DCX2496 Upgrade Board - Objectively Tackling the Improvement of a Stock DCX2496 - diyAudio
Go Back   Home > Forums > Source & Line > Digital Line Level

Digital Line Level DACs, Digital Crossovers, Equalizers, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 13th June 2013, 04:31 PM   #1
opc is offline opc  Canada
diyAudio Member
 
opc's Avatar
 
Join Date: Oct 2004
Location: Ottawa, Ontario, Canada
Default DCX2496 Upgrade Board - Objectively Tackling the Improvement of a Stock DCX2496

Hi All,

Being a long time owner of a DCX2496, I have always wanted to tackle a long list of needed improvements to some of the common issues I have had with this slick little XO.

I finally have a need for this crossover in a more critical application than the usual development role it has found itself in for the past few years, so I've finally got some good motivation to get a board done!

The goal here is to start by summarizing all the things I've noticed while listening to the DCX over the years, and then hopefully quantifying those impressions with measurements.

The measurements should provide the baseline for performance, and help to zero in on the most critical issues with the DCX. I want every change to provide a tangible improvement that can be measured and qualified by listening.

So where do we start? How about a list of subjective things I don't like about the DCX as it is now:

1. Noise - this is one of my biggest issues with the DCX. The outputs suffer from a very high level of output noise which is further exacerbated by the gain in whatever amplifier stage is being used after the DCX. I have also noticed that there is significantly more noise when the input is done via AES/EBU, which is my preferred input type.

2. Output levels - This unit was clearly designed to be used with pro level inputs and outputs. The input voltage required to hit 0dBFS on the ADC is 10VRMS, and the output from the DCX at 0dBFS is a whopping 10VRMS as well! In a normal home HiFi, I think aiming for 2VRMS on both ends is a little more realistic and makes much better use of the 24-bit processing by using more of the available dynamic range.

3. It sounds dull. There is something very wrong with the way the upper mid and high frequencies sound and it's hard to pinpoint exactly what is causing it. I find the issue to be glaringly obvious in some recordings, and not as noticeable in others, but it always seems to be there to some degree. I noticed it most while working on a speaker using a pair of NeoPro5i ribbons which sounded astonishing with a passive crossover, and completely bland and flat when crossed over with this unit.

4. It's fatiguing to listen to. Again, I don't know exactly what causes this, but I find when it's in the signal chain, I'm reaching to turn the volume down more frequently than when it's not in the chain. There's something a bit grating to the way it sounds, and I don't like it. This is the main reason I've been using it primarily for development.

5. It clicks and pops. I have noticed this come and go under different use cases, but I find it occasionally pops on startup and clicks when the digital signal locks or unlocks. This is made worse no doubt by the obscene amounts of output gain, but it's also generally poor design.

Enough about the negatives, let's talk about why this little box is worth putting some effort into. Here's what I like about it:

1. It's an incredibly powerful and user friendly active crossover. It can provide pretty much any crossover type, slope and frequency with great ease, and I find it very intuitive to use. The built in EQ is also extremely powerful and allows the end user to achieve a flat response almost regardless of the drivers chosen in the system.

2. It's self contained. No dragging around a PC with 8 DACS and buying expensive software that has to run in the background, limiting your source to pretty much only your PC.

3. It's incredibly cheap given how powerful of a tool it is. You can buy the DCX for less than the price of your average 3-way passive crossover, which is an impressive feat from Behringer. It's also a fraction of the cost of competing solutions, and in my opinion, offers one of the most powerful DSP sections of the competing products I've seen.

4. The processing itself happens at 24/96 which most other options don't support. The last thing I want to do is downsample my 24/96 recordings back to 16/48 for processing - like other solutions do.

5. It supports both analog and digital inputs which is nice if you have multiple sources (a turntable and a music server for example)

In the next post, I'll provide all the baseline measurements and discuss exactly what might be causing some of the above issues, and what we can do to fix it.

Cheers,
Owen
  Reply With Quote
Old 13th June 2013, 06:40 PM   #2
diyAudio Member
 
jan.didden's Avatar
 
Join Date: May 2002
Location: Great City of Turnhout, Belgium
Blog Entries: 7
Default Chuck the output board!

DCX2496 6-channel vol control

jan
__________________
Any sufficiently advanced troll is indistinguishable from a genuine kook - Clarke paraphrased
Check out Linear Audio!
  Reply With Quote
Old 13th June 2013, 06:52 PM   #3
qusp is offline qusp  Australia
diyAudio Member
 
qusp's Avatar
 
Join Date: Oct 2009
Location: Brisbane, Australia
pretty sure thats on the list chucking the output board anyway
  Reply With Quote
Old 13th June 2013, 06:59 PM   #4
SY is offline SY  United States
diyAudio Moderator
 
SY's Avatar
 
Join Date: Oct 2002
Location: Chicagoland
Blog Entries: 1
I went passive on the output (adaption of Jan's circuit)- ended any issues with noise or gain, and (talking out of my rear because I don't have solid ears-only data) I think it sounds better. Certainly a lot of measured distortion artifacts at treble frequencies went away. No remote, but I didn't spend what his remote-controlled board costs!
__________________
“The short explanation is always dull. It generally includes the word ‘just.’ The explanation only becomes beautiful when you immerse yourself in every nuance.”
  Reply With Quote
Old 13th June 2013, 07:07 PM   #5
opc is offline opc  Canada
diyAudio Member
 
opc's Avatar
 
Join Date: Oct 2004
Location: Ottawa, Ontario, Canada
Alright... on to the measurements!

Let's start with the noise floor of the outputs. I set all 6 channels to bypass the crossovers, set the gain in each channel and the inputs to be 0dB, and measured with both analog inputs and AES/EBU.

Measurements 1 and 2 show the noise floor with analog input, and with AES/EBU respectively. As I had thought, the noise floor is significantly worse using digital input than it is using the analog inputs. My best guess is that the AES/EBU signal is polluting the analog outputs since they all run together on the same un-shielded ribbon cable. The only other possible culprit is the CS8420 which is present on the digital input but not on the ADCs. I doubt this is the source of the noise, however, since that level of noise is significantly worse than what's specified in the datasheet for the CS8420.

The remaining noise present in both the digital and analog input setups is almost certainly lower harmonics from the SMPS switching frequency.

There are large spikes present at 88Hz, 180Hz, 270Hz, 440Hz etc... which are not related to normal linear line rectified supplies. This noise is clearly present right up into the 1-3kHz region.

Based on the above, it looks like a great deal of the noise could be reduced by doing the following:

1. Wire the AES/EBU input directly from a dedicated input connector to the isolation transformer on the digital board using a shielded TP cable.

2. Remove the questionable SMPS and replace it with a fully linear supply using toroidal transformers and ultra-low noise regulators for all rails.

Next up is THD and THD+N. The last 8 measurements show THD and THD+N on the outputs using both analog and digital inputs.

Overall, distortion is reasonable at 1kHz, but it still falls short of the spec for the DAC which would imply that the limitation is the output stage design. Distortion also climbs to rather high levels at higher frequencies which, although probably not audible, points towards an output stage that isn't all that well designed.

These measurements are as follows:

3. Digital Input - THD @ 1kHz
4. Digital Input - THD+N @ 1kHz
5. Digital Input - FFT 1kHz, 2VRMS
6. Digital Input - THD VS FR 2VRMS
7. Analog Input - THD @ 1kHz
8. Analog Input - THD+N @ 1kHz
9. Analog Input - FFT 1kHz, 2VRMS
10. Analog Input - THD VS FR 2VRMS

Based on the above, it's probably best to redo the output stage to maximize the performance to at least the same level as what the DAC is capable of. It's a shame to have the analog output stage being the bottleneck when it doesn't cost much to do it right!

More to follow...

Cheers,
Owen
Attached Images
File Type: png FFT Spectrum Monitor - Noise Floor.png (180.1 KB, 1284 views)
File Type: png FFT Spectrum Monitor - noise floor 20kHz.png (215.5 KB, 1254 views)
File Type: png THD Ratio.png (86.4 KB, 1224 views)
File Type: png THD+N Ratio.png (86.6 KB, 1217 views)
File Type: png FFT Spectrum Monitor 1k 2VRMS.png (217.9 KB, 1212 views)
File Type: png THD Ratio VS FR.png (92.4 KB, 120 views)
File Type: png THD Ratio - 2VRMS.png (86.8 KB, 104 views)
File Type: png THD+N Ratio - 2VRMS.png (90.4 KB, 77 views)
File Type: png FFT Spectrum Monitor 2VRMS.png (190.3 KB, 100 views)
File Type: png THD Level - No XO - 2VRMS.png (83.5 KB, 108 views)
  Reply With Quote
Old 13th June 2013, 07:21 PM   #6
opc is offline opc  Canada
diyAudio Member
 
opc's Avatar
 
Join Date: Oct 2004
Location: Ottawa, Ontario, Canada
Quote:
Originally Posted by SY View Post
I went passive on the output (adaption of Jan's circuit)- ended any issues with noise or gain, and (talking out of my rear because I don't have solid ears-only data) I think it sounds better. Certainly a lot of measured distortion artifacts at treble frequencies went away. No remote, but I didn't spend what his remote-controlled board costs!
I would imagine the significantly lower gain (almost 21dB less!) would certainly help reduce the output noise, but given the poor PSRR of the DAC itself (50dB), I think the 5VA rail for the DAC would also have to be looked at to address the noise from the PSU.

The DAC itself also has a rather low differential output of 0.85VRMS at 0dBFS, so I think I would opt for a single OPA1632 per channel with a gain of 2.4 to get the differential output up to 2VRMS and the SE output up to 1VRMS.

Are you using balanced or SE outputs? If you're using SE then you're giving up the CMRR needed to maximize the DAC performance and you're only getting 0.425VRMS at 0dBFS

I think a well executed output stage would still be beneficial compared to just running straight out of the DAC, although running straight out of the DAC is almost certainly better than running it stock

Cheers,
Owen
  Reply With Quote
Old 13th June 2013, 08:07 PM   #7
SY is offline SY  United States
diyAudio Moderator
 
SY's Avatar
 
Join Date: Oct 2002
Location: Chicagoland
Blog Entries: 1
Balanced. I'm a big advocate of that. See Jan's article on the passive RC mod. Noise floor on my unit is down below -100dBV. Some people are dropping in output transformers, which also work well, but (IMO) not any better and perhaps not as well as a dollar's worth of R and C.
__________________
“The short explanation is always dull. It generally includes the word ‘just.’ The explanation only becomes beautiful when you immerse yourself in every nuance.”
  Reply With Quote
Old 13th June 2013, 09:10 PM   #8
opc is offline opc  Canada
diyAudio Member
 
opc's Avatar
 
Join Date: Oct 2004
Location: Ottawa, Ontario, Canada
Next up is frequency response...

Everything here looks pretty good, and inter-channel gain matching is actually much better than I was expecting.

Using the digital input, all six channels are within 0.1dB of each other which is quite good. Each respective channel is down about 0.13dB at 20Hz and 0.2dB at 20kHz.

Using the analog inputs pushes the channel matching a little broader, but still within 0.2dB which is pretty good. Each respective channel in this case is down about 0.2dB at both 20Hz and 20kHz.

Overall, no problems with the FR. I still prefer DC coupled outputs, but there are no serious red flags here, and certainly nothing to explain the poor high frequency performance!

1. FR and Inter-channel matching with Digital input
2. FR and Inter-channel matching with Analog input

So we've covered most of the basics... noise, distortion, frequency response, and we still don't have an explanation for the dull sound and the poor high frequency performance. After performing all the above measurements, I figured it would be best to see if I could get some meaningful jitter measurements from a system level to see if maybe that was the culprit.

It turns out this thing happens to have by far the worst jitter artifacts I have ever seen in anything I've ever measured. I can't be sure until I fix it, but I think this might have to something to do with the less than desirable sound.

3. Jitter Artifacts - Digital input at 44.1kHz
4. Jitter Artifacts - Digital input at 48kHz
5. Jitter Artifacts - Digital input at 96kHz
6. Jitter Artifacts - Analog input at 96kHz

Unfortunately I can only measure for artifacts at 96kHz using the analog input because that's the only frequency it operates at. I was able to measure at each of the common frequencies for the digital input.

Overall, this is very poor performance. As a reference, the NTD1 and the ESS9018 in the same measurement setup only have two tiny artifacts at -155dBr. The DCX produces a plethora of artifacts, some at a whopping -86dBr which is higher than most of the distortion harmonics!

Interestingly, the issue is mostly confined to the digital input, which tells me it probably has something to do with the CS8420. The fact that it gets worse as we approach a 1:1 ration also hints at this.

This issue definitely needs to be addressed, either through the use of a better ASRC, or possibly a better clock, although the more reasonable jitter levels on the analog input side would imply that the existing clock isn't really to blame.

I was really hoping not to have to alter the digital board, but it looks like that might not be an option.

More to come!

Cheers,
Owen
Attached Images
File Type: png FR Gain Relative.png (71.1 KB, 109 views)
File Type: png Gain - All Channels - Relative.png (72.7 KB, 97 views)
File Type: png Jitter 441.png (152.5 KB, 113 views)
File Type: png Jitter 48k.png (159.8 KB, 117 views)
File Type: png Jitter 96k.png (155.2 KB, 103 views)
File Type: png Jitter - 24k.png (180.6 KB, 93 views)

Last edited by opc; 13th June 2013 at 09:13 PM.
  Reply With Quote
Old 13th June 2013, 09:13 PM   #9
tomtom is offline tomtom  Slovakia
diyAudio Member
 
Join Date: Dec 2006
Im in!.

I have balanced passive output - and coaxial straight to trafo.
I think that passive output has 1,7v but im not sure. For sure it is enough.

When i went to passive - i start to have problem with digital signal - so i went to coaxial direct to transformer - but it didnt help. It was very strange grounding problem - i must cut some grounds in ribbon cables and left some. Just one combination works. I use spdif /i change resistor behind trafo to 75ohm/ and with some sound cards - coaxial connector must be connected to DCX chassis otherwise it did not play and with other sound card it must float...

Basicaly grounding problem behind my comprehension.

Do you guys thinking about Najda board? For ca 250E maybe better point to start. Also it has 8 channels... /but only SE/

Back to DCX. One of ultimate digital input solutions is GOOD new clock in DCX
optical in and optical out module back to source to slave source to DCX clock.

Many sound cards has this option /f.e. ESI july/ this way get total isolation from PC /optical/ and utilize good local clock in DCX - without need for PLL clock reconstruction. You also bypass ASRC in DCX - so you must send 24/96 from PC. But you can configure f.e. WIN7 to do that. And use very hiquality resampling in PC f.e. SOX

Other solution is Asynchronous USB module
  Reply With Quote
Old 13th June 2013, 09:32 PM   #10
diyAudio Member
 
Join Date: Jan 2002
Why not replace the dac chips with the es9018 and add usb input???

I can dream, can't he?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
FS: DCX2496 + 6-Channel Jan Didden Volume Control (Active Upgrade) mrwireless Swap Meet 4 19th July 2012 08:15 PM
WTB Jan Didden DCX2496 board peterhenk Swap Meet 0 11th March 2012 07:58 AM
MiniDSP vs DCX2496 vs DCX2496-Modded vs DEQX gainphile miniDSP 20 14th September 2011 09:24 AM
WTB: DCX2496 digital board revh/1.17 nonsub Swap Meet 4 24th March 2010 03:24 PM
DCX2496 upgrade problem sfdoddsy Digital Line Level 9 24th May 2004 02:36 AM


New To Site? Need Help?

All times are GMT. The time now is 05:09 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2015 DragonByte Technologies Ltd.
Copyright ©1999-2015 diyAudio

Content Relevant URLs by vBSEO 3.3.2