Dear Leon,
In that case, build a nice enclosure with 3 amplifiers and place it right behind the speaker. I would suggest to use the in my opinion very nice Speakon connectors from Neutrik. You can put multiple speaker wiring into one connector. The connection is very tight and firm, and those connectors ensure a very solid contact.
Neutrik - Audio - speakON STX Series
With kind regards,
Bas
Are you a jeweller, a galvaniser or did a few kilos of silver just fall off a lorry in front of your house?
Hi,
It has got both analogue and digital inputs. Note, the analog outputs have a level of nearly 8V for digital full scale and way too many Op-Amp's to sound good. There are however tons of documented modifications for the DCX and quite a bit of support as well.
I remember these, they where not bad at all. Pretty hard to get into and modify though.
Yes, this is an option. However it makes controlling the volume more difficult. As suggested before, i would consider still controlling the volume in the analog domain, after the DCX2496.
An extreme option would be to use the Volume Control Modules I recently designed with the DS1666 (this digitally controlled potentiometer is also used by Audio Research in all their Preamplifiers including the top of the line Reference ones). For this module I included the option to add a J-Fet buffer, so the whole thing basically becomes a buffered passive preamp. Six channels can be easily build up.
Not wanting to "advertise product" (esecially as the cost is not low), but the details are here, in case you are interested:
Ultimate Volume Control
I am currently listening to a Preamp using this and one of the Tube stage boards (Universal Stereo Tube Output Stage and it does not seem to give up much to the passive with the Noble Pot on immediacy and transparency, but it sounds a little more dynamic and "musical" (da toobz sound).
Ciao T
It has got both analogue and digital inputs. Note, the analog outputs have a level of nearly 8V for digital full scale and way too many Op-Amp's to sound good. There are however tons of documented modifications for the DCX and quite a bit of support as well.
I remember these, they where not bad at all. Pretty hard to get into and modify though.
Yes, this is an option. However it makes controlling the volume more difficult. As suggested before, i would consider still controlling the volume in the analog domain, after the DCX2496.
An extreme option would be to use the Volume Control Modules I recently designed with the DS1666 (this digitally controlled potentiometer is also used by Audio Research in all their Preamplifiers including the top of the line Reference ones). For this module I included the option to add a J-Fet buffer, so the whole thing basically becomes a buffered passive preamp. Six channels can be easily build up.
Not wanting to "advertise product" (esecially as the cost is not low), but the details are here, in case you are interested:
Ultimate Volume Control
I am currently listening to a Preamp using this and one of the Tube stage boards (Universal Stereo Tube Output Stage and it does not seem to give up much to the passive with the Noble Pot on immediacy and transparency, but it sounds a little more dynamic and "musical" (da toobz sound).
Ciao T
Hi,
What Bas did for the power supply is excellent. But I noticed he did not seem to have done much on the analog output side. A friend of mine uses a methode I suggested to him for modifying these:
DCX2496
It is really quite easy to do. The whole analog stage is replaced by transformers. Then you can build a smaller PSU that only feeds the DAC's (especially if only digital input) and a second small one for the DSP part.
Yes, I would suggest keeping it a seperate unit.
Many parts are quite microphonic. Capacitors are the worst offenders, but transistors and amp-chips also pick up and generate vibrations (you can hear the 'chirping' testing the amp with a swept sinewave). Any such microphonics will alter the parts properties to a degree, so now the electrons still behave as before, but the parts that process them no longer do.
The Bronze Heatsink thing is recommended by my friend Dieter Ennemoser. His views and approaches are HIGHLY controversial, but I find they work. You can read some more here:
The Freakzoid Tweekaloid Strikes Again by Thorsten Loesch
I may have originally recommended Bronze to Peter. I normally make do with copper because I have links to factories that work the stuff.
Probably not. It is highly magnetic (much more than steel - hence it's goodness in shilding transformers fields), not what you want close to where signal currents flow.
Silver tends towards a characteristic sound that is slightly lean, forward and bright. In tube based systems this is rarely a problem, in many cases the opposite is true. By comparison Copper sounds a little warmer, fuller and with a very slightly subdued top-end. This applies to solid core wires of comparable diameter (or foils of comparable size) and with reasonable purity (>99.97%).
Soft annealed Silver has less of the "shine", but retains some.
So it is a matter of synergy. When I working with Tube circuits I tend towards using components and surrounding items that tend towards a bright sound (but not as bright as Black Gate capacitors) to offset some of the "slow and laid back" sound that tubes often produce. For solid state the reverse hold, you want to use components that give a warm, full and relaxed sound to offset the usually bleached out and bright sound of solid state circuits.
In the end either designs can arrive at a pretty neutral sound that is close to reality, sounding neither "typical tube" nor "typical solid state", but much more "like music". With a neutral tonality and a sound that does not become stressful it is also much easier to immerse one self in the emotional content of the music.
Ciao T
What Bas did for the power supply is excellent. But I noticed he did not seem to have done much on the analog output side. A friend of mine uses a methode I suggested to him for modifying these:
DCX2496
It is really quite easy to do. The whole analog stage is replaced by transformers. Then you can build a smaller PSU that only feeds the DAC's (especially if only digital input) and a second small one for the DSP part.
Yes, I would suggest keeping it a seperate unit.
Many parts are quite microphonic. Capacitors are the worst offenders, but transistors and amp-chips also pick up and generate vibrations (you can hear the 'chirping' testing the amp with a swept sinewave). Any such microphonics will alter the parts properties to a degree, so now the electrons still behave as before, but the parts that process them no longer do.
The Bronze Heatsink thing is recommended by my friend Dieter Ennemoser. His views and approaches are HIGHLY controversial, but I find they work. You can read some more here:
The Freakzoid Tweekaloid Strikes Again by Thorsten Loesch
I may have originally recommended Bronze to Peter. I normally make do with copper because I have links to factories that work the stuff.
Probably not. It is highly magnetic (much more than steel - hence it's goodness in shilding transformers fields), not what you want close to where signal currents flow.
Silver tends towards a characteristic sound that is slightly lean, forward and bright. In tube based systems this is rarely a problem, in many cases the opposite is true. By comparison Copper sounds a little warmer, fuller and with a very slightly subdued top-end. This applies to solid core wires of comparable diameter (or foils of comparable size) and with reasonable purity (>99.97%).
Soft annealed Silver has less of the "shine", but retains some.
So it is a matter of synergy. When I working with Tube circuits I tend towards using components and surrounding items that tend towards a bright sound (but not as bright as Black Gate capacitors) to offset some of the "slow and laid back" sound that tubes often produce. For solid state the reverse hold, you want to use components that give a warm, full and relaxed sound to offset the usually bleached out and bright sound of solid state circuits.
In the end either designs can arrive at a pretty neutral sound that is close to reality, sounding neither "typical tube" nor "typical solid state", but much more "like music". With a neutral tonality and a sound that does not become stressful it is also much easier to immerse one self in the emotional content of the music.
Ciao T
So did i and found it audibly lacking in many ways but very convenient. Almost unbeliavable that ARC are using a mediocre $5 part in the most critical part of their designs but i guess their devoted customers expect no less.
Hi,
There are a few crucial issues to deal with, using the DS1666, to get good results. The Dual J-Fet buffer on that PCB is not for show. Low source and ultra-high load impedance are pretty much mandatory. Power supplies surprisingly (or not) anlso have an impact.
In testing we have for comparison TVC/AVC Preamps, ones with normal Pots (Alps Black, Noble & PEC Carbon) and with switched attenuators. I would not call the DS1666 lacking in comparison (especially to the pots and switched attenuators - the TVC's & AVC's are another story and group together and are generally a step up from all the resistive alternatives including DS1666).
We actually designed the DS1666 based Volume Control into the latest linestage for diyhifisupply, a kind of "statement" product DHT based linestage.
We would not have done this if it was not very good in it's own right... Actually, the first prototype had no input switching and an Alps Black Pot. The final version with the DS1666 and input switching is better sounding. So i do not think that the DS1666 itself is neccesarily as limiting sonically as you suggest.
For interest, the thread about this is here (also shows the earlier prototype):
http://www.diyaudio.com/forums/diy-hifi-supply/156854-dht-otl-linestage-tram-2-a.html#post2147566
Ciao T
There are a few crucial issues to deal with, using the DS1666, to get good results. The Dual J-Fet buffer on that PCB is not for show. Low source and ultra-high load impedance are pretty much mandatory. Power supplies surprisingly (or not) anlso have an impact.
In testing we have for comparison TVC/AVC Preamps, ones with normal Pots (Alps Black, Noble & PEC Carbon) and with switched attenuators. I would not call the DS1666 lacking in comparison (especially to the pots and switched attenuators - the TVC's & AVC's are another story and group together and are generally a step up from all the resistive alternatives including DS1666).
We actually designed the DS1666 based Volume Control into the latest linestage for diyhifisupply, a kind of "statement" product DHT based linestage.
We would not have done this if it was not very good in it's own right... Actually, the first prototype had no input switching and an Alps Black Pot. The final version with the DS1666 and input switching is better sounding. So i do not think that the DS1666 itself is neccesarily as limiting sonically as you suggest.
For interest, the thread about this is here (also shows the earlier prototype):
http://www.diyaudio.com/forums/diy-hifi-supply/156854-dht-otl-linestage-tram-2-a.html#post2147566
Ciao T
I think the DS1666 is broadly comparable to a conventional pot but with much better channel tracking and repeatable settings. It's sonic deficiencies are only obvious in comparison to a a good switch like a Shallco even when used with cheap metal film resistors.
In my design i used a reasonable but not "audiophile" PS: Muse filter caps and a 3-terminal regulator with sufficient local polyprop decoupling. No input buffer. Pretty much the way ARC use this and similar electronic pots in their preamps.
The suggestion for an input buffer is interesting although it would probably benefit any other type of pot.
Aren't the actual resistive elements inside electronic pots the real issue?
Hi,
I would say that the dficiencies need a pretty high end switch with more than cheap metal film resistors to be obvious, applyi ngthe "all else equal" dictum.
Try adding some chokes. You can see how I implemented it all in this picture:
Basically LC from the main raw supply, optimised 317/337 regulators and then RC filtering with Electrolytics and Wima caps at the chips.
i do not use an input buffer, though that too may not be a bad idea. I provision space for a LSK389 as output buffer (basically the old original 2 J-Fet buffer originated in the 70's and recently popularised by Nelson Pass as "B1", but in this case no coupling caps).
One may think so. But I think the Fet Switches are the bigger issue. And these switches like to see very high load impedances.
Ciao T
I would say that the dficiencies need a pretty high end switch with more than cheap metal film resistors to be obvious, applyi ngthe "all else equal" dictum.
Try adding some chokes. You can see how I implemented it all in this picture:
Basically LC from the main raw supply, optimised 317/337 regulators and then RC filtering with Electrolytics and Wima caps at the chips.
i do not use an input buffer, though that too may not be a bad idea. I provision space for a LSK389 as output buffer (basically the old original 2 J-Fet buffer originated in the 70's and recently popularised by Nelson Pass as "B1", but in this case no coupling caps).
One may think so. But I think the Fet Switches are the bigger issue. And these switches like to see very high load impedances.
Ciao T
Thank you for your opinion and advice T, very constructive as usual. Btw, what do you think of Mr Khomenko's white paper on attenuators?
Hi,
The one where they praise shunt mode use of electronic attenuators?
The PCB I reverenced above has a jumper that allows selection between shunt and standard attenuator mode and has through hole resistors for the job of the series resistor.
So everyone is free to try.
Due to the limited resistance of the Fet switches (always a few 100 Ohm it seems) the maximum attenuation is lower than in ladder mode. Measured distortion is up quite a bit. I am unsure about the sonics, to be honest it seems I slightly prefer standard mode in the tube preamp I have here using this volume control and the universal tube stage. It seems a little more open, dynamic and detailed, the shunt mode sounds in direct comparison very slightly "sat upon", but I suspect preferences will depend on both personal taste and prejudices.
Ciao T
The one where they praise shunt mode use of electronic attenuators?
The PCB I reverenced above has a jumper that allows selection between shunt and standard attenuator mode and has through hole resistors for the job of the series resistor.
So everyone is free to try.
Due to the limited resistance of the Fet switches (always a few 100 Ohm it seems) the maximum attenuation is lower than in ladder mode. Measured distortion is up quite a bit. I am unsure about the sonics, to be honest it seems I slightly prefer standard mode in the tube preamp I have here using this volume control and the universal tube stage. It seems a little more open, dynamic and detailed, the shunt mode sounds in direct comparison very slightly "sat upon", but I suspect preferences will depend on both personal taste and prejudices.
Ciao T
Actually, i find his claim that cmos switches are transparent, whereas diffusion type resistors are not, more interesting. An integrated cmos switch and discrete resistors seems easy and doable, no idea why i still haven't tried it. Laziness?
The shunt attenuator has always been an easy choice for me as a compromise between cost and quality - i only use one "nice" resistor. S/N obviously suffers though.
The shunt attenuator has always been an easy choice for me as a compromise between cost and quality - i only use one "nice" resistor. S/N obviously suffers though.
It could be an easy choice, but I don't think there is much advantage to it if less nice parts are being used. In other words, it seems like shunt section is equally important as the series for best sonics: http://www.diyaudio.com/forums/everything-else/94506-shunt-attenuator-myth.html
That would be about right.
Although, last time I checked Noble pot in a shunt application (with some fancy series resistor) I recall that pot by itself seemed to sound better. 😉
Although, last time I checked Noble pot in a shunt application (with some fancy series resistor) I recall that pot by itself seemed to sound better. 😉
Hi,
All such things are debatable. Some CMOS Switches are better than others. Relays are still better. If I really want to beat a DS1666 or similar I'd use either my "relay matrix" attenuator as implemented in the big AMR Amp or a TVC with a nice switch.
For a long time "Single Series Singe Shunt" resistor attenuator a'la Audio Synthesis "Passion" was the benchmark for volume controls in the circles I move in. It still provides good performance and one needs only a few really good resistors.
With most pots though shunt attention is worse as they do not like wiper current any better than the DS1666.
In fact, the CMOS Switches also do best into infinite load, making me doubt they are really good for shunt mode systems where their appreciably non-linear impedance is given the maximum ability to impact sound quality.
Ciao T
All such things are debatable. Some CMOS Switches are better than others. Relays are still better. If I really want to beat a DS1666 or similar I'd use either my "relay matrix" attenuator as implemented in the big AMR Amp or a TVC with a nice switch.
For a long time "Single Series Singe Shunt" resistor attenuator a'la Audio Synthesis "Passion" was the benchmark for volume controls in the circles I move in. It still provides good performance and one needs only a few really good resistors.
With most pots though shunt attention is worse as they do not like wiper current any better than the DS1666.
In fact, the CMOS Switches also do best into infinite load, making me doubt they are really good for shunt mode systems where their appreciably non-linear impedance is given the maximum ability to impact sound quality.
Ciao T
I didn't mean electroplating 🙂 For that I would need silver cyanide salts - and that stuff is very dangerous, even by my standards 😀 I was talking about this sort of thing. The resulting plating is extremely thin and not very hard-wearing - but makes stuff look gorgeous all the same.
Is it really not feasible to control volume while the signal is still in digital stage? I thought there would be a way to at least feed the processed audio bitstream from crossover to an external digital volume control and then back to crossover's DAC? Digital signal shouldn't suffer from such a divertion (I think?) and analogue signal would not have to deteriorate from going through volume control. And this would make things much easier further down the signal path. Or am I missing something important?ThorstenL said:
In fact, wouldn't the best solution be to use external DACs? I was thinking TDA1541, with tube stage? Although that is an expensive option...
Also, thanks for explaining the resonances/microphonics principle to me - I have never looked at it this way! But it certainly makes sense.
Thanks for joining the thread Peter! I remember that you once said (well, wrote) Noble was so good as a preamp that it took some seriously expensive Mark Levinston to significantly improve over it. Now, guess why I decided to use Noble in my GC? 😀
Hi,
Well, what you are missing are resolution and SNR.
Let us take the DCX2496. It uses (IIRC) AKM AK4393 DAC's. These give an "A Weighted" SNR of around 117dB and a dynamic range (due to distortion) of around 113dB. So their performance is well below 20Bit accuracy levels.
If you attenuate in the digital domain you are throwing away 1 bit resolution for each 6dB attenuation. So by the time you are past 18dB attenuation this is likely to reduce the effective wordlength of your CD Signal.
On the other hand, a LM3875 has around 114dB SNR/DNR, so one might argue that if you match clipping on the LM3875 to 0dBfs on the DAC you have system that has about the same resolution through the bank.
BTW, if we look at a 87dB/2.83V/1m Midrange and a bridged LM3875 Amp with 25V Rails (around 28V RMS max out) we get a maximum 110dB peak SPL @ 1m from this midrange and hence our whole noise and resolution issue seems not very problematic. If our midrange and treble drivers are much up on this efficiency we may have possible issues though.
In such a case the volume control could be inside the CD-Player or in any other suitable device ahead of the DCX2496. NPC and BB/TI make sample rate converters that include volume attenuation.
I would also add that using ANY volume control ANYWHERE (digital domain or analog domain) degrades performance. Ideally the whole electronics systems would be build gain-matched so that 0dBfs is just below clipping and then attenuation would take place in "power TVC's" right ahead of the speakers, as this avoids any of the other degradations.
This is a possibility, however if you use the TDA1541 be aware it is strictly a 16 Bit DAC, so you must place the volume control after it.
My own system right now uses such a DAC, but of course an analogue volume control after it (even though my PC source could handle the volume as well).
So, I would repeat my suggestion for a chipamp and digital crossover based system as follows:
1) Modify digital crossover to take output directly from the DAC Chip directly, no added op-amps, preserve the balanced signal. Provide a good linear supply for the DAC's and a seperate (including transformer) linear supply for the rest of the digital stuff.
2) Apply volume controls per channel after the DAC. This could use Chips such as the DS1666 or a relay/resistor based solution, preserve the balanced signal.
3) Build balanced, bridged Chipamps to drive each driver in your speaker (so one amp for each woofer in case of the MS Towers which have two) with a power level suitable to the driver. Use seperate mains transformers for MF/HF and LF Amplifiers.
I personally would likely buy a pair of Schroff Multipac Alu 19" Enclosures and use the sides as heatsinks (for expediency) with copper or bronze interface blocks between chips and sides.
The MF/HF stuff definitly LM3875 and with "Class A Bias" on the output (2.2K/5W resistor will do) and on around 24V rails. For the Bass likely LM3886 or LM4780 with 35V Rails.
Volume controls "active" parts I'd build into the enclosures with the Amp's, the actual control section could be somewhere else.
I would suggest that the result, once the crossover is set up correctly the results will be quite spectacular and may not require any further extremism.
The US Importer of 47 Labs (I still remember hid DIY days with tube amps and high efficiency speakers before he decided to become an importer) has a similar system setup, look here:
Yoshi Segoshi's system
As you can see, the custom 6-Channel Gaincard has a volume control.
Ciao T
Well, what you are missing are resolution and SNR.
Let us take the DCX2496. It uses (IIRC) AKM AK4393 DAC's. These give an "A Weighted" SNR of around 117dB and a dynamic range (due to distortion) of around 113dB. So their performance is well below 20Bit accuracy levels.
If you attenuate in the digital domain you are throwing away 1 bit resolution for each 6dB attenuation. So by the time you are past 18dB attenuation this is likely to reduce the effective wordlength of your CD Signal.
On the other hand, a LM3875 has around 114dB SNR/DNR, so one might argue that if you match clipping on the LM3875 to 0dBfs on the DAC you have system that has about the same resolution through the bank.
BTW, if we look at a 87dB/2.83V/1m Midrange and a bridged LM3875 Amp with 25V Rails (around 28V RMS max out) we get a maximum 110dB peak SPL @ 1m from this midrange and hence our whole noise and resolution issue seems not very problematic. If our midrange and treble drivers are much up on this efficiency we may have possible issues though.
In such a case the volume control could be inside the CD-Player or in any other suitable device ahead of the DCX2496. NPC and BB/TI make sample rate converters that include volume attenuation.
I would also add that using ANY volume control ANYWHERE (digital domain or analog domain) degrades performance. Ideally the whole electronics systems would be build gain-matched so that 0dBfs is just below clipping and then attenuation would take place in "power TVC's" right ahead of the speakers, as this avoids any of the other degradations.
This is a possibility, however if you use the TDA1541 be aware it is strictly a 16 Bit DAC, so you must place the volume control after it.
My own system right now uses such a DAC, but of course an analogue volume control after it (even though my PC source could handle the volume as well).
So, I would repeat my suggestion for a chipamp and digital crossover based system as follows:
1) Modify digital crossover to take output directly from the DAC Chip directly, no added op-amps, preserve the balanced signal. Provide a good linear supply for the DAC's and a seperate (including transformer) linear supply for the rest of the digital stuff.
2) Apply volume controls per channel after the DAC. This could use Chips such as the DS1666 or a relay/resistor based solution, preserve the balanced signal.
3) Build balanced, bridged Chipamps to drive each driver in your speaker (so one amp for each woofer in case of the MS Towers which have two) with a power level suitable to the driver. Use seperate mains transformers for MF/HF and LF Amplifiers.
I personally would likely buy a pair of Schroff Multipac Alu 19" Enclosures and use the sides as heatsinks (for expediency) with copper or bronze interface blocks between chips and sides.
The MF/HF stuff definitly LM3875 and with "Class A Bias" on the output (2.2K/5W resistor will do) and on around 24V rails. For the Bass likely LM3886 or LM4780 with 35V Rails.
Volume controls "active" parts I'd build into the enclosures with the Amp's, the actual control section could be somewhere else.
I would suggest that the result, once the crossover is set up correctly the results will be quite spectacular and may not require any further extremism.
The US Importer of 47 Labs (I still remember hid DIY days with tube amps and high efficiency speakers before he decided to become an importer) has a similar system setup, look here:
Yoshi Segoshi's system
As you can see, the custom 6-Channel Gaincard has a volume control.
Ciao T
Last edited:
Isn't that the point of attenuation?
..and create other degradations of their own.
The looks of that amplifier case are totally different from what one would (should?) expect from somebody in his position.
- Status
- Not open for further replies.