[mnaglik]

Hi All,
First of all forgive me that I'm complete beginner if it comes to circuit design.

Because I wanted to have good quality DAC for my Alessandro MS-1 (Grado) I've built usb DAC based on PCM2702 / OPA2134 / BUF634 from circuit that I got from one of my friend (attached).

The problem is that because of lack of OPA2134 I used Analog AD8066 and the result is that in my opinion the output is just to bright and harsh (It's a little better on Sennheiser).

Do I need in fact tune the circuit to play properly with AD8066 instead of OPA2134?
Maybe sth is completely wrong?
Thanks for any hints.

Marcin

 

[MASantos]

is that supposed to be connected to a amp after or drive theheadphones directly? I think you could use a simple resistor based I/V. Much easier and cheaper than OPAMP+buffer.

 

[mnaglik]

Quote:

Originally Posted by MASantos is that supposed to be connected to a amp after or drive theheadphones directly?

Directly to headphones.

 

[MASantos]

I'm not sure if I'm correct but you should increase C7 to about 4.7uF because otherwise it creates a low pass filters. In the datasheet of the BUF634 there is a schematic for a headphone driver Using and opamp plus buffer configuration. It is a simple configuration but simple to implement and should sound good. There was a picture of such an amp built by a member here and he stated that it sounded bery good. You could also go with a 3 channel design similar to PIMETA or PINT.

 

[jcx]

leave C7 alone - it is part of the "multiple feedback" 2nd order filter - you can't change just one part without changing the filter Q and corner frequency - you really want to know you're doing here


a likely problem is that the AD8066 is simply too fast for this circuit - I would expect oscillation in the 10s of MHz when wrapping a 140 MHz GBW op amp around a 30 MHz BUF634 without any frequency compensation

even with suitable loop compensation the AD8066 would need superior local bypass and possibly gnd plane cutouts and isolation R to work by itself at such low gain - not a "drop in" replacement for a 8 MHz GBW OPA2134

the simplest compensation that might work with the AD8066 is to add a ~20 pF cap between pins 6,7 to give a local feedback that slows down the output servo amp - I would still want to look at this with a 100+ MHz oscilloscope with a low C probe

 

[tomb]

I have an 8066/8065 combo in a PIMETA. It definitely tends toward a little on the harsh side, even in a good setup. It can be sharp enough to have a "bite," you might say.

I can't scope it, but as far as I know, they are not oscillating. I have them in a double buffer mode at 18-24V, with everything else stock PIMETA. My DC offset is practically zero. However, I can read surface temp of the chips, and they seem to consistently run at about 50 deg. C. Whereas an 8620/8610 combo in the same exact setup runs at about 30-35 deg. C.

From what I've read, 50 deg. C. is not that unusual, so I'm thinking my impressions of the 8066/8065 may be accurate: scalpel sharp. With great headphones it can sound good. With others and certain types of music, it can sound very harsh.

 

[cetoole]

Quote:

Originally Posted by MASantos is that supposed to be connected to a amp after or drive theheadphones directly? I think you could use a simple resistor based I/V. Much easier and cheaper than OPAMP+buffer.

I wouldnt recommend using a resistor I/V here. Actually, I wouldnt advise it anywhere, but especially not on a vout dac.

 

[jcx]

Quote:

Originally Posted by tomb ...I can't scope it, but as far as I know, they are not oscillating.

In Wide Bandwidth mode BUF634 @ >150 MHz and with a higher overall gain around the whole loop this is probably pretty safe in the PIMETA using a AD8066

Low Bandwidth mode leaves the BUF634 at only 30 MHz bandwidth - the faster AD8066 is likely to "oversteer" the slower buffer into oscillation/massive overshoot with the posted circuit with a loop gain of only +2

for buffer-in-the-loop feedback amplifiers the op amp GBW/gain < Buffer bandwidth - preferrably by factors of > 4x

as designed:
8 MHz(opa2134)/2 = 4 MHz
30 Mhz (low bandwidth mode buf634)

30/4 = 7.5 an OK speed ratio

with a 70 -"145" MHz AD8066 and 30 MHz buffer you're looking for trouble at low gains like +2

 

[tomb]

Well, on my PIMETA, the bandwidth resistor is at 100ohms, which should ensure the double-stacked BUF634 is in high bandwidth mode. Also, stock gain = 11 is used, so it should be OK.

All of this is to say that some harshness from the 8066/8065 may be completely normal, I think.