Re: Modding The DCX2496
Hi all,
Just a short note to let you know that I replaced the "el cheapo" caps as described above by Oelhrich and I have to say it makes a BIG difference
Effect is exactly as described. I'll add that sound is more relaxed and natural, specially with female voices.
Only thing that bothers me a bit is that the BG-Nx are not polarized , unlike the original caps, and their voltage is 6,3v instead of the 25v original "el cheapo" caps. I hope the unit wont catch fire
Thanks Charly, big bang for the bucks
.
oehlrich said:
Hi all,
Just a short note to let you know that I replaced the "el cheapo" caps as described above by Oelhrich and I have to say it makes a BIG difference
Effect is exactly as described. I'll add that sound is more relaxed and natural, specially with female voices.
Only thing that bothers me a bit is that the BG-Nx are not polarized , unlike the original caps, and their voltage is 6,3v instead of the 25v original "el cheapo" caps. I hope the unit wont catch fire
Thanks Charly, big bang for the bucks
.
Davey said:All the speculation in this thread is......amusing.
good for you!
the VCA tracking is IMO the only reason to prefer the VCA solution above passive control.
Or you must be deeply amused by building VCA circuits....
Anyway, the "dave reite" unit looks very well made....
It could use some post VCA padding (-20dB switch), to be in a more substantial VCA voltage-range for nominal listening levels...
dokter dB said:
It could use some post VCA padding (-20dB switch), to be in a more substantial VCA voltage-range for nominal listening levels...
More speculation?
I don't believe a post-vca pad to be necessary in most cases, and certainly not 20db. Since the VCA is already operating in the optimum voltage/distortion range (assuming the input levels are correctly optimized) any output attenuation won't improve the situation, but it will add considerably to the possibility that the SSM2018 internal output buffer might be driven into clipping.
Users with high efficiency speaker systems and/or power amps with higher than normal voltage gains might find it necessary to change the range of the control or attenuate (externally) slightly the output (maybe 6db,) but I don't think others will need to.
Source levels vary quite a bit with various CD's and vinyl via preamp so if you're a person who obsesses about these kinds of things you'd be looking for (in each case) an optimum setting for relative levels on the DSP-crossover, pre-VCA padding, post-VCA padding, etc. That would be a tedious pain-in-the-butt for most folks so I selected a nominal range and offset for the VCA control that should work well in the majority of situations and still prevent clipping at all stages within the gain structure of the system.
Cheers,
Davey.
Davey said:
I thought (no speculation, this was in favour of resolution) we would be going into the dcx at full level, preferably spdif/aes. In this case the level differences between cd's are in a way that we can't do anything about it, besides staying right under input clipping. This point is fixed for every played CD. turntable must have more safetyheadroom off course
Outputwise I agree with the possible clipping because the max outputlevel is uncertain/unfixed. It depends on how much of that frequencyband is present in the material.
Still the dcx has a fixed outputmaximum (about +20dBU), and if the VCA is set-up for attenuation only (wich I think is the right way to do it), the VCA can't add any clip, if there is enough supply voltage.
To me the padswitch of 20dB is a very right value to keep the volumepot in a more middle range with low listening levels.
Another option is to make a switch for a different potmeter-resolution (quasi 1/3log to 1/10 log with parallel resistor, or just a simple dc bias shift) instead of a pad.
Your gainstructure asks for quite a low gain in the power amps(wich can be absolute right too), or else you will mostly be using the volume pot practical closed.
I think we mean sort of the same or is that speculation
Is there available a design including pcb layout- for a passive relay based attenuator along the the lines of the apox for example? -and using readily available components.
I do not understand digtal electronics and the mention of binary encoders makes my eyes glaze over.
A question- , with all the digital controllers driving the relays , is any sort of programming required or do these things just work when they are assembled correctly (much like analogue circuits for example)
Thanks in advance
I do not understand digtal electronics and the mention of binary encoders makes my eyes glaze over.
A question- , with all the digital controllers driving the relays , is any sort of programming required or do these things just work when they are assembled correctly (much like analogue circuits for example)
Thanks in advance
Doc,
I used a +/-12 volt supply for the VCA's which yields a clipping point on their input portion of about 4.2Vrms. Obviously the DCX can output output more voltage (especially if a XLR/RCA converter is used because it increases gain 6db in the DCX) than this so the user has to be careful with gain settings, EQ boosting, etc, to not exceed this voltage.....or else insert pads in between.
My power amps have 28db gain, my speakers are medium efficiency, and I'm generally always above the -20db (12 o'clock position) and sometimes close to 0db (fully CW) with normal listening on a variety of sources. A 20db pad on the output of the VCA's would never be used in my situation.
I generally use the analog inputs on the DCX because I have a number of SACD's, but switching to the AES input does increase the levels inside the DCX somewhat. 0dbFS recorded signals have the potential to cause clipping inside the DCX and/or at the input of my VCA control so I use some attenuation of the input slider to prevent this.
Davey.
I used a +/-12 volt supply for the VCA's which yields a clipping point on their input portion of about 4.2Vrms. Obviously the DCX can output output more voltage (especially if a XLR/RCA converter is used because it increases gain 6db in the DCX) than this so the user has to be careful with gain settings, EQ boosting, etc, to not exceed this voltage.....or else insert pads in between.
My power amps have 28db gain, my speakers are medium efficiency, and I'm generally always above the -20db (12 o'clock position) and sometimes close to 0db (fully CW) with normal listening on a variety of sources. A 20db pad on the output of the VCA's would never be used in my situation.
I generally use the analog inputs on the DCX because I have a number of SACD's, but switching to the AES input does increase the levels inside the DCX somewhat. 0dbFS recorded signals have the potential to cause clipping inside the DCX and/or at the input of my VCA control so I use some attenuation of the input slider to prevent this.
Davey.
@davey:
Seems like a fair setup...
You mention the 4.2VRMS max input value:
This is indeed what I remember about the 2018, (used it once about 6 years ago...)
This is also why I did not continue using it in the design that time (bal input stage with limiter, impossible to clip etc).
To be clear: this does not mean that I find it not a good choice for other designs (like this one)
If the bal input stage is dual inverter (both 18k resistors connected both to a virtual earth on the chip-side) and you double the two input resistors (this means make them 36k), the max input voltage will go up to 8.4Vrms, and input level goes 6dB's down. It will lead to a bit of extra noise (resistors raised), but I think it will not be substantial (thus audible) compared everyting else that is pre VCA.
I don't remember I tried this, but I suspect the input of being an ordinary single opamp stage, with a fixed feedback and +ground resisitor (both 18k) wich forces the design into using 18k input resistors. If not done so, CMRR degrades heavily, because bal. transmission impedances are not equal, and there is no use of the bal connection anymore.
If the input is indeed single opamp, it would IMO be ok to use it unbalanced on the inverted input. The above resistor raise/gain down/headroom-up will work fine that way. The VCA-Output will be inverted but that can be fixed by rotating the dcx outputs. It would be interesting to find out what kind of input stage the 2018 has, data sheet doesn't mention as far as I can see, and I don't remember....
In my opinion The VCA part/input should have a clipping point of +28dBU, if connected balanced. Now it is +14dBU.
This personally would make me want to design my own bal. input, and use VCA's without one....
I am not aware of the nominal dist. figures of the 2018, it is possible that they are better if used on lower levels like 1Vrms, and that is maybe the reason of AD for the low headroom....
Since it is used unity gain here, to me +14dBU clip is too low.
Seems like a fair setup...
You mention the 4.2VRMS max input value:
This is indeed what I remember about the 2018, (used it once about 6 years ago...)
This is also why I did not continue using it in the design that time (bal input stage with limiter, impossible to clip etc).
To be clear: this does not mean that I find it not a good choice for other designs (like this one)
If the bal input stage is dual inverter (both 18k resistors connected both to a virtual earth on the chip-side) and you double the two input resistors (this means make them 36k), the max input voltage will go up to 8.4Vrms, and input level goes 6dB's down. It will lead to a bit of extra noise (resistors raised), but I think it will not be substantial (thus audible) compared everyting else that is pre VCA.
I don't remember I tried this, but I suspect the input of being an ordinary single opamp stage, with a fixed feedback and +ground resisitor (both 18k) wich forces the design into using 18k input resistors. If not done so, CMRR degrades heavily, because bal. transmission impedances are not equal, and there is no use of the bal connection anymore.
If the input is indeed single opamp, it would IMO be ok to use it unbalanced on the inverted input. The above resistor raise/gain down/headroom-up will work fine that way. The VCA-Output will be inverted but that can be fixed by rotating the dcx outputs. It would be interesting to find out what kind of input stage the 2018 has, data sheet doesn't mention as far as I can see, and I don't remember....
In my opinion The VCA part/input should have a clipping point of +28dBU, if connected balanced. Now it is +14dBU.
This personally would make me want to design my own bal. input, and use VCA's without one....
I am not aware of the nominal dist. figures of the 2018, it is possible that they are better if used on lower levels like 1Vrms, and that is maybe the reason of AD for the low headroom....
Since it is used unity gain here, to me +14dBU clip is too low.
Doc,
I think you're getting confused on the capabilities of the DCX and how it works internally. It's not the analog output section that is determining its maximum capabilities, but rather the internal DSP processing.
Assuming 0dbFS on the digital input the DCX will output 4.26Vrms (8.5V balanced) from either plus/minus output relative to ground before (digital) clipping within the DSP architecture. This is the maximum level the DCX can be programmed for.
The analog output stage is where the increased voltage capability takes place so you can end up with 8.5Vrms balanced or single-ended if you short pin 1 to 3 with a XLR/RCA adaptor. So, you have a hard limit of about +21dbu that the DCx can output in any configuration.
With my +/-12V supply the VCA chips can't tolerate more than about 4.2Vrms on either input pin (relative to ground) without clipping. Applying 8.5Vrms to a single input unbalanced will clip hard the input portion of the VCA chip. Assuming a balanced connection (which I use) you can see that the maximum output capabilities of the DCX and maximum input capabilities of the VCA chips are fairly well matched. A pad could be used between DCX output and VCA input, but it would only have to be about 1db to certainly prevent clipping.
With my current +/-12volt configuration the DCX is presenting the maximum voltage possible to the inputs of the VCA chips. I'm not sure what is unclear about my description. You appear to be under the impression that there is more signal voltage available from somewhere......there isn't.
I suppose if I added an external signal amplifier between the DCX and VCA's, and raised my VCA supply voltage to +/-15volts a higher signal voltage level could be maintained, but I don't see the point because it doesn't accomplish anything. The distortion performance of the VCA chips would not improve and they're already easily capable of providing a voltage output to drive any power amplifier to its rated output.
I investigated all of these voltage/distortion/padding/etc concerns before building the prototype initially. Also, there are a whole series of distortion performance graphs in the SSM2018T data sheet. As you would expect, distortion performance varies somewhat depending upon gain setting, supply voltage, input signal level, etc. The input toplogy is also clearly shown. Please read the SSM2018T data sheet before commenting further.
Ouroboros,
You're correct about the cost.....not cheap. $7.10 each here in the States.
Cheers,
Davey.
I think you're getting confused on the capabilities of the DCX and how it works internally. It's not the analog output section that is determining its maximum capabilities, but rather the internal DSP processing.
Assuming 0dbFS on the digital input the DCX will output 4.26Vrms (8.5V balanced) from either plus/minus output relative to ground before (digital) clipping within the DSP architecture. This is the maximum level the DCX can be programmed for.
The analog output stage is where the increased voltage capability takes place so you can end up with 8.5Vrms balanced or single-ended if you short pin 1 to 3 with a XLR/RCA adaptor. So, you have a hard limit of about +21dbu that the DCx can output in any configuration.
With my +/-12V supply the VCA chips can't tolerate more than about 4.2Vrms on either input pin (relative to ground) without clipping. Applying 8.5Vrms to a single input unbalanced will clip hard the input portion of the VCA chip. Assuming a balanced connection (which I use) you can see that the maximum output capabilities of the DCX and maximum input capabilities of the VCA chips are fairly well matched. A pad could be used between DCX output and VCA input, but it would only have to be about 1db to certainly prevent clipping.
With my current +/-12volt configuration the DCX is presenting the maximum voltage possible to the inputs of the VCA chips. I'm not sure what is unclear about my description.
You appear to be under the impression that there is more signal voltage available from somewhere......there isn't. I suppose if I added an external signal amplifier between the DCX and VCA's, and raised my VCA supply voltage to +/-15volts a higher signal voltage level could be maintained, but I don't see the point because it doesn't accomplish anything. The distortion performance of the VCA chips would not improve and they're already easily capable of providing a voltage output to drive any power amplifier to its rated output.
I investigated all of these voltage/distortion/padding/etc concerns before building the prototype initially.
Also, there are a whole series of distortion performance graphs in the SSM2018T data sheet. As you would expect, distortion performance varies somewhat depending upon gain setting, supply voltage, input signal level, etc. The input toplogy is also clearly shown. Please read the SSM2018T data sheet before commenting further.Ouroboros,
You're correct about the cost.....not cheap. $7.10 each here in the States.
Cheers,
Davey.
Davey said:
With my current +/-12volt configuration the DCX is presenting the maximum voltage possible to the inputs of the VCA chips. I'm not sure what is unclear about my description.
Cheers,
Davey.
You are absolutely right, I thought there was about 12dB extra headroom to win in your setup (for free), but it seems that there isn't!
Your explanantion means that everything is well matched. With +/-15V on the VCA you are probably done for that missing dB...
How do you manage the level diminishing with filtered outputs, just gain them up digitally I suppose?
Cheers...
Re: Re: Modding The DCX2496
Hello again,
I started the moding of the caps. Thmartin also did it with the same good effects. He asked if the low voltage of only 6.3V using the BGs will lead to problems.
I ran the modded unit some wheeks now and the low voltage of the caps REALLY LEAD TO PROBLEMS! During regular operation there is no DC at the caps. But during power on there are max. +-15volts at the caps. This seems to make a problem over the time. My caps begin making a lot of noise at startup. I have to switch the power on and off more than once till the noise is gone. But it returns during operation. It is not only one cap. The noise moves between the channels.
From Thmartin I got the schematics of the DCX2496 and I studied it. The good news: The caps are not really necessary! They block a theoretical offset voltage from the D/A converters. This voltage is that low (can not measure ist with my multimeter) that I can live with it. So I inserted WIRES instead of the caps. It works! And no cap is even better than a BG cap ;-))
The very litte DC offset at the output of the DCX2496 which can appear, is blocked by the caps in my amplifier so this mod works well for me.
I believe the mod is a must. It has a big effect at NO COST (*gg*)
Charly
Thmartin said:
Hi all,
Just a short note to let you know that I replaced the "el cheapo" caps as described above by Oelhrich and I have to say it makes a BIG difference
Effect is exactly as described. I'll add that sound is more relaxed and natural, specially with female voices.
Only thing that bothers me a bit is that the BG-Nx are not polarized , unlike the original caps, and their voltage is 6,3v instead of the 25v original "el cheapo" caps. I hope the unit wont catch fire
Thanks Charly, big bang for the bucks
.
Hello again,
I started the moding of the caps. Thmartin also did it with the same good effects. He asked if the low voltage of only 6.3V using the BGs will lead to problems.
I ran the modded unit some wheeks now and the low voltage of the caps REALLY LEAD TO PROBLEMS! During regular operation there is no DC at the caps. But during power on there are max. +-15volts at the caps. This seems to make a problem over the time. My caps begin making a lot of noise at startup. I have to switch the power on and off more than once till the noise is gone. But it returns during operation. It is not only one cap. The noise moves between the channels.
From Thmartin I got the schematics of the DCX2496 and I studied it. The good news: The caps are not really necessary! They block a theoretical offset voltage from the D/A converters. This voltage is that low (can not measure ist with my multimeter) that I can live with it. So I inserted WIRES instead of the caps. It works! And no cap is even better than a BG cap ;-))
The very litte DC offset at the output of the DCX2496 which can appear, is blocked by the caps in my amplifier so this mod works well for me.
I believe the mod is a must. It has a big effect at NO COST (*gg*)
Charly
Are you fellows also aware there are six capacitors (two each on A, B, and C) that can probably be removed? This is assuming your source has minimal DC offset.
There is also another capacitor in the input section. DON'T remove that one since it blocks the 2.5vdc level shifting for the A/D converters.
Cheers,
Davey.
There is also another capacitor in the input section. DON'T remove that one since it blocks the 2.5vdc level shifting for the A/D converters.
Cheers,
Davey.
Davey said:
Hi Davey,
looking at the schematics this mod may be possible but I will not really recommend it. The six caps (guess you're talking of C26, 27, 41, 42, 8 and C9) isolate the input OPAmps from the input connectors. I didn't have a good feeling to bridge them. It's possible yes. But any DC or high voltage on the input will go directly into the OPAmps and may destroy them.
There are another tree coupling caps in the input stage (C6, 13, 28) but from the schematics I would say they carry a DC offset same as C91. Do NOT bridge them!
Charly
Judging by the schematic, there appear to be 3 47uF caps each for A,B,C. For input A they would be C26, C41 and C6, that last one coupling two stages together.
Personally what I think is really needed is a whole new input/output board, directly coupled, with on board volume control for the output, probably single ended output is all we'd need. Looking at the Behringer output section compared to the AKM data sheet, theres a lot of extra stuff going on. After the holidays are over, I'm going to talk to AKM tech support and see what they really suggest is needed after the AK4393. In one schematic they show a very simpler summer, in another a more complicated multi opamp thing for balanced out. The simple summer doesn't appear to have a low pass filter in it, but I don't understand something about the actual chip, it appears that it might have some sort of on board filtering, a switched capacitor? Anyone know about this?
Ron
Personally what I think is really needed is a whole new input/output board, directly coupled, with on board volume control for the output, probably single ended output is all we'd need. Looking at the Behringer output section compared to the AKM data sheet, theres a lot of extra stuff going on. After the holidays are over, I'm going to talk to AKM tech support and see what they really suggest is needed after the AK4393. In one schematic they show a very simpler summer, in another a more complicated multi opamp thing for balanced out. The simple summer doesn't appear to have a low pass filter in it, but I don't understand something about the actual chip, it appears that it might have some sort of on board filtering, a switched capacitor? Anyone know about this?
Ron
I thought I was very clear when I said "asuming your source has minimal DC offset." And I also said DO NOT replace the other capacitors (C6, C13, C28 as you noted) because they have a 2.5 volt DC potential across them. Obviously there is a concern with bridging capacitors. I think everyone is aware of that.
Ron,
There is definately more circuitry in the analog sections of the DCX than most purists would be happy with, but it is what it is and modificationi/replacement/etc are all options.
Cheers,
Davey.
oehlrich said:
Hello,
what will be your opinion on replacing complete analog output part with tube pre amps such as Bottleheads Foreplay. I have California Audio tube DAC and I have to admit that my biggest obstacle in full switching to Behringer would be the loss of that great tube part within the DAC. My feeling is that digital portion of the 2496 is quite fine, but analog circuitry could be improved. Is full preamp gain what is needed or just a buffer stage?
AR2
what will be your opinion on replacing complete analog output part with tube pre amps such as Bottleheads Foreplay. I have California Audio tube DAC and I have to admit that my biggest obstacle in full switching to Behringer would be the loss of that great tube part within the DAC. My feeling is that digital portion of the 2496 is quite fine, but analog circuitry could be improved. Is full preamp gain what is needed or just a buffer stage?
AR2