TRA / FFT window flickering - clock ∆ unstable

[redmountain]

Macbookpro 9,1
Macos 10.14.6
REW 3131
96khz clock
firewire stream
I get inconsistent behaviour on the fft window...
See clock ∆ in the movie
Best
Klaus

 

[John Mulcahy]

Looks like dropped samples/lost blocks.

 

[redmountain]

This is with Firewire
I tried with USB too
tried 48khz clock
same thing...

 

[redmountain]

Is there a system log or REW log that shows something ?
Klaus

 

[John Mulcahy]

You could try at 48 kHz to see if it is any better, but the appearance of the plot and the change in clock delta are both typical of lost samples.

 

[redmountain]

Sorry > edit post > re 48khz

 

[redmountain]

Buffer(s) ?

 

[sam_adams]

I get inconsistent behaviour on the fft window...

That's a dropout in the data stream. Remove Menu Meters, the clock, the battery, and all the other widgets from the Menu Bar. Check in the Users and Groups System Pref for login items that start when you login and remove them. Might be easier to create a new account on the device without all that stuff running than to turn it on-and-off when you want to run REW.

 

[John Mulcahy]

For reference, I did a quick check just now with my ADI-2 Pro on a mac mini at 384 kHz, no issues.

 

[redmountain]

> new account
nothing in the menu bar no login items
> still clock ∆
?
Macos / coreaudio ?

 

[John Mulcahy]

If the output and input are both from the same interface and so using the same clock it would be odd for there to be a clock delta. Java gets its audio data via coreaudio, the OS will not resample unless the sample rate for the interface(s) in Audio Midi Setup is not the same as the rate in REW.

 

[redmountain]

...if such dropouts happen, I would hear them for sure, not ?
I can listen to the loopback and can't hear anything...
I can re-check but am alsmost sure the clocks are the same everywhere...

 

[redmountain]

Soundpath :
Motu analog out 1+2 out ( defined as output channel mapping 3+4 in REW )
Motu analog in 5+6 as input
timing ref analog out 2
loopback input analog 6
wrong ?

 

[John Mulcahy]

If there are dropouts in the source they would be audible. If data is being missed because something pushes REW into the background long enough to miss an entire sample block then nothing would be audible.

 

[redmountain]

How can we find out ?

 

[John Mulcahy]

On macOS I don't know, sorry. Windows has tools like LatencyMon but I don't know if there's any equivalent for Mac.

 

[redmountain]

Hmmhh...
If I start the generator and start RTA after, the clock ∆ goes down to ca. 1.8 ppm and then struggles
If I start RTA first then the generator, the clock ∆ goes down all the way to 0.2 ppm and no flickering...
Is this only on my rig ?

 

[phofman]

@redmountain : what REW buffer size do you use?

 

[redmountain]

64k in and out

 

[sam_adams]

Is this only on my rig ?

You can test the Round Trip Latency (RTL) of your setup with RTL Utility.

 

[redmountain]

and now it seems to work with all the widgets, former user account...

 

[John Mulcahy]

If I start the generator and start RTA after, the clock ∆ goes down to ca. 1.8 ppm and then struggles
If I start RTA first then the generator, the clock ∆ goes down all the way to 0.2 ppm and no flickering...
Is this only on my rig ?

There have been other reports of odd behaviours on Mac for some interfaces if the generator is started before the input.

 

[redmountain]

You can test the Round Trip Latency (RTL) of your setup with RTL Utility.

Thank you
For real world results I use Smaart...until now
Maybe REW does that too ?
( from XLR out to XLR in ? )

 

[redmountain]

There have been other reports of odd behaviours on Mac for some interfaces if the generator is started before the input.

I didn't find them searching this forum...
Thanks for confirming !

 

[John Mulcahy]

There was also a report of odd behaviour once that was tracked down to different sample rates for the input and the output in Audio Midi Setup. Wouldn't think that could occur if they are the same interface though.