Negative IR Delay

oliver

New Member
Thread Starter
Joined
May 24, 2017
Messages
108
Location
Italy
Hi John,
I am on Mac OS 10.13.6, with the latest REW build.
I am testing my system, but I found something that I really do not understand, I do have a negative IR delay estimation, even when I test a wired Input/Output.

I have connected the output 1 of my Fireface UC to the Input 1, these are the settings:

Schermata 2022-12-02 alle 12.24.22.png


Schermata 2022-12-02 alle 12.24.31.png



Schermata 2022-12-02 alle 12.24.57.png


After performing the measurement I calculate the IR delay and this is the result:

Schermata 2022-12-02 alle 12.25.18.png


How it can be?
Oliver.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
Easier to comment if you post the mdat, but there is a small influence from the implicit windowing from measuring a range that doesn't extend to half the sample rate. It is also important to check that the sample rate for the Fireface in Audio Midi Setup is the same as the rate in REW to avoid any OS resampling.
 

oliver

New Member
Thread Starter
Joined
May 24, 2017
Messages
108
Location
Italy
Hi John,
here the mdat.
The sampling rate for the Fireface is set at 96K the Midi Setup, in the RME USB setting and in REW.
In the mdat file the first measurement was taken with the acoustic timing reference, the second one with a loopback timing reference.
 

Attachments

  • Soundcard .mdat
    4 MB · Views: 7

oliver

New Member
Thread Starter
Joined
May 24, 2017
Messages
108
Location
Italy
I have also made the measurements at 44.1K, with loopback REW says the delay is 0, with acoustic timing there is a 0,0013ms of delay but positive.
 

Attachments

  • Soundcard .mdat
    7.2 MB · Views: 3

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
The measurements all look normal, interesting that the card appears to use a different reconstruction filter for the 44.1k measurements. To understand the delay estimate it helps to understand the process. REW estimates delays by comparing (cross-correlating) the measured impulse response with a minimum phase version of the response, since a minimum phase response has no time delays. This works well for most acoustic measurement purposes, but it will typically give small negative delays for electrical measurements since DAC outputs are most often linear phase rather than minimum phase, so their response starts before the impulse peak, hence the negative figure for delay.

1669984364853.png
 

oliver

New Member
Thread Starter
Joined
May 24, 2017
Messages
108
Location
Italy
Hi John,
thanks for the explanation.
I do measure Balanced Armature drivers using a Gras 0402 Ear coupler. The distance from the driver's output and the microphone is very small, 12.3mm. I did some measurement with the same driver at 44.1K and 96K, and when using the acoustic time reference I can see a negative delay.
Here the mdat file:
 

Attachments

  • BA measurements.mdat
    6.4 MB · Views: 3

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,212
The acoustic timing reference shows delays relative to the arrival of the timing signal at the microphone. The timing signal is a 5 kHz to 20 kHz sweep, if the system response drops off significantly within that range, as it does for those measurements, that will affect the detection of the timing signal. Those responses are almost completely minimum phase, however, so the IR delay estimation gives an accurate figure to use to shift the response for meaningful phase. I'm not sure what benefit a timing reference could bring in that measurement situation.
 
Top Bottom