A lot. A matter of days actually. Try to diple it down to 5-7 um at least... I etched quite a few samples on different ion mills, and they really make a difference (regarding diameter of the ion-beam)... what's your setup? And watch out for re-sputtering of W back onto the sample... otherwise you'll discover nano-W in TEM :-)
It is hard to analytically to tell the time because the W is a refractory/hard material. The experience of running the Ar ion milling on the W will only tell when the W becomes transparent for the TEM examination. I have run Ar ion milling on of several Fe-B-Si metallic glasses with an average thickness of 20 um in my previous investigation. It took from a half day to one day to make the samples transparent for the TEM observation.