Sorry about that. Wouldn't want anyone to think I am a heretic

T52, I've update the schematics on my website.

Quote:



I'd use the AD8065 before I'd use the AD8610. If you can stand a +/- 6V supply instead of 12 or whatever you were planning, you can try the AD8057. Currently with the PCM1702 in my DAC, I'm using the LM6171 for I/V after trying 15 or so chips. I liked the THS4051 and AD825 too, though. AD8065 was more lively than the 825 but sacrificed some depth and balance and tended to get some congestion in the mids on some music.

You could also try the AD844 which I found did a pretty good job. Just make sure you use the correct feedback resistor value as it's CFB.

Quote:



The OPA134 will often not be sufficient for this purposes. Opamp for I/V conversion creates a problem. Finite slew rate combined with negative feedback creates slew distortion in I/V conversion. This is one of the reasons that discrete NFB stages often sound better (something i'll be testing this weekend if i bought the right parts ).

OPA627 yes, 637 better, but OPA134 is really pushing it for slewrate.

What about an OPA121ku, what kind of performace would I get from that in that type of I/V? Or how about OPA124?

Slew rate is not the main figure to look at in order to find opamps suitable for I/V conversion. Gain bandwith is. Barrie Gilbert explained how the limited bandwith of the opamps creates distortion and thus why discrete I/V could sound better. But of course look also for settling time, noise and slew rate. There's a very good discussion here : http://www.diyhifi.org/forums/viewtopic.php?t=262

But of course, all this doesn't matter here. The I/V is done by the resistors in a passive fashion. The opamp is just a balanced to unbalanced converter. An opa134 is all that's needed

How about an OPA551 in the I/V stage?

The slew rate is kinda low (15V/us), isn't it?

Slew rate is absolutely everything in opamp I/V. Go for the LM6171, just be careful of the layout with an opamp that fast.

C35 in particular is pretty far away... I'd move R28 if possible so C35 can get closer. Maybe make the feedback resistors SMD?

Quote:



There is no opamp in the I/V stage.... read my post again.

And for slew rate... it's not everything. It is one parameter among others. You just need an amplifier "fast enough". From memory, it has been calculated that for the PCM1704, a 50V/us opamp was more than enough (the analog output of the DAC isn't that fast and no musical signal has square waves anyway). The LM6172 is a good choice because of the mix of good specs (though some other opamps can compete now).

Quote:



The images on the output of a DAC are flat to 20Mhz or more. Put a spectrum analyzer on a DAC and say the 50V/us is enough. I understand that the GBW is the important figure, but... A) no opamp with 50v/us is going to have enough GBW anyway, B) The lead inductance of a bypass cap on the feedback resistor of a nonidealized I/V circuit is going to get you into trouble with slew rate if your opamp can't keep up.

just out of curiosity:
if you have a passive i/v conversion with a resistor and a opamp buffer afterwards - will this have the same slew rate demands as compared to the opamp i/v?

Quote:



A capacitor bypassing the I/V resistor can make a first-order filter before an opamp (or whatever) based buffer/filter. This'll take a big whack at the particularly high freq images and massively reduce the requirements of the opamp as long as you're very careful of lead inductance. Still the LM6171 is pretty good sounding so I'd go with that for further filtering unless there was a particularly good reason not to.

I know resistor I/V has compromises, but more and more I think outside of spice models it may be king.

Quote:



With digital filtering and a capacitor accross the I/V resistor ?

The figure of 50V/us comes from there btw : http://www.diyaudio.com/forums/showt...749#post401749

Quote:



i'm pretty sure that digital filtering will not influence this...

Quote:



Digital filtering doesn't change the high frequency noise, in fact the faster the DAC itself runs (including it's internal oversamping for anything less than straight 18-20-24 multibit) the higher the noise spectrum goes. Lynn Olsen measured a PCM63 straight multibit as utterly flat to 20 Mhz and it didn't fall into the noise floor until 50 (!) Mhz. Modern multibit sigma deltas w/ thier internal oversampling probably go even higher.

The cap across the feedback looks like its working until the opamp gets close to unity GBW and then everything falls apart. You need incredible speed and GBW. 1-2kv/us & 100Mhz+, and even then...

EDIT: Whoops, Did you mean the resistor in resistor I/V stages, or the feedback resistor in opamp I/V? Above text pertains to caps across feedback resistor in opamp I/V. I believe those measurements were taken with a small resistor I/V with no bypass cap. The funny thing is the lower the I/V stage's input impedance, the harder it becomes to filter the noise. So I suspect there's a balance between noise and high-i/v-impedance distortion that is different for every DAC chip & system & listener. Discrete active I/V stages could have both very low input impedance and handle the filtering, but we're talking about HF/VHF stuff that has to be amplified & filtered without issue. It would probably mean no feedback and very very good RF layout design work.