Behriner DCX 2496 transformer output mod.

Design and use of the various types.
Post Reply
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Behriner DCX 2496 transformer output mod.

Post by dave slagle »

hey,

A fellow list-member sent me his DCX to install transformers into.
Since this was somewhat uncharted grounds I figured is best to be
safe and make the mod easily reversible. From the application sheet
of the dac chip it appeared that all that i needed to do was to hang
the transformer off the + & - outputs which both float at around
2.5VDC. Since they are at the same potential no DC would flow. In the
event that there would be a slight DC offset, a transformer design
that is happy with a ma of current should more than suffice.

In looking at the schematic for the stock output stage there are a
pair of series 2K resistors feeding the stock analog output stage.
In order to keep the whole thing reversibly i simply soldered the
transformer primary directly across the Dac side of those 2K
resistors and hung a pair of RCA connectors off the secondary and the
mod was done (at least for the initial listening session) The added
benefit of this method was that the existing analog outputs still
worked for an easy A/B comparison.

I'll be brief with my subjective description of the sonics. For my
tastes the transformers just sound right. The Opamp outputs just
sound harsh and hyper-detailed to me and using the transformer output
takes an edge off the music. Everything seems clearer and it as
almost as if a haze is lifted from the sound. It was one of those
situations that had me initially thinking it sounded recessed with
the transformers but after a small amount of time it quickly became
apparent to me that the opamp output was the situation that
artificially detailed and harsh.

I didn't add a filter to the output and on the scope i could see some
noise at the 96K frequency, but i didn't hear any artifacts.
Measurement with a test CD showed me flat frequency response from
10-20K and the output 2.5VAC with the transformer. Since i knew the
noise was there I figured a simple cap would help attenuate it and a .
01uf dropped the output at 20K about 1/10th a dB and cleaned up the
waveform considerably (from memory and now that i think of it i'll
have to look a bit closer since that seems wrong)

I need to play with the cap a bit more and since the transformer goes
out to 100K when used this way my guess is that it isn't providing
any out of band filtering. SInce i do not like the idea of providing
unnecessary loading on transformers i'll probably toy with a pair of
capacitors to ground off the dac output to stop the noise before it
gets to the transformer. I'll give the various possibilities a
listen and let you guys know what i find. My gut feel is the best
sound will be from no additional filters added but my ears will need
to confirm that. The one other thing i'll need to measure is what
the output impedance of the chip is just for curiosity sake.

dave
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

did some crude measurements of the Z-out before and after the chip and it seems that the output Z of the chip is in the 50-60 ohm range. This makes me really want to do a 12 step 1dB per step TVC output for level adjust so the digital attenuation would only be used for a 0 to -1dB trim.

the TVC's could be custom would for the respective frequencies to assure extended bass and perfect highs.

Time to wind a test dummy.

dave
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

OK.

test dummy #1 wound and tested. The thing that worries me most is what happens with the TVC in the first few steps. Also how the whole thin behaves wrt the balanced to SE conversion. I shot for 25hy's of primary inductance which assuming a 60 ohm source should give me response flat to well below 10hz and looked at the HF behavior on all of the settings. (0dB to -10dB)

frequency response looked better than the autoformers (boy you can get away with murder at 50 ohms :-) with a gradual slope up to a 4-5dB peak in the 300-400Khz range. this all unloaded and generally the following capacitances typically found after the units will take care of that.

i'm reasonably sure that given a known frequency range i could get the bandwidth out into the mhz for the HF portion so for now i see it as a no-brainer to tap the units for level adjust.

Maybe the best idea would be to o for 2dB steps for a 20dB range and then rely on the DCX digital attenuator to trim?

this will at least get all of your levels matched without relying too much on the digital attenuation, then the question becomes how to do the overall attenautation. My gut feel is to just hang 6 ganged autoformers off the secondary of the transformers (or a 6deck pot) and them be done with it.

I must say that given my initial brief listening sessions, i'm tempted to get one of these units for when i get back to playing with my azuras to get the ballpark ranges set, then i can design the system with a passive linelevel crossover.

dave

dave
JeffreyJ
Posts: 320
Joined: Fri Jan 28, 2005 2:53 am
Location: stuck in the USA for now...
Contact:

good job...

Post by JeffreyJ »

I completely agree that tapping would be great... with these you really need lots of thought on gain structure to keep the digital sugnal up... it took me a while to get mine just right... but with the tapped tranformers, it is easy..

and, yeah, once you get rid of the output stage, they sound good...

Peace,
Me
jeff mai
Posts: 15
Joined: Sat May 06, 2006 11:57 am

Post by jeff mai »

Hi Dave,

What's the size of the transformers used? No chance of them residing inside the DCX case, I suppose. I bought the DCX for tweaking, but the convenience and the delay feature make it likely to be relatively permanent.

What do you need to know about the system to get this happening for me? I've been suffering through the op-amp output stage of the DCX and now a silicon-based 6 channel attenuator for a few weeks.

The system is 3 way with crossover points of 400Hz and 5000Hz. All output levels are currently set within 6db of each other on the DCX. I reckon that 10db of adjustment on the output trans would be plenty for me since I DIY all the amps.

Cheers,

Jeff
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

got everything wired up today including a zif socket for mounting a capacitive load to eliminate the digital noise i see.

a 10nf cap got rid of it but also slightly phase shifted a 20K sine a 4.7nf cap did less damage to the phase and still did the trick on the noise. going smaller in cap size started ot let the noise through so i chose those two values for a brief listening session.

I first started with the naked transformers and the sound was as i remembered. Going to the opamp out was again as i remembered so i inserted the 10nf caps and went back to the trannies. The difference wasn't as profound, the sound again was harsh and there was an edge to the sound. quickly yanking the caps brouhgt back the sound i preferred. moving down to the 4.7uf caps did the same but maybe to a slightly lesser degree. I would have to say on the scale from right (the trannies) to wrong (the opamps) the cap loading roughly split the difference.

Again I want to note that i hear no artifacts with no filtering in place even though it looks a bit ugly on the scope. This got me thinking of why the 96K noise would be a problem? I know of many people heating with 20Khz and above filaments so how is this any different? I know that HF artifacts can fold back down into the audio band, but this doesn't seem to be happening.

The idea of draining all of that HF info into the ground on the secondary side really doesn't appeal to me anyways and the pollution of ground on the analog side seems like the worst thing a person can do. Maybe tomorrow, i'll try to keep the hash on the digital side and use a pair of caps to the Vref in an attempt to send that noise back where it belongs.

I was able to get 4 transformers pretty easily under the hood and another two should be possible, but not in a tapped format.

One other thin that kicked my ass for the bulk of the day was the less than logical placement of the 2K smt resistors. All 6 outputs appear to have the same layout, yet the phase didn't seem to follow the resistor placement and on output #4 i had to grab signal from the far side of the 2K's opposed to the near side on outputs 3,5 & 6.

I do suspect that some sonic improvement might be heard from removing power from the existing analog outs and removing the 2K series resistors thus severing the unused connections from loading down the dac output.

dave
jeff mai
Posts: 15
Joined: Sat May 06, 2006 11:57 am

Post by jeff mai »

I'm not opposed to removing the entire output PCB from the unit in order to get the transformers in there. Would need to redo the input section, but the digital input circuitry is awful and should be redone anyway.
JeffreyJ
Posts: 320
Joined: Fri Jan 28, 2005 2:53 am
Location: stuck in the USA for now...
Contact:

hello guys...

Post by JeffreyJ »

hello guys...

I love Jeff's comment about buying it for tweaking, but it has become permanent... that is *exactly* my situation... can't let it go... crossovers belong in front of the amps... every time I do this I am amazed..

but the digital input? how can that be terrible?!?!.... all they have to do is route the digital to the correct place... how can that be hard? I guess what I am saying is keep me posted on what you find about modifying the front end... sigh...

Peace,
Me
jeff mai
Posts: 15
Joined: Sat May 06, 2006 11:57 am

Post by jeff mai »

Hi Jeffrey,

The digital input circuitry should maintain the 110ohm characteristic impedance of the AES/EBU digital standard. This is nearly impossible to do with the connectors used and then there is the relay that switches input A between analog and digital. This screws up the reflected impedance and causes jitter. Try a long digital cable (6 feet or more) if you aren't already doing so and you may find this improves things.

If you use an RCA to XLR adapter to convert from from SPDIF to the AES/EBU input it makes things even worse. The standard SPDIF output voltage is marginal and indeed some DCX units fail to lock on to the digital signal when used with such an adapter.

The partially good news is that the sample rate converter is always 'on' and so this makes the internal clock worth upgrading as the incoming digital signal is always converted to the internal clock. My Tentlabs clock is going in as soon as it comes back from Tentlabs. A special version is needed for the DCX. It's only partially good news because some people aren't so keen on sample rate conversion, but you can't have everything.

I'll keep you posted on what I end up with.

Jeff
jeff mai
Posts: 15
Joined: Sat May 06, 2006 11:57 am

Post by jeff mai »

Sitting here listening this evening... Where do I send my money? :D I need a master attenuator solution too. I've got this sounding pretty good, but I know it could be lots better.
DuoBruce
Posts: 91
Joined: Mon Mar 07, 2005 12:17 am

Post by DuoBruce »

http://www.sound4sale.com/products/MVC.php

I'm getting ready to buy one of these as my volume control solution, I hope.

JJ pointed out it has input and output buffers but looks like you could easily bypass the input one. The output may need the opamp since part of the ladder it is the feedback for the opamp but I need to look closer at the circuit. Maybe you can just use the resistor ladder.

It appears they just used the dual programmable attenuator from page 9 of this data sheet

http://www.linear.com/pc/downloadDocume ... 1571,D4081
Bruce Bosler
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey guys

jeff,

any suggestions for the input circuit? I simply hacked the input in and it sounded good. The trannies really made a big difference, you know the whole music vs. hi-fi thing.

Bruce,

another option is here

one final option might be to contact john chapman and see if he has any ideas for a 6 channel remote. I think the important feature would be to be able to adjust levels and then have all 6 channels track each other.

dave
Last edited by dave slagle on Thu Sep 13, 2007 11:46 pm, edited 1 time in total.
DuoBruce
Posts: 91
Joined: Mon Mar 07, 2005 12:17 am

Post by DuoBruce »

can you post a schematic of your input hack? I understand you have to maintain a 2.5v bias on both input pins.

My phono stage is transformer coupled out so I thought this might work. Straight to the ADC and bias through added resistors chosen to get the load you want on the phono. Get the 2.5V from a regulator added to one of the 5V supplies.

Then I see that the Zin of the chip is pretty low, 2.4K min and 4K typical so I don’t think the phono will like that since it's Zout is probably about the same. I'll need a buffer to make that work.
Attachments
phono.JPG
phono.JPG (10.8 KiB) Viewed 28601 times
Bruce Bosler
jeff mai
Posts: 15
Joined: Sat May 06, 2006 11:57 am

Post by jeff mai »

dave slagle wrote: any suggestions for the input circuit? I simply hacked the input in and it sounded good.
Something like what is discussed here is where I was going to start:
http://www.diyhifi.org/forums/viewtopic.php?f=2&t=1191
jeff mai
Posts: 15
Joined: Sat May 06, 2006 11:57 am

Post by jeff mai »

Hi Dave,

I'll need to do all six output channels of this thing - I'm using a 1 inch compression driver as a tweeter. Not sure if it would be better to do them one pair at a time and use caps for the time being in the other channels. Do you think there will be a learning curve?

The current crossover points are 400Hz and 5KHz. I may want to change those a bit, particularly the HF point.

One thing I'm not sure about is if or how many secondary taps I'll need. What is the tradeoff in tapping the secondary? I have a lot of control over the amps and speakers since I DIY everything, and there is always the attenuation in the DCX in a pinch. What do you think?

Cheers,

Jeff
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I assume you are talking about the ideas attributed to JH. looks simple enough. what about the remaining two analog inputs?

the three input thing is nice and for all things digital, keeping it in the digital domain works since transport-dac-adc-eq-dac seems like the department of redundancy department at work. the JVC DCX i have does just that and it kills me. the sheer value of it and the fact it is on loan makes it to cherry to chop even if i have permission. The $200 DCX on the other hand is ripe for the choppin'.

for the analog (as bruce wants to do) the input to the ADC needs to be addressed if that whole board is going to be scrapped. Actually my intent is to revisit the digital X-over as a tool to see about where my passive line level needs to be set and that process should teach us a bit if i ever get my analog rig running. Until that point I am 100% digital so keeping thins there only makes sense. (plus the DCX will do a 6dB slope where my JBL will not)

dave
DuoBruce
Posts: 91
Joined: Mon Mar 07, 2005 12:17 am

Post by DuoBruce »

I am interested in both inputs actually. I plan to use a DEQ2496 in front of the DCX for room correction. I'm pretty sure the analog in on the DEQ is the same as the DCX so it will work on either. I'll feed the balanced digital out of the DEQ to the DCX so a better digital in on the DCX would be good.

I'm going to stream iTunes wirelessly to an Aiport Express and use the toslink to the DEQ

I also see people using jitter reduction boxes like the Monarchys between Behringers so it could be as complicated as the diagram
Attachments
deq dcx block.JPG
deq dcx block.JPG (29.24 KiB) Viewed 27206 times
Bruce Bosler
jeff mai
Posts: 15
Joined: Sat May 06, 2006 11:57 am

Post by jeff mai »

I'd likely start with something similar to the last schematic in that thread I linked. The video op-amp in the first schematic is probably asking for trouble. Use 75ohm BNC connectors for the external interface and proper 75ohm cable. Bypass the relay on the input board and the pulse transformer and 110R resistor on the DSP board of the DCX.

As for the analog inputs, I hadn't thought about those. I have no analog sources. You could use something else to do the ADC (a soundcard?) and go digital in to the DCX. I'd likely keep/duplicate the microphone preamp on the C input so I can still use the auto-align function.

If you were going to do an analog input, you'd want more gain than the stock unit - it expects pro level signals. You might get away with a completely passive input. I can't immediately tell from the schematic, but I doubt the ADC operates at pro signal levels. The thing probably throws away a lot of gain. All of the level controls are in the DSP section, so you'd want to get the right level at the ADC input for maximum resolution.
DuoBruce
Posts: 91
Joined: Mon Mar 07, 2005 12:17 am

Post by DuoBruce »

Yes, pro levels at the input but the ADC gets much less, I believe only a 5Vpp swing is max at the chip


As for the digital in, I skimmed through that link and saw 2 opposing views, the guy who proposed it who admitted he was trying to do RF design work with only a multimeter, and Gordon Rankin who didn't think much of it. I am not an RF engineer or digital expert but I would defer to Gordon.
Bruce Bosler
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

had the unit back for a few days and did some more listening this morning. I knew the difference in gain between the two outputs was around 9dB and the thing that kept bothering me was i could not "match" levels since my attenuator has only 2dB steps.

I decided to attempt to simply match levels by ear and consistently came up with the opamp outputs needing 10-12dB less attenuation that the transformer outs. Remembering the difference was less, i measured the outputs with some pink noise at -20db and got .39V for the opamps and .15 for the transformers. This comes out to 8.3dB difference, so what i heard with music was about 3dB off from what is actually happening.

This is the same type of thing i hear going from the resistive to the autoformer attenuators. Even though you are playing it louder it doesn't seem like it. When i set a max volume level with the transformers, and then attenuated 8dB and went to the opamps the sound drove me from the room. Another 4dB of attenuation seemed to bring things back in line.

One other thing of interest to note is that the output with the transformers and the -20dB of random noise is just about perfect. Add the 20dB and your peak output is 1.5V. feeding my 437-2A3 amp wiht lowthers puts the 0dB attenuation peaks just past clipping. I was typically listening at -6dB (which was loud) and thins sounded great!

dave

dave
kumbal
Posts: 1
Joined: Mon Dec 25, 2017 7:10 pm

Post by kumbal »

[quote="DuoBruce"]Yes, pro levels at the input but the ADC gets much less, I believe only a 5Vpp swing is max at the chip


As for the digital in, I skimmed through that link and saw 2 opposing views, the guy who proposed it who admitted he was trying to do RF design work with only a multimeter, and Gordon Rankin who didn't think much of it. I am not an RF engineer or digital expert but I would defer to Gordon.[/quote]

You scum of the earth, you pedophile. You think you can get away with it??
Post Reply