Aaargh!
Such a listening tests with DAC and different opamps have one weak point - we doesn't know high-frequency noise and glitch from DAC chip.
In such a conditions an opamps with relatively high unit gain freq and deeper feedback in the audio band (usually beter in terms of measurements and formal specs) can be easily outperformed with a relatively cheap and primitive NE5534. This fact can be easily explained. Low GBW opamp exhibit lower input overloading effect from the described input signal and can have no issues with a high-freq feedback loop.
Such a listening tests with DAC and different opamps have one weak point - we doesn't know high-frequency noise and glitch from DAC chip.
In such a conditions an opamps with relatively high unit gain freq and deeper feedback in the audio band (usually beter in terms of measurements and formal specs) can be easily outperformed with a relatively cheap and primitive NE5534. This fact can be easily explained. Low GBW opamp exhibit lower input overloading effect from the described input signal and can have no issues with a high-freq feedback loop.
I'd be interested to know how you can setup blind, ABX tests for this, since typically the opamp is socketed and someone's got to go in and pull/replace a chip each time - ideally you just have this switched at random by a computer as you run through the test.
What is the justification for using ABX as verses some other blind protocol? Blind is blind, isn't it?
That would depend on the details of setup. Your statement is equivalent to saying, "Loudspeaker is loudspeaker, isn't it?" 🙄
Pick your favorite then, its the generic blind comparison test, that's all. My issue is swapping opamps in I/V converter for a DAC is not a trivial task for unbiased testing.
https://en.wikipedia.org/wiki/ABX_test
If you are designing your own DAC you can assemble more than one prototype and then switch between them.
Just wanted to input a subjective testimonial with both opa1611 and 1656 as I/V of an AD1862 DAC, NOS, no further filter or buffer either on miro1360 member pcb.
The 1656 is sounding to me subjectivly better than the Bjt input 1611. Bass behavior is disapointing from what I expected from the Bjt. Here I suspect decoupling problem as layout cause diy subjective choices.
I/V resistor is through hole so a little high with extra lead length when I use bigger resistors. I assumed maybe because far from the ground return signal there is an added extra capacitance hence no few pico or nano added extra cap needed in the feedback loop. But I dunno. Maybe the 1611 needs more pF or nano F.decoupled Rf loop. I have no measure so subjective of course.
But quite happy on how you can have a great sounding dac today with a minimal parts choices. Thanks Ti and others for the discussion about that opas'.
So layout is a concern. I will try later opa164x family which I hope is less demanding as I/V for the layout in a minimalist design (no buffer or further filter).
Any new opas on the bench today ?
The 1656 is sounding to me subjectivly better than the Bjt input 1611. Bass behavior is disapointing from what I expected from the Bjt. Here I suspect decoupling problem as layout cause diy subjective choices.
I/V resistor is through hole so a little high with extra lead length when I use bigger resistors. I assumed maybe because far from the ground return signal there is an added extra capacitance hence no few pico or nano added extra cap needed in the feedback loop. But I dunno. Maybe the 1611 needs more pF or nano F.decoupled Rf loop. I have no measure so subjective of course.
But quite happy on how you can have a great sounding dac today with a minimal parts choices. Thanks Ti and others for the discussion about that opas'.
So layout is a concern. I will try later opa164x family which I hope is less demanding as I/V for the layout in a minimalist design (no buffer or further filter).
Any new opas on the bench today ?
I made three preamps (actually Vunce built them for me 🙂): one with OPA1656, one with OPA1642, and one with LM4562. IMO, the 1656 and 1642 were very similar and great sounding - natural and very enjoyable. The LM4562 was “drier” but still very good. Given a choice, I would pick either 1656 or 1642, depending which one is in stock. But given the parts shortages, the 4562 is not a bad substitute. I did use OPA1637 balanced line drivers on all of them.
check out post #16 and #19 under
https://www.diyaudio.com/community/threads/i-v-and-analog-output-stage-for-ad1862.299414/
and this thread:
https://www.diyaudio.com/community/threads/diy-d1-i-v-stage.181449/
It would be interesting to know, what about the sonic character with use of AD844 successors and the discrete I/V version from Nelson Pass' D/A converter "D1".
I do like the opa861 introduced by Rogic as a successor of the ad844 and op660 but have not compared them in the same dac VS the 1656 ot 1612.
100% Dissrete has its own layout problems I surmise. More H2 and that is what is liked by people ? Ido not know...still not convinced by 100% discrete. We are in the XXI century after all.
100% Dissrete has its own layout problems I surmise. More H2 and that is what is liked by people ? Ido not know...still not convinced by 100% discrete. We are in the XXI century after all.
H2-H9 isn't an audible problem in general - check out post #2 and the attachments
Check out this buffered opamp module i use for my dual mono preamp. Ive tried 5534, 1655, 1641. With the latter two its simply nirvana. The module has pretty much an ideal opamp layout (taken from the AD guideline) and it beat all of the dozen other circuits ive tried for this preamp including buffers and discretes. 16xx series ftw 😁👍
Attachments
Nice. Similar to the three I tried (but mine were the dual opamp variants). I agree the 165x and 164x sound fantastic.
Its nothing special really.
Opa> 4 transistor classic diamond buffer and mark levinson style global feedback. 47pf from pin 5 to 2 for opamp stabilizing.
The traces are extrememly short- almost p2p in most junctions and i think thats why this sounds as good as it does.
Bravo,
Rvas member also showed me in 1666 John's Thread a well made pcb layout around the 1656 and any nowadays wide bandwith opas. Guard rings, power supplies traces not too close to input and output pins, short traces for better inductance, very close decoupling with small inductance caps, and well managed return paths.
Inspiring...thanks guys.
Rvas member also showed me in 1666 John's Thread a well made pcb layout around the 1656 and any nowadays wide bandwith opas. Guard rings, power supplies traces not too close to input and output pins, short traces for better inductance, very close decoupling with small inductance caps, and well managed return paths.
Inspiring...thanks guys.
Hello, I want tu buy some OPA1642, in mouser.com apears two options OPA1642AIDR and OPA1642AID, what's the difference, wich one should I buy? Another option could be AliExpress, somebody now a reliable store? Thanks.
