Soundcard Spurious Signals

[brian_mk]

I have been evaluating the Virtins spectrum analyser using a laptop PC connected to a 'Creative Soundblaster x-fi USB'.

Initially I was using it to look at the frequency spectrum from a home constructed Wien Bridge Oscillator. As well as the expected oscillator harmonics, I was puzzled to see some frequency components about 60-70dB below the fundamental that were not simply harmonically related. I was initially unsure if these were originating from the oscillator or from the soundcard.

To investigate, I removed the Wien Oscillator and looped the soundblaster ‘line out’ phono connectors directly back into the ‘line in’ sockets. I used the built in Virtins signal generator to generate a 1 kHz sinewave. The spurious signals were still present.

The spurious signals are not fixed in frequency – they shift depending upon the frequency of the fundamental. Here are some examples:-
The first column is the sig gen frequency. The second column is the lowest frequency spurious signal. The third column is the next higher frequency spurious signal. These are some 60-70dB below the fundamental. There are other similar spurious signals at higher frequencies, but they are lower in amplitude.

800 Hz 3.1 kHz 4.7 kHz
1kHz 2.9 kHz 4.9 kHz
1.2 kHz 2.7 kHz 5.1 kHz
1.5 kHz 5.4 kHz 6.3 kHz

Can anyone suggest where these spurious signals may be originating?
The soundblaster x-fi USB noise and distortion specifications suggest they should not be there.

I can see nothing obvious using the Virtins or an external oscilloscope.

I have also tried connecting a 3rd order 1 kHz low pass filter (18dB / octave) between the line out and line in connectors to ensure the spurious signals were not coming from the signal generator.

[brian_mk]

Some additional info...

1) The amplitude of the signal generator was set well below the clipping threshold during my tests.
2) The amlplitude of the spurious signals relative to the fundamental remained unchanged even if I changed the level of the signal generator. i.e. they are always between 60-70dB down. They also seem pretty stable.

I am wondering if the spurious signals could be AM sidebands caused by some interfering signal(s) in the soundblaster?
Anyone seen anything like this in other soundcards?

[VirtinsTech]

Have you tried it with the ASIO driver? The automatic Sampling Rate Conversion by Windows Sound Card MME driver might cause some artifact.

[brian_mk]

I'm not sure if ASIO drivers are available for the sound blaster x-fi USB.

I did some more tests today. The results are even more puzzling:-

First I installed the Virtins sofware on another PC: A desktop machine running 32bit Windows XP.
This has a sound blaster audigy 2 PCI card.

I looped back the line out and line in connectors and enabled the Virtins signal generator as I did when using the laptop and x-fi USB.
I got exactly the same frequency spurious signals with about the same levels as before.

Next I tried removing the loopback cables and connected the x-fi USB output into the audigy 2 input.
When I enabled the Virtins sig gen on the x-fi USB, again I get the same spurious signals on the audigy card.

However, now if I insert the 1kHz low pass filter into the signal path, the spurious signals are attenuated.
What is more, if replace the Virtins sign gen with my Wien Bridge Oscillator feeding the audigy card I do not see any spurious signals - only the expected oscillator harmonics.

So now I am even more baffled: The last test using the audigy 2 suggests the signals may be coming from the Virtins sig gen but the earlier test with the x-fi USB produced the same spurious signals when using the Wien Oscillator. Work that one out!

[brian_mk]

To recheck my findings, I have carried out the test in the opposite direction...

This time using the Virtins sig gen in the PC with the audigy 2 card feeding into the x-fi USB connected to the laptop.
The spurious signals are there as before. Inserting the 1kHz low pass filter has no effect on the spurious signal levels relative to the fundamental.

Just to make sure I had not made a mistake, I repeated the original test feeding the Wien Oscillator into the x-fi USB.
I got the same result as before i.e. spurious signals on the analyser. As before, these are unaffected by inserting the low pass filter.
Note that the Wein Oscillator harmonics are attenuated by the filter. So this time it would appear that the spurious signals are originating in the x-fi USB input / ADC.

[VirtinsTech]

If the sound card does not come with an ASIO driver, you can download a universal ASIO driver from: http://www.asio4all.com. ASIO driver is bit-perfect as it bypasses the automatical Sampling Rate conversion by Windows. In your tests, the sampling rate was not mentioned. You can upload pictures as attachment here.

[brian_mk]

I discovered the ASIO4all drivers the other day.
I downloaded and installed them on the laptop.
However, I couldn't get them to work with the x-fi USB :-

I followed the instructions in the PDF that came with the drivers.
After selecting ASIO under the Vitirns ADC and DAC settings and clicking 'Run', the ASIO4Aall control panel appears and shows both the internal Realtek hardware and the x-fi USB. I selected the latter.
Both the 'scope and spectrum analyser showed no signal at all when the sig gen was enabled.
I'm not sure if I was doing something wrong or had an incorrect setup?
Is there something else I need to do?

[VirtinsTech]

The selection of a sound card on the ASIO4ALL panel may be a little confusing. Select the left most button of the entry instead of just hightlighting the name of the sound card. After this, stop the oscilloscope and signal generator, and then restart them. You might need to select the input endpoints, such as mic input / line input, through the sound card's own control panel, or [Setting]>[ADC Devices]>"Analog Channel Configuration".