Build thread for Diyinhk ES9018 DAC on Ebay

[KlipschKid]

Hi,

I've migrated the info I posted here to my blog ;

http://myl8test.wordpress.com/audio/...18-32-bit-dac/

cheers, Tom

[abraxalito]

Does a schematic exist for the I/V stage? I'm just curious as nowadays I'm designing I/V stages I rather suspect that if the LME49990 is being used for that function its almost certainly not the best choice for SQ owing to having an undegenerated low noise LTP input stage which IME are very susceptible to HF overload.

[KlipschKid]

Hi,

There's one in the datasheet, looks like this, without the balanced parts (link to GLT's page):

[abraxalito]

Thanks - the problem there for the sound is connecting the DAC output directly to one side of an LTP (long-tailed pair - the opamp input stage). That's a recipe for some serious intermodulation distortion which doesn't show up on sinewave testing. But you'll hear it as digital harshness, loss of soundstage depth, glassiness in massed strings and general loss of dynamics.

I reckon that changing the input opamp to something that will overload less easily will pay dividends in SQ, I'll check out some suggested parts. The amps which sound best will avoid the LTP and will hence be CFB architecture which would mean some slight circuit board tweaking...

[KlipschKid]

With other DACs, I found some prefer FET in the I/V so for them, ADA4627 are my current favourites, with LME49990 as buffers. If you have the time to go through op amp datasheets to find the optimal ones for this circuit, I'm sure I'm not the only who'd be grateful !

[abraxalito]

Yes using FETs is indeed a step in the right direction as they're not so non-linear when overloaded as bipolars are. I know that ADI has at least one part which has improved input stage linearity - one of Barrie Gilbert's innovations a 'multi-tanh input stage' - so I will see if that one is possibly compatible with this circuit.

[abraxalito]

Here's my first suggestion to try - this one doesn't need any circuit changes at all.

http://www.ti.com/lit/ds/symlink/lm6171.pdf

Notice on page13 its showing a simplified schematic - not the normal LTP input stage, rather internally its a current-feedback (CFB) amp. Hence no overload problems - lots of internal current available to handle huge slewing demands.

Here's a similar (looks better but only SO8 package, no DIP) part - LT1818 is the single - http://cds.linear.com/docs/Datasheet/18189fb.pdf

Ooops, forgot to mention that LT1818 is limited to 12V total supply, max.

[glt]

Hey abraxalito,

Thanks for sharing your knowledge. Truly appreciated :-)

[KlipschKid]

I've listened to LM6171 - I used them with LM6172 in AMB's Mini3. The signature is okay, if a little bright and over-detailed. However, their performance is below that of the DAC IC so they won't do the DAC justice. I preferred the sound from LM7322/LM7321. So ltp or cfp matters, but so does thd+n, signature, etc. I've grown to seriously dislike op amp rolling - compromises...

I have a list of op amps that are "good enough" according to their specs :

OPA627, OPA827, OPA1641, OPA211, ADA4627, AD8597, LME49990, LME49710

and I'm sure there are more.

[abraxalito]

For +/-5V supplies - perhaps using TL431 local shunts on each chip - then OPA690 looks rather hard to beat. Again no DIP version but the price looks highly reasonable ($1.42, 1k) for the level of performance on offer.

http://www.ti.com/lit/ds/symlink/opa690.pdf