Accurate DF'ing is done using Time Difference Of Arrival (TDOA) because timing can be measured extremely accurately and can yield far more accurate results.

Thus it takes three receiveing stations that are synchronized in operation to DF an emitter when using the TDOA technique; however, the results can be very accurate.

I've recently read a little more explanation since my last post, and I think I may still have some details wrong, but I think what I wrote is close to the way at least some systems work to generate immediate DF results.  It appears to me there are two primary methods, one using angle of arrival of the signal to each receiver antenna, the other using time difference of arrival of the signal to pairs of receiver antennas.  But I've read that some systems used for long-range detection (of aircraft, for example) need three receivers just to get a two-dimensional fix in azimuth and range, and actually require four if you also want to determine altitude of the emitter.  Furthermore, while some systems can get quite precise in azimuth and range to the emitter (as in within a degree and a few thousand feet when it's two hundred miles away), altitutde is more difficult to pin down since it can also be off by many thousands of feet (which is probably only going to varying over a range of maybe 50,000 feet to begin with).