Ok, the controlled listening for the grill on/off/switch for the 560 was a very mixed result.
I was consistently able to tell when the grills were on and off, without much problems - about 90% - so theoretical difference is there, indeed!
In practice, however, the difference is smaller. I listened to 8 instrumental and one vocal track.
In 6/8, I was mostly guessing and the correct/wrong ratio never really went over 50%, so in these songs, the difference was either too small or it did not affect it to a desirable margin.
In the other two instrumentals, I fared a bit better - somewhere around 70% of success so there seemed to be a distinguishable difference, but I really had to strain my ears hard to hear it. A bit more air to it slightly better imaging, though the stage did not seem much wider... about 5%, probably.
With the vocals, the 8 - 9khz dropped seems to help by reducing slight sibilance from very little to less than very little... and the vocals did sound a bit more open... and with more instruments, they seemed to image a bit better and stretch a bit wider... The consistency was around 80% and the difference might have been a bit more obvious, say 10% with vocals and somewhere north of 5% with instruments.
It's worth mentioning that I chose 30 second segments of each track, which I thought would be desirable for this particular task.
I'll probably modify it next time to longer, like 60 seconds, segments and either with, without or half with/without to give a more reasonable chance to my ears to pick the differences. I'll do more controlled tests this week.
I'll be also doing the grill mod on the 560 and 400i tomorrow [mostly because it is a possibility] and compare it to without the grills and with the modded grills against the stock grills on the pink-noise track, where the difference is biggest now... Then I'll be using both headphones with modded grills for a week and then switch back to stock.
So far the difference really is very small though and it is easily the least effective modification I've done so far.