[LAD] A little quiz about audio measurements...

Robin Gareus robin at gareus.org
Fri May 28 17:22:24 UTC 2010


On 05/28/2010 05:34 PM, fons at kokkinizita.net wrote:
> On Fri, May 28, 2010 at 05:05:19PM +0200, Robin Gareus wrote:
> 
>> Well, interpreting an 'experiment' that one has not conducted
>> him/herself without knowing _all_ parameters of the setup is not
>> something to encourage doing..
> 
>> Did you use symmetric (XLR) connections?
> 
> Where possible, yes. Note that we are seeing a *modulation*
> effect, the level of the spurious signals is proportional
> to the wanted signal. It's not additive.

a 2nd measurement showing a measurement at - let's say 1215Hz would have
made that clear in the first place. Anyway I take your word for it.

Thanks for /organizing/ this little quiz; it's quite fun!

>> Are the sample-rate settings of both cards identical?
> 
> Shouldn't matter, as the connection is via an analog signal.
> 
>> what was the physical distance between the devices
>> (cross-talk)?
> 
> It's absolutely not crosstalk.
> 
>> Was there any ground-lift equipment (built-in?) in place
>> or maybe you simply forgot to switch off phantom-power?
>  
> All those would create additive effects, which is not what
> we see. Phantom power, where available, was off. 

good; those charge-pumps that are used to create 48V from 5 or 12V
often cause ripples in the supply voltage.. which in turn can influence
the oscillator.. resulting in signal modulation.

> Card X
> has both line and mic inputs, all show the same result.
> Mic inputs are driven via a balanced passive attenuator
> with an output impedance of 50 ohm.
> 
>> Anyway, good that you've labeled it 'quiz'; so here's my guess:
>>
>> You're encountering reflections (or even standing-waves) in the cable.
>> (did you uncoil it? or was it a very short one?)
>> Did you repeat the measurement? using different cables for example.
> 
> No differnece seen between a cable of 3 m and on of 18 m.
> 
> This will answer most other posters, except Chris:
> 
>> Why 1015 Hz ?
> 
> To clearly separate the signal from any 50 Hz harmonics.
> 
> 
> One more hint: card X is a very high end one. It should
> be 'perfect', or at least much better than card A.

OT: I would not trust any consumer-grade audio-interface for signal
differences larger than 80dB anyway. I'm amazed that card A shows such a
low noise-floor in the loop-back test.

> Ciao,
> 



More information about the Linux-audio-dev mailing list