Xonar STX soundcard spectrum analyzer help needed

[boborich]

Hi,

I'm new to this forum and I just recently got into electronics by watching Dave's video blog. I'm doing digital marketing for PayPal at Shanghai China, and electronics is just an hobby for me recently. I started revamp a power amplifier using a circuit I put together by looking on the internet. I managed to have the prototype start to work, but I want to do some measurement of the performace, especially the distortion.

I read a lot about distortion measurement and decided to go with the soundcard + ARTA solution. I bought an ASUS Xornar STX and set it up on a win7 X64 desktop PC, and played with it for several nights ended up blew the input opamp for the left channel(RC4580 From TI).  I spent another 2 nights repaired it and everything seems to be fine, until I found some loopback test in this forum indicating the THD for 10Khz on this card can do down to 0.000024%

I have a couple of questions for those who use the same setup to verify that I'm using the tools correctly. Like I said, I'm new to this domain. The exact setup I'm usings:
1. ASIO driver of Xonar under Win7 64bit with UNi-Xonar-1822-v1.75a-r2 (the recommended one), I also used the latest (non-beta) driver from ASUS with the same result. ASIO set to 32bit and 20ms delay.
2. Loopback test using a homemade RCA to TRS cable with left channel input grounded. 96kHz sampling rate with FFT size at 65536, Kaiser7 windowed as recommended by ARTA. Using linear averaging I was able to get 0.00074% THD and 0.0011%, which is much higher than IvoS posted in one of his replies.
3. If I change the sampling rate to 192kHz, I will get a lot of noise over 50kHz, which leads to 0.0019% THD and 0.018% THD+N. Is this expected result on this sound card or there is something wrong? From what I learnt, in order to get an accurate THD reading for 10kHz, I should at least be able to get the harmonics at least 4~5 orders, which is ~50kHz, and 96k sampling rate doesn't give you 50kHz. Am I right?

I have other questions related to this but I think it is probably too much for my 1st post here at EEVBlog.

BTW, I considered buying a HP 8903B so that I don't have to deal with software based issues, but that device seems can't mesuare anything beyond -100db THD due to a high noise floor. An AP device is simply too expensive for me to afford. QA400 is not available in China and international shipping is not my preference (although this device was manufactured in China). I also found panasonic 2722 is an excellent device but I can't find any user manuals for it, let alone software written for that device(I assume they need a PC in order to get a frequency response chart, unless I do it manually on a peice of paper).

Some may ask why I need a over -140db dynamic range. Well, because the amplifer is designed to deliver 0.002% THD with 260W into 2ohm load at 20kHz. Well at least the simulation result says so, and I'm expecting much better results at much lower power with 8ohm load at 1kHz. (The simulation software just shows 0.000% and that is it.)

Thank you for reading this long post!

[boborich]

Aha! I found the answer!

The 192kHz noise was due to the nature of the ADC CS5381! I found this chart on the QA400 page, which has the same issue when using 192kHz sampling rate:


Also I found this in the ADC data sheet (picture attached). Not sure I understand this correctly, but the datasheet do single out 40kHz bandwidth mode out.