21 Nov
2016
21 Nov
'16
4:42 p.m.
On 11/21/2016 12:33 PM, Gerhard Zintel wrote:
On Monday 21 November 2016, Tino Mettler wrote:
Additional filtering in this
case would only be necessary when
converting the sample rate between the audio file and the target output
sample rate. If both are the same, we should be able to get "precise"
output. (neglecting eventual numerical issues from converting floating
point to integer samples and vice versa, which are not that relevant in
the square wave case anyway)
On Fri, 2016-11-18 at 11:30 -0600, Chris Caudle
wrote:
Be aware that a rectangular wave has infinite frequency components in
its edges. In any real physical system you have to lowpass this
signal (either the reconstruction and / or the anti-aliasing filter).
After filtering into the frequency range of interest (smaller than
half sampling frequency) the wave looks like what you show in
recording.png.
This is in each case true for a analog out- and input. I assume that a
digital connection as well has to filter the signal as well to stay
in the allowed frequency band according to your sample rate. Others
might skim in with more knowledge on the digital side. On Fri, November 18, 2016 8:59 am, Tino Mettler
wrote:
Hi,
the recording.png is the result from recording the playback from
playback.png. It should be identical for bit-transparency.
Actually, a colleague of mine did exactly that in
Windows,
using another RME card.
Here are example screenshots of the original and the recording:
https://tikei.de/playback.png
https://tikei.de/recording.png
That looks correct. Do you have pictures of your
result? 
