Any changes in waterfall calculation ?

[Vorreux]

Hi John,

I measured a mid transducer in october 2018 (outdoor half space measurement), and I did exactly the same measurement last week.
As you can see, the frequency responses are exactly the same :

Hereunder the waterfalls :

As you can see, the waterfalls are completely different, while using exactly the same parameters !
Can you explain it ?
I am currently using the v5.20 Beta 6 version of REW. In october I had a former version, I do not remember which one.
I attached the IR text files.
Thanks in advance for your help.

Pascal

 

[John Mulcahy]

I can't do very much with text files Pacal, can you attach an mdat file with the measurements? Would ordinarily use a much shorter time span when examining a drive unit's behaviour, by the way, with that 100 ms time span the plot is mostly showing the background noise.

 

[Vorreux]

I never encountered any problem with those settings before.
I attach the mdat for both measurements.
To appreciate the temporal behaviour of a mid driver, which settings would you recommend ?

Pascal

 

[John Mulcahy]

For such a short rise time (0.2 ms) I'd use a time span of 10 ms or less. I don't see any differences in the waterfalls of those measurements that don't reflect the differences in the frequency responses or the differing noise content of the measurements, which parts are you referring to when you say they are completely different? With a 50 ms time span the 2nd slice is already showing little more than the noise in the measurement, there is a bit more to see with a 10 ms span but there are no resonances in the span selected.

 

[Vorreux]

Hi John, and thank you for the quick reply.
The transducer I'm working on is the 8NM610 from 18 Sound. It has its own closed rear load :

Concerning the measurements I sent to you, both have been done in exactly the same conditions (amplifier, B&K microphone, audio interface) :

So the waterfall plots should be identical, and it is not the case on the pictures you sent.
I modified the settings to take your recommendations in account :

There is a resemblance for zone 1 above, but not at all for zone 2.
What can explain this ?
I'm especially interested in studying diaphragm resonances, with typical 40 dB depth.
Can you recommend better settings for this kind of driver analysis ? The ideal settings should probably be different when studying bass, mid or high drivers ?

Pascal

 

[jtalden]

Vorreux, expanded thoughts FYI:
> As John pointed out, the ETC overlay shows significantly more noise in one of the 2 traces. Possible the mic gain was different in the 2 measurements or ambient conditions were different?
> Note that the scale on the chart clips visibility of the first slice by a lot (~30 dB off the top of the chart?). This makes the second slice trace look like the first and thus the information at the bottom of the chart looks more significant than it is.

Try rescaling the chart so the first slice is visible and set the range to a smaller number such to hide the noise of the noisier one.

 

[Vorreux]

Vorreux, expanded thoughts FYI:
> As John pointed out, the ETC overlay shows significantly more noise in one of the 2 traces. Possible the mic gain was different in the 2 measurements or ambient conditions were different?
> Note that the scale on the chart clips visibility of the first slice by a lot (~30 dB off the top of the chart?). This makes the second slice trace look like the first and thus the information at the bottom of the chart looks more significant than it is.

Try rescaling the chart so the first slice is visible and set the range to a smaller number such to hide the noise of the noisier one.

Hi jtalden,

Maybe there was a difference in the measurement conditions : one of the measurements could have been made with 1 sweep and the other one with 4, explaining the noise difference, but I'm not sure.
I do not understand the rest of your message, can you be more specific ?

Pascal

 

[jtalden]

I just downloaded the file again and took a more carful look. I misread the small type on the scaling the first time. In reviewing this again, I don't really understand that chart at all given the mdat provided. Regardless, given the careful setup shown in post 5, I can see you are looking for subtle detail >30dB down. This is beyond my level of experience.

 

[John Mulcahy]

They were both made with one sweep, the details of the measurements are in the Info panel. The differences you are pointing out look like noise in the measurements. If you look at the Distortion graph you can see the noise floor is much higher in the February measurement. It is hard to control the ambient noise level outside, particularly the contribution due to air movement. If the driver has a resonance it should be visible from the very first slice and would also make itself known in the impedance measurements.