Hello DCX2496 users. I have a rather simple question regarding this device.
Simulating speakers using WinISD Pro, the program shows that high pass filters introduce a very high group delay approaching their frequency. Is this a property of analog filters, or will this always be the case?
What I'm considering here is a ~19Hz high pass rumble filter with the DCX2496, and I'm worried that this will introduce rather high group delays at 30Hz and below. Can somebody shed light on the issue? Will I avoid this problem using a digital equalizer/filter?
Simulating speakers using WinISD Pro, the program shows that high pass filters introduce a very high group delay approaching their frequency. Is this a property of analog filters, or will this always be the case?
What I'm considering here is a ~19Hz high pass rumble filter with the DCX2496, and I'm worried that this will introduce rather high group delays at 30Hz and below. Can somebody shed light on the issue? Will I avoid this problem using a digital equalizer/filter?
This is 'always' the case.
DCX2496 does mimics a passive filter by IIR algorithms, so the behavior is close to identical.
A change in amplitude will always change phase/GD, this is called miniphase behavior.
A lowpass at 20Hz will cause added GD, is it something to worry about? Not that low in frequency.
DCX2496 does mimics a passive filter by IIR algorithms, so the behavior is close to identical.
A change in amplitude will always change phase/GD, this is called miniphase behavior.
A lowpass at 20Hz will cause added GD, is it something to worry about? Not that low in frequency.
Thanks guys, you're very helpful.
While I'm at it I have another question - what's the lowest frequency that the filters can be applied at with the DSX? I'm thinking about getting one for distributing signals to my subwoofers....
I had the software for it installed on my windows machine, but my motherboard stopped working, and I can't install it on my Apple I don't think.
An 8th order highpass at 20Hz causes (according to WinISD) in my application 80ms of GD at 20Hz, and 16ms of GD at 41Hz. I have read that any GD in the double digits of milliseconds is audible. Perhaps this is not the case in low frequencies?
While I'm at it I have another question - what's the lowest frequency that the filters can be applied at with the DSX? I'm thinking about getting one for distributing signals to my subwoofers....
I had the software for it installed on my windows machine, but my motherboard stopped working, and I can't install it on my Apple I don't think.
Last edited:
You cannot use Stef's PS in DCX for the very reason you described. Stef's 9V is not regulated and when fed it drops bellow the level that is insufficient for Franks board to operate. I know since I tried it, so no reason to do it. I believe if you use +- 15 V you will have way too much heating on the Frank's board, or I am not sure that his regulator could handle that big of a voltage drop.
As explained many times, it is not necessary to exchange the PS.
Stef's boards are made for DEQ or for DCX without Frank's mod. They are really well made, but as explained before switcher in DCX is not bad.
The perceptibility of GD is frequency dependent. Think about it this way: A 20 Hz wave needs 50 ms to complete one wavelength. 80 ms would be about 1,5 wavelength. But your brain needs a couple of wavelength to recognize a frequency.
Rudolf
is this why backloaded horns seem to work with the MF & HF that comes from the front?
In very general terms: yes. But while the wavelength of 20 Hz is 17 m, the traveling distance between the sound from the front and the back of a horn is typically about a tenth of that. It only becomes an issue if there is too much midrange coming through the horn.
How often does "Dull Sound" problem actually occur
I don't think I've ever noticed the "dull sound" problem, even though I have fed DCX 96 Khz on quite a number of occasions. I always feed it digital in AES balanced from my Tact 2.0 RCS. Mostly I feed it 44.1Khz, but occasionally I feed it 96Khz when I am playing a DVD-Audio or SACD (resampled from analog outs thorugh a MSB Pad-1 into 96Khz, then through the Tact, bypassing Tact's analog input, or straight-through 96Khz from a DAD or my Masterlink). I have frequently tested hearing (better than average for 54 year old) and very transparent top end with Acoustat 1+1's augmented with Elac omni ribbon tweeters (response to 35Khz), and verified flatness to 20Khz.
I was also wondering if DCX doesn't auto-reset digital input if "dull sound" problem begins to occur. I have noticed that sometimes DCX hangs on power restoration after power outage, then I manually restart DCX.
I see people say that DCX input is "useless" with 96Khz digital input, well that hasn't been my experience, though maybe I missed it, or maybe it's less frequent with AES input.
I don't think I've ever noticed the "dull sound" problem, even though I have fed DCX 96 Khz on quite a number of occasions. I always feed it digital in AES balanced from my Tact 2.0 RCS. Mostly I feed it 44.1Khz, but occasionally I feed it 96Khz when I am playing a DVD-Audio or SACD (resampled from analog outs thorugh a MSB Pad-1 into 96Khz, then through the Tact, bypassing Tact's analog input, or straight-through 96Khz from a DAD or my Masterlink). I have frequently tested hearing (better than average for 54 year old) and very transparent top end with Acoustat 1+1's augmented with Elac omni ribbon tweeters (response to 35Khz), and verified flatness to 20Khz.
I was also wondering if DCX doesn't auto-reset digital input if "dull sound" problem begins to occur. I have noticed that sometimes DCX hangs on power restoration after power outage, then I manually restart DCX.
I see people say that DCX input is "useless" with 96Khz digital input, well that hasn't been my experience, though maybe I missed it, or maybe it's less frequent with AES input.
<<I was also wondering if DCX doesn't auto-reset digital input if "dull sound" problem begins to occur. I have noticed that sometimes DCX hangs on power restoration after power outage, then I manually restart DCX<<I see people say that DCX input is "useless" with 96Khz digital input, well that hasn't been my experience, though maybe I missed it, or maybe it's less frequent with AES input<<<
Yes It's because of CS8420 (originally bug) and bad placement in DCX .If You use another digital driver, as CS8414 or DIR9001 and better impulse transformer You can use AES/EBU without any problems or SPDIF If You choose the right transformer working for both . The unique working mods that selectronic proposed is for that way with reclocking as well. (All other selectronic' mods are not good).
But, I always find using DCX is much better with analog input signal,BUT, with some heavy mods that I can describe, my choice,if interested.
Yes It's because of CS8420 (originally bug) and bad placement in DCX .If You use another digital driver, as CS8414 or DIR9001 and better impulse transformer You can use AES/EBU without any problems or SPDIF If You choose the right transformer working for both . The unique working mods that selectronic proposed is for that way with reclocking as well. (All other selectronic' mods are not good).
But, I always find using DCX is much better with analog input signal,BUT, with some heavy mods that I can describe, my choice,if interested.
1. The ‚dull sound’ problem (10 kHz cut-off) occurs at power up of DCX or digital source ONLY. Also disconnecting and connecting the digital cable might cause the problem. There is NO auto-reset. You must power down and restart DCX again in case of ‘dull sound’ error.
My 3 DCXs all had this ‘dull sound’ problem. It occurred at about every 10-20th power up. There seem to be DCX users who don’t have this problem at all (or do not hear it?) which might depend on power-up behaviour of digital source. The ‘dull sound’ problem is independent of the sample rate and transformer! It’s caused by the erroneous CS8420 only.
To get rid of the problem you need to replace the CS8420 by CS8416 + AD1896 which also provide improved sample rate conversion (SRC). Some of the CS8420 mods provide an additional low jitter 24.576 MHz clock which improves sound quality of DCX significantly not only using digital but also analog input. The attached Word-File shows the different mods.
Based on my measurements and experience sound quality of digital input is better than analog one. That’s because you avoid a D/A- and an A/D-conversion.
2. Although the ‘huge cap’ believers will disagree to my point of view replacing the switched DCX power supply will cause no significantly sonical improvement. I’m not aware of any measurement which shows the opposite.
The DCX supply provides 5 output rails (http://www.awdiy.com/uploads/pdf/DEQ2496-PSU-1.5.pdf ):
A. The +3.3V and +5V rails are used for digital devices only. The believing modifying these two rails causes any sonical improvement is pure NONSENSE (sorry)!
B. The 8-9V rail is used by two 7805 regulators which supply the ADCs and DACs. To really improve PSRR (110 dB) and noise (18 µV) it’s best to replace the 7805 regulator for the DACs and if you use analog input also the one for the ADCs by Vreg .
C. On the +/-15V rails there is 50 kHz noise (out of sonical band) caused by the switched supply. If you use digital input and passive outputs +/-15V rails are not used at all. For all other applications noise can simply be reduced by adding pi-filters before the 78/7915 regulators as it is realized by L2+C10 and L3+C11 for the +3.3V and +5V rails. The costs for three 100 µH (0.65 ohm) coils and 220 µF (25V) caps are about 2 dollars only.
My 3 DCXs all had this ‘dull sound’ problem. It occurred at about every 10-20th power up. There seem to be DCX users who don’t have this problem at all (or do not hear it?) which might depend on power-up behaviour of digital source. The ‘dull sound’ problem is independent of the sample rate and transformer! It’s caused by the erroneous CS8420 only.
To get rid of the problem you need to replace the CS8420 by CS8416 + AD1896 which also provide improved sample rate conversion (SRC). Some of the CS8420 mods provide an additional low jitter 24.576 MHz clock which improves sound quality of DCX significantly not only using digital but also analog input. The attached Word-File shows the different mods.
Based on my measurements and experience sound quality of digital input is better than analog one. That’s because you avoid a D/A- and an A/D-conversion.
2. Although the ‘huge cap’ believers will disagree to my point of view replacing the switched DCX power supply will cause no significantly sonical improvement. I’m not aware of any measurement which shows the opposite.
The DCX supply provides 5 output rails (http://www.awdiy.com/uploads/pdf/DEQ2496-PSU-1.5.pdf ):
A. The +3.3V and +5V rails are used for digital devices only. The believing modifying these two rails causes any sonical improvement is pure NONSENSE (sorry)!
B. The 8-9V rail is used by two 7805 regulators which supply the ADCs and DACs. To really improve PSRR (110 dB) and noise (18 µV) it’s best to replace the 7805 regulator for the DACs and if you use analog input also the one for the ADCs by Vreg .
C. On the +/-15V rails there is 50 kHz noise (out of sonical band) caused by the switched supply. If you use digital input and passive outputs +/-15V rails are not used at all. For all other applications noise can simply be reduced by adding pi-filters before the 78/7915 regulators as it is realized by L2+C10 and L3+C11 for the +3.3V and +5V rails. The costs for three 100 µH (0.65 ohm) coils and 220 µF (25V) caps are about 2 dollars only.
Frank, you are correct on all, of course, but just one correction here. 15V rail is also used for the operation of the front display board so it cannot be eliminated. If you use passive output you do not need this rail obviously for the output but you still need to use it for the display of DCX. As far as I remember it is just +15V and not negative rail, but do not quote me on this, check for yourself.
Hi Vladimir, you are right. The frontpanel uses LCDVEE for the LCD display which depends on the +15V rail but this is sonically not relevant. I have overseen this because I do not use the frontpanel at all (disconnected it). I’m using the DSP-PCB and the supply only in a housing with the power amps. Therfore I prefer the PC-software via RS-232. Regards, Frank
Ha, I know it is easy to oversee it. 
I am talking out of personal experience.
I was trying something and I was thinking since my output is passive... oh, I do not need this 15V any more... until I got the black front panel.
Yes I do prefer as well the PC connection and control. With the laptop in my lap in the listening position... nothing could not beat that.
I figure out to mention, for anyone not comfortable of loosing that feature of front control.
I am talking out of personal experience. I was trying something and I was thinking since my output is passive... oh, I do not need this 15V any more... until I got the black front panel.
Yes I do prefer as well the PC connection and control. With the laptop in my lap in the listening position... nothing could not beat that.
I figure out to mention, for anyone not comfortable of loosing that feature of front control.
Describing 'Dull Sound' problem
If running a single digital source(s) continuously, will DCX sometimes go into dull sound mode spontaneously, or does it only happen on transitions between digital sources and/or power on?
I switch between digital sources 44.1Khz - 96Khz on my Tact RCS 2.0. Tact syncs to incoming clock, whatever it is, then feeds result with same clock (after EQ processing) to Behringer through AES/EBU interface. There's usually a blip on switching from one source to another, but it's surprisingly small.
The official AD description of dull sound problem is pretty confusing. Do you mainly hear a 10Khz cutoff, or spurious tones, harshness or something else?
If running a single digital source(s) continuously, will DCX sometimes go into dull sound mode spontaneously, or does it only happen on transitions between digital sources and/or power on?
I switch between digital sources 44.1Khz - 96Khz on my Tact RCS 2.0. Tact syncs to incoming clock, whatever it is, then feeds result with same clock (after EQ processing) to Behringer through AES/EBU interface. There's usually a blip on switching from one source to another, but it's surprisingly small.
The official AD description of dull sound problem is pretty confusing. Do you mainly hear a 10Khz cutoff, or spurious tones, harshness or something else?
I generally concur with Frank's post above. Expect the dull sound part. It can happen with a rupture in the digital input signal. A glitch, a change of sample rate, or break in the signal can trigger it. I've heard it many times. Maybe some are more sensitive than others, so it doesn't happen as often on some units?