During our 5-6 years of developing an DAC, we have seen that the DAC-chip itself isn't that important.
Clearly the PCM1704 is a high performer. But followed by a op-amp based analog stage, it is all ruined.
We have testet PCM1702 (Same design as 1704) with a discrete non feedback analog stage, with good results. We have done PCM1794 and other DAC's. Anyway... We ended up with the CS4398 from Crystal in the end. This DAC performs very well, you give it optimum working conditions.
Follow our project here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=138230
Clearly the PCM1704 is a high performer. But followed by a op-amp based analog stage, it is all ruined.
We have testet PCM1702 (Same design as 1704) with a discrete non feedback analog stage, with good results. We have done PCM1794 and other DAC's. Anyway... We ended up with the CS4398 from Crystal in the end. This DAC performs very well, you give it optimum working conditions.
Follow our project here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=138230
Re: Re: Re: Re: PCM1704 or newer chips?
That is what it looks like, but it doesn't appear to be what is actually implemented beyond the processing - the DA section is still the same sigma delta type (from the limited information in the datasheet). The back-end is the same DWA pipeline of identical current sources - shared between the different processing topologies. Thus, there appears to be reasons such as perhaps efficiency (i.e cost or other) which has promted TI to select such a topology
Petter
suchtgutenklang said:
That is what it looks like, but it doesn't appear to be what is actually implemented beyond the processing - the DA section is still the same sigma delta type (from the limited information in the datasheet). The back-end is the same DWA pipeline of identical current sources - shared between the different processing topologies. Thus, there appears to be reasons such as perhaps efficiency (i.e cost or other) which has promted TI to select such a topology
Petter
Hi,
Very thanks suchtgutenklang for explanation, very appreciated🙂
I am a rookie here and I am trying to learn🙄
IMHO maybe Denon now save some money with those hybrid DACs,
but I remain of my idea that old "pure" 1704 was more audiophile
Cheers,
Paolo
Very thanks suchtgutenklang for explanation, very appreciated🙂
I am a rookie here and I am trying to learn🙄
IMHO maybe Denon now save some money with those hybrid DACs,
but I remain of my idea that old "pure" 1704 was more audiophile
Cheers,
Paolo
Hi Chris,
I agree 100%. My venerable Aragon D2A sound still good despite his giurassic digital section! 🙂
Agree again, but IMHO it depends because the Alpha Processor
Perfectly agree again
I am Skeptical here. IMHO you can modd the S-10 improving his weack point: the analog output stage.
I have just purchased a DCD1450AR and I want swap all those ICs
with triodes!😀
Only not sure about what tipe of I/V conversion : passive or active?😕
Time to experimenting.........
Cheers,
Paolo
Anyone want my Assemblage Dac 2.7?
Anyone want my Assemblage Dac 2.7? I should be able to dig up the box and manual which included the schematic.
Anyone want my Assemblage Dac 2.7? I should be able to dig up the box and manual which included the schematic.
Hi inertial,
Should be interesting to follow your project. The tube idea sounds interesting. I also have a Carver SDA 390t CD player. It uses 6922s in the analog section. I believe it's a "bitstream" type of DAC preceding. They needed that tube in the signal path I think.
I think the biggest problem might be the SNR of the finished analog section, unless the first stage is solid state. Something along the lines of J-Fets or BJTs. You can get very creative with these devices and end up with quite a good analog section. A pure tube effort may be disappointing unless everything is just right.
The DCD S-10 does have a few changes that can be made. I was thinking of working on the analog power supply first, then carefully look at the analog section. A straight op amp switch is not what I would call a worthwhile effort. It uses OP275s in there, and they are not that bad. The film capacitors are decent as well. Your 1450 should turn out to be a satisfying build.
Hi sendler,
What is wrong with your DAC? It would be interesting to study it, and possibly use it. That 390t could definitely use help in the DAC section. At the very least, I would very much like to see the schematic for it.
-Chris
Should be interesting to follow your project. The tube idea sounds interesting. I also have a Carver SDA 390t CD player. It uses 6922s in the analog section. I believe it's a "bitstream" type of DAC preceding. They needed that tube in the signal path I think.
I think the biggest problem might be the SNR of the finished analog section, unless the first stage is solid state. Something along the lines of J-Fets or BJTs. You can get very creative with these devices and end up with quite a good analog section. A pure tube effort may be disappointing unless everything is just right.
The DCD S-10 does have a few changes that can be made. I was thinking of working on the analog power supply first, then carefully look at the analog section. A straight op amp switch is not what I would call a worthwhile effort. It uses OP275s in there, and they are not that bad. The film capacitors are decent as well. Your 1450 should turn out to be a satisfying build.
Hi sendler,
What is wrong with your DAC? It would be interesting to study it, and possibly use it. That 390t could definitely use help in the DAC section. At the very least, I would very much like to see the schematic for it.
-Chris
anatech said:
Hi Chris🙂
Yup, maybe I will open a new thread on this.......
S/N ratio is the last of my problem now, I am in doubt on wich tipe of I/V conversion employ !😀
I am accustomed with MC cartridge thus S/N is not a problem for me😀
IMO "Digital sound" is the problem!
What do you think of the 1450 stock version about jitter? I understand nada of digital but seems to me it is a possible point of
make improvments🙄
Thanks,
Paolo
Assemblage Dac2.7 was pretty trick
Nothing wrong with the Dac 2.7. I said DF1704 above but now I remember it was actually the PMD200 HDCD filter. It was considered fairly high up towards the cutting edge in it's day. Originally offered as a kit by Assemblage which was the diy branch of Sonic Frontiers and Parts Connection back in their hey day. Guys used to refer to it, and the DAC 3.0 which featured dual dacs for each channel running full balanced all the way from the digital filter, as a "parts graveyard" because of the high quality Wima through hole caps bypassing all of the power supply pins. It uses the green Wima foils to form the second order ultrasonic active filter which I modded to move higher as the original filter was too close to 20k for my tastes. The opamps are all socketed so it was a great test sled for comparisons. I/V, servo, active filter, output buffer. One at a time I listened to different opamps and it is interesting to find that the LT1360 drops into any application and sounds way better than whatever you took out. Who would have though changing an amp running as a servo would make such a sonic difference. It was a great dac in it's day but now I find that the digital filter, dac, and output sections of the newest AKM dac chips have pushed the boundaries to new levels. The super low levels of ultra sonic noise and the voltage output sound great running in a purist "direct out" configuration with nothing more than a high quality cap or transformer to block the dc between you and your music.
anatech said:
Nothing wrong with the Dac 2.7. I said DF1704 above but now I remember it was actually the PMD200 HDCD filter. It was considered fairly high up towards the cutting edge in it's day. Originally offered as a kit by Assemblage which was the diy branch of Sonic Frontiers and Parts Connection back in their hey day. Guys used to refer to it, and the DAC 3.0 which featured dual dacs for each channel running full balanced all the way from the digital filter, as a "parts graveyard" because of the high quality Wima through hole caps bypassing all of the power supply pins. It uses the green Wima foils to form the second order ultrasonic active filter which I modded to move higher as the original filter was too close to 20k for my tastes. The opamps are all socketed so it was a great test sled for comparisons. I/V, servo, active filter, output buffer. One at a time I listened to different opamps and it is interesting to find that the LT1360 drops into any application and sounds way better than whatever you took out. Who would have though changing an amp running as a servo would make such a sonic difference. It was a great dac in it's day but now I find that the digital filter, dac, and output sections of the newest AKM dac chips have pushed the boundaries to new levels. The super low levels of ultra sonic noise and the voltage output sound great running in a purist "direct out" configuration with nothing more than a high quality cap or transformer to block the dc between you and your music.
Re: Ak4395 beats pcm1704 - I dont think so
I think that you compared apples and oranges.
Because you compared converters built of several parts, not the DAC chips themselves.
I give you a hint: compare the Esoteric X-01 or X-03 (PCM1704) vs the X-05 (CS4398).
You may not know that roughly 80-90% of the sound is merit of the power source and the output stage. The DAC chip itself and the way it operates do not play a big factor as people believe.
Edit: I didnt read the following posts. I think that in the Assemblage DAC the pcm1704 was killed by the use of opamps, not sure to which level it could be possible to mod it, could you send me the schematics by mail?
It's very interesting because of the pcm200 hdcd receiver chip.
sendler said:
I think that you compared apples and oranges.
Because you compared converters built of several parts, not the DAC chips themselves.
I give you a hint: compare the Esoteric X-01 or X-03 (PCM1704) vs the X-05 (CS4398).
You may not know that roughly 80-90% of the sound is merit of the power source and the output stage. The DAC chip itself and the way it operates do not play a big factor as people believe.
Edit: I didnt read the following posts. I think that in the Assemblage DAC the pcm1704 was killed by the use of opamps, not sure to which level it could be possible to mod it, could you send me the schematics by mail?
It's very interesting because of the pcm200 hdcd receiver chip.
Hi Cobra2,
Yeah, it would have to be really cheap for me. I'm outta loot!
Hi sendler,
Just think. My wife worked for Sonic Frontiers in the office. I never was able to afford their stuff. She split after it was taken in house.
Hi Paolo,
-Chris
Yeah, it would have to be really cheap for me. I'm outta loot!
Hi sendler,
Just think. My wife worked for Sonic Frontiers in the office. I never was able to afford their stuff. She split after it was taken in house.
Hi Paolo,
You should. Sounds like a really interesting project. There are about as many different ways to perform I / V conversion as there are people who are sure their way is the only way! You could easily compare some promising methods in your project, then use the one that seems to work the best for you.
I wouldn't think so. This will be affected by all your choices in power supply and output circuits. S/N ratio is designed in from the start.
-Chris
Several parts not needed with the AK4395
.
No I/V as it has a line level voltage output.
.
No analog ultrasonic filter as the ultrasonic noise is very low. Check out the graphic.
.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=137703
.
No output buffer. It has plenty of power and low enough output impedance to drive my 4k stepped attenuators directly.
.
All that is needed is a high quality cap or transformer to block the 2.6v dc. I like a 4.5uf stack of Dayton foil caps. Better sonics than any film cap I have heard and only $15.00. And better than my Sowter transformers. $200.00 boutique trannies may be great as well, especially as the balanced legs of the chip have 0v dc between them, but I'm not interested in going there.
.
The AK4395 also seems to be pretty immune to power supply issues as the stock regulation, filtering and bypassing in the Behringer 2496 units (which is actually pretty well executed for the $300 price) take the AKM chip to a level of final performance that is not improved at all with additional bypassing with oscons and ceramics or even (on the digital supply pin) a dedicated regulator and filters.
.
I'll let you know about it's sensitivity to clocks as I have an Oettle receiver/ clock kit coming for my DCX.
Of course this will always happen in these discussions unless we can compare your best implementation of your favorite chip with my best of mine. The thing that has me so jazzed over the AK4395 is that many of the "several parts" are non existent. Most of the circuits you guys have been fine tuning for years are not needed.Telstar said:
.
No I/V as it has a line level voltage output.
.
No analog ultrasonic filter as the ultrasonic noise is very low. Check out the graphic.
.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=137703
.
No output buffer. It has plenty of power and low enough output impedance to drive my 4k stepped attenuators directly.
.
All that is needed is a high quality cap or transformer to block the 2.6v dc. I like a 4.5uf stack of Dayton foil caps. Better sonics than any film cap I have heard and only $15.00. And better than my Sowter transformers. $200.00 boutique trannies may be great as well, especially as the balanced legs of the chip have 0v dc between them, but I'm not interested in going there.
.
The AK4395 also seems to be pretty immune to power supply issues as the stock regulation, filtering and bypassing in the Behringer 2496 units (which is actually pretty well executed for the $300 price) take the AKM chip to a level of final performance that is not improved at all with additional bypassing with oscons and ceramics or even (on the digital supply pin) a dedicated regulator and filters.
.
I'll let you know about it's sensitivity to clocks as I have an Oettle receiver/ clock kit coming for my DCX.
Hi Scott,
You can use an op amp to get rid of your DC offset. Just make sure that the 2.6 VDC is within the common mode range of your IC.
What are you thinking on your Sonic Frontiers DAC?
-Chris
That's only because these elements are internal to the DAC. They still exist, you just can't get at them.
You can use an op amp to get rid of your DC offset. Just make sure that the 2.6 VDC is within the common mode range of your IC.
What are you thinking on your Sonic Frontiers DAC?
-Chris
anatech said:
Yup! 🙂
I'll do it. Because I need "guidance " of some experts guys here!
I am intrigued by transformers ( Sowter 8347) but maybe OTL is better ? 🙂
I have read that I/V converters with a triode ( transimpedance-config) is not so well because the input impedance can't be so low .....Is it true?
Probabily a transistor can make better work here ( one order of magnitude more lower ?) , but the sonics?
Or more "puristic" again only a little resistor, but you need very high gain so one triode can be not sufficent.....Maybe two triodes in fall, like a phono MM ? 🙂
Hard decision here............
You are right, I have to considerate every aspect in order to obtain real top performances ! 🙂
This is reason because I need precious suggestions 🙂
Do you think the standard jitter of the stock DCD1450AR is sufficent/ OK or do I have to program also an improvment in this area?
Thanks,
Paolo
Re: Several parts not needed with the AK4395
Yes, I agree. and keep me updated on the clocks 🙂
Already sigma-delta dacs are easier to implement and do not need extreme care in just about everything. The AKM chips and in particular the 4395 and 4396 seem the easiest of all to implement.
With it you can achieve "8" with 1/20 of the cost needed to get the same (which i'm not sure will be the same sound, but lets make things simple) on a pcm1704 based DAC.
It's a pity that I sold my deq2496, so i cannot try a mod and compare. i heard the AKM 24 bit dacs only in Echo and RME firewire devices, without mods.
Besides that, there is always something in sigma-delta DACs that hurts my ears, while I'm basically immune to HF noise from NOS D/A.
My reference player is the Naim cd555 (I let you guess which DAC chip it mounts).
sendler said:
Yes, I agree. and keep me updated on the clocks 🙂
Already sigma-delta dacs are easier to implement and do not need extreme care in just about everything. The AKM chips and in particular the 4395 and 4396 seem the easiest of all to implement.
With it you can achieve "8" with 1/20 of the cost needed to get the same (which i'm not sure will be the same sound, but lets make things simple) on a pcm1704 based DAC.
It's a pity that I sold my deq2496, so i cannot try a mod and compare. i heard the AKM 24 bit dacs only in Echo and RME firewire devices, without mods.
Besides that, there is always something in sigma-delta DACs that hurts my ears, while I'm basically immune to HF noise from NOS D/A.
My reference player is the Naim cd555 (I let you guess which DAC chip it mounts).
Naim CD 555?
4xPCM1704, digital filter is PMD200?
I am very curious about comparing with LINN CD12.
4xPCM1704, digital filter is PMD200?
I am very curious about comparing with LINN CD12.
naim555:
"Discrete components handle the current-to-voltage conversion, seven-pole analog filtering, and analog output stages."
7 pole!!! hmm 😉
, also, -90 dB sinewave test suggests they get some special (insert sales pitch) "grade" of pcm1704 you guys wont lay your hands on anytime soon... (not that -90 matters at all, but still, nice feat! )
"Discrete components handle the current-to-voltage conversion, seven-pole analog filtering, and analog output stages."
7 pole!!! hmm 😉
, also, -90 dB sinewave test suggests they get some special (insert sales pitch) "grade" of pcm1704 you guys wont lay your hands on anytime soon... (not that -90 matters at all, but still, nice feat! )Re: Naim CD 555?
Yes, pmd200. The cds3 + 555ps comes very close too (same chips).
I haven't done the comparison... Do you remember the specs of the cd12?
AD1865 said:
Yes, pmd200. The cds3 + 555ps comes very close too (same chips).
I haven't done the comparison... Do you remember the specs of the cd12?