Yep, I was thinking the same thing. Just take the signal after the 1st opamp. Maybe even change the gain of that stage. It's all we really need for consumer gear.
I have one of those remote multi-gang pots, too. Never used it because it's 100K. They don't seem to come in other values.
Gary, thank you for the detailed description. Quite possibly that I did exactly as you explained - too fast and furious. 
I will try with my next project this method. I was really disappointed, after I finished soldering with hot air gun, that something was wrong. So I went back to unsolder and start from scratch and to find out balls all over under the DACs belly. So I gave up, which pushed me to become very precise and effective with soldering and it worked very well. Working just with hot air gun will be much easier, cannot wait to try!
Thank you
I will try with my next project this method. I was really disappointed, after I finished soldering with hot air gun, that something was wrong. So I went back to unsolder and start from scratch and to find out balls all over under the DACs belly. So I gave up, which pushed me to become very precise and effective with soldering and it worked very well. Working just with hot air gun will be much easier, cannot wait to try!Thank you
You know, I never understood why Behringer has such a complicated circuit, where they go from balanced to unbalanced and back to balanced output circuit. Add on the top of that electrolytic cap?
I am sure that this mod would crete big improvement over the existing circuit. Now I am not fond of opamps, but the fact is replacing opamp and simplifying the whole thing on the existing board is relatively easy and rewarding task. Any other way requires separate boards, tiny wires, maybe another case...
You know I would be interested in learning what do you find after these improvements. My experience is that any work on this PS doesn't make a big difference. I tried very good analog PS instead of switcher and I didn't find any improvements worth mentioning. What I believe will make impact is working on regulators and caps after the switcher, the ones that are on the DSP board. Replacing basic 7805 regulator with something more sophisticated would be good. Next I replaced all bypass caps, 10uF 25V for equal value OSCONS. OSCON doesn't recommend adding any additional lower value baypass cap next to the main 10uF and that would be hard to do anyways on that crowded board. So that should be good improvement to try. I cannot say that I could hear improvement since I didn't do A/B listening tests, but while I was working it was easy to do that as well. I also did that with all other improvements, so I didn't have isolated step by step progress in order to pinpoint and value single change.
I mentioned before, there are two mods that are very audible and that are making a big difference. The output circuitry mod and new digital input circuit with SRC and new clock with dedicated low noise regulator which changes your DAC into asynchronous DAC. That is hard mod but worth every bit of effort.
Thanks for the info. I will contribute my findings once I can get to work.
Any good drop ins for 7805 regulator that one could sample or buy? Dexa uwb would be great but IMO they are too expensive for regulators.
Are the two in the PSU (agains the aluminum socket with thermal paste) also voltage regulators? Any good drop ins for those?
There is a lot to consider about interfacing two circuits that do not share (exactly) same mass potential.
Best is to work through those "easy to read" papers:
http://www.analog.com/static/imported-files/data_sheets/ssm2142.pdf
http://www.analog.com/static/imported-files/data_sheets/ssm2141.pdf
http://www.audiodesignline.com/howto/audioprocessing/196604119
Bottom line :
In the context of the DCX analog board, the topology chosen by Behringer is pretty much "state of the art" – meaning :_
– they first reference the DAC output to the supposedly less noisy mass potential of the analogue board in order to not transmit any unnecessary common mode HF noise form digital board mass over the cables to the amps (this is done by going asymmetric first)
– then they implemented a floating differential out which has some benefits about non floating topology.
Very clever and unbeatable in its performance if implemented well.
There's some irony that the beauty of that approach is widely misunderstood in this thread.
Sadly, Behringer missed to do adequate analogue filtering at first, use closely matched resistors and to work out adequate trace layout in the analogue section.
This done (plus solving the 'lytics and IC's PSRR issue) I'm sure – no one ever would have felt a need to mod the analog section.
Michael
Last edited:
Hi Gary and Michael,
100k is quite common in this applications, why do you consider it too high in this case?
Ciao!
Paolo
100K works fine inside a tube amp. But in this case (and I think Gary's), we are driving medium impedance solid state amps thru a bit of cable. My amps have an input impedance of 20K and 10K. The output impedance of that 100K volume control does not allow much current to drive what is after it.
It will work, but I've found that lower impedances work much better for me. I'd be happy with 5K, but 10K seems to work OK.
It will work, but I've found that lower impedances work much better for me. I'd be happy with 5K, but 10K seems to work OK.
If you did a cap-coupled direct out mod from the DAC, you could then run it into this volume control, which would allow pretty small and comfortable film caps to be used, and follow it up with three LME49720's as buffers (unbalanced). Could be a nice end result.
cap coupled direct out mod
Hi!
so the high pot value is meaningful for the effects in combination with the following load. That's my feeling too, for this reason I am going to install a buffer after the pot.
Now my problem is how to connect the balanced output of the dac to the pot in a single ended mode using a coupling capacitor. The only way I see now is to use only the positive output of the dac, but I don't see this solution as the best, this means I would loose the benefits of the balanced output of the dac. Is there any other feasible way to connect them?
Ciao
Paolo
Hi!
so the high pot value is meaningful for the effects in combination with the following load. That's my feeling too, for this reason I am going to install a buffer after the pot.
Now my problem is how to connect the balanced output of the dac to the pot in a single ended mode using a coupling capacitor. The only way I see now is to use only the positive output of the dac, but I don't see this solution as the best, this means I would loose the benefits of the balanced output of the dac. Is there any other feasible way to connect them?
Ciao
Paolo
Just using the + side of the DAC, maybe with a light load resistor on the - side, is fine... and I believe is what a lot of guys doing "direct out" mods do.
Another option, which is a lot more complicated, is to use a differential amplifier to convert the DAC output to single-ended. This would be a good quality op amp and high tolerance (like .1%) resistors to allow the op amp to strip the 2.5V common DC offset.
I suppose you could also install two pots and do your volume control in balanced mode, then convert to unbalanced in your output buffer. There you would still probably want to decouple the DC with caps, since the tracking error of the pots would make it difficult to differentially strip the DC after the pots.
One thing to note for any of the "direct out" implementations is that you should, I believe, switch to 4395's or 4396's, as these DACs do not really need an external LP filter. The 4393's have more conversion noise which should be filtered off and is filtered off in the stock DCX output topology.
Another option, which is a lot more complicated, is to use a differential amplifier to convert the DAC output to single-ended. This would be a good quality op amp and high tolerance (like .1%) resistors to allow the op amp to strip the 2.5V common DC offset.
I suppose you could also install two pots and do your volume control in balanced mode, then convert to unbalanced in your output buffer. There you would still probably want to decouple the DC with caps, since the tracking error of the pots would make it difficult to differentially strip the DC after the pots.
One thing to note for any of the "direct out" implementations is that you should, I believe, switch to 4395's or 4396's, as these DACs do not really need an external LP filter. The 4393's have more conversion noise which should be filtered off and is filtered off in the stock DCX output topology.
Regarding analog volume control I again highly recommend to get the point of the papers linked in my previous posting.
The good old SSM-xxx units are pretty "ancient" designs but the outlined theory of operation is covering each and any aspect of audio interfacing in a understandable way.
The paper of the InGenius units from Bill Whitlock tells the same story (in other words) plus emphasis on improvements possible in the presence of source imbalances.
You may have to read twice or three times - but once you got it - it definitely and permanently will change your view about a very important point in audio electronics design .
I bring it up again as it seems there is no wide spread understanding what is done in the stock unit and on the other hand, many like to improve exactly on that point of the DCX.
In *any* case of implementing analog volume control in the DCX you first will have to understand *why* actually it is, that there are DAC's with complimentary output in order to not compromise signal ?
The simple reason is that signal integrity can be preserved way better with complimentary-out than with single out - *if* handled right.
Imagine the DAC board to have veeeery bad mass traces - and from that the relatively small out-voltages would be veeeeery much messed up by all the digital currents floating around on that board.
This way you immediately can see the benefits of dual out - if processed properly by either a differential receiver (stock circuit) or a transformer (would do almost equally well on a single out tho).
What does *not* work that well is a simple pot attenuation referenced to ground for one or both outputs - meaning you nevertheless will be in need for a balanced topology afterwards.
As for the value of the cheap 100k pot I would have to do a quick calculation - but my rough guess would be that you add noise more than you possibly like (given the levels involved)
#######
Just had a lookup on the resistor noise diagram and it turns out that for 50kOhm (in the middle position) you "earn" a whoppy ~ 20nV /Hz^-2
Well no need for LME49720 (2.7 nV /Hz^-2) then ??!!
Michael
The good old SSM-xxx units are pretty "ancient" designs but the outlined theory of operation is covering each and any aspect of audio interfacing in a understandable way.
The paper of the InGenius units from Bill Whitlock tells the same story (in other words) plus emphasis on improvements possible in the presence of source imbalances.
You may have to read twice or three times - but once you got it - it definitely and permanently will change your view about a very important point in audio electronics design .
I bring it up again as it seems there is no wide spread understanding what is done in the stock unit and on the other hand, many like to improve exactly on that point of the DCX.
In *any* case of implementing analog volume control in the DCX you first will have to understand *why* actually it is, that there are DAC's with complimentary output in order to not compromise signal ?
The simple reason is that signal integrity can be preserved way better with complimentary-out than with single out - *if* handled right.
Imagine the DAC board to have veeeery bad mass traces - and from that the relatively small out-voltages would be veeeeery much messed up by all the digital currents floating around on that board.
This way you immediately can see the benefits of dual out - if processed properly by either a differential receiver (stock circuit) or a transformer (would do almost equally well on a single out tho).
What does *not* work that well is a simple pot attenuation referenced to ground for one or both outputs - meaning you nevertheless will be in need for a balanced topology afterwards.
As for the value of the cheap 100k pot I would have to do a quick calculation - but my rough guess would be that you add noise more than you possibly like (given the levels involved)
#######
Just had a lookup on the resistor noise diagram and it turns out that for 50kOhm (in the middle position) you "earn" a whoppy ~ 20nV /Hz^-2
Well no need for LME49720 (2.7 nV /Hz^-2) then ??!!
Michael
Last edited:
As a follow up, and to provide one other option, you could tap in to the stock circuitry at or around points C43, C44, C45, C46, C47, and C48. These caps are there to block any DC passed by the diff amp / LP filter that converts the DAC diff outputs to single-ended. Ideally there shouldn't be any DC in the output of the op amp that feeds these caps, but, since the parts aren't high tolerance, differences in the actual resistor values will allow for some DC here, which is then blocked by the caps.
Anyway, you could install better coupling caps here and feed this single-ended signal into your pot. You could also pick up the signal just the other side of the 499 Ohm resistor fed from these caps, if you want to keep the power-on analog muting in place.
By picking up the signal here, you don't need to swap DAC's, and you preserve the common-mode noise benefits of the DAC differential output that Michael was (I think) talking about.
Not trying to peddle my wares here, but in case you guys missed it, a lot of people use my unit, which includes a level display, remote control and a couple of other goodies:
http://www.linearaudio.nl/6chan-2.htm
jd
http://www.linearaudio.nl/6chan-2.htm
jd
Yes, that would exactly be the right point to easily insert a pot and not compromising the benefits of stock topology!
But there is a drawback:
The two 499 Ohm resistors (plus mute) are best to be seen as the input impedance of the following stage. Also – this resistors set the gain of the following stage (together with the feedback ressitor)
So – you would have to simulate first to get the curve of attenuation – or use a pot thats resistance is pretty low compared to the 2x499Ohm (which is not really feasible)
Michael
But there is a drawback:
The two 499 Ohm resistors (plus mute) are best to be seen as the input impedance of the following stage. Also – this resistors set the gain of the following stage (together with the feedback ressitor)
So – you would have to simulate first to get the curve of attenuation – or use a pot thats resistance is pretty low compared to the 2x499Ohm (which is not really feasible)
Michael
I remember you having developed these – I love the digital volume display !
Have a schematic somewhere?
Michael