a way I could never touch with (my current knowledge of) passive designs. In my case, the components I chose (and I chose them for this very reason) play extremely well together. So a very respectable passive solution should be pretty straightforward. I think one of things that makes these forums so amazing are the differing philosophies and experiences that people bring to the table. It's a veritable treasure trove of knowledge and experience. I remember the days before the internet. Conversations at this level just weren't possible -- at least not for most of us non-research/engineering community types. So I really appreciate all the responses I've gotten so far. Thank you, everyone!
Dave and fedde, I'm definitely interested in the possibility of PC-based DSP solutions -- particularly since my entire music library consists solely of FLAC. I'll check out the OktoDAC 8. Any particularly strong recommendations for PC DSP? Right now I'm just using PC USB out to a Pro-Ject Pre Box S2 DAC. I'm guessing PC-based DSP would require a multi-channel sound card? And I would go analog out to the amps? Is that a desirable home-brew solution? Are there be problems with output quality?
I believe there are 2 cases where passive components come into their own.
1. Tweeter protection. I always use a series capacitor in my active systems, just incase some DC or low frequency energy goes into the tweeter. This is not a sound quality matter, purely protection
2. Series notch filters for driver breakup. I can't recall the specifics, but I recall reading somewhere that notching the breakup via passive means has a damping effect not achievable via line level DSP (i.e. active)., that can be seen by reduced harmonic distortion. This somewhat negates my claim above (that either approach would sound the same with the exact same electrical transfer function). Sorry I can't find the reference / measurements. This of course opens up the "how much distortion is audible?" debate, but a point worth considering.
1. Tweeter protection. I always use a series capacitor in my active systems, just incase some DC or low frequency energy goes into the tweeter. This is not a sound quality matter, purely protection
2. Series notch filters for driver breakup. I can't recall the specifics, but I recall reading somewhere that notching the breakup via passive means has a damping effect not achievable via line level DSP (i.e. active)., that can be seen by reduced harmonic distortion. This somewhat negates my claim above (that either approach would sound the same with the exact same electrical transfer function). Sorry I can't find the reference / measurements. This of course opens up the "how much distortion is audible?" debate, but a point worth considering.
Definitely check out CamillaDSP (on this site - look at the link in the PC Based forum:
https://www.diyaudio.com/community/...ne-for-crossovers-room-correction-etc.349818/
Here's my current audio chain if it helps:
(Local media on SSD -> MPD) OR (Sony CD Player -> Toslink -> USB 5.1 Soundcard -> USB) -> DellOptiplex 3050 (Archlinux) -> CamillaDSP (switchable sources) -> Topping DM7 -> various amplifiers -> DIY 3 way speakers
I was tossing up between the DM7 and Okta and went DM7 due to price and I only need USB input. Okta certainly has more flexibility, but I've been happy with the DM7 in terms of the basic features I need and sound quality.
PC options are easily:
1. HDMI out (e.g. to an AV receiver)
2. USB (to a DAC - like the topping DM7)
3. Mutli-channel sound card as you note
Don't be tempted to "gang up" multiple 2 channel devices. these won't be clock sync'd unless the manufacturer has a way of doing that.
I'm glad you chimed in with that, Dave! Because I was originally planning to break the passive crossover into 2 parts. Part 1 inside the speaker cabinet (i.e. permanent): zobel and notch filters for impedance/breakup. And Part 2 outside the cabinet: the LR2 filters. But then I started second guessing myself, and was considering making it all external so that when I changed over to DSP/active, the notch/zobel networks would not be used. So your response was very helpful.
And thanks for giving me a direction to move in with the PC DSP! I like USB out solutions. They strike me (intuitively) as better/simpler. Though I have absolutely knowledge-base to back that up that opinion! 🤣
Edited to add the PC DSP remark at the end.
Last edited:
Some great anecdotes in this thread, but what strikes me is the expense and kerfuffle of going active with DSP for what? At the end of the day; a Discovery woofer, a vintage Vifa mid and similar level tweeter; all budget drivers with what looks like not-problematic responses in the first place. I don't see the point of getting e.g. 2 x FA123s for such a system, and then the beloved SET amp becomes uselesss? Seems dubious.
Not doubting the benefits of active, nor the convenience of DSP when tuning the system, but I don't see much value for this intended project when you could swap the Vifa for a Bliesma middome or 12MU Illuminator mid instead, swap that 8' Disco for the 10", and still come in less than the DSP + active gear.
Not doubting the benefits of active, nor the convenience of DSP when tuning the system, but I don't see much value for this intended project when you could swap the Vifa for a Bliesma middome or 12MU Illuminator mid instead, swap that 8' Disco for the 10", and still come in less than the DSP + active gear.
Last edited:
Active crossovers don't have to be expensive. An old PC, an old AVR, a £20 8 channel sound card and bunch of wires will get you going for little money. A dedicated Raspberry Pi plus parts would be a step up. Top of the range AVRs from 20 years ago can be pretty cheap because the lack of current connections makes them undesirable as AVRs. There are other approaches given a bit of thought and DIY since digital electronics is cheap and widely used these days with audio processing requiring only a modest amounts of resources compared to video and general computing.
MiniDSP, Hypex and friends are convenient, partially proprietary and expensive. They are certainly an option to consider if not particularly computer savvy/interested and intending to keep and use the hardware for many future speaker projects. Spreading the cost over many speaker projects is another advantage over passive crossovers.
MiniDSP, Hypex and friends are convenient, partially proprietary and expensive. They are certainly an option to consider if not particularly computer savvy/interested and intending to keep and use the hardware for many future speaker projects. Spreading the cost over many speaker projects is another advantage over passive crossovers.
I've thought a few times of going active, but I also have a SET DHT amp I love and it has comfortably seen off all good quality solid state and Class D amps that I've thrown at it. The SET is optimised and quite big, in 4 boxes. What stops me going active is that I would need at least 2 of my SET amps, and that's just too big in size to contemplate. I'm not willing to go Class D or use plate amps after 15 years of optimising my SET amp builds. So in my case I feel unable to go active because in my experience the amp is at least as important as the speakers, if not more important.
Hard to think of something more ugly, and will sound about the same as the horrible early MiniDSP units from 10-15 years ago? Nah.
Until we get transparent turnkey ADC/DAC and multichannel amp boxes (knock a zero off the current MiniDSP, Hypex, Danville, DEQX stuf), I dunno, save the money or spend it on room treatments and better measuring equipment, or dual active subs.
For little outlay it will provide full control and flexibility and enable wholly evidence based development and upgrades. The lack of audiophile magic and prestige parts may not be everyone's idea of a fun hobby but should work for some. Possibly not the OP though given the value he places in his SET amplifier.
All audio systems have a weakest link. Finding and improving it for little financial cost is a fun part of the hobby for some. I get more satisfaction from assembling and fiddling with cheap electronic parts to put together something audibly neutral in use compared to buying expensive quality parts. Others prefer the higher cost of quality prestige parts (and often the lower cost in messing about).
Another advantage with DSP solutions is instantaneous crossover blind AB testing. I use a script when choosing between two "voicings" that randomizes the A and B crossovers, then alternates them over a repeating section of music (or continuous play). With a PC there is a lot you can do. Including dynamic EQ (that is loading an EQ for certain awfully mixed albums). I had this running on my Raspberry Pi system then adapted to switch CamillaDSP crossover configs.
A pair of FA123s will less than 1000 euro. For that money you get 6 good amps, 6 DACs, 2 DSP, 2 ADC and supplies. Only need to add a bit of MBF and a remote control kit. I do not think this expensive, but yes cheaper (and much more expensive) options exist...
Doing high quality passive crossovers will be also expensive, especially with big coils and higher end caps. An alternative could be an active crossover with passive (low voltage) parts and buffers by the way.
I'm questioning only the current cost of entry for quality, convenience and without a ton of ugly boxes. So let's say FA123 is the baseline, which it kind of is. In USA, 2 x FA123 + remote control accessories is more like 1200 USD without tax. System budget as stated here is 1700 USD without tax, for which you can swap the Vifa P17s for 12MUs for 482 USD, and so on.
The Purifi guys are on the case:
https://purifi-audio.com/blog/app-notes-2/low-distortion-filter-for-ptt6-5x04-naa-11
Thanks for the Purifi link, DSP_Geek! That was probably one of the most succinct reads on the subject that I've ever seen.
Quality can be a tricky subject if one moves away from objective requirements like sufficient to be audibly neutral in use. Not to say less objective quantities aren't valid but it will tend to mean the assessment of quality is going to vary from person to person.
I would disagree with the Hypex FA123 being a baseline for DIY folk. A handful of biquads would mean additional DSP hardware to move beyond mimicking passive crossovers and using the strengths of signal processing to control beamwidths, pull linear distortion below audibility levels, switchable directivity, room control,... and no doubt plenty of other things that DSP enables. The packaging is for one-off use with a particular speaker rather than for a sequence of speaker projects. Might be addressable with alternative Hypex modules (if one wishes to stick with the brand) and/or some metalwork DIY. 100W is likely to be insufficient for the woofers in most high quality 3 way speakers. Hypex has higher power options but they are expensive given what is required for a woofer amplifier.
I am not intending to knock Hypex which designs and manufactures good quality prestige products and sells them at reasonable prestige product prices. I am simply pointing out that this is not what speaker DIY is about for many hobbyists and it doesn't represent the entry price for using DSP with good quality speakers if one "DIYs" the signal processing side as well as the speaker side.
One of the biggest benefits of DSP is what can be done below 150Hz. My current system is flat in-room down to about 10Hz. If you haven't heard a system where the bass is actually flat down to sub-20Hz, you really don't know what you're missing. Everything else sounds broken in comparison.
Chris
Hi tktran. You can find my posts in the "How to build spherical speakers" https://www.diyaudio.com/community/threads/how-to-build-a-spherical-speaker.374278/page-23 I have pictures of my first version there 6" diameter with 20 drivers. Seeking to reduce the size the new version 4" diameter. I will have to create a new thread to document the new design. The DSP is very helpful to correct response in the high frequencies. The directivity plot is amazing. I need to work on improving the range above 4 kHz.
IMO, DSP has huge limitation in terms of tinkering with amps, preamps and dac's.
I have no doubt that it really works well when done properly, but one will loose flexibility in system other than loudspeakers.
Compromises, again...
I have no doubt that it really works well when done properly, but one will loose flexibility in system other than loudspeakers.
Compromises, again...
The series notch filter is raising the drive impedance. It has been demonstrated that high output impedance amplifiers can reduce the distortion. Sound magnitude is proportional to current, so speaker motors with variable impedance with motion produce distortion. The transconductance amplifier works very well with DSP to produce this effect for all drivers in the system. I built some LM3886 based transconductance amps with about 5 kOhm output impedance to run with my DSP crossover. If you want a very clear understanding of how speaker drivers actually work contrary to what magazines and sales people say, read this book. https://www.current-drive.info/