Coupling caps on CD player output?

[Pars]

I've been doing a bit more work on my Rotel RCD-855 (TDA1541A, SAA7210/7220) player. When I have LM6172s in the analog output section, the sound gets a touch "tizzy" in the highs. The stock opamps were NE5532s. I have a 0.33 stacked film soldered onto each of the 6172s from the V+ to V- pin, and just added 0.01uf ceramics (Kemet) to the V+->gnd and V-->gnd (I haven't done any listening to it since I did this). The player has 100uf BG F std. coupling caps (as well as same for V+ and V- bypass caps). I cannot detect any signs of oscillation... scope view of both the PSUs and the output pins look good. Also, no heat.

Someone on AA commented that the BG stds. can give a "tizzy" high end sound. I was toying with replacing the coupling caps with some 2.4uf BC 383 Polypropylenes (because I have them laying around), but was conerned about what effect these might have on frequency response. The player has 22K to ground, with the 100uf in series, so this forms a low pass filter. Substituting the 2.4uf would give a cutoff of ~3Hz. My preamp has an input impedence of 20K (tube preamp, so I assume its somewhat higher than normal?), and I assume that this 20K would be effectively in parallel with the player output impedence? Any thoughts?

One other thing: I am thinking about building one of Jocko's Simple Discrete IV stages for this. Do I need anything after this stage? Any high pass filter from the DAC? Buffer? Anything?

Thanks

 

[Garbz]

If the 20k is on the far output then yes it is in parallel, and there are very good reasons for this. Firstly it makes the circuit behave more predictably. Input impedance can vary from 10k to 660k in my case. Paralleling with a 20k resistor reduces this variance to 6.6k - 19.4k. Secondly it provides current to charge the capacitor so you do not turn the unit on and have it charge the cap through your preamp.

The use of Jocko's I/V stage would depend on the circuit and the DAC. As far as I can remember the circuit was never fully complete although I have not followed it too closely. Can you provide a link? The output of a DAC typically needs I/V conversion and a lowpass filter to remove the high frequency crap left over in the DAC process. The value of this filter again depends on the DAC. For my PCM1730 I chose a single pole filter tuned at 40khz

 

[Pars]

Quote:

Originally Posted by Garbz If the 20k is on the far output then yes it is in parallel, and there are very good reasons for this. Firstly it makes the circuit behave more predictably. Input impedance can vary from 10k to 660k in my case. Paralleling with a 20k resistor reduces this variance to 6.6k - 19.4k. Secondly it provides current to charge the capacitor so you do not turn the unit on and have it charge the cap through your preamp.

Thanks for the reply Garbz. Yes, the 22K follows the cap. Verified by measuring it.

Quote:

Originally Posted by Garbz The use of Jocko's I/V stage would depend on the circuit and the DAC. As far as I can remember the circuit was never fully complete although I have not followed it too closely. Can you provide a link? The output of a DAC typically needs I/V conversion and a lowpass filter to remove the high frequency crap left over in the DAC process. The value of this filter again depends on the DAC. For my PCM1730 I chose a single pole filter tuned at 40khz

Pretty much a standard Philips circuit with Rotel tweaks and better parts here and there. TDA1541A DAC chip... this is an R2R, correct? I though I had read somewhere that R2Rs don't need the LPF on the output? This one does have the more or less standard (it would seem, but I'm just starting to get into digital and have much to learn) 2.4k resistors and a 470uH inductor in between the first opamp stage (1/2 of a dual chip) and the second stage which is just a gain of 1 buffer (the other half). I can post schematics if you or anyone else wants to see them (Rotel gives them out for free if you ask them).

I don't have a link to the "real" jocko circuit, though the one he posted on DIYaudio is supposed to sound very good (just not quite as good as the completely optimised one). I thought I would throw something together with transistors I have laying around (BC328/338 or Gilmore 2SA1015/2SC1815) on a breadboard and see what happens

 

[Garbz]

It is an R2R ladder based DAC but they also still digital noise on the output. The analogue filter depends a lot on the digital filter.

Yeah post the schematics. I'm working on a different I/V stage myself. It seems the circuits I've seen on DIYAudio still rely on passive I/V but the discrete active stage appears to isolate the I/V from the DAC and provides sufficient compliance that it actually works. In all theory an opamp is "ideal" for I/V but you know how the non global feedback fans get.

Can you also post a link to the DIYaudio page you were referring to. I never was able to find jocko's original post but I did find a few adapated posts.

I am in the same boat as you. Trying to whip something together out of the Gilmore transistors for my current DAC. I am essentially going to destroy my DAC as a lab experiment for my next DAC to give me a bit more of a clue as to how things work.

 

[Pars]

Garbz,

Thanks for the reply. I only have the Rotel schematics as a .pdf in 3 parts (schematic section is 1 part of 2.7Mb). You can access it here.

As far as Jocko's original:

http://www.diyaudio.com/forums/showt...&threadid=6121

And an interesting thread (of many) on diyhifi

http://www.diyhifi.org/forums/viewtopic.php?t=190

Clutz has a thread or two related to the original or I/Vs in general as well. Nicke apparently figured out Jocko's full circuit.

 

[Garbz]

Have you had a look at Hawkfords paper that Nicke mentions? I'll email it to you in a sec.

I have also decided to try something else which has been discussed in length at diyaudio. The Passlabs D1 I/V stage which uses IRF610s of which I still have oodles laying around.
Jocko's full circuit is worth trying at some point as well. I've built my DAC in a modular way so I can strap on any stage I wish at this point. I just need to make a powersupply for it. My current DAC's analogue PSU has only +/-12V :S

 

[Crowbar]

The Hawksford paper is a gem, as usual. But cut down on the current feedback (I wouldn't completely remove it, however); error correction decreases the need significantly, without the negative side-effects of NFB.