Hmm the pic seems to be gone, damn imageshack. My and Napalm's worklog: http://www.xtremesystems.org/forums/...d.php?t=182354
Originally Posted by DeepDayze
oh wow. pigs do infact fly.
Does the digital out actually work now?
Hope this is not a rather late April Fools' joke
Originally Posted by L33F3R
I can take a gander at these...as I got an older X-Fi in my junk parts box now.
Originally Posted by hax0r
Okay, I take that joke back. Didn't know people were this serious about it.
Sound card mods
No offense to Napalm, but that capacitor mod on xtremesystems has no grounding in reality. By changing the values of capacitors, he likely changed the frequency response of filters placed by the original engineer. So, even though the new capacitors have better electrical characteristics, it's entirely possible that the "better" sound Napalm experienced was just an increased treble and bass response caused by haphazardly replacing capacitors. Further, the sound card may now be emitting RF energy in excess of legal limits.
Originally Posted by hax0r
For example, if I were designing an analog to digital input section, I might place a high pass filter with a low corner frequency to block DC (this would likely involve a capacitor in series with the signal), and a low pass filter to block signals above the ADC's limits (this would probably involve a capacitor to ground or an inductor in series). An output section will also likely require high-pass filtering because the DAC probably uses a 0V-xV range (rather than -x/2V to +x/2V), and low pass filtering for controlling RF. Changing the values of capacitors will change the frequency response of these filters.
Secondly, the -150dB noise floor reported in one of the sound test screen shots is physically impossible to attain at room temperature and audio voltages. It looks like the program generated a test tone, then analyzed the test tone without passing it through the sound card. Also, the cross-talk spectrum shown has little or nothing to do with actual channel frequency response.
That said, we all have to learn somehow, there are lots of things I don't know, and I don't want to discourage anyone from modding their hardware. I just don't want to see people spending hundreds of dollars on capacitors without a well-measured baseline and a clear goal for improvement. If I were to offer advice to sound card modders, it would be something like this:
- If you don't know what a particular capacitor is doing in the circuit, replace it with the same capacitance value (you don't change the capacitance if you change voltages -- the frequency response of a simple filter depends only on R and C, not the input voltage).
- Always measure before and after!
- Make sure you know how to use your measurement software. RoomEqWizard is a very good cross-platform tool.
- Your target should be a flat frequency response from the sound card.
- Clean the signal path first. Find the components that directly touch the signal (you may have to look up pin diagrams of the chips on the card, or carefully probe the card with a scope), and replace those first. This would include the op-amps.
- Next, look at your power supply. Changing capacitor values here might make sense, if you know what you're doing. You essentially want a very low frequency low pass filter to block anything that isn't DC from going to the power input lines of the ADC, DAC, clock generator (if any) and op-amps. Chip manufacturers will often specify what capacitance to use and where to put it.
- Finally, you might be able to improve the clock circuit to reduce jitter, though the power improvements would probably have already helped here.
- Don't put capacitors in series to reduce capacitance. It works, but an former coworker who's been designing pro audio equipment for years said that it can lead to one of the capacitors failing.
- Final thought: increasing the output voltage of the sound card (by changing the resistors that govern the op-amp gain) may reduce the relative noise level picked up by long cables to your power amplifier, but may also overdrive other components. Make sure you're well within the voltage supply rails of the op-amps.
Recommended Google search terms for learning: "analog filter design", "RC filter design", "RC filter cutoff frequency", "RC time constant", "2nd-order filter design", "RC filter frequency response". Read the Wikipedia pages about filters, resistors, capacitors, and op-amps.
Napalm could save a lot of listening time by mathematically analyzing the circuit (or at least measuring it) and ensuring a flat frequency response and suitable noise+distortion measurements (if you can hear it, you can measure it!). The easiest way to measure a sound card's frequency performance is to hook the line out to the line in, and run a frequency sweep with Room EQ Wizard. You can also build the circuit in a simulator, like Qucs, and measure its expected ideal response there. Other than that,you can look for a dedicated hardware audio measurement system, such as the one my previous employer used: the Audio Precision line of test tools, which can run in the tens of thousands of dollars per unit (there are cheaper tools available). If you have an oscilloscope that will display a frequency spectrum, you could output white noise from the sound card and feed it into the scope, but that won't provide as much detail as a nice, long sine sweep.
Originally Posted by hax0r
As for RF energy, the various clocks on a sound card operate at frequencies up to the MHz range. Without proper filtering, these clocks (and their harmonics and subharmonics) can leak through the DACs into the output signal, and then be emitted into the air by your cables (which act as transmitting antennas). You can check for unwanted frequencies on the wire using a high bandwidth spectrum analyzer. You can also go to the expense of building a shielded room (my previous employer had one), complete with RF test equipment, if you're feeling a bit crazy.
Finally something good from Creative.
I Wonder does it work with Auzentech's cards, they are based x-fi chips...?