Behringer DCX2496 digital X-over

[tnargs]

Yes that's right. The DEQ feeds the DCX. The DEQ is an equaliser not a crossover.

[oettle]

Hi AR2,

I’ve reproduced the following of your mods:

1. Replace DACs AK4393 with AK4396
My expectations have been rather low because the AKM data sheets only show two relevant differences. Power consumption is reduced from 310mW to 200mW http://www.asahi-kasei.co.jp/akm/en/...k4396_f00e.pdf (page 4) and out of band noise is somehow reduced http://www.akm.com/images/ak4396%20graphic.htm . Measurement diagrams don’t show substantial differences http://www.asahi-kasei.co.jp/akm/en/...3/akd4393e.pdf and
http://www.asahi-kasei.co.jp/akm/en/...6-sbw2-02e.pdf .
The mod is pretty easy. You just have to disassemble the three AK4393s and to assemble the AK4396s.

Although not expected I would say there is a small improvement. To my opinion sound became more accurate and realistic. So I would say it’s worth doing this mod.

2. Replace digital input transformer
I replaced the original DCX transformer by a low jitter SC947 transformer http://www.scientificonversion.com/catalog.html . I compared the different transformers to my second optical digital input. I haven’t been able to hear any difference before and after the mod. That might be different with higher sample rates e.g. 96 kHz and other input sources.


Hi Ergo,

I made some measurements with a scope at the original DCX supply http://www.awdiy.com/uploads/pdf/DEQ2496-PSU-1.5.pdf . There is an about 400 mVpp noise on each of the three analog +/-15V and +9V rails but not on the two digital +3.3V and +5V rails. Looking at the schematic reason is that the two digital rails have a pi-filter (CLC) at their output and the analog rails not. The used regulators 7815/7915 and 7805 can’t reject this HF noise and transmit it to the ADCs, DACs and opamps. This noise is an out of band noise (>20 kHz). I can’t say whether this HF noise is sonical relevant.

Regarding your RF shielding I’m still wondering which components could be transmitter and which receiver of this RF noise. Sampling frequency of the front panel is about 150 Hz. So it would be interesting whether RF noise can be reduced by simply disconnecting front panel (flat cable).

Regards, Frank

[theob]

I know this a dxc mod thread but I've look far and wide and haven't found one.

[ergo]

I hope to do the STOCK versus MODIFIED DCX2496 measurements tomorrow - have both units at home and awaiting.

The plan is to do at least following

noisefloor
freq resp
THD
IMD
Dunn Jitter signal

if time allows also 44.1kHz and 48kHz comparison. Due to odd complications I can not create 96kHz SPDIF for DCX with neither soundcard I have

So hopefully I will have these results up tomorrow.

Ergo

[stef1777]

Originally posted by oettle
[B]
I made some measurements with a scope at the original DCX supply http://www.awdiy.com/uploads/pdf/DEQ2496-PSU-1.5.pdf . There is an about 400 mVpp noise on each of the three analog +/-15V and +9V rails but not on the two digital +3.3V and +5V rails. Looking at the schematic reason is that the two digital rails have a pi-filter (CLC) at their output and the analog rails not. The used regulators 7815/7915 and 7805 can’t reject this HF noise and transmit it to the ADCs, DACs and opamps. This noise is an out of band noise (>20 kHz). I can’t say whether this HF noise is sonical relevant.

Hi Oettle!

Not sure that the PI filter is solving all noise problem. On my linear PSU, I first included a PI filter on the analog rail and after more testing, the noise floor was lower without the PI filter. 7815/7915 are not the best regulators for noise too.

With the linear PSU, +-15v and 3.3v are around 15mVpp noise, the +5V is below 30mVpp. If you replace the 7805 on the DSP board with a low noise regulator, you go further too.

With the linear psu, the difference is not only lower noise. Dynamic and high frequencies are also better. I received 3 reports from diyers with the same opinion.

For the one who use the DCX in analog way, I would also suggest to change the op-amp before going with the passive solution. It's a pain to do but very very efficient. I suggest LM4562 for input circuits and buffers, and OPA2134 for the DAC output (or better a double OPA827). Passive solution is in theory very fine but is not very dynamic and bandwidth not really flat.

Stephane

[oettle]

Hi Stephane,

You are right you can’t reduce all noise with a pi-filter but you can reduce noise depended on its frequency. With the proposed 100µH coil and a 220µF cap noise reduction @ 100 kHz is about 40-50 dB. So the 400 mVpp noise can be reduced to something about a few mVpp and it’s getting difficult to measure the remaining noise with a scope at all.

PSRR of the AK4393 is 50 dB and that of a standard 7805 is about 50-60 dB within sonically frequency range. That might be the reason why Ergo can’t see any improvements @10Hz-20kHz. Above 20 kHz this should be different because PSRR of the standard 7805 regulators get worse. Remaining question would be whether this out of band noise is sonically relevant at all?

Regarding PSU modding I didn’t understand benefit of improving +3.3V and +5V rails so far. These rails aren’t used by a single analog component. They supply digital components only. I’m using 3 DCX for several years and I’m not aware that DSP failed nor have I heard about such a problem.
But noise is an issue for the +/-15V and +9V supplies which are used mainly by analog components (DACs, ADCs, opamps).

Regards, Frank

[AR2]

Originally posted by oettle
Hi AR2,

I’ve reproduced the following of your mods:

1. Replace DACs AK4393 with AK4396
My expectations have been rather low because the AKM data sheets only show two relevant differences. Power consumption is reduced from 310mW to 200mW http://www.asahi-kasei.co.jp/akm/en/...k4396_f00e.pdf (page 4) and out of band noise is somehow reduced http://www.akm.com/images/ak4396%20graphic.htm . Measurement diagrams don’t show substantial differences http://www.asahi-kasei.co.jp/akm/en/...3/akd4393e.pdf and
http://www.asahi-kasei.co.jp/akm/en/...6-sbw2-02e.pdf .
The mod is pretty easy. You just have to disassemble the three AK4393s and to assemble the AK4396s.

Although not expected I would say there is a small improvement. To my opinion sound became more accurate and realistic. So I would say it’s worth doing this mod.

2. Replace digital input transformer
I replaced the original DCX transformer by a low jitter SC947 transformer http://www.scientificonversion.com/catalog.html . I compared the different transformers to my second optical digital input. I haven’t been able to hear any difference before and after the mod. That might be different with higher sample rates e.g. 96 kHz and other input sources.



Regards, Frank

Hello Frank, from stormy Northern California. I just got power after 1 and 1/2 day of outage. Regarding DACs that is exactly how I felt about exchange of DAC chips. There is difference, slight and mostly heard in high frequencys. Not a day or night but worth doing.

Regarding digital transformer, I believe it is not such a mod that could be heard with all other mods installed. Since Jon and Scientific Conversion is literary in my neighboroughood I had a opportunity to visit them and to have a demonstration on very sophisticated HP jitter measurement unit. Jon has big collection of various digital transformers for comparison. It was an eye opener. With simple replacement of various transformers in the same test board, instrument was showing jitter reading. The difference was tenfold.

Just exchanging digital transformer is not going to make a bad design a good one. In the case of Behringer it most likely lowers jitter at one step, and than everything else contribute to the sound.

I am glad that you found improvement with new DACs.

Cheers
AR2

[oettle]

Hi AR2,

You are right my experience with the S/PDIF transformer is based on my own SRC/clock mod. With the original CS8420 it might be different although I think jitter before the SRC isn't such important as long as PLL is able to lock properly because of reclocking.
That seems to be completely different on the secondary side of the SRC. The DACs obviously need to have a low jitter clock.

Frank

[stef1777]

Originally posted by AR2


Hello Frank, from stormy Northern California. I just got power after 1 and 1/2 day of outage. Regarding DACs that is exactly how I felt about exchange of DAC chips. There is difference, slight and mostly heard in high frequencys. Not a day or night but worth doing.

Have you tried to play with CKS pins? On the original design, the setting is pins 26, 27 and 28 connected to ground. It can be interesting to try disconnecting pin 28 and 27 from ground with AK4396.

[oettle]

Hi Stephane,

I'm a little bit confused. LRCK = 96 kHz and MCLK = 12.288 MHz in the DCX. These clock inputs need a 128 Fs setting. How can you modify Fs setting without modifying clock inputs?

Regards, Frank