Thanks Krivium.
When I was comparing DSP / LMS a few years back I came across this, it concludes even within the DSP sector there are significant variations of "Reference Curves" and settings... This makes transferring / importing filters from brand to brand or system to system tricky.
https://soundforums.net/articles/measurement-and-optimization/a-tale-of-two-processors-190580/
When I was comparing DSP / LMS a few years back I came across this, it concludes even within the DSP sector there are significant variations of "Reference Curves" and settings... This makes transferring / importing filters from brand to brand or system to system tricky.
https://soundforums.net/articles/measurement-and-optimization/a-tale-of-two-processors-190580/
How yes, eq are all different in digital and analog! It change the way they respond to manipulation...
This is why you have people raving about SSL 4000 E eq ( 'brown eq') over 'J' revision ( 'pink eq' ... E are constant gain design, J constant Q...), or Ams/Neve or,...
Constant gain will sound 'fatter' as when you use them the base of the bell curve will enlarge/widen when you boost or cut, constant Q won't so they are more precise/accurate and if you want to have same fatness as constant gain you have to modify Q parameter...
And it doesn't take into account the electronic implementation either: parallel eq cell doesn't sound the same as serial ones... gyrators doesn't sound the same as inductors,... a lot of thing to explore!
In digital it's the same, results depend from implementation...
This is why you have people raving about SSL 4000 E eq ( 'brown eq') over 'J' revision ( 'pink eq' ... E are constant gain design, J constant Q...), or Ams/Neve or,...
Constant gain will sound 'fatter' as when you use them the base of the bell curve will enlarge/widen when you boost or cut, constant Q won't so they are more precise/accurate and if you want to have same fatness as constant gain you have to modify Q parameter...
And it doesn't take into account the electronic implementation either: parallel eq cell doesn't sound the same as serial ones... gyrators doesn't sound the same as inductors,... a lot of thing to explore!
In digital it's the same, results depend from implementation...
Hi krivium, I've seldom see any PA self-powered speaker manufacturer tout steep slope xovers (by steep I mean over 24dBoct)
Maybe sometimes between CD and next section down, but not very often at all.
I think the reason Is how dang important SPL is to PA, and steep xovers cost some max SPL due to lost summation capability thru xover region.
Every dB is precious to PA.
I find the benefit of steep xovers is that they get to use just the sweet bandwidth region...where mag and phase are naturally flat.
This of course requires sufficient multi-way count to take advantage of, and it does come at the expense of a small reduction is SPL.
I've come to realize steep linear phase xovers really are the domain of over zealous home audio nuts, not much suitable elsewhere 😀
There is a lot of truth in that... When I first got a DEQX I thought lets go, all my worries are over 96dB per octave slopes and high Q notch filters with no "side effects".... How wrong I was!
I never got the same musical realism and natural flow I was getting with a DBX PA2 with IIR 24dB slopes!
I ended up selling the DEQX and buying a used Danville Sounds DSP... 24 dB slopes with Audio weaver and some gentle Eq with Q below 2.8 and the odd shelf filter covers 90% of domestic music and movie needs.
I never got the same musical realism and natural flow I was getting with a DBX PA2 with IIR 24dB slopes!
I ended up selling the DEQX and buying a used Danville Sounds DSP... 24 dB slopes with Audio weaver and some gentle Eq with Q below 2.8 and the odd shelf filter covers 90% of domestic music and movie needs.
Hi Mark,
I hope you are fine.
Well i wasn't thinking about self powered system, more modular ones with 'fly over' box, low mid and sub, with dsp xover either standalone or incorporated within poweramp, etc,...
Maybe this is fading away? I don't know, the only live systems i've played with in the last decade are diy sound systems. I'm not really on the edge of what is used atm and this is of secondary interest to me.
But yes i agree except for high freq it is of low interest and might bring more trouble than solutions ( try to play any instrument with low mid/ low with a 25ms delay/ latency... so fun!).
It was a surprise with my Lake too: in 4 way Fir, you don't have choice of steepness: frequency range in which xover is located determine this parameter ( it's not a system wher you define a number of taps explicitly - you do this by allowing a latency setting but it's not said as such). Higher than 100db/octave is possible but only at high freq, the lower you get in freq, the less steep the filters becones...
And yes PA is all about spl. And ability to withstand this over long period of time hence limiters, compressors, feedback from the amp to monitor output and counteract some nasty's... complicated situations which need complex answers.
Home audio doesn't need all that imho. With careful planing and sizing of components it's definitely not needed.
I hope you are fine.
Well i wasn't thinking about self powered system, more modular ones with 'fly over' box, low mid and sub, with dsp xover either standalone or incorporated within poweramp, etc,...
Maybe this is fading away? I don't know, the only live systems i've played with in the last decade are diy sound systems. I'm not really on the edge of what is used atm and this is of secondary interest to me.
But yes i agree except for high freq it is of low interest and might bring more trouble than solutions ( try to play any instrument with low mid/ low with a 25ms delay/ latency... so fun!).
It was a surprise with my Lake too: in 4 way Fir, you don't have choice of steepness: frequency range in which xover is located determine this parameter ( it's not a system wher you define a number of taps explicitly - you do this by allowing a latency setting but it's not said as such). Higher than 100db/octave is possible but only at high freq, the lower you get in freq, the less steep the filters becones...
And yes PA is all about spl. And ability to withstand this over long period of time hence limiters, compressors, feedback from the amp to monitor output and counteract some nasty's... complicated situations which need complex answers.
Home audio doesn't need all that imho. With careful planing and sizing of components it's definitely not needed.
Last edited:
In IIR crossover filters, Only LR4 sums flat electronically. LR2 creates a peak and higher ones create dips. I wonder what they did when they were using LR8(48dB/oct).
Hi, i wouldn't give up on steep xovers (unless you were using IIR or course).
High Q- PEQ's & notches I won't touch period, but complementary linear phase xovers cause zero problems and are very easy to implement.
I think good implementation, whatever strategy we are using, is the key. I'm inclined to think achieving really nice mag and phase easily, consistently, is the sonic advantage, phase audibility or not.
fwiw, the DEQX despite touting an ability to handle steep linear phase xovers, was pretty anemic as far as FIR capability. I'd not base any strong opinions on it.
Oh, I saw elsewhere (if I remember correctly) you had a Linea ASC-48. That is one clean measuring DSP. It's LIR filters, which are essentially linear-phase LR24 db/oct, are what I used for live sound. Only about 12ms latency for a 100Hz xover which for most stages worked fine. Plus, it's the only linear-phase xover I've ever seen that can adjust xover frequency on the fly.
Oh I'm sorry jheoaustin, you must not know me on the forums...
I'm the audio nut I was talking about....because I incessantly extoll the virtue of steep linear phase xovers..
Hi again krivium, yes doing fine thanks 🙂. Hope you are too..
Yeah, I have no idea what goes on in big flown line arrays, especially ones like Martin's MLA, or EAW's Anya, with their incredible beam steering.
The only boxes with public processing info tend to be smaller speaker on a pole type boxes...those were what i was referring too.
That said, I do read Meyer's Bod McCarthy talk about using low slopes to milk all the SPL possible on their big boys. But Meyer is often in a world of its own.
I always used 24dB / octave with the DBX PA 2 as they sounded the best. From memory the PA 2 only had 48 dB slopes on Butterworth or maybe Bessel ... I cant remember.
True. Back in the good old days of analogia, 18-24dB/octave were typical PA crossover slopes. With DSP these days, anything goes. We don't lose as many tweeters (horn+CD) as we used to. 😉
All LR's sum to flat fluency magnitude response, with the odd multiples of 12dB/oct needing a polarity reversal.
Here's an example of a whacked out, never to be used, LR 84...just to show good summation (and the needed polarity reversal)
Attachments
Yea I bought a used (well used) Linea Research ASC 48 on Ebay.... It was great sounding, for about 6 months before it literally went up in smoke! I will never know if it was performing at its best before it went bang... I would love to borrow a new one and compare to my Danville Nexus (also used, but gently!) which has great A to D and DACS. It can implement 96dB slopes using its Dream Weaver software and its all very easy to use. I tend to use 24 or sometimes 48 dB FIR slopes and have got used to them I guess.
Hi Pano, do you know of any PA examples of steep slope CD high-passes currently being used?
The FIR time needed would be insignificant for live sound.
One thing I've seen with my stuff, when trying to get the CD to reach real low, a sharp knee steep makes can ironically increase excursion risk ...
Of course, no problem if not trying to dig too low.
At any rate, I figure better limiters have more to do with today's CD survival rates than any type xover.
Sure @jheoaustin :
You are right when you say that a transition frequency around 1kHz is a bad idea - I'm not a pro too, but I experimented that it is better to avoid the 500-1500Hz area...
No they are not locked at all - REALLY not at all ! I was at one of the last Live Audentity Concerts that Klaus Schulze performed in Lyon in February 1983 at La Bourse Du Travail to say this...
For EM, I often use a "multi-stereo" system - if I can say so - with my 375L 3-Ways driven by one amp (Accuphase E-560), plus my SMGb driven by another amp (MC-7100), and the subwoofer (with its integrated LPF/amp module) hooked to the SMGb as an infra-bass complement (<40Hz) :
Once balanced, the SMGb and the SW acts as an "Ambient Enhancer" to the 3-Ways 375L, providing a large sound in my small auditorium... For example with this :
Or that :
Among so many others !
On the picture above (taken from the listening position), my 4-ways project would take the place of my Heresy I HBR... Fortunately, the acoustics in my Tiny (dedicated) Auditorium shows no notable flaws and is fine for near-field listening (due to its size). No active filter, no DSP required...
T
You are right when you say that a transition frequency around 1kHz is a bad idea - I'm not a pro too, but I experimented that it is better to avoid the 500-1500Hz area...
No they are not locked at all - REALLY not at all ! I was at one of the last Live Audentity Concerts that Klaus Schulze performed in Lyon in February 1983 at La Bourse Du Travail to say this...
For EM, I often use a "multi-stereo" system - if I can say so - with my 375L 3-Ways driven by one amp (Accuphase E-560), plus my SMGb driven by another amp (MC-7100), and the subwoofer (with its integrated LPF/amp module) hooked to the SMGb as an infra-bass complement (<40Hz) :
Once balanced, the SMGb and the SW acts as an "Ambient Enhancer" to the 3-Ways 375L, providing a large sound in my small auditorium... For example with this :
Or that :
Among so many others !
On the picture above (taken from the listening position), my 4-ways project would take the place of my Heresy I HBR... Fortunately, the acoustics in my Tiny (dedicated) Auditorium shows no notable flaws and is fine for near-field listening (due to its size). No active filter, no DSP required...
T
Oh, I am sorry. I fully misunderstood it. I also saw your comment on FIRs. Will add mine when I get time.
Thank you for this info. I must have done something wrong when I simulated these in MATLAB...