Getting Started - Weird Calibration Frequency Response

[lange.benjamin]

System:

• Windows 10
• REW 5.20.13
• FocusRite Scarlet 4i4
• Loopback cable SS-BPBQMX2 - 2 FT TRS-XLRM Cale

Sound Card Calibration:

1. I'm following the tutorial videos to get started.
2. Instructions say to connect a loop back cable so I connect the XLR port on the front to a TRS output port on the back.
3. I modify the gain for the XLR port, so that the levels approx. match.
4. Eaven with the loopback cable, there's a bar that flutters around within the calibration dialog, if I was using a speaker, it makes sense for it to flutter, not sure why that's happening when it's a loopback cable.
5. I run the calibration sweep. This is litterally the first time I'm getting used to seeing a response curve and it doesn't look like anything I'm seeing online.
6. Physically, there's not that many things that I could have done wrong.
7. Is the loopback cable balanced vs unbalanced or something?
8. Software, there's a whole world of possible misconfigurations.
9. ASIO vs Java drivers just begins to scratch the surface.
10. Just confused.

 

[John Mulcahy]

Looks like just noise. Are you sure the loopback is connected to the channel you are using as the measurement input?

 

[lange.benjamin]

Fair point, getting the ports straight helps. There's a lot of things that can be misconfigured at the Scarlet layer. But this doesn't look that much better.

 

[John Mulcahy]

Disable all input monitoring, the only thing going to the output should be the signal from REW.

 

[lange.benjamin]

Starting from the bottom knowledge rung here, I'm not sure what "Disable all input monitoring" means.
1. I've turned the monitor gain dial to 0. It didn't really clean up the window.
2. I found an option within the FocusRite software to create a virtual loopback. That didn't seem to make a difference.
3. I zoomed the plot out a bit and things still look pretty muddled. Like many new concepts, I need to see many possible ways to fail before the path of goodness opens.

Really appreciate the feedback.

 

[John Mulcahy]

Input monitoring is routing input signals to the outputs so they can be heard through the speakers. Pretty essential for musical instruments, but a big problem for measuring equipment. Sometimes the interface has a control, labelled "Mix" or similar, with "Input" at one end and "DAW" or "Computer" at the other - for measurement the control needs to be all the way around to DAW, which is telling the interface the only thing that should go to the outputs is signals coming from the DAW (i.e. the host computer). Other interfaces have their own mixer software for more flexible, albeit typically more complicated, control of input routing to outputs. Disabling input monitoring in that case means delving into the mixer controls. It can sometimes just be a case of muting the inputs in the mixer panel.

If turning off monitoring is proving elusive a temporary option is to use different channels for input and output. For example, input on channel 1, output on channel 2. If the input is being mixed back to output 1 that won't matter as it isn't going anywhere. That at least lets you confirm things are working, but the input monitoring will still need to be sorted out.

 

[lange.benjamin]

oks like just noise. Are you sure the loopback is connected to the channel you are using as the

 

[lange.benjamin]

I had a good conversation with my audio interface vendor. The details are above in my first post in this thread. I need a few tweaks on that side but more importantly I needed to get REW's setting configured correctly. The Timing Reference Output and Loopback Input needed to match the corresponding Output and Input fields. By default it was like L->R and 1->2. I don't know if the tutorial said this and I missed it, but it's worth calling out to others.

Now I'm doing measurements and they look like poo. Based on the SPL calibration warning, I suspect I need a real SPL meter rather than trying to get by with an app on my iPhone.

 

[sm52]

Input and loopback input must be different. Select an output other than 'default'.

 

[John Mulcahy]

The Timing Reference Output and Loopback Input needed to match the corresponding Output and Input fields.

That's not the case. The soundcard calibration only uses the input and output channels selected for measurement, the timing ref output and loopback input don't matter for that. When you come to make measurements the use of the timing ref output and loopback input fields depends on what you are using as a timing reference, if anything.

 

[lange.benjamin]

Ok, I got my first measurement. Thx for pointing me in the right direction.

Next step(s)

Compare my measurement microphone and signal to a DUT microphone.
Do I need to register REW to enable multi-channel recording?

 

[John Mulcahy]

Capturing from multiple mics simultaneously requires the Pro upgrade, but if you are doing substitution measurements the mics would be measured individually. More on that here.

 

[lange.benjamin]

Ok, I've got the Pro Upgrade now. I'm attempting to compare the behavior of my reference microphone to a DUT microphone. I have a calibration file for the measurement microphone but not the DUT. Is that needed? How do I create that? I performed a multi-channel measurement and their a long ways apart. What does that mean?

 

[John Mulcahy]

If it is the microphones you are comparing you don't need any cal for the DUT. The responses look quite noisy but it is difficult to offer much comment as the fragment of graph you posted doesn't include the x axis, which also looks to be set to linear rather than the usual log frequency. There is a button for that near the top right of the graph. The Capture icon at the top left of the graph is a good way to get graph images, but for comments on measurements it is best to post the mdat file of the measurement.

 

[lange.benjamin]

Got it, learning each step of the way. I've included an mdat. I expect the DUT to have a low pass filter applied, so rubbish at the high end is about right.
What else are these curves telling us?
Thx,
Ben

 

[John Mulcahy]

The channel 2 measurement looks OK, channel 1 looks like mostly noise and perhaps crosstalk, something wrong there. Missing phantom power? Gain too low?

If you want to compare mic responses it would be better to use a source with more bandwidth.