[Nordenstam]

Am not saying the floor HAVE to be the source of the particular problem here. What we do know is that there is a floor there, there will be a reflection from it and with the distances involved and the levels I had in that setup, it'll look like the graphs I posted. He may have other levels of the reflections.

Originally Posted by DanDan

SAC proposed that a broadband absorber in the reflection path would diminish the reflection enough to alleviate the dip somewhat.
I simply cannot see how this would happen. I do not think a MiniTrap or other panel in the path of a 135Hz wavefront, would drop the level by the required 10dB or so.

Well, it does! As shown in my test. That particular reflection (not the total sound field) drops 10dB in level with the measly 2x2' panel. What's being measured is a broadband reflection, not a 135Hz wavefront. If you move the mic a bit the comb filter dip frequency will change, yet the signal is still the exact same old signal. It doesn't have a single specific frequency.

Here are two excerpts from Bob Golds absorption coefficients tables:

Product	thickness	mounting	density	125hz	250hz	500hz	1000hz	2000hz	4000hz	NRC
703, plain	4" (102mm)	on wall	3.0 pcf (48 kg/m3)	0.84	1.24	1.24	1.08	1.00	0.97	1.15

Product	thickness	density	125hz	250hz	500hz	1000hz	2000hz	4000hz	NRC
RHT 40	4" (100mm)	3.5 pcf (56 kg/m3)	1.07	1.01	1.07	1.06	1.07	1.16	1.05

The rockwool I use should be similar to the latter. The most interesting thing is that it worked so well with the wee size!

Originally Posted by DanDan

My test appears to confirm this view. I would welcome any views on that one way or the other.

Such a test needs a dead steady excitation signal (hard to get even with a sine wave) and a SPL measurement device with at least one digit significant precision. As good as your SPL meter is, it's hardly made for measuring zero dot something differences with any accuracy. You'll be hard pressed to have such a steady excitation signal to begin with in a real room with a human in it. The RMS level differences as cited above was calculated from normalized files.

Originally Posted by DanDan

A fascinating issue has arisen now. Lupo's test shows an alleviation of the dip, in the case of the fibre, but strangely not the foam.

The foam does something, but not that much. Waste of money..

Originally Posted by DanDan

Could we extrapolate that a mere 10cm of Rockwool, no gap, is enough to diminish SBIR dips?

This test was for a reflection, not SBIR, although I think the results translate. Problem isn't exactly gone, but it's sure much better than it was!

Originally Posted by DanDan

I will try to duplicate Lupo's test.

Great!

I do each measurement twice to check for correlation (avoiding random stuff to make much influence) and usually do every measurement series twice to confirm the results. Sometimes pays off!

Originally Posted by DanDan

Lupo, nothing was changed when generating those graphs right? e.g. You have previously shown how manipulation of the window can result in different FR graphs.

Right. Playing around with the windowing can make the rockwool look worse than the bare floor in the FR graphs. :D