Delay variability

thothsong

Member
Thread Starter
Joined
Jun 29, 2021
Messages
161
I'm measuring the Center channel of a soundbar that's composed of 3 drivers. I'm using a sound mode that doesn't upmix to other channels, and the sub is disconnected. I'm measuring all within a single REW session, and the UMIK-1 remains in the same position throughout. I'm using 48 kHz sweeps all output through the same HDMI port on the laptop, with timing reference always on the Center channel. I measure using the REW-generated sweep using ASIO4ALL (7.1 channels), as well as file playback in various formats produced by taking the REW-produced sweep file and encoding it as the center channel, with silence on all other channels. The file formats measured are 7.1 LPCM, DD 5.1, DTS 5.1, DD+ 5.1, and DD+ 9.1.6 (the Dolby ones are bitstream output over HDMI). When I look at the Delay that REW notes for each measurement, it's pretty much all over the place. For example, 3 sweeps with ASIO4ALL, the Delays are 0.0288, 0.1385, -0.0001; the corresponding Clock adjustments are 0.3, 117.4, 1.2. With a 7.1 LPCM file, example Delays are -0.0031, 0.0429, -0.5497; the corresponding Clock adjustments are 1.4, -1.1, 2.4. With DD 5.1, example Delays are 0.0016, -0.0656, -0.2180; the corresponding Clock adjustments are 33.3, -6.9, 7.8. If look at some DD+ sweeps where no clock adjustment is reported (some do have clock adjustment), the Delay is fairly consistent, but non-zero: 0.0235 to 0.0252.

What's the likely explanation for this variability?
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
5,833
You didn't include a unit with your delay figures, presumably they are in milliseconds. It is normal for the delay to be non-zero but small, the response to the timing signal 5 kHz - 20 kHz sweep can differ from a full range sweep. I don't know what causes the outliers in your measurements, perhaps the encoding or decoding is a factor, but the replay clock does seem to be a little variable.
 

thothsong

Member
Thread Starter
Joined
Jun 29, 2021
Messages
161
Sorry, yes, they are in milliseconds. What would you expect as a normal range for the delay in this case?
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
5,833
One sample at 48 kHz is 0.0021 0.021 ms, variability would usually be less than depending on background noise levels. Compressed content might affect timing, but the effect should be consistent. If the replay clock variability is an issue shorter sweeps may give more stable results.
 
Last edited:

thothsong

Member
Thread Starter
Joined
Jun 29, 2021
Messages
161
One sample at 48 kHz is 0.021 ms, or am I confused? The sweeps above were 1M length. I ran some sweeps generated from REW with 128k, 256k, and 512k length, and I'm not sure any were really any more consistent. The variability isn't a problem for anything I'm doing, just trying to understand things.
 
Top Bottom