I own two audioinjector octo-s.

In 2022,.some months ago I tested the signal to noise ratio  and  ground noise 

for empty, not attached devices I got the expected results better than -90dB

Now, in May 2023- best Imy results are widely worse about  -55dB !!!

It looks that I do no more  use the full resolution of  24 bits.

On diyAudio it  shows up to -95 dB 

REMARK: I use KDE5 plasma with Pulseaudio., same as in 2022

QUESTION: Are there known  problems with the 6.x.y kernels or has anybody an  idea?

the last kernel was somethng around 5.1x.y

I now use  kernel 6.2.1 (64bits)

Recording audio via Audacity I get obvious crappy signals (see attachments.)

the input  signal should be that large and somehow wit digital steps?

Using  gnuradio to calculate a  FFT-analysed  spectrum I get distortion  and noise between frequencies below  22kHz above the distortion strangely is reduced .

I tried sampling at 48kHz, 96kHz and even the highest 192kHz.
Same results for all of Rates.!!!
stangel: Ouside the 22kHz bandwidth, the signal /nose is almost as expected.  

The card seems to be detected properly.

I  attached the audacity-recording, the gnuradio spectrum and
 the  result of alsa-info. and my /boot/config.txt

Anybody any Idea?

$play -l

----------------- here is my /etc/asound.conf

pcm.octo {
    type route
    slave{
     type lfloat             # Linear<->Float conversion PCM
        slave {                 # Slave definition
            pcm "plughw:CARD=audioinjectoroc"
            format s24le
        }
       
        channels 8
    }
    ttable {
           0.0 1
           1.1 1
           2.2 1
           3.3 1
           4.4 1
           5.5 1
           6.6 1
           7.7 1  
    }
    hint {
        show {
           @func refer
           name defaults.namehint.extended
        }
        description "AudioInjector Octo Card (upto 6 channels input, upto 8 channels output)"        
    }
}

ctl.octo {
    type hw
    card "audioinjectoroc"
}