virtins.com

Hello
My sound card (Asus Xonar Essence STX) lost its good performance as you could see in the picture. Its bandwidth is also reduced to about 63KHz. Is noisy and the THD is big. I suspect that its ADC converter Cirrus Logic CS5381 suffers from something. Have you any idea or to suggest anything? This SC is still considered the best on the market.
Thanks

Not sure what happened about the bandwidth. But for THD measurement, usually a longer Record Length and FFT Size should be used.

Thanks for the reply.
Yes, I know. Simply I used these settings to compare with an old measurement obtained with MI v3.1 I have stored on my PC. I can't upload it because the original picture is in jpeg. I tried to convert it to png but the size is still big.

Here are two screenshots of the performance, taken with the best possible settings. SC in ASIO mode at 24bit/192KHz. The FFT size is 32678 and the window is Kaiser 6. The THD etc. range setting is from 20-20,000Hz. As you could understand for a range setting up to 96,000Hz the values are dramatically worsening. The measurements show two main issues 1) the bandwidth is limited and 2) the sound card is noisy. Please note that for the hardwired loopback test, I use a very short signal cable, equipped with a high-quality stereo TRS jack and 2 phone jacks of Neutrick, all gold plated. The power supply of PC is also carefully selected for the best possible noise figure, it is a Zalman ZM750-XG of 750 Watts.

Thanks for the information. Not sure what is the cause of the problem, probably hardware is too old?

Yes, I bought this sound card 10 years ago. I removed it from the PC and did a thorough cleaning of the PCI tracks on PCB, as well as of all I/O phone jacks. I also replaced the 3 ICs in the DAC section, with new LME49720s, without any performance improvement.
Look, I think the THD is very close to factory specifications. The problems are two, the limited bandwidth, and the noise level. Regarding the bandwidth, I'm not sure which is the cause. Regarding the noise, I did two new measurements, both in No Loopback mode, the one with inputs and outputs disconnected to make an estimation of input noise. The noise floor I found is 12uV which equals -98.4dBV. The second was with the input hardwired to output to estimate the noise of output. The noise floor was increased to 16.5uV and so is 4.5uV which equals -106.9dBV. Such levels of course are not for practical use. The best audio preamplifier I have built in the past had a noise floor of -99dBV. However, I will try to replace the sound card electrolytic capacitors in all power supply sections, because it is 10 years old and maybe they are derated. I have stocked enough Low ESR Panasonic electrolytic caps which in my tests have proven better than the so much advertised Silmic and Os-Con.

Thank you for your elaboration. Maybe you can also measure the bandwidths of the input and output separately. The bandwidth of the output can be measured using a true RMS multimeter or an ordinary oscilloscope point by point using frequency stepped sine.

Thanks for the advice.
To the present a test of sound card ADC converter, with an external Hameg HM8130 function generator in sweep mode from 20Hz-96KHz. The ADC -3dB bandwidth looks better but not close to 96KHz. Please see the attached picture and the notes.

According to the specifications of Xonar Essence STX, the -3dB upper frequency limit is 90kHz. Your result seems quite close to that.