Difficulty measuring with Optical interfaces, and or, multiple USB soundcards

JonPike

New Member
Thread Starter
Joined
Nov 10, 2017
Messages
12
OK, coming back after a long while to measuring with REW. Very nice to see how much this software has grown over the years!

Along with refamiliarizing with the basics... (I formerly did a lot of speaker driver and room measuring) I'm trying a few new things, like measuring optical interfaces, and trying to evaluate if they have impact on downstream devices. I have run into two kinds of problems, one of which I'm not sure is a capability issue or not.

1. I try to measure something between two USB sound devices, and get wildly erratic results. Example: A Schiit Modi USB DAC, analog out into a SB Xfi-HD soundcard analog in. I get FR traces, but extremely distorted. Seems like there's issues with timing? Maybe I have a timing setting wrong for this case? Maybe you can't run two separate USB sound devices at the same time?

2. I'm also trying to evaluate optical inputs. I find issues where my receiving device cant always recieve the digital audio. Example, the aforementioned XFi-HD, which has both optical in and out, can happily loopback, but with some inputs, for example my TV optical out... HDMI audio in -> TV TOSLINK out, to XFi TOSLINK in, I get zero audio. If I plug that into the Modi, I hear it fine.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
It's fine to use separate USB interfaces for input and output, but make sure they are running at the same sample rate. On Windows that is best done by using the Java WASAPI exclusive device entries (names starting EXCL).

TOSlink inputs may not provide any data until the output device driving them is actively generating a signal. The same caveat on making sure sample rates match applies.
 

JonPike

New Member
Thread Starter
Joined
Nov 10, 2017
Messages
12
Hmmm... wonder if there are "tricks" to trick the input to work, such that they will be providing data when the sweep occurs? But, if this is the source of my problem #2, the XFi card not seeing anything on its input till a sweep starts... why do I always have a successful self loopback? Maybe it's that the TV or whatever is at a lower/different sample rate. I know it's 16b not 24, but it might be 44.1 or 48khz, instead of higher... Will try REW set at lower sample rates...

For my problem #1, what would be the "correct" settings for timing reference and such for digital in to digital out? Same question for optimum settings for a USB mic like the UMIC-1? I've used analog mics and actual input/output jumpers for reference before, I guess they wouldn't be the same...
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
You can use the acoustic reference for most measurements. It is the only option for a USB mic, and is convenient for digital paths where it may be difficult to have a loopback channel. There often wouldn't be a need for a timing reference on a digital path measurement, depending on what it is you are trying to measure.
 

JonPike

New Member
Thread Starter
Joined
Nov 10, 2017
Messages
12
Well, it seems that unless you explicitly set the input device and output device in Windows to the same sample rate, it won't work. That seems to be my prior problem, though I haven't checked it on all paths I was having problems with.

I was under the impression that REW would handle "talking" to the windows settings somewhat, and you didn't always have to set the soundcard devices exactly. Is needing to do so expected, and I've forgotten about that, or is this different now? (Using a ? .89 beta ver)
Example: I was set to 24b/96k, and one device can only do 48k. Setting REW to 48k didn't do it, had to go into Win sound devices and set both ends to 48k.

Hmmm again... Just tried acoustic reference (I assume you meant in the "make a measurement" panel?) and it did not work at all. It doesn't see and waits forever for the timing reference, despite showing a -12db measurement sweep as normal in the level display, gives 3 error boxes upon cancelling, one of which says it saw nothing on the input, all data was zero. Any of the other loopback modes work perfectly.

I had "adjust clock with acoustic ref" checked in the Analysis page, thought maybe that's it and unchecked it (all of the boxes there) I see the sweep, it gets timing reference this time (late, reporting a delay of -8570ms!) and I get a measurement but it's extremely messed up, as you'd expect for not catching a timing reference till 8.5 seconds into the sweep!
Did a round of those adjust clock boxes and the loopback reference is the IR peak, but they all give you the never seeing the ref failure mode.

Hope this all sheds some light on something. Will go try the exact Win audio settings and using loopback mode, on my original problematic test case.

Edit: Went forward to try my intended loopback test of Laptop HDMI -> AV Receiver -> TV HDMI in -> TV optical out -> XFi soundcard optical in. As before, I have no detection of any digital into the SB XFi card, during the Set Levels test. Tried manually setting the sample rates in Win. 96k, 48k, 44,1k, nothing. Pull the fiber and plug it into a separate DAC, yup there's data there.

Still scratching my head.
 
Last edited:

JonPike

New Member
Thread Starter
Joined
Nov 10, 2017
Messages
12
Second edit... I played music thru that path, listening to digital extracted at the beginning of the chain and also to the AV rx putting it out over the speakers. There is a good fraction of a second delay in there I can hear.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
The standard Java drivers are abstracted from the hardware. REW asks the Java runtime for audio data at the requested rate and the OS provides it, resampling the data if the interface is running at a different rate. That isn't the case for the WASAPI exclusive Java drivers (device names starting EXCL) or ASIO drivers, which set the device to the rate requested in REW.

If you use the acoustic timing reference on a wired connection the timing reference output needs to be the same as the measurement output, since REW listens for the timing reference on the measurement input.

The 5.20.14 early access build has an option to "Fill silence with dither" which can help wake up interfaces ahead of the measurement signal arriving.
 

JonPike

New Member
Thread Starter
Joined
Nov 10, 2017
Messages
12
So, does that mean there shouldn't be a rate setting conflict that would prevent things working? I have been mostly using the EXCL versions of the Win drivers. Not clear to me if expected behavior is for REW to be able to control and change to a different than Win driver sample rate setting. I seemed to not see that happening.

Not sure if the above is really my issue, though. I haven't been able to get back to this for a while, will try to make sure I don't have a mismatch between ref and measurement settings when I do.
 
Top Bottom