Anomalies with Linux and E-MU 0404 USB

Interference

New Member
Thread Starter
Joined
Apr 28, 2019
Messages
5
Hi all,
first of all let me express my appreciation for the degree of maturity this application has reached.

Then, I come to my problem.

I have recently become in possession of a glorious E-MU 0404 USB, but I am experiencing odd behaviour under Linux:
- card works as expected up to 48 kHz
- at 88.2 and 96 kHz something strange happens: one channel (always the same) reads a floor of -54 dBFS and is somehow responsive to the input signal, the other stays low and apparently almost unresponsive (but not dead);
- at 192 kHz I had some problem in a Ubuntu LTS machine (acquisition works ok but loopback tests crashes) but apparently it works OK on my more up-to-date Fedora.

Despite the outdated driverx, everything looks OK in Windows.

I am using the official JRE.

Now, since the card is working at 192 I could live with it, but I am very puzzled about the strange behaviour at 96. It's like the channels are in some strange operational mode. Reproduction and acquisition with aplay / arecord do not show any issue.

Any idea on how to further troubleshoot the issue?
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
2,131
Other Linux users will hopefully jump in, I'm afraid I can't help with that. The REW sample rate setting is a request to the operating system for audio data at that rate, it typically results in resampling in the OS if the device is not configured to run at the requested rate in the native OS audio configuration tools.
 

Interference

New Member
Thread Starter
Joined
Apr 28, 2019
Messages
5
Other Linux users will hopefully jump in, I'm afraid I can't help with that. The REW sample rate setting is a request to the operating system for audio data at that rate, it typically results in resampling in the OS if the device is not configured to run at the requested rate in the native OS audio configuration tools.
Thank you John, no worries. I understand the amount of factors at play and this is likely impossible to reproduce without the same interface and OS.

But let me clarify a bit: alsamixer has a slider for the samplerate of the soundcard from which I can see the response of the hardware is consistent (when a different sample rate is selected, either by REW or by aplay/arecord, the slider moves into the right position). What is puzzling me is that even without any playback, the acquisition gets in some strange state. Simply activating the "levels" widget I get this noise floor at any rate between 48 and 192 (extremes excluded) that should be supported by the soundcard. I will try some more detailed troubleshooting and report some more detail :)
 

Interference

New Member
Thread Starter
Joined
Apr 28, 2019
Messages
5
I have some more details, not sure if useful though.

What happens is that channel mapping and functionality gets screwed up:
- ch. 1 reads a sinc function with a 2 kHz repetition (!)
- ch. 2, 3 are now R, L but with high noise floor and high gain (guess the former is consequence of the latter)
- ch. 4 some other kind of signal, not identifiable as a specific function.

This issue seems possibly related. I think the problem appears only with the interface is driven in full duplex, as REW does, but of course not a specific REW defect (sorry for bothering here).
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
2,131
Sounds a bit like a sample format (word length) mixup so the interleaved channel data is being incorrectly deinterleaved.
 

Interference

New Member
Thread Starter
Joined
Apr 28, 2019
Messages
5
Thank you John, I did some more testing.

Apparently the pattern I get in REW at 96 kHz s the same I get with arecord whenever I try to capture 4ch at samplerates >= 96 kHz. So it seems the Linux driver does not support properly the 4 x 24/96 capture, but does not "forbid" it to applications either. Strangely REW accesses automatically the card in the correct way (2ch) at 192 kHz but in the unsupported way (4ch) at 96 kHz. At the same time 4 x 24/192 is not forbidden to arecord, but yields the same wrong data as with 4 x 24/96.

Probably there's some error either in how the driver reports the supported configuration or either in how Java reads/infers them (I am not so familiar with the mechanism).
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
2,131
I'd suspect the driver, since REW is getting the available channel count at the requested sample rate from JavaSound, which gets it from the driver. REW offers as many channels as are advertised as being supported. If you use the Generate debug file button on the REW soundcard preferences it will produce a file showing what the various interfaces say they support, worth seeing that.
 

Interference

New Member
Thread Starter
Joined
Apr 28, 2019
Messages
5
I can confirm now that using stereo access to the board allows to exploit the two analog inputs at 96 kHz. Thanks!
 
Top Bottom