Well, I heard a difference with different digital filters crossing at the same frequency.
And normally I'm quite insensitive to those (i.e. on top of a DAC).
Let's try to sum up pros and cons of both approaches (i'd like to split active into digital and analog-line level though, electronic for short)
*passive XO
+requires a single amp
+makes the speakers easily portable
-expensive if using boutique parts
-time-consuming and harder to implement it right
*active digital XO
+more power-efficient
+separates amplifiers to a dedicated narrower band
-requires multiple amplifiers ($)
+allows for phase linear filters (or whatever is your preference)
+allows for endless adjustment (free)
+allows for higher orders without degrading sound
-introduces jitter
-harder to setup (at least first time
)-in some cases requires a computer source
-requires an A/D if using analogue source
-requires multiple DACs ($)
*active electronic XO
+more power-efficient
+separates amplifiers to a dedicated narrower band
-requires multiple amplifiers ($)
-usually not adjustable and no DIY electronic XO exist easy to make
-harder to do time align
-does not allow for slopes steep as digital XO
Let's see if i missed something
Last edited:
My analogue electronic crossover allows for 50 different xover points between 50Hz and 20kHz. Plenty enough for me and I get the choice of quality of components to achieve those. I'm fine with 24dB slopes btw. All I need to adjust them is a soldering iron and a handful of components, can't think of many diy projects simpler than that and it's easier than wiring a passive from scratch.
I bought my amps s/h which worked out cheaper than passive using boutiquey components.
I bought my amps s/h which worked out cheaper than passive using boutiquey components.
You forgot one of the most important pluses...
In a passive system with a well implemented series XO there is automatic dynamic compensation, Something no active XO can do (well at least without some rube goldberg feedback mechanism)
Unless implemented as a PLLXO, takes a whole lot of work to make transparent... a digital unit has at least one pre-amp stage, and an analog one is like multiple cascaded pre-amps.
dave
sounds interesting
I have never been attracted to series xo
is there a way to determine when the dynamic auto compensating actually happens ?
Andrew, this is very interesting. Can you expand on how compression drivers bring more distortion than regular direct radiators? If that's what you're saying.
IME, CDs sound much cleaner in the mids compared to cones in the home environment.
Horn-loaded woofers also sound like they have more headroom and sound much cleaner than when they are conventionally mounted.
I like symphonic music too. And regular radiators just don't do the dynamics right, i.e., it doesn't sound real.
ahh, you mean when it sounds good
Not the thot i had. All but the most compressed music has dynamics in it that start shifting the drivers paraeters around.
dave
It takes gobs of unclipped amplifier output to get a symphony out of a direct radiator. And is takes well implemented multi-way speaker.
I've no idea what your listening to, or how much time with live, unplugged symphonies.
Lot of good acoustics info for the searching on compression drivers and horns.
Mechanically air is driven to non-linear state, causing system to create harmonic distortion, and intermodulation distortion. With pure tone, 2nd harmonic creates difference tone with fundamental that is exactly at the fundamental frequency. Summation occurs and sound is louder at fundamental, plus 2nd harmonic also added on top. Now 2nd real tone, say a note or two higher/lower than 2nd harmonic of first tone is added, difference tone of 1st and 2nd fundamental are not masked.
Waveguides and horns that avoid high compression tend to lower distortion.
Sealed subs of small volume to high driver displacement have higher 2nd harmonic distortion than higher volume enclosure. Much of perceived bass from this sort is distortion.
Here is example from High Frequency Compression Driver Evaluation #7
of "82dsHi" which is response recording of B&C DE82TN driver driven with dual sine waves of 2093Hz & 3729Hz:
Optimally, only peaks would be 2093Hz and 3729Hz. Instead, each of these has 2nd harmonics as seen at 4186Hz (-34dB, 2.0%) and 7458Hz (-30.4dB, 3.0%). IMD components: 1636Hz (-40dB, 1.0%) 5885Hz (-25dB, 5.6%). When two pure tones are played on two separate drivers side by side no unmasked IMD exists and it sounds very clean. Likewise, playing single sine on one driver, and it's 2nd harmonic on a second driver at 1.0% is difficult to tell apart for most, but when 2nd harmonic get's to be much more than 2%, many can pick it right out. 2nd harmonic at 5-6% compared to <1% is night and day for most people. IMD at 5-6% sticks out like a quiet accompaniment that is completely out of tune. IMD difference component at 1% is often easy to pick up; the closer IMD components are to fundamentals, typically results in rougher sound for given magnitude of IMD component.
For comparison, cone driver I tested with results in #794http://www.diyaudio.com/forums/multi-way/221643-active-vrs-passive-80.html#post3243395 has all harmonic and intermodulation components <1.6% at similar SPL, but using gobs of power:
Regards,
Andrew
ah, the coherency issue, right
yes, exstremely important
Im lucky not to have any problems with that
Hard to do AB if you have to solder/desolder even one part to change a frequency.
I'm looking to do something along the lines of Pass XVR1/B4 or Bryston 10B.
Strangely there isnt any diy projects of such sort. There is plenty with fixed frequencies.
Sorry, but I dont understand what is "automatic dynamic compensation"?
(yes, i'm a noob)
I should have distingued digital from source and digital using A/D, I think that digital filtering when done from a digital source with a single D/A conversion after it is more transparent than any filtering done in analogue (including line-level ofc). There are other problems, though such as the quality of the DAC, how to implement attenuation and the ringing of the filters.
I think like Pano, a Passive XO if done really well and with good driver units can make u not desire anything better.
Yes because the digital filtering is essentially lossless, but the following DACs aren't perfect by any stretch of the imagination. All digital processing generates wordlength expansion so then dither is needed upon truncation. Given that a decent DAC (i.e. a multibit one) benefits greatly from an analog post-filter then it becomes an interesting question as to how much digital processing to do.
Yep, that's true enough. But I'm ever curious - if active done really well can do even better still? All the actives I've come across, including my own designs to date, suck in some significant way
Actually I prefer it without analog post-filter. A/B that was ears-opening.
Only literature I have found is from DAD about mastering with DXD. And i strongly believe that the future of digital audio is with very high sample rate, either native (possibly) or integer upsampled. But i'm afraid that they'll push DSD as much as they can.
But I'm ever curious - if active done really well can do even better still? All the actives I've come across, including my own designs to date, suck in some significant way
I think it can, the engineering side is undoubtedly at its favour, but it won't be cheaper nor easier to do than a SOTA passive XO.
I dont think my last active XO sucked, but the speakers had different (=not related to the XO) issues due to room and HF drivers limitations, which I hope to overcome in the next ones (semi-active, i.e. woofer-mid section with line-level filtering for cost and efficiency reasons).
Like John K said in its Note2 news, midrange drivers benefits from a steep highpass (3rd-4th order usually enough, but it wasn't enough for the fullrangers i was using). YMMV.
Was the DAC multibit? Was the filter passive or active?
I'm skeptical of very high sample rates, for the same reasons Dan Lavry is. 2X OS of redbook is far enough for me
As you play music the drivers change in a dynamic manner, a series XO can compensate for those changes.
dave
....has a degree of self-compensation ...sound more 'moderate', and right
but Dave also this time said a well adjusted...., which then would be true for any system
maybe you can achieve acceptable result with 1.order series xo because of this
but I know nothing about it
I will have to give it try
I do not find sweries xo very interesting, in general
but if a 1.order series xo actually works......
Multibit ofc, but the filter F3 (1st order) was too low, I think if that such filter is cornered at 200 or 300khz won't be as harmful (i.e. the classic veil on the sound), but I couldnt experiment that. I can do the same thing at amplifier input*, but I'm not ready to develop the amp yet.
(* only ONE HF filter is required in your chain, be it at DAC output or better at amp input, and I'm assuming no preamp with volume control done in a smart way either digital or within the amplifier like Ayre or Bakoon or ASR)
m skeptical of very high sample rates, for the same reasons Dan Lavry is. 2X OS of redbook is far enough for me
Mind you, I dont agree with Lavry's papers. They work only for sigma-delta DACs which he makes.
Use also your ears to judge. I mean this, as some wise man said, "if it mesures bad, it's bad; if it measures good and sounds bad, it's bad; if it measures good and sounds good, then it's good".
I'll see if i can make it work, but I have some doubts
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.