Hi to everyone, i am considering building a 24/96 capable DAC.
I am not an expert in digital circuits, but if memory serves, back in the late 90s, everyone stated that R2R dacs are - technically speaking- superior to Delta-sigma designs.
Almost every expensive player-dac back then, used multibit dacs (exept from meridian, i think).
In the last 5 years, i have not seen a new chip of this sort, every new design is some kind of bit-stream architecture (i am aware of the difficulty and cost to built a good R2R dac).
Is the built of a DF1704/06 --> PCM1704 worth the extra effort (boards almost immposible to obtain-have to make my own) and cost?
For me, even small impovements in perfomance justify the cost/trouble overhead, but are the PCM1704 really (measurable) superior to newest Delta-sigma like PCM1794, wolfson, AKM designs (all other things considered equal, e.g best possible supllies, clocking, e.t.c for the chip of choice) ?
I am not talking about "sound", just about measuring better in the lab. I just want to have a top-notch piece of equipment.
Every answer from the digital experts hier would be much appreciated.
Greetings from Athens (Greece)
Konstantinos
P.S
Is it true, that Delta/sigma is inherently flawed, because it is impossible to operate at such a high clocking frequency, to fully compensate for the truncation of the 16-24 bits to just one?
Unfortunately i don't remeber where i found this statment, it is long ago..
I am not an expert in digital circuits, but if memory serves, back in the late 90s, everyone stated that R2R dacs are - technically speaking- superior to Delta-sigma designs.
Almost every expensive player-dac back then, used multibit dacs (exept from meridian, i think).
In the last 5 years, i have not seen a new chip of this sort, every new design is some kind of bit-stream architecture (i am aware of the difficulty and cost to built a good R2R dac).
Is the built of a DF1704/06 --> PCM1704 worth the extra effort (boards almost immposible to obtain-have to make my own) and cost?
For me, even small impovements in perfomance justify the cost/trouble overhead, but are the PCM1704 really (measurable) superior to newest Delta-sigma like PCM1794, wolfson, AKM designs (all other things considered equal, e.g best possible supllies, clocking, e.t.c for the chip of choice) ?
I am not talking about "sound", just about measuring better in the lab. I just want to have a top-notch piece of equipment.
Every answer from the digital experts hier would be much appreciated.
Greetings from Athens (Greece)
Konstantinos
P.S
Is it true, that Delta/sigma is inherently flawed, because it is impossible to operate at such a high clocking frequency, to fully compensate for the truncation of the 16-24 bits to just one?
Unfortunately i don't remeber where i found this statment, it is long ago..
Hello,
The only two chips I would use are the pcm1704 and the ESS sabre. The latter does delta-sigma in the best possible way, with the current technology, so it's worthy of recommendation.
The latest AKM dacs such as the 4397 should be very good sounding too, but very few people in the diy world uses them, so i cant be as sure.
My choice is clearly the 1704, in a number no less than 4 (2 per channel, they are mono).
Of older chips I like the pcm56, pcm63k, tda1541a, which are IIRC all R2R.
The only two chips I would use are the pcm1704 and the ESS sabre. The latter does delta-sigma in the best possible way, with the current technology, so it's worthy of recommendation.
The latest AKM dacs such as the 4397 should be very good sounding too, but very few people in the diy world uses them, so i cant be as sure.
My choice is clearly the 1704, in a number no less than 4 (2 per channel, they are mono).
Of older chips I like the pcm56, pcm63k, tda1541a, which are IIRC all R2R.
The 1704's are awesome, but several folks have reports that the sound quality has slipped when TI transitioned to ROHS compliance for the 1704. I don't if this is true, but a common refrain.
The TI 1792/4 or AD1955 are of the same caliber. These delta sigma DAC's resolve greater detail and present a crisper image. The 1704 seems to present a smoother aand more relaxed sound.
The AKM4396 is delta sigma, and a great compromise between the best of both of both the TI/AD and 1704. If you are making your own DAC the AKM4396 is very forgiving and easy to put together. This chip has an on board I/V and does not require software control.
I've posted pic of the ad1955 and AKM4396 eval DAC's on the projects page on my website. I'm working on a review of all the major DAC eval boards. I have the 4397, and I'm not a fan of the 32-bit DAC's. The 4396 actually sounds quite a bit better than the 4397.
-David
The TI 1792/4 or AD1955 are of the same caliber. These delta sigma DAC's resolve greater detail and present a crisper image. The 1704 seems to present a smoother aand more relaxed sound.
The AKM4396 is delta sigma, and a great compromise between the best of both of both the TI/AD and 1704. If you are making your own DAC the AKM4396 is very forgiving and easy to put together. This chip has an on board I/V and does not require software control.
I've posted pic of the ad1955 and AKM4396 eval DAC's on the projects page on my website. I'm working on a review of all the major DAC eval boards. I have the 4397, and I'm not a fan of the 32-bit DAC's. The 4396 actually sounds quite a bit better than the 4397.
-David
Maybe we have a language mix-up, and maybe we are saying the same things? The AKM4396 current to voltage conversion is performed on chip and presents only an output voltage on it's pins. AKM definitely has an I/V, but it is internal.
My prior post said the akm4396 I/V was "on board" meaning "inside", but could be interpreted as external on a circuit board perhaps.
-David
My prior post said the akm4396 I/V was "on board" meaning "inside", but could be interpreted as external on a circuit board perhaps.
-David
Thanks guys, understood
Thus, this "native" voltage output can be seen as a "pure natural process "here ( exerpt the cap )?
This because I love R-2R multibit dacs and particulary when I/V conversion is made in "audiophile" terms ( ie no crappy ICs )
as like resistor + triode , etcetera
Cheers,
Paolo
Thus, this "native" voltage output can be seen as a "pure natural process "here ( exerpt the cap
)?This because I love R-2R multibit dacs and particulary when I/V conversion is made in "audiophile" terms ( ie no crappy ICs
)as like resistor + triode , etcetera
Cheers,
Paolo
well you saw those out of band noise plots, those switched caps must work hard for an AKM ...
-bad remark cause, switched cap filters said to be THE MAIN culprit plaguing cd format in 80's, sony used insane attenuation and integrated "murata" switch cap filter IC-s were a commonplace in digital FX units... This lead to the so called apogee fame , when they started out making "944-G" replacement filters with cool impulse response and lesser steepness. These were installed in many Mitsubishi and Sony multitrack recorders.
-bad remark cause, switched cap filters said to be THE MAIN culprit plaguing cd format in 80's, sony used insane attenuation and integrated "murata" switch cap filter IC-s were a commonplace in digital FX units... This lead to the so called apogee fame , when they started out making "944-G" replacement filters with cool impulse response and lesser steepness. These were installed in many Mitsubishi and Sony multitrack recorders.
tritosine said:as far i know each have capacitor inside whereas the pcm1704 doesnt have one and no, no number crunching either
V-out delta sigma DAC have switched capacitor low pass filters , I rather would be aware of that fact, this makes them more and more "CODEC"-ey in my eye.
The very first sigma delta DACs in the early 90's did not have switched capacitor filters. Two examples are the Cirrus Logic CS4303 and the NPC SM5872, both of which had 1 bit pure digital continuous-time outputs.
Current output sigma delta DACs such as the PCM1792 and AD1955 are of the continuous-time current segment type, and also have no switched capacitor filters.
yes, here it is : http://www.grimmaudio.com/whitepapers/discrete ad converter.pdf
completely different philosophy, than, say, a pcm1704 operating at maximum oversampling (16x).
Then if someone decides to throw oversampling out of the multibit cdp and stay 44.1 / build non-os dac, the output spectra (courtesy of Pedja R) :
completely different philosophy, than, say, a pcm1704 operating at maximum oversampling (16x).
Then if someone decides to throw oversampling out of the multibit cdp and stay 44.1 / build non-os dac, the output spectra (courtesy of Pedja R) :
Oh thanks tritosine
To be onest I am too much inexpert to understand those graphs
There are not the "notes" to explain right vs left or below vs above comparisions
Anyway, I have always asked myself if is it possible to see an impulsive and indicial response of a NON Oversampling DAC..
Are they similar to those of Wadia ? Excuse my rookies questions!
Thanks,
Paolo
To be onest I am too much inexpert to understand those graphs
There are not the "notes" to explain right vs left or below vs above comparisions
Anyway, I have always asked myself if is it possible to see an impulsive and indicial response of a NON Oversampling DAC..
Are they similar to those of Wadia ? Excuse my rookies questions!
Thanks,
Paolo
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.