Are digital x-overs flawed??

[terry j]

I wasn't sure whether to post this here or in the digital forum, we'll see.

Our little aussie hi fi forum just had a tornado pass through, and many locals are still reeling in shock. The weather people have yet to classify the phenomena, but for all intents and purposes could call it Hurricane Romy.

The day started off rather uneventfully, and the unsuspecting locals had little warning of the force of nature about to strike and innocently went about their daily business.

It was not too difficult however to pinpoint the catalyst for the amazing events about to unfold, and took the form of an innocent enough looking post, an excerpt follows


""""""there was a discussion about integrating multiple drivers in an open baffle design and the idea of using a digital crossover with steep slopes came up.

One of the posters was the infamous Romy the Cat, who maintains his own highly opinionated website.

Romy said some very negative things about digital crossovers in this particular thread.....

"....by nature digital can not filter. When you introduce a crossover slope in digital domain you are effectively loose volume with frequency. Dropping volume at digital domain is made by tossing away bits, so in d-croosivers you have full resolution (15 bit or 20 bit for instance) only at full passband and as volume goes down (in a crossover slope) your sound is cared by much lower resolution. At minus 30-40dB the d-crossovers sound like a telephone in 1920. This is a fundamentals basis and no modification can cure it, particularly the Behringer. In context of any more or less capable playback the nastiness of d-crossover are well auditable…""""""


a bit of rather desultory discussion followed, mainly as we didn't really know one way or the other.

Then Hurricane Romy hit us, and after a few 'interesting' comments it was over as soon as it began.

Anyway, all of us 'morons' on the forum were in the end left completely unenlightened by the GOD of audio, but we were left feeling very chastised for being complete idiots and either having cloth ears or completely unrevealing and substandard systems (or even worse...both!), as we are unable to fully comprehend the infinitely wise words of the audio GOD.

Are any of the more knowledgable people here able to first decipher Romy's holy writ, and secondly explain in such a way that we morons in Aus can understand??

He did not deign to explain the simple yokel questions we asked in trembling awe, such as 'can you then give us a rundown of the relative advantages/disadvantages of both passive and digital' ??




So, what is true/untrue about digital x-overs, can we be enlightened please???

We antipodean "Indomitable Morons ™" from the aussie forum that breed idiocy ... "" If you folks at this forum breed those cretins then good luck to all of you. I too much know all those games to have any inters to play them""" would like to thank any help you give to enlighten us.

[Iain McNeill]

Romy's analysis is right in a way but his conclusions are flawed. When he says that the resolution is reduced when a digital signal is attenuated he is right. This is a fundamental property of quantization, the process of converting an analog signal with an infinite number of possible levels to a digital signal with a finite number of levels.

However, he's missing the concept of dynamic range. 16 bits gives you almost 90dB of dynamic range when skillfully deployed. If you set 0dBFS (loudest digital signal) to be the loudest you want to listen, 116dBSPL for arguments sake, then the noise floor is around 26 dBSPL which is recording studio quiet. The 16 bit system will faithfully reproduce all signals in this range.

Now, if you were to apply 24dB of digital attenuation (shift word right 4 times, backfill with 0's) and turn up your amp to compensate then yes, it will sound pretty crummy, like a 12 bit system in fact. Thats because you just turned up the noise floor by 24dB not because the digital reproduction broke down.

BTW, if you think my example is rather marginal, you're right. This is why we have 24bit audio.

Full disclosure: I'm running a DSP crossover into a triamped speaker and I love the sound. Noticeable improvement over my Rane AC23. But then the DSP uses a 40bit multiply-accumulate for all the FIR filtering and only quantizes back to 16 bit for output.

So rest assured that there are good digital audio solutions out there. Just like everything audio, it's very easy to mess it up.

[ttan98]

Oops! someone should have told me earlier I just bought a digital x-over. maybe I should return it.

seriously I have confident that it will improve the sonic quality of my system if installed correctly. Furthermore the audio system must be reconfigured to take into account of the present of the x-over, eg sound level across all drivers, etc . BTW I am designing my system from scratch.

I visited Romy the Cat's website, this subject was discussed. I have not read all the discussions only a few. I don't know the Cat's background or his system setup, is there a possibility that his X-over is faulty or system incorrectly setup? Just ask those guys from Behringer DCX2496's thread, they will tell you a different story.

Furthermore the Cat has a section on horns for beginners, honestly I learn very little after reading 7 pages of so call notes....

there is more I can add but irrelevant here.

[Javachip]

Iain, where/how did you obtain your DSP crossover? Is it commercially available? How does its 40 bit internal resolution compare with, say, the DEQX? Thanks.

[Iain McNeill]

it's a Blackfin evaluation board from Analog Devices that I coded myself. I'm trying to find a DEQX to compare it to - do you have one? I looked at the DEQX but wanted more control over the filter responses.

P.S. sorry for threadjacking Terry!

P.P.S. I think the DEQX uses TWO Blackfins!

[terry j]

Originally posted by Iain McNeill
it's a Blackfin evaluation board from Analog Devices that I coded myself. I'm trying to find a DEQX to compare it to - do you have one? I looked at the DEQX but wanted more control over the filter responses.

P.S. sorry for threadjacking Terry!

P.P.S. I think the DEQX uses TWO Blackfins!

hey, it's all for learning Iain!!

I use a deqx myself, and am curious about the dacs in it. I'm a complete idiot with all this stuff, so am very interested indeed to learn more.

A common 'point to be wary of' re the deqx is the dacs, they can't be as good as this or that, I've no idea about the truth or otherwise.

Is a blackfin good?? Indeed, what IS a blackfin?

I'd lend you a deqx, but that may be a bit impractical!!

Romy certainly spices up things eh??

[Tenson]

Originally posted by terry j

Our little aussie hi fi forum just had a tornado pass through, and many locals are still reeling in shock. The weather people have yet to classify the phenomena, but for all intents and purposes could call it Hurricane Romy.

lol!

[Javachip]

No, I don't have a DEQX or any other active crossover. I've had an idea for an active 3-way with DSP crossover, so I've been keeping an eye on the DEQX, along with offerings by Juice HiFi and Ground Sound.

[MaVo]

Since both CD players and digital crossovers work in the digital domain, this "experts" deduction has to apply for both. Do cd's sound like cheap telephones on quiet passages in the music?

Since digital crossovers have even more resolution than cd technology (24 bits with the behringer, 16 with cd), they should modify the sound less than recording to cd does.

Second thought, this analogy of a cheap telephone is flawed in itself. The telephone works in the analogue domain. He should have said "like an old video game". 8 bit style

Third thought, what is a filter other than a device that decreases volume? So digital can filter. The problem is the noise floor. But this is so ridicously low on modern technology that i dont mind it at all. I would rather say, the noise floor in 24 bit gear is lower than with old analogue stuff like vinyl players, only the mechanism that induces noise is different. on vinyl its mechanically introduced on the recording and the playback, in digital technology it is due to the quantisation.

I really wonder why people stop to listen to the music once they have found an sophistic argument, why it should sound the way they want it to.

[jackinnj]

Originally posted by MaVo


Third thought, what is a filter other than a device that decreases volume?

Well, a filter also introduces group delay.