It was suggested to me today that comparing analogue v digital XOs is like analogue v digital sources, ie analogue sounds smoother . .
As possible context, this guy has a long broadcasting studio background, who has made or played around with systems including Peerless, the 2nd top Scan tweeters, Aurum Cantus; Bryston, etc – he only uses audiophile drivers in hos home.
He implemented his home three way with a passive two way (tweeter & mids), then splits the bass to the sub off with a digital XO. This is even though his digital XO (a dbx driverack) is capable of doing all three. He claims that the (2nd top Scan) tweeter sound smoother that way.
Anyone think that a digital XO can add harshness?
Thanks
As possible context, this guy has a long broadcasting studio background, who has made or played around with systems including Peerless, the 2nd top Scan tweeters, Aurum Cantus; Bryston, etc – he only uses audiophile drivers in hos home.
He implemented his home three way with a passive two way (tweeter & mids), then splits the bass to the sub off with a digital XO. This is even though his digital XO (a dbx driverack) is capable of doing all three. He claims that the (2nd top Scan) tweeter sound smoother that way.
Anyone think that a digital XO can add harshness?
Thanks
Well no, not really.
Properly done, digital crossovers are essentially distortion free and have exceptionally low noise. Cascading many digital processes is possible in situations where cascading the same set of analogue processes would result in too much noise and/or distortion.
Digital crossovers are also much more flexible and can do things not possible with analogue circuits. Examples of that are pure time delay and linear-phase filtering.
Badly-implemented digital processing (common in cheap equipment) can have nasty distortion signatures and can sound very bad. But of course the same is true of analogue.
Both types are capable of excellent performance. It just depends on what you want to do.
I agree There are lot's of models on the market. If for example you need time alignment , and parametric equalizing . You need
a digital XO. DPS systems are used in cinemas with RTA evrywhere.
If it's for a two way home system a Rane with time alignement analogue can do the trick.
But what I want to Warn you guys against is the cheap ones.
The converter is one of the most important component of these units.
So if you go with Behringer take their top of the line at 300$.
As for DBX ,Klark Teknik ,TC ELECTRONICS AND SYMITRIX . These
are more dependable.
I could go on and on . There are so many different applications.
Cinema, Touring, Disco. Home intertainment.
So to conclude I'll give you in a nut shell the prices range : 20,000$ for a Dolby Cinema Theater Processor to 150$ for a Behringer.
...........................................................
a digital XO. DPS systems are used in cinemas with RTA evrywhere.
If it's for a two way home system a Rane with time alignement analogue can do the trick.
But what I want to Warn you guys against is the cheap ones.
The converter is one of the most important component of these units.
So if you go with Behringer take their top of the line at 300$.
As for DBX ,Klark Teknik ,TC ELECTRONICS AND SYMITRIX . These
are more dependable.
I could go on and on . There are so many different applications.
Cinema, Touring, Disco. Home intertainment.
So to conclude I'll give you in a nut shell the prices range : 20,000$ for a Dolby Cinema Theater Processor to 150$ for a Behringer.
...........................................................
It's not true, anyway.
Properly-implemented digital systems are essentially THD-free. The only error they produce is white noise (or shaped, if you're that way inclined. Either way it's at a very low level).
No real-world active analogue system is capable of this - there are always distortion products.
Properly-implemented digital systems are essentially THD-free. The only error they produce is white noise (or shaped, if you're that way inclined. Either way it's at a very low level).
No real-world active analogue system is capable of this - there are always distortion products.
Wingfeather said:
You do realize that digital crossovers have an analogue output stage right after it, right? So it still has to go through a bunch of opamps, which IMHO colors the sound. Whereas opamps in analogue filters are transparent because they are playing a different role. I don't have measurements to prove it, but this is what listening tests tell me.
Well of course. But your analogue system will have already gone through a DAC so I don't see how that's a disadvantage. The analogue system will have the analogue DAC output stage followed by another stage of analogue processing, which would surely compound the problems.
The opamps in an analogue crossover filter are playing much the same role as those in the reconstruction filter in a DAC (i.e. filtering!), although I concede that the frequencies are much lower - which always helps.
I suppose you could argue that an all-analogue (perhaps vinyl) system would get around this. I would personally suggest that in such a system the colouration caused by a precision opamp halfway down the chain will be the least of your worries.
If listening tests are your only criteria for the statement that analogue is better, you have to beware that there are a lot of badly-executed digital systems around - and these are most definitely not representative of what a digital system is capable of.
The opamps in an analogue crossover filter are playing much the same role as those in the reconstruction filter in a DAC (i.e. filtering!), although I concede that the frequencies are much lower - which always helps.
I suppose you could argue that an all-analogue (perhaps vinyl) system would get around this. I would personally suggest that in such a system the colouration caused by a precision opamp halfway down the chain will be the least of your worries.
If listening tests are your only criteria for the statement that analogue is better, you have to beware that there are a lot of badly-executed digital systems around - and these are most definitely not representative of what a digital system is capable of.
Convergence
I’m talking here for home systems. Even within Behringer there are differences – are you referring to the 8024 and the 2496 (top of the line or better)?
Jeff
I’m a way off measurements, and was more after listening experiences, which followed someone else's listening experience, hence my initial question
“The main difference would be with latency. The digital processor has to convert the analogue signal into digital and back. We're talking if mili seconds.”
not if you take digital out from a CDP, XO, then into analogue.
m0tion
While the two approaches are very different, with flexibility implications etc, they are choices that it needs to be known if there are SQ losses/ benefits either way.
cotdt your listening favoured analogue – what were the Xos you compared?
Thanks
I’m talking here for home systems. Even within Behringer there are differences – are you referring to the 8024 and the 2496 (top of the line or better)?
Jeff
I’m a way off measurements, and was more after listening experiences, which followed someone else's listening experience, hence my initial question
“The main difference would be with latency. The digital processor has to convert the analogue signal into digital and back. We're talking if mili seconds.”
not if you take digital out from a CDP, XO, then into analogue.
m0tion
While the two approaches are very different, with flexibility implications etc, they are choices that it needs to be known if there are SQ losses/ benefits either way.
cotdt your listening favoured analogue – what were the Xos you compared?
Thanks
If you're starting with an analog signal, I can see the argument for keeping it analog all the way. If you start with a digital signal, and process it digitally, then DAC-Amp-Driver, makes a strong case for digital with proper processing. I could never give up my PC based system. The convenience and flexibility tips the scales for me by a wide margin.
Wingfeather said:
Properly implemented, yes. I spent much of last year chasing down distortion residuals in a new design for $DAYJOB. It worked: the thing has the best numbers of our entire product range.
Here's a small secret for DSPers out there: 32 bit floating point is useful but it's not a panacea. It turns out single precision floats have better noise performance than 24 bit fixed point, but extreme IIRs could still have small distortion problems because the mantissa itself is only 24 bits, so you run into some finite word length issues in feedback filters with large loop gains. Filters lower than 0.005 of the sampling frequency [1] are best done in double precision as a result if you want really good performance.
[1] For example, 48 kHz * 0.005 = 240 Hz.
Cheers,
Francois.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.