I have checked this and generated three wav files which you should find attached to this post. They are a 0db sine wave, a 0db square wave (+3b), a -5db sine wave, and a -10db sine wave.

It transpires that the visual was using 1.5 times the rms for its display to expand the needle range for normal sounds into the region above 0db. Actually this should be a multiply by (square root of two) but 1.5 is close enough. This brings the 0db point to the far right of the display.

I have now altered the algorithm so that the scale is as discussed with 0db being the sine wave reference point at around 64 degrees. However, it is up to you, the users which you would prefer. You should not see much activity in the region above 0db unless your audio is clipping.

Tony's latest scale is pretty accurate, even -5db and -10db are nearly in the right place, and he didn't have any theory to work from for that scale.