[raoultrifan]

For I/V 2 x MUSES8920 will have a lower noise than 8820, same applies for LPF. In PRE-AMP stage there's no much difference between the op-amps in regard with the output noise, so I care most about the output-DC here, this is why I went with MUSES8920 with DIY dual-single adapters or AD797 (also NE5534 is a low-output-DC option).

 

[Eiffel]

You may try also OPA132/134 - it has better SNR than Muses ( about 3-4dB ), but it might alter a bit the soundstage. Very slight as the Pre-Gain does not affect too much the sound.
I've tried the dual OPA2132 in LPF combined with V6Vivid in I/V and the soundstage got narrow. On my STX I use it with V5i and it was ok.

This has been discussed already.
P.S.: LME49860 was noisier than LME49720 in every configuration and DAC and headamp I ever tried (maybe I got a strange batch of LME49860). I'm not speaking about night and day difference, but it was a difference unexpected by me, especially that these 2 op-amps are sharing the same datasheet, just power voltage is different (+/-22V instead of +/-18V).

49860 needs higher voltage for better sound. It's the same 49720 but slightly altered to work at higher voltages. Some op-amps give their best at a specific voltage, even if thy have a voltage range.

What is interesting about TI op-amps is that they use the same op-amp with different codes. Even LME49720 is in fact the old LM4562NA. Funny, some people swear they hear a difference.
They also have the "Audio" range of op-amps - like the OPA2134 which is an OPA2132 who failed to run properly at high frequencies. If an industrial wide use OP-AMPS fails to run at higher frequency put pass a 100kHz test they are labeled as Audio and sold a bit cheaper under other code. Even the OPA1612 mentioned in an earlier post seems to be the "Audio" version of the OPA2211.

 

[raoultrifan]

I would ignore the datasheet for the moment, I've tested op-amps having a lower noise spec against less noisier op-amps and results were mixed, depending on the configuration used on the PCB.

One thing's for sure: LME49720 has are lower noise than LME49860; tested in I/V, LPF, VAS stages in ASUS E1 MKii and MATRIX M-Stage HPA-3B. Hence, I can't recommend LME49860 over the LME49720 unless higher voltage are required by the design.

I personally started to think more technically and I mostly care about bit-resolution of an audio equipment more than soundstage and else. This is why I care most about the background noise, linearity and THD and this why I settled down in my PLAY to 3xAD8599 and 2xAD797.

 

[Eiffel]

You're right. It happened that OP-AMPs with worst specs to sound far better than ones with great specs. The Muses 01 and 02 don't have great specs. but they are damn good when using them. As long as you have a good power filter.
Also some OP-AMP's sounds good on a device and not so good on another - depends on producer implementation.

Me, I try to keep a balance - if they sound good I care less about SNR and stuff as long I don't hear the noise and the sound is as natural as possible,

 

[raoultrifan]

[...]The Muses 01 and 02 don't have great specs. but they are damn good when using them. As long as you have a good power filter.
[...]

This is probably caused by their mediocre CMRR/PSRR specs. There's always a tread-off when building good audio op-amps. Also, who has the money for, can jump up to MUSES03.

 

[Eiffel]

They don't have a good power noise rejection. Around 90dB if i remember right.
On my STX I run some tests and I got several noises from the PSU. Changing the PSU gives completely different results.
Muses02 are sensitive to hum - 50 Hz and harmonics. Muses01 are sensitive to higher frequency noises - got some 1kHz and 8kHz on first PSU tested - noise from other components of PC.
Also they have quite low Slew Rate.
On STX with 2x01 in I/V and 1x02 in buffer I got 110 dB SNR in RMAA. With original Op-Amps I had 117db. With OPA2132 I got 123,6dB SNR - almost as ASUS claims.
What's great with 01 and 02 - they have an incredible channel separation - about 140dB.

Muses03 seems to be better, but it's single OP-Amp.

 

[metalicblue]

I too am coming from XD-05 and yes, I have DT 1990 and indeed 3x OPA1692 2x OPA209 is amazing. I recommend try 1x LME49860 in I/V

The bass shy comparison to XD-05 is the coloration from OPA1612. It has a huge sub bass. If you put OPA1612 in I/V you will get somewhat similar kind of sub bass and bass presence as XD-05.

I just tried
2x OPA 2197 and 1X V5I in the I/V-LPF
and 2x v5i single in the preamp
The OPA2197 will roll off the Treble and will give the impression of a more warm sound with my AKG K812 Headphone but I think the SABRE signature can not be tamed.
This is a bright dac-amp and I wish it would have a bass boost switch like the xduoo xd05 has.
I'm waiting for another OPA 2197 for the LPF , maybe this will solve the bass problem I have with my AKG K712
BTW I think the OPA 2197 has a less wide sound stage and also less deep too but the AKG will partially compensate for this
Next I will try the OPA 1612 but I only have 2 for now. Just ordered a new batch of opa's. Will comment after receiving them.

 

[Eiffel]

The I/V OP-AMP's are the one who gives the sound signature - 80-90% of the sound characteristics - bass, treble, details...
The LPF affect only 10-20% of the sound and mostly the soundstage with little or no effect on the sound. - bass or treble.
The pre-amp - even less - may be 1% but I doubt it.

 

[Wiljen]

Should have my review of the Play done in another day or so. I am loving it but not getting enough time to sit still and listen.

 

[metalicblue]

The I/V OP-AMP's are the one who gives the sound signature - 80-90% of the sound characteristics - bass, treble, details...
The LPF affect only 10-20% of the sound and mostly the soundstage with little or no effect on the sound. - bass or treble.
The pre-amp - even less - may be 1% but I doubt it.

Not wanting to start any dispute but how did you come to this conclusion? Do you have any backup info on this? Also, what is the role of the LPF OPA in the sound chain?
Sorry, just asking so I would have a better understanding and not just try endless combinations of OPA's .
Thanks

 

[raoultrifan]

It is true, indeed, this is why some DACs are using passive I/V stage, Lundahl transformers or FET transistors instead of op-amps. Basically, the output sound of a DAC-chip could get easily "altered" by an improperly designed/implemented I/V stage. Most of the time SNR and THD are the ones that are easily to measure, but also in LPF things are about the same (feel free to search on the web about it).

Speaking about measurements, here's a decent proof that NE5532 is a better option than OPA2134 when used in the I/V stage: http://nihtila.com/2017/01/08/pcm1794a-output-stage-opamp-measurements-lm4562-ne5532-and-opa2134/. Always trust the measurements, because this is the only way to compare 2 or more electronic devices/components installed on a piece of PCB. .

A good thread exists here: http://www.diyaudio.com/forums/digital-line-level/34324-opamp-conversion-dac.html.

Also, a perfectly designed and implementation of the PCB combined with very good component selection will increase performance (measurements) of the output sound.

 

[raoultrifan]

Now speaking about the PLAY itself, like I said before: using 2xAD8599 in the I/V stage and 1xAD8599 in LPF lowered the output noise (also 3xMUSES8920 is very close). Test conducted with sensitive 16-ohms IEMs and O2 headamp @3.7X gain, volume to the max. with no input source playing from PLAY, but USB connected and computer powered ON. Difference in background noise with such a high volume can be heared around 1AM in the morning when silence is dominating my bedroom. Also, with a good ADC this can be easily proven even with RMAA tests (with my cheapo ASUS U7 ADC I was able to get a 2dB improvement between NE5532 and AD8599).

 

[Eiffel]

And about the LPF - his name says what it does - Low Pass Filter - his role is to eliminate the frequencies over the audible range - there are some residual high frequencies at the output of the DAC. And usually it also act as a buffer - impedance adapter.

 

[WilliamLeonhart]

This thread has been among the most informative I’ve seen in all dac/amp threads I’ve been in.

 

[imran27]

I agree with @raoultrifan that indeed measurements are the quantitative way of comparing electronic devices.

I personally think measurements are just one more aspect. For me personally other apsects like tonality, the clarity and resolution of transients in music is more important.

For example, OPA2209 is definitely more noisier than most of the opamps I've tried, it is audibly noisy. But, the transient resolution or to say the attack and decay is so insanely good it always makes me think that I can ignore the noise floor for the sake of other good things it can do.

Same goes for OPA1692 but it has pretty low noise floor too, in fact audibly it sounds much darker. But I won't say it with confidence unless I try it in an amp while driving an IEM.

And then there's rail to rail decoupling which in my listening is extremely effective compared to rail to ground decoupling. It not only affects the character of sound but improves on the character of the particular opamp.

After testing multiple combinations and rail to rail caps, 1x LME49860 with Nichicon PW, 2x OPA1692 without any caps (not yet tried on OPA1692) and 2x OPA209 with Nichicon VZ sounds the best.

I've tried PW caps in gain stage with OPA209, incredibly analytical to the point of being harsh. VZ caps typically have shown tube'ish character and provide amazing soundstage.

So both of these combined in different opamps and different stages properly give just the perfect sound for me as of yet. I'm on it to try more, like Panasonic FM on I/V stage for eg. Won't try any other caps on gain and lpf (why fix if it ain't broke?).

Peace