[Sapientiam]

At the risk of going uber-geeky, what inductor did you find that gives you 50dB attenuation at 5kHz?
Or have I misread what you're saying?
 
There are different types of ceramic caps - I have found NP0/C0G to be very neutral sounding caps.
But then I also listen via some X7Rs as coupling caps in my current setup which kinda makes me a heretic.

 

[project86]

Quote:

 
project86, youre safe with Demian, the man knows his stuff. i'm running one of his regulator/ isolated clock designs (diy) with my i2s fifo buffer (which completely isolates/decouples the dac from the source, so the clock jitter and power supply/ground noise are all thats left to tweak). well just finished the reg, will initiate testing with fifo soon. Probably after the sydney meet though, I dont want to be stuck with my dac in pieces on the bench the week before the meet again.

 
Thanks for the reassurance. One can Google for internal pics of the PK90 compared to others (Cary MS-1 for example). You'll see that the Auraliti is clearly a purpose built machine for audio playback, while the Cary is simply a SFF PC in a custom enclosure. Nothing terribly wrong with that by itself, but it won't play in the same league as the PK90. Which happens to sell for half the price.... what's wrong with that picture?

 

[Viper2001]

Quote:

At the risk of going uber-geeky, what inductor did you find that gives you 50dB attenuation at 5kHz?
Or have I misread what you're saying?
 
There are different types of ceramic caps - I have found NP0/C0G to be very neutral sounding caps.
But then I also listen via some X7Rs as coupling caps in my current setup which kinda makes me a heretic.

 
The inductance is 1.5mH.  But the effectiveness also depends on the resistance of the choke, the capacitance and the ESR of the cap.
 
Yes, I'd only use NP0/C0G as far as ceramic is concerned.  Try a SMD film cap for coupling, it should sound better than X7R.

 

[Sapientiam]

With a 1.5mH choke, watch out for the SRF - 'self resonant frequency' - its unlikely to be much over 1MHz, if that. So putting it in series with a lower valued choke will help the HF rejection. Do they make SMD film caps in 47uF? I'm using several paralleled X7R to make that value, but being 0805 they take almost no space. Any specific manufacturers to recommend?

 

[veritas aurum]

See if they make nobium oxide in 47uf. They are a higher quality replacement for tantalum. Regards

 

[veritas aurum]

That should be niobiium. AVX is the manufacturer, but only up to 10v. Regards

 

[Viper2001]

Quote:

With a 1.5mH choke, watch out for the SRF - 'self resonant frequency' - its unlikely to be much over 1MHz, if that. So putting it in series with a lower valued choke will help the HF rejection. Do they make SMD film caps in 47uF? I'm using several paralleled X7R to make that value, but being 0805 they take almost no space. Any specific manufacturers to recommend?

 
One can also 'extend' the freq range by adding a 1K resistor parallel with the choke.  But of course, with an expense on having lower impedance at the entire freq range.  What I'm trying to filter is the noise from the SMPS and DC-DC converters coming from the SBT via the USB 5v line.  The DC-DC converters usually operate around the 500KHz freq.
 
No one makes SMD film caps that large.  The largest is 1uF.  In what signal coupling situation do you use this for?  With film caps, you only need about 2-4uF for signal coupling b/w say a DAC chip and the op-amps.  Even at 2uF, you have enough corner freq to cover down to 20Hz.  I use Panasonic ECPU 1uF stacked to make a 2uF signal coupling cap b/w AKM4396 and the op-amps in my Transporter, and they sound great.  FWIW, the datasheet calls for 100uF polarized lytic caps in that position.

 

[Viper2001]

The LC filter is now installed in my D2 (U2).  It only filters the high freq noise for the power supply to my clock circuit, and not the rest of U2.  The reason is that U2 reclocks the S/PDIF output of the XMOS chip, so it only makes sense to have clean power to the clock and the reclocking circuit, and the rest can still feed on the noisy USB power and it won't matter.  So far only my clock circuit is feeding on this LC filter, but the reclocking circuit is still feeding on the USB power line.  But I did change the reclocking circuit's voltage regulator to an ultra low noise one (9uV RMS from 10Hz - 100KHz).  The reclocking circuit will tap its power on the LC filter once I find time to make that change.  I'll need to think about how to do this best without cutting any circuit trace on U2.
 
So far, the sound is the most analog and the most liquid I've heard.  All those high freq harshness from a stock D2 is pretty much gone.  There is an increase in reverb/decay and a better contrast, due to a darker background.  Transient/attack are as impactful as they can be.
 
More to come next week.

 

[Sapientiam]

Quote:

 
One can also 'extend' the freq range by adding a 1K resistor parallel with the choke.  But of course, with an expense on having lower impedance at the entire freq range.  What I'm trying to filter is the noise from the SMPS and DC-DC converters coming from the SBT via the USB 5v line.  The DC-DC converters usually operate around the 500KHz freq.
 
No one makes SMD film caps that large.  The largest is 1uF.  In what signal coupling situation do you use this for?  With film caps, you only need about 2-4uF for signal coupling b/w say a DAC chip and the op-amps.  Even at 2uF, you have enough corner freq to cover down to 20Hz.  I use Panasonic ECPU 1uF stacked to make a 2uF signal coupling cap b/w AKM4396 and the op-amps in my Transporter, and they sound great.  FWIW, the datasheet calls for 100uF polarized lytic caps in that position.

 
I can't follow how a shunt resistor 'extends' the range. It certainly damps down any resonances though.
 
You might find this LT appnote interesting in getting rid of noise from DC/DC converters, by the analog guru
Jim Williams http://cds.linear.com/docs/Application%20Note/an101f.pdf. Ferrite beads are your friend.
 
Where I'm using 47uF is in the input to a very low impedance preamp (AD605) which I'm using as a volume
control - I'm concerned about group delay at low frequencies so I like to set the corner freq down below 4Hz.

 

[Viper2001]

Ok, the first U2 clock mod just rolled off the assembly line

  LC filter to the clock power supply is also in place.
 
Have not replaced the output pulse transformer yet.  Want to hear how it sounds first before committing the transformer change.
 
Still plenty of clearance to fit the whole thing back in.
 
Lets burn-in for a few days and see how I like it.  But so far so good

 

 

[cky8]

Wow! Nice one Viper. I wonder whose U2 board that is 

 
Yeah it is good to have multiple reference points of how it sounds throughout modding, before changing the output pulse transformer.

 

[Bubu1]

Great pic Viper.  Please keep us posted.  I have been using my stock U2 since March and have been quite happy with it.  If the mods go well perhaps I'll send you mine 

Keep up the good work and thanks for sharing!

 

[Viper2001]

Finally changed the output pulse transformer to an ultra low leakage inductance one.  The sound is a tiny bit more relaxing and smooth.  Overall, it is a bit more analog.  There is still plenty of space after fitting all those components

 

 

 

[danik97]

I'm choosing between D2 and Violectric V800. 
Which would be better with HD800 in pair?

 

[Viper2001]

Quote:

I'm choosing between D2 and Violectric V800. 
Which would be better with HD800 in pair?

 
I'd think D2 is the better one, after all, this is the Anedio D2 thread

 
Ok, seriously, by looking at the internal pics, D2 is using much higher quality components.  Better DAC chip, better clock module, etc.  Also, D2's published specs measured spectacularly, including the headphone section.  The only problem is the long wait before their webstore opens again for D2 sales