While the instruments seem to handle most USB sticks just fine during regular operation (as long as they are FAT16/32 format), the boot loader is a completely different piece of software that uses more basic device drivers and might be more picky.
(I also believe it's a USB stick problem...)
Let's try to understand this in a more scientific way:
As Performa01 suggests, it definitely can be (or is) a problem in the way the sticks's USB controller handles the scope's requests. The scopes usually use several raw-access routines which may not be correctly handled by the controller.
For example, in the DS1054Z, Rigol tries to read explicit boot sectors to see if the stick is a "special vendor USB stick" which triggers special options/access rights. Those raw boot sector access routines are not used anywhere else in the scope, only on USB boot/insert.
I suggest you all to use the ChipGenius (
http://www.usbdev.ru/files/chipgenius/) software to see which is the controller/fw version of your stick and to establish a compatibility table with these scopes.
And, when you mention sticks that work or not work, include that controller reference so that we try to map the valid ones.