Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Sound card loopback with and without electrolytic DC blocking capacito (Read 5932 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Sound card loopback with and without electrolytic DC blocking capacito

This test is also discussed at another forum. The samples in the package cap_abx.zip are loopback recordings from the headphone output jack of a Xonar Essence STX sound card, captured with the line input of the same device.

An approximately 600 ohms load was simulated using 680 ohms resistors (which were in parallel with the ~4.3 kiloohms impedance of the line input, and in series with the 10 ohms output impedance of the headphone amplifier). The full scale voltage was about 1.7 Vrms. Three test samples were used (abx1*, abx2*, abx3*), of which the second is the same as the first one, but at 12 dB higher level by shifting the sample values to the left by 2 bits. For each test sample, there are three files, abx*o96.flac are from the original sample used for playback, abx*n96.flac were recorded without capacitors, and abx*c96.flac were recorded with 10 uF/50 V electrolytic capacitors (of this model) added in series with the headphone output. The capacitors were mesured to add 0.01% THD at 20 Hz and full scale level, which should not be audible, as well as a low frequency roll-off with a -3 dB corner frequency of about 28 Hz. The roll-off has been equalized in these samples with a FIR filter.

The playback and recording sample rate was originally 96 kHz, but this was downsampled to 48 kHz in the packaged files to reduce the download size. However, the original files did not have any significant content above 24 kHz, so the conversion should not really be an issue.

It is possible to make a comparison of the 'c' and 'n' files to test the non-linear effects of the capacitors only, or either of those versus the 'o' files to test the entire loopback chain. In the latter case, there is a very slight (0.1-0.25 dB) low and high frequency roll-off, which is also unlikeliy to be audible.

A more complete package of all original source and recorded samples is available as well in cap_test.zip. This includes all scripts and source code necessary to build the final test files from the first package without FLAC compression, and also some graphs and all the intermediate files. On a Linux system with C/C++ development tools, libsndfile, and sox installed, it is enough to run 'make'.

In the full version, there are also samples recorded with a 44100 Hz playback sample rate. This increases the noise floor of the DAC by a few dB to a level that should still not be audible, and uses the hardware 22.05 kHz lowpass filter of the PCM1792 DAC instead of a software resampler. Therefore, it is possible to use the samples for ABX testing the two different filters.

Other than the obvious low frequency roll-off, I do not think any of the effects should be successfully detected in theory, however.

Sound card loopback with and without electrolytic DC blocking capacito

Reply #1 –
Quote from: stv014 on
In the full version, there are also samples recorded with a 44100 Hz playback sample rate. This increases the noise floor of the DAC by a few dB to a level that should still not be audible, and uses the hardware 22.05 kHz lowpass filter of the PCM1792 DAC instead of a software resampler. Therefore, it is possible to use the samples for ABX testing the two different filters.

Other than the obvious low frequency roll-off, I do not think any of the effects should be successfully detected in theory, however.


Figure 4 in this document suggests that in some cases removing output capacitors from some headphone amplifiers can make a potentially audible difference due to the objectively measurable reduction in  low frequency nonlinear distortion:

Maxim Semiconductor headphone amp IC data sheet

The counterpoint is that a Tantalum electrolytic coupling capacitor was required to elevate nonlinear distortion to the point of possible audibility. Due to their expense and well-known nonlinear performance it is unlikely that this would show up in an intelligently designed commercial product.

Sound card loopback with and without electrolytic DC blocking capacito

Reply #2 –
Semi-related:  heres a neat link I saw recently about distortion in capacitors:

http://stephan.win31.de/capdist.htm

Some interesting examples of poorly chosen caps.

Sound card loopback with and without electrolytic DC blocking capacito

Reply #3 –
Quote from: saratoga on
Semi-related:  heres a neat link I saw recently about distortion in capacitors:

http://stephan.win31.de/capdist.htm

Some interesting examples of poorly chosen caps.



The specific examples of production gear with what appears to be bad cap choices are interesting, but people need to know that in general, they are rare exceptions.

Sound card loopback with and without electrolytic DC blocking capacito

Reply #4 –
FWIW, in Linear Audio #1, "Self-Improvement for Capacitors", Douglas Self noted a measurable and apparently permanent break-in phenomenon with polyester caps (though he advised using polypropylene instead).

Sound card loopback with and without electrolytic DC blocking capacito

Reply #5 –
Quote from: 4season on
FWIW, in Linear Audio #1, "Self-Improvement for Capacitors", Douglas Self noted a measurable and apparently permanent break-in phenomenon with polyester caps (though he advised using polypropylene instead).



What sort of numbers are we talking about here?

Sound card loopback with and without electrolytic DC blocking capacito

Reply #6 –
Quote from: Arnold B. Krueger on
Quote from: 4season on
FWIW, in Linear Audio #1, "Self-Improvement for Capacitors", Douglas Self noted a measurable and apparently permanent break-in phenomenon with polyester caps (though he advised using polypropylene instead).



What sort of numbers are we talking about here?


0.002% THD for a new capacitor, ultimately falling to 0.00030% over a number of hours.
 

Sound card loopback with and without electrolytic DC blocking capacito

Reply #7 –
Quote from: 4season on
Quote from: Arnold B. Krueger on
Quote from: 4season on
FWIW, in Linear Audio #1, "Self-Improvement for Capacitors", Douglas Self noted a measurable and apparently permanent break-in phenomenon with polyester caps (though he advised using polypropylene instead).



What sort of numbers are we talking about here?


0.002% THD for a new capacitor, ultimately falling to 0.00030% over a number of hours.


Presumably there are other preconditions including things like:

(1) Relatively large signal voltages.  Self is a big power amp man, and the signals there are dozens of times higher than elsewhere in the signal chain.

(2) An application where the capacitor has a relatively high AC signal dropped across it. IOW the transition band of a filter.

(3) Relatively high impedances, which comes from use at audio frequencies and the application of film capacitors.

Long story short, just another thing that makes me glad we now do these things in the digital domain with DSPs and the like!