Is now quite long time since I did those Xonar STX modifications presented here. I thought it was finished... but in the last time I found out about a new oscillator from Crystek. My old Ebay 0,3ppm oscillator is now replaced with this last 24,576Mhz Crystek one. This oscillator is an 3,3v one. So I had to go down with the regulated tension from 5v to 3,3v, and I used an low noise dedicated 3,3v regulator. The new oscillator is now supplied from 3 regulation levels: 12vDC from Molex is down regulated to 7v, then to 5v, and at last to 3,3v (shunt and voltage regulations).
It seems that the signal output level from this 3,3v supplied Crystek is enough to well run the clock of CMedia processor.
For those who can buy a such last edition Crystek oscillator I strongly recommend to do it so and use it for the ST/STX board. It cost 26$ at Digikey. The result is fantastic. I just was not aware about how high fidelity can come out from an PCM1792 DAC with this last high quality oscillator (and the rest of my modifications on STX card).
It seems that the signal output level from this 3,3v supplied Crystek is enough to well run the clock of CMedia processor.
For those who can buy a such last edition Crystek oscillator I strongly recommend to do it so and use it for the ST/STX board. It cost 26$ at Digikey. The result is fantastic. I just was not aware about how high fidelity can come out from an PCM1792 DAC with this last high quality oscillator (and the rest of my modifications on STX card).
Last edited:
The xonar st does provide phantom power but only 5 volts as is per standard in PCs. The input is switchable via relays, in line in mode you get what you'd expect, when you swap to the mic, via the software, the relays click in and turn the input into signal, phantom power and common ground. I don't think the mic input goes through the same ADC as the line input though.
Hi, I'm modding and trying to improve Xonar Essence ST since a few months. I use it mostly as an DAC and headphone amp for various DIY headphones. I tried a lot with amp power decoupling using amp socket. Also tried a lot of different opamps with different feedbacks. Headphone driver, cables and plugs made also change to how the card works.
However no changes were made to the original card. I noticed the I/V amps have electrolytic capacitors near them and I may try to utilize them as they may cause some resonanses with all the setups. Low ESR cap inserted into the RCA output amp can change how the I/V amp work even if I use headphone amp as output. That's why I may try to bypass existing caps and change them as they may interfere.
Here I'm asking if I should desolder these Nichicon Muse FG caps or cut their legs and use them to solder different one? There are always 2 caps near important chip. Do they utilize ground and use +12V/GND/-12V or the ground is not used? When I decouple using amp socket I cannot access ground plane and use at least 35V caps. 25V tantal cap put here was blown away in a few seconds after computer start.
Also this could let get better access to the TPA amp power.
Proper power decoupling is very important for these opamp as they can behave very unstable and their settling time can be very large. Different cables and plugs causes a lot of changes in the settling time and feedback delay. Large headphone amp feedback with very good cables and plugs can give very good results but increases the power quality requirements. When it lacks power, op-amp can behave very strange causing oscillations depending on the components used in headphones or just saying "I'm off, I'll be slower". The option is lowering feedback and putting 1kohm serial resistor into the output but that decreases the resolution but makes op-amps more stable.
However no changes were made to the original card. I noticed the I/V amps have electrolytic capacitors near them and I may try to utilize them as they may cause some resonanses with all the setups. Low ESR cap inserted into the RCA output amp can change how the I/V amp work even if I use headphone amp as output. That's why I may try to bypass existing caps and change them as they may interfere.
Here I'm asking if I should desolder these Nichicon Muse FG caps or cut their legs and use them to solder different one? There are always 2 caps near important chip. Do they utilize ground and use +12V/GND/-12V or the ground is not used? When I decouple using amp socket I cannot access ground plane and use at least 35V caps. 25V tantal cap put here was blown away in a few seconds after computer start.
Also this could let get better access to the TPA amp power.
Proper power decoupling is very important for these opamp as they can behave very unstable and their settling time can be very large. Different cables and plugs causes a lot of changes in the settling time and feedback delay. Large headphone amp feedback with very good cables and plugs can give very good results but increases the power quality requirements. When it lacks power, op-amp can behave very strange causing oscillations depending on the components used in headphones or just saying "I'm off, I'll be slower". The option is lowering feedback and putting 1kohm serial resistor into the output but that decreases the resolution but makes op-amps more stable.
That's how it looks now. I noticed that power decoupling caps change may change the output impulse and step response. It may be the not the best record precision, but these changes are audiable in the same way. Some impulse peaks change amplitude up to 20%. Wish to change TPA power caps.
An externally hosted image should be here but it was not working when we last tested it.
You may replace those caps with quite large capacities tantal or ceramics. If you will go back in this thread, you may find some pictures of my mods as inspiration (and quite much information about). I used caps of 1000µ or more to decoupling the opamps. The big clue with this card is to be powered from an better (external) power supply for opamps (+/-) and 5v for DAC and clock (power on it must be synchronized with the computer power on sequence - details in the thread). A better oscillator it improve very much the sound. Removing the filters in the output audio signal path, it will bring an big improvement too. DC coupling on output is better (must be compatible with the power amp used).
A very simple mod is to use the headphone out to connect an amplifier to. The headphone chip is much better than the original opamp used. In such case an adequate resistor must be connected on this output to simulate the headphone (set the power amp volume to 0 and then go up carefully when testing this way).
A very simple mod is to use the headphone out to connect an amplifier to. The headphone chip is much better than the original opamp used. In such case an adequate resistor must be connected on this output to simulate the headphone (set the power amp volume to 0 and then go up carefully when testing this way).
Should I desolder these Nichicon FG caps or do I have to move them to broke their legs?
I'm not keen on using tantal. Ceramics gave me large resonanses and the sound was very, very grainy with less details. Monolithic ceramics were better, but it was still the same kind of distortion. MKT - smooth but not fast. MKP - not tried yet. Electrolitycs sound different depending on the brand. The sound of different parallel capacities made of different types is the worst.
Coris, your ideas and pictures are very helpful!
I'm not keen on using tantal. Ceramics gave me large resonanses and the sound was very, very grainy with less details. Monolithic ceramics were better, but it was still the same kind of distortion. MKT - smooth but not fast. MKP - not tried yet. Electrolitycs sound different depending on the brand. The sound of different parallel capacities made of different types is the worst.
Coris, your ideas and pictures are very helpful!
The best way to remove that so called Nichicon caps is to broke their legs. Desoldering will may destroy the traces/board.
Here is about to use those caps as decoupling, not in signal path. I got best results with tantal/ceramics. Electrolytics have high ESR/ESI and are very big, comparing with the other ones. If you get bad sound, distortions, resonances, and so on with ceramics/tantal, this it may be because a quite bad setup for the opamp involved, grounding, and not at last because of the main power supply quality. Use lower as possible resistors for opamps setup (feedback, gain), and less as possible capacities in the signal path.Right grounding is also very important, too. Decoupling this way as I have done, bring distortion free sound, very deep bass, very high dynamics, and only beautiful sound... But decoupling is only a part of the whole rest of right using of opamps and many other design details involved in such analogue stage setup.
You may find a picture with the board and all my mods on it just in the first page of the thread (as many other in my following posts here). I may suggest you to take a look over this thread from the beginning...
I may say that I`m since quite long time ago now, both surprised by the results and very satisfied with the outputted sound from my modded STX... All done so as you can see here.
Not to be forget it the right setup of the rest of the computer software for sound stage. I use Foobar - ASIO out on win7
Here is about to use those caps as decoupling, not in signal path. I got best results with tantal/ceramics. Electrolytics have high ESR/ESI and are very big, comparing with the other ones. If you get bad sound, distortions, resonances, and so on with ceramics/tantal, this it may be because a quite bad setup for the opamp involved, grounding, and not at last because of the main power supply quality. Use lower as possible resistors for opamps setup (feedback, gain), and less as possible capacities in the signal path.Right grounding is also very important, too. Decoupling this way as I have done, bring distortion free sound, very deep bass, very high dynamics, and only beautiful sound... But decoupling is only a part of the whole rest of right using of opamps and many other design details involved in such analogue stage setup.
You may find a picture with the board and all my mods on it just in the first page of the thread (as many other in my following posts here). I may suggest you to take a look over this thread from the beginning...
I may say that I`m since quite long time ago now, both surprised by the results and very satisfied with the outputted sound from my modded STX... All done so as you can see here.
Not to be forget it the right setup of the rest of the computer software for sound stage. I use Foobar - ASIO out on win7
What I see is that this card lacks in power in sense of amount and quality. I still play with the analog part power decoupling. I remember much better headphone amp signature and improving xonar card power closes the gap. The problem is the 10mm caps diameter. This limit most caps in reasonable capacities.
First the TPA chip power has to be improved to minimize difference between raw I/V output and headphone output. Then improving I/V amp piece helps give the deeper sound: smoother and more detailed trebles, more natural sounding mids and deep bass. Current driver has lesser SPL and totally different diaphrag and suspension. But I could get almost the same results by improving card.
I wonder how many farads will I need. I would like to try 2x1000uF, but the cap diameter is large.
First the TPA chip power has to be improved to minimize difference between raw I/V output and headphone output. Then improving I/V amp piece helps give the deeper sound: smoother and more detailed trebles, more natural sounding mids and deep bass. Current driver has lesser SPL and totally different diaphrag and suspension. But I could get almost the same results by improving card.
I wonder how many farads will I need. I would like to try 2x1000uF, but the cap diameter is large.
Yes, is right, the original power system of STX is very poor, and bad designed. So this stage it may be improved first of all. But is not an easy task, as an external power supply it may be used, mechanical solutions have to be founded to mount that PSU and its trafo near the board, and so on. You can get some ideas from the pictures here...
The I/V stage is also important, but it have to be implemented together with the opamp used. Try higher as possible (you may find) slew rate of that opamps. The resistor for conversion it have to be very thermal stable (lowest ppm - quite expensive).
I use on my card 1200µF but ceramics (6v) and tantal (20v), as you can see in the pictures. Is very important to decouple very well the DAC chip on both digital and analogue rails.
The I/V stage is also important, but it have to be implemented together with the opamp used. Try higher as possible (you may find) slew rate of that opamps. The resistor for conversion it have to be very thermal stable (lowest ppm - quite expensive).
I use on my card 1200µF but ceramics (6v) and tantal (20v), as you can see in the pictures. Is very important to decouple very well the DAC chip on both digital and analogue rails.
Last edited:
Xonar ST is very sensitive to the the headphone driver, cables and plus. I get totally different result on each setup and I can't clarify the best overall setting. However for the given headphone setup, the proper amp solution can be given. Right now I use AD797BR as you can see on the picture. I decouple them with existing FG + elna silmic II. The TPA opamp is decoupled with FG + Nichicon Muse KZ. The TPA works at the highest gain available on the stock card. The driver is 32ohm, mylar membrane with no additional suspension made from different material, not too big magnet, some ventilation in the driver for easier bass response: increases stability thru lesser power requirements on bass. Cables and plugs pretty fast for lowest settling time.
Tested opamps AD797BR, OPA627AU, LM6172, JRC2114 (stock), LME49860, LME49720, LM4562, NE5532, TL072, TL082.
Why is it crucial for power decoupling to use 2 serial caps for ground/+12V and ground/-12V? Will it decrease it's capacitance as serial connection or improve impedance and so the speed of decoupling?
If I decide to modify the card I may play with other decoupling caps like near DAC. However first I know I have to improve this analogue stage. I heard the results with external amp and this is a must. This card may sound good. I compared mine not so best xonar setup (the opamp and decoupling setup was not transparent enough) with the Xonar Essence One using both as a source for tube headphone amp. They were very close.
Making external power supply to fit on the card requires more work as a proper design gas to be found. The available space is limiting. I read the whole thread two times. I will it third time
Tested opamps AD797BR, OPA627AU, LM6172, JRC2114 (stock), LME49860, LME49720, LM4562, NE5532, TL072, TL082.
Why is it crucial for power decoupling to use 2 serial caps for ground/+12V and ground/-12V? Will it decrease it's capacitance as serial connection or improve impedance and so the speed of decoupling?
If I decide to modify the card I may play with other decoupling caps like near DAC. However first I know I have to improve this analogue stage. I heard the results with external amp and this is a must. This card may sound good. I compared mine not so best xonar setup (the opamp and decoupling setup was not transparent enough) with the Xonar Essence One using both as a source for tube headphone amp. They were very close.
Making external power supply to fit on the card requires more work as a proper design gas to be found. The available space is limiting. I read the whole thread two times. I will it third time
Yes, they are coupling caps. It should not be there... If you look closer you will see too there, some ferrite beads and small caps in the signal path. Is best to remove all those things, and connect directly (DC coupling) from final opamps out to power amplifier. But you have to be sure about the amplifier input stage is compatible for DC coupling. As usually on the amplifier input is an resistor or just the potentiometer of that amplifier to the ground. In DC coupling the computer ground is direct connected to the amplifier ground.
Those coupling caps are not used when you are coupled to headphone output. Anyway, my suggestion is to remove it.