I've read that its possible not to use any opamp on this kind of DAC, so I plan to connect it directly to RCA out on my Elite Pro, but of course I need to use a capacitor to block DC, but how do I counter the out-of band frequency, I read from Prodigy HD2 (AK4396) thread here that I need to use 560ohm to remove it, but does this applies to CS4398 too?

The CS4398 is a differential output DAC. The opamp provides filtering as well as differential to single ended conversion. If you want a balanced output, you can skip the opamp.

This is what I did to the Zero DAC. Maybe the cutoff frequency could be higher. Maybe there's no need for a 2nd order filter. I don't know. This filter is quite transparent as it is. The drawback is hum and buzz . I think it's the ground layout that's not as good as it should.

The most elegant, and maybe best sounding, way is to use output transformers. The downside is the cost. There are several threads about it at diyaudio.com

there are good technical reasons for the filter, diff-to-single ended conversion and buffering in the eval circuits of V out delta-sigma DACs

the output analog output is a switched C filter running at MHz so there is high frequency that needs to filtered out even though the chip does a good job filtering below a few Hundred KHz

also as mentioned these chips use differential output to get around the large internal ground noise in the chip from the digital processing

common mode+differential filters with RC as the 1st stage after the DAC are highly recommended to get rid of the MHz noise before sending it into your audio analog stages

its not hard to choose good quality audio op amps that won't degrade the signal relative to the DAC chip's internal CMOS op amp that has to be compatable with the 100K+ digital CMOS on the same substrate

anyone who thinks they're "increasing accuracy" by going for one of the low/no active component or no op amp outputs is simply delusional - you may hear differences but its not because your "simple" circuit is more accurate

if you need "elegance" why not look inside audio op amps for it? LT1469 complementary bipolar, unity gain stable implementation of Scott Wurcer's A797 topolgy - that's elegant and accurate

Yeah but opamps are never desirable for high end audio. I wonder how well a really simple RC filter would work. Follow that by a simple discrete buffer for very low output impedance.... Pass B1 style.

except for low noise front ends and power amp outputs that is a way outdated opinion, in the last 20yrs we're talking Billions of dollars and tens of thousands of man years improving op amps for DSL and other high speed signal processing as well as precision industrial/instrumentation apps

on the discrete front for audio apps we're only losing the best devices as they're discontinued - no real motion other than die shrink as older parts have to be moved into newer processes as the old lines are replaced or become unmaintainable

but the discrete vs op amp is completely silly when you look at whats inside one of these delta-sigma DACs - we're looking at over 1/2 dozen op amps internal to the 4398, implementing >>120 dB of feedback to achive the noise shaping - delta-sigma converters have the highest amount of negative feedback of any audio chip - and the op amps internal to the chip are implemented in a compromise semiconductor process with the fact that they also have >100K digital gate equivalents on the same chip - seems ver foolish to argue that a highly optimized modern analog process op amp is somehow going to compromise the DAC output

But we are trying to make it sound good. I'm not disputing the technical performance of modern op amps, just saying there is a lot to be said for keeping the audio path uncomplicated.

Top of the line DAC chips tend to sound better than cheaper parts so it doesn't matter that they have 1/2 a dozen op amps in them. That part is out of our hands. We can only use them and control what we put on the outputs and how the design is implemented.

Quote:



Actually I have used AD797 and it's doing well both IRL and in sim. No oscillation or ill behaviour. It sounds good. I'm afraid I haven't heard LT1469.

I'm quite sure that the passive circuit is less accurate when it comes to THD, FR and noise. I'm not sure if it sounds worse or not.

I can always mod the stock opamp to a better one, I have some SOIC opamp like LME49720, LT1364 and AD8066, but the implementation on Elite Pro is a bit different from evaluation datasheet, the DC blocking caps are supposed to be after the opamp but its done in reverse in Elite Pro, doubling its capacitor to four, I only have two 47uF Blackgate NX HQ.

Should I tap it to external PCB and follows the evaluation board analog out, or is there any good circuit to follow? I have more DIP8 opamp, the LME49720, 2xLT1363 (single version of LT1364) and LT1361. I'm plan to separate the left and right channel so they have separate opamp by using 2 single instead of one dual. Any advice?

Anyone?

One more question, if I decide to bypass the opamp just to try, do I need to combine both + and - signal to the output or just one of them?

just one.

i used some transformers with mine. massive improvement to my ears.

Finally I have time to do the mod, I just did connect the plus of both channel to Rubycon Blackgate NX HQ 47uF on both with 1k ohm resistor before the output (I might got the resistance wrong, it supposed to be 560ohm) direct to RCA output. Initial sound is quite nice, quite airy and the vocal sounds excellent, it does opens out the details in the background

But I got the same result as using the 3.5mm jack, both the right and left channel is mixed! I haven't got a clue how this happens. Maybe 7 driver, I don't know, I have used ASIO, and bit matched playback, but still the same.

EDIT: How stupid can I get? I accidentally put the mono button to ON on the amplifier. Damn, I just noticed it just now after for about some weeks....

Hello to all ,
 
Is there anyone who tried to couple a differential out from CS4398 to a differential input amplifier ?
Maybe via good capacitors or direct coupled ?

Ok.So today I tried the next experiment.
Connected the differential outputs of dac direct to a Lm1875 in differential topology.This means I don't have any capacitor in the signal path.Due to the fact that 1875 is in differential topology the 2.5 volts from dac outs are substracted between them so I have 2 mV on speakers.
Now the results:
-huge soundstage,localisation of each instrument easily and precise.
- more instruments can be heard in some tracks
- a lot of micro details
Big improvement in piano sound
And in all frequencies .
- each instrument seems to be more natural .
- musical or more analog sound

Tomorrow I will listen louder.In this evening I couldn't.

jcx said:

except for low noise front ends and power amp outputs that is a way outdated opinion, in the last 20yrs we're talking Billions of dollars and tens of thousands of man years improving op amps for DSL and other high speed signal processing as well as precision industrial/instrumentation apps

on the discrete front for audio apps we're only losing the best devices as they're discontinued - no real motion other than die shrink as older parts have to be moved into newer processes as the old lines are replaced or become unmaintainable

but the discrete vs op amp is completely silly when you look at whats inside one of these delta-sigma DACs - we're looking at over 1/2 dozen op amps internal to the 4398, implementing >>120 dB of feedback to achive the noise shaping - delta-sigma converters have the highest amount of negative feedback of any audio chip - and the op amps internal to the chip are implemented in a compromise semiconductor process with the fact that they also have >100K digital gate equivalents on the same chip - seems ver foolish to argue that a highly optimized modern analog process op amp is somehow going to compromise the DAC output

Click to expand...

Thanks for the info! You're right.