dac I/V convertion with very low distortion

[smms73]

Ken i have also made a listening test with 50 ohms and 100 ohms, and i noted differences, with 100 ohms the sound was more loose with a wider stage, but with 50 ohms the sound was more focused specially in the details, personaly i prefer the 50 ohms version. But only listening for about 4 hours and only with my beyerdynamic dt880.

[staccatiss]

Fantastic!!!
Then I will build your IV converter and hook it up to one of my many Buffalo 2 and 3s.

The ESS9018 makes circles around the PCM1794/92, IMHO.

If you can, try the Twisted Pear Buffalo or the AquaDAC made by someone on this forum.


/S

Originally Posted by smms73

staccatiss, yes it can be use with the ESS9018 and others current output dac´s, I will apreciate if you guys tell me the requirements, (as you did) for DAC´s that you guys like. I want to try different dacs in the future.

So for the ES9018 the dc current is 16ma and 16ma p-p out .
PCM1794 dc current is -6.2ma and 7.8ma p-p

[smms73]

staccatiss , how do you know that is necessary less than 1 ohm for lower THD ?
Im just asking because i have discovered that the pcm1794 can have very good THD even with 50 ohms. But i don´t know (yet) the ess9018.

[zenelectro]

Originally Posted by staccatiss

Fantastic!!!
Then I will build your IV converter and hook it up to one of my many Buffalo 2 and 3s.

The ESS9018 makes circles around the PCM1794/92, IMHO.

If you can, try the Twisted Pear Buffalo or the AquaDAC made by someone on this forum.

/S

With the ES9018 the distortion will be significantly higher due to the DAC's
low OP impedance. You can simulate this. To get around this you need to
run more quiescent current in the first complimentary pair.

One of the virtues of the 1794 is that it has a very high OP Z. As such these
super low distortion (semi) open loop IV's can be realised more easily.

I've built a few very similar to these. IME, it's worth playing with the
Baxandall loops. Disabling them will see a increase in distortion due to miller
effects but can actually sound a bit more natural. In the end it's subjective.

The design I'm using atm gets almost as low distortion as these but doesn't
use any Baxandall (+feedback) loops. It does however use SS / tube hybrid.

Maybe even TI don't know what the ultimate capabilities of the 1794 are,
especially in mono mode.

Impressive results.

[MiiB]

looks like we have some good guys here... why not get a schematic on the table, then we all can chip in.... more attention could also lead to a possible group effort and buy...like the one we had in Gerhards MPP thread with the Paradise.

[smms73]

Ok, now i only have to choose a current source.

in this pictures there are 3 current sources the first one does not have a higher output impedance, but have low noise, the second have high output impedance but have more internal noise because R3 is only 18 ohms (this is the current source that i have in the prototype ), i want R3 to be 100 ohms, so the soluction is maybe the third current source where a low noise voltage reference (v1) is used . what you guys think? Suggestions are welcome.

[Joachim Gerhard]

You could use 2 red Leds in series for V1 in version 3. You could shunt the Leds with a high value cap.

[MiiB]

for the best current source use a casodede Jfet.... oh... do you need lots of current then you can parallel more.. the casode makes the impedance of the current source much much higher Tera ohms, but only to a fen hundred KHz than the capacitance swamps it..

[smms73]

Originally Posted by zenelectro

With the ES9018 the distortion will be significantly higher due to the DAC's
low OP impedance. You can simulate this. To get around this you need to
run more quiescent current in the first complimentary pair.

One of the virtues of the 1794 is that it has a very high OP Z. As such these
super low distortion (semi) open loop IV's can be realised more easily.

I've built a few very similar to these. IME, it's worth playing with the
Baxandall loops. Disabling them will see a increase in distortion due to miller
effects but can actually sound a bit more natural. In the end it's subjective.

The design I'm using atm gets almost as low distortion as these but doesn't
use any Baxandall (+feedback) loops. It does however use SS / tube hybrid.

Maybe even TI don't know what the ultimate capabilities of the 1794 are,
especially in mono mode.

Impressive results.

Hi Terry,

I have test this I/V converter with input impedance of 1 Kohm with very good results.

This is not a baxandall pair, it is a complementary folded cascode it serves the same task as baxandall pair ( avoid the distortions caused by the early effect and miller effect ).

This topology don´t use feedback loop, to have feedback you need to have a mechanism that can compare the output with the input, if you inject a signal in the output of this circuit the input is not affected at all. the only thing that is near a feedback loop is the emitter degeneration resistors of the input.

The second current source of post #486 is a good exemple of a feedback loop as Q3 compares the voltage of R3 with his own Vbe and affects Q2 to try to mantain V(R3) Vbe(q3) equal.

[Ken Newton]

Originally Posted by smms73

Ok, now i only have to choose a current source. Suggestions are welcome.

Sergio, while not a direct answer to your question, here are the hyperlinks to a two part article you may not have yet seen which explores various audio current source circuits. It is written by the esteemed Walt Jung.

http://waltjung.org/PDFs/Sources_101_P1.pdf

http://waltjung.org/PDFs/Sources_101_P2.pdf