Home Theater Forum and Systems banner

1 - 6 of 6 Posts

·
Registered
Joined
·
7 Posts
Discussion Starter #1
Hi,

I wonder if there is a bug in soundcard calibration in the current beta of REW (very cool software btw).

Indeed, when I use 44.1k or 48k sampling rates, bandwitdh is truncated around 10k and 12k.
However, using 88.2k and 96k, bandwidth goes over 20k.
I tested with Arta and it doesn't exhibit this behavior.

Souncard is a Focusrite Saffire Pro26. Latest ASIO Saffire drivers are used.
Please find attached measurement files at each sampling rate.

Maybe another finding (not directly related) : it seems when "Merge A+B" is used, the resulting curve is smoothed. I expected the merged curves are put "as-is" at specified frequency.

Regards.

PS : is there a "feature request" topic for REW ? For example, I would propose 24bit resolution and "target filter" curves.
 

Attachments

·
Premium Member
Joined
·
6,711 Posts
Odd that in each case the response stops at 1/4 sample rate. Someone on AVS had a problem with a Focusrite 2i2 this week that he speculated may be a Focusrite driver issue. I've rechecked the soundcard cal using ASIO on a UR22 I have, all working normally, ditto for WDM operation on another USB card. Worth trying to increase the ASIO buffer size to see if that has any effect and check any other settings offered in the ASIO control panel.

Smoothing is not applied as part of arithmetic operations, unless one of the traces being operated on was smoothed in which case that will be carried across to the result. Can you provide more details and perhaps an mdat of what you are seeing?

To request features best to post a topic for the feature you are interested in. ASIO drivers will use the bit depth configured in the ASIO control panel, that is generally 24-bit (sample rate and bit depth are shown in the bottom left corner of the main REW window). WDM drivers go through Javasound which only offers 16-bit data on Windows, though that isn't a limitation for acoustic measurements.
 

·
Premium Member
Joined
·
54 Posts
hello John I use a focusrite 2i2 mac OSX, I upgraded to the latest Java release in the coming days I try to do some measure and see if the problem is resolved
 

·
Registered
Joined
·
7 Posts
Discussion Starter #4
Hi John,

Thanks for your reply.
Very strange it seems to work now, even at 44.1k...

I modified some settings in REW... but don't know which one had an impact.

- First I remarked that doing measurement outside REW cal procedure works fine
- Tried again cal procedure, still not working
- Tried to alter IR decimation or truncation
- Tried to alter IR window calculated automatically or manually
- Tried with and without loopback
- Tried to modify FFT length : the first time I tried 256k, the measurement was very jerky. Tried all other length as well, but the result was ok.
- Did new measurement inside and outside soundcard cal proc... and it's fine

Doing a comparison between measurements, I see this hint in notes :
- The souncard cal that is strange : -3 dB points: <2,0 Hz, 11,024 kHz
- The one that is ok : -3 dB points: 3,4 Hz, 22,049 kHz

BTW : 24bit resolution is used with ASIO.
I put the meas in attachment.

Regards.
 

Attachments

·
Premium Member
Joined
·
6,711 Posts
In the file you attached, the first measurement looks fine (though the right window is set to 5ms, measurement is flat if that is increased and the mic cal is removed). The second has the problem of stopping at 1/4 sample rate. The third is just bizarre, I don't really know what to make of that. The fourth measurement, the soundcard cal, looks fine again. All quite odd, good that it seems to work at the moment but not sure why it wasn't working.
 

·
Registered
Joined
·
7 Posts
Discussion Starter #6
Hi John,

- The first is a measure outside the calibration proc
- The second is a measure with the calibration proc
- The third is the jerky after modifying the FFT length
- The fourth is a measure with the calibration proc, after alter settings and back
 
1 - 6 of 6 Posts
Top