[superrobin]

Hello all, I'm new here, this is my first post
 
I have bought a XONAR STX for my home system, and after numerous tweaks and mods, I finally reached the sound signature that I like
 
If someone is interested I can post later the list of my mods
 
The last one was replacing the crystal clock with a Vanguard TCXO 0.3 ppm, available from a Hong Kong vendor on Ebay
 
I saw many post on the Net about the quality of this oscillator, and the benefit it brings on this sound card vs the stock oscillator
 
I have to say in my case the result is not satisfing, a step backward in the overall sound quality
 
The sound becomes dull, sterile, with harsh vocal sibilant, not immediately noticeable, but in the end I found the music is turned gray, uninvolving, and fatiguing after a while
 
I tried to supply the oscillator with a dedicated very low noise PSU instead of the onboard 7805 , but things doesn't improve
 
Restoring in place the stock oscillator gives definitely a better result
 
I can't understand the reason for this behaviour, I can't measure it, but it seems the vanguard XO exhibit an higher jitter than stock one
I noticed also the stock XOon my STX is marked IT405, and all pictures found on the NET show IT475, maybe ASUS has replaced this component with a better one in recent items?
 
Any idea, suggestion, please
 
Robin
 
 

 

[Groundloop]

Or your Vanguard was fake

 

[superrobin]

Yes, I though the same
 
But over the Net there are a lot of good reviews about this component, just only one had a similar experience
 
There is only one vendor AFAIK, so could be a matter of random quality
 
 

 

[superrobin]

The inferior performance of the Vanguard is visible also through the oscilloscope
 
Playing a test tone, sine wave 11.025 Hz, while with the stock quartz the sinusoid plotted on the screen is not very stable, peaks seems to oscillate in the horizontal axis, with the vanguard the sinusoid is moving crazy like a snake...

 

[Groundloop]

Have you modded the STX somehow other than the XO?

 

[superrobin]

Yes
 
1) Replaced NJM 2114 in I/V with LM4562
2) Replaced 2.7n film caps in output stage feedback loop with wima MKP 3.3n
3) Replaced electrolytic output caps with 150 ohm resistor
4) added wima FKP 3.3u to opamp bypass caps 
5) added 2 tantalum cap 10uF in parallel to 3.3V and 5V bypass
6) added 2 nichicon FpCap 220uF in parallel to molex input 12V and 5V
7) added ferrite core coils in series to molex plug
 
All these mods changed dramatically the sound signature for the better, more detailed sound stage and instrument texture, refined treble, punchier, tighter, deeper bass (who says lm4562 is bass shy? try to filter out all HF quantization noise from I/V section)
 
In my mind the last important mod should have been the clock upgrade, but something went wrong ....

 

[superrobin]

IMHO the biggest fault of the XONAR STX, is the output stage cut frequency: they copied the first implementation example schematic from PCM1792 datasheet, with a fc of 217 Khz !!!
 
Far too high to remove the quantization noise in the 400Khz region
 
I measured a lot of HF noise coming out from the RCA, and my amp doesn't like it at all
 
After all listed modifications the noise decreased from 6-7 mV to 1.4 mV peak to peak for a 44.1 Khz, and under 0.1 mV for higher sample rates
 
Yes, I discovered this sound card (or just the PCM1792?) gives the worst noise performance so far in 44.1 Khz sample rate, thus upsampling 44.1 material to 176.4 with Sox under Foobar gives another noticeable step up in sound quality

 

[Groundloop]

Are you sure you have not something that is oscillating? Adding caps chaotically can create resonances.

 

[superrobin]

No, not chaotically, I follow a method
 
I add caps once a time and I carefully measure ripple before and after with oscilloscope
 
Sometimes I had to come back and remove the cap or change for a different type or value
 
In some places I found improvement with stacked film, in others tantalum gave better results
 
In the end now I have reduced ripple and oscilaltions on PS lines, a lot !!

 

[superrobin]

But anyway this is not the case, I supplied the Vanguard with a dedicated PSU taking power directly from 12V on the molex
 
My only doubt is about ground
 
 

 

[leeperry]

 

IMHO the biggest fault of the XONAR STX, is the output stage cut frequency: they copied the first implementation example schematic from PCM1792 datasheet, with a fc of 217 Khz !!!

 
IMHO, get the STX back to stock and move on. With all this money and time spent, you could have gotten something far better sounding right off the bat.

 

[superrobin]

No, few money really, few euros in caps, the biggest expense is the Vanguard, 25 euro shipped
 
A lot of spent time of course, but this is part of the fun
 
Anyway I owned other dacs and CDP, and none of them approached the SQ reached by my modded STX, nor the same stock card
 
 

 

[leeperry]

Well, the STX sells for 160€ + the 25€ Vanguard...you still got better options like I dunno, the NFB-12? I didn't like it too much when I tried it but it was the version w/o the oversampling jumpers and it still sounded heaps better than an ST(which an improved STX w/ a reclocker) to my ears. What you're trying to do is about polishing a turd really, that hardly ever works IME.

 

[superrobin]

I agree with you, when I bought the STX, one year ago, I was disappointed; harsh and aggressive, not very pleasant
 
The return policy in Italy doesn't include money back, so I had two options: resell immediately, loosing at least a third of the value, or try to address the card faults
 
There were so many people swearing for STX, claiming it was the best value for money available on the market with few mods
 
And I like so much modding everything .....
 
NFB-12 ? I will check for it 
 

 

[superrobin]

Wow, that NFB-12 looks really interesting ...
 
what about feedback?
 
I heard about a lot of faulty items from this manufacturer