Odd issue with impedance measurement

ajs123

Registered
Thread Starter
Joined
Dec 1, 2023
Messages
4
Hoping for ideas here...

I'm using a TASCAM US-1x2HR USB interface for impedance measurement.

I've verified that the Direct Monitor is off, by sending a signal to the line input that I'm using for input and reference, and verifying that it does NOT feed through to the headphone output that I'm using to drive the measurement. Calibration goes fine, including clean-looking sweeps from each of the three calibration steps. But impedance measurements look like the examples of input-to-output feedback, unless I reduce the sweep length to the absolute minimum.

I've verified that the Direct Monitor is off, by sending a signal to the line input that I'm using for input and reference, and verifying that it does NOT feed through to the headphone output that I'm using to drive the measurement. I've also verified that inverting the input (checkbox in Sound Card Preferences) makes no difference.

This is with REW 20.13 under MacOS Ventura. I was running the latest beta but uninstalled and re-installed with release 20.13 in case that was an issue.

Any ideas appreciated!

Captures from cal and testing follow...

Calibration works just fine. Here's the final step with a 45.9 Ohm calibration resistor:

cal.png


But here's an impedance sweep (96 KHz, 512K sweep length):
test1.png


Reducing the sweep length to the minimum (REW prompts to increase it) results in a clean measurement. For example, this is at 96 KHz and 256K sweep length:

test2.png
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,146
I'm using a TASCAM US-1x2HR USB interface for impedance measurement.
Presumably you are using the 2 line input connections on the rear? It looks like the loopback function may be active:

A loopback function allows two audio channels from the computer to be merged with the input signals and send this mix to the streaming or communication software
 

ajs123

Registered
Thread Starter
Joined
Dec 1, 2023
Messages
4
Presumably you are using the 2 line input connections on the rear? It looks like the loopback function may be active:
I am. Yet the Direct Monitor switch is off, and the loopback function is turned off in the US-1x2HR app. Feeding an audio signal to the rear panel inputs, I verified that as long as the Direct Monitor switch is off, the signal does not feed through to front panel headphone output.

Most confusing to me is that I get a clean sweep when doing the fixed resistor cal, then without touching anything get the modulation in the impedance measurement. And I get a clean sweep when doing an impedance measurement at the lowest possible sweep length.

One related additional bit of info: When I run the fixed resistor cal with Direct Monitor turned on - expecting a bad result - the signal on the little scope in the measurement window looks noisy but the sweep shows a clean, constant resistance. So perhaps the way REW is sampling when at minimum sweep length makes it immune to feedback. But where the feedback is coming from remains a complete mystery.
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Messages
7,146
Stick with the latest beta.

If there is some feedback path active it may be getting calibrated out for the cal measurement, but that would only work when measuring if using the same sweep settings and load.
 

ajs123

Registered
Thread Starter
Joined
Dec 1, 2023
Messages
4
Thanks, John. Speaker impedance measurements look clean and make sense as long as I use the minimal scan length. Do the cal measurement use a fixed (perhaps minimal length?) scan length?
 

ajs123

Registered
Thread Starter
Joined
Dec 1, 2023
Messages
4
This seems to be resolved, and rooted in an error I must have made along the way. In a brute force approach :-) I uninstalled with deletion of all preferences and logs, and re-installed (latest beta). I suppose that what happened is that I had the loopback or direct monitor turned on at some point and had cal(s) saved with the feedback.

Thanks, John, for your guidance and your patience with user error.
 
Top Bottom