Stepped waterfall

[Fledermaus]

Hi,

Recently changed/moved furniture in my listening room and gave it a REW (v.5.19 beta 5) measurement out of curiosity, and found this strange looking waterfall (threshold level intentionally lowered, and time window increased the same)
http://hpics.li/ec3c83c



with this triangular, comb filter-like looking stepped decay above 5k

Whereas it formerly looked like this
http://hpics.li/d40d4f8


Similarly, the unfiltered IR shows a lot of late noise, associated with a very dirty 8k filtered IR (see attached file).

Can't figure out the origin of this change, as I doubt the new couch could ring this way.

Thanks for any insight you'd be willing to share !

 

[John Mulcahy]

The measurements show high frequency noise that appears at about 400 ms into the impulse. Quite odd. Try a shorter sweep to see if it looks the same. Not sure where that could be coming from, something in the room vibrating, perhaps?

 

[Fledermaus]

Hi John,

Thank you for quickly answering.

Here are my latest attempts, same time, same conditions except sweep time :

10.9 s :



5.5 s :



It looks better in the highs department - but it worsens what looks like modal decay in the lower range - how come ?

Could it have to do with that kind of repetitive echo sweep that manifests when the main sweep reaches the upper range ?

I must confess I'm still a bit confused ...

Thanks again

 

[John Mulcahy]

Could it have to do with that kind of repetitive echo sweep that manifests when the main sweep reaches the upper range ?

What do you mean by that? If the sweep sounds odd that would indicate something is wrong with the setup.

 

[Fledermaus]

What do you mean by that? If the sweep sounds odd that would indicate something is wrong with the setup.

Ha, this gets interesting
I mean, on making a measure, when the sweep reaches 5-10k, one can hear a series of fainter, echo-like sweeps that go shortening, say, from 1 s to 1/10 s, while the main sweep fades (at least to aging ears like mine) and ends.
Didn't give it much attention until now since it didn't seem to interfere.
And, yes, furniture aside, I changed one other setting , i. e. checked the "frequency dependent window width" box in the settings menu.
I'll give it a go tomorrow with box unchecked.
Otherwise, which setup feature should I look into ?

 

[John Mulcahy]

Make sure there is no form of audio processing active on the PC, and that whatever device you are using for output is set to 48k sample rate and not some lower rate. You will not get proper measurements until the sweep sounds clean. What equipment is in the replay path?

 

[Fledermaus]

Right !
REW runs on a W10 Dell PC, USB connected to a Totaldac d1-six DAC (FIR treble compensation off) and then analog through preamp, amp and speakers.
When I checked as indicated, the DAC displayed USB 44.1 as input, instead of 48 ; so I switched it to 44.1 in REW preferences, and everything is fine again, as can be seen below ; I just have to figure out how to tune the PC output to 48 k...
Thanks for your help !

 

[John Mulcahy]

Looking much better, but if you are using a UMIK that should run at 48k - if you run it at 44.1k it will be sample-rate converted in Windows. If you look in the Windows properties for the TotalDAC (on the Advanced tab) you may find it has been set to use 44.1k, if so better to change that to 48k to avoid resampling.

 

[Fledermaus]

Done !