It's the salts, and it happens with Q kits. Also true about the low concentrations, the are always "some" salts (in phenol purification "some phenol"), and the less NA you have, the worse the ratio.
Sometime increasing the final spin step in spin protocol to 3 minutes is as an optimization for getting as much ethanol out as possible, but you already did an excess of this.
But what kind of kit you use? I don't recognise the buffers.. PE is the common ethanol was buffer, for gel extractions you use either QIAQuick/MinElute or QIAEX and they do not have QT or PH buffers (it's QG for column kits). What do you use?
And there is just one important note about salts in the column protocol, after adding PE you should let it stand for 5 minutes before spin/vacuum.
And about your digest, you should be limited by the volume first. Depending in what volume you want to digest, you need to keep the concentration high enough to fit that volume. Since the highest volume you can get from QIAquick column is ~45 ul, then you can't have concentration below 22 ng and that is still horrible volume.
Sensitivity of RE to salts differ, some are quite sensitive, some not at all (they already use high-salt buffer). Bu amount of salts also increases with the ratio of DNA/reaction volume, logically, so you just want to have as high concentration as possible (i.e. using MinElute columns if your vector is below 4kb or what is in the manual..).