My old extruder nozzle was made from a solid brass rod with a 0.5mm hole in the end.
It was drilled 3.2mm from the other end to accept the filament. The problem was gauging how far to drill from the back. Ideally the hole in the end should be as short as possible to provide less resistance to the filament flow. Drilling too far would write it off so I erred on the side of caution.
I suspected it was a bit on the long side as I got more die swell than I was expecting. Since it is now scrap I sectioned it to find out exactly how long the hole was. It turns out it was about 0.6mm.
The latest RepRap design uses an acorn nut which gives much better access to the back of the hole. I used the smallest drill I have, which is 0.3mm, to start with, I might open it up later.
The dome of a brass acorn nut is quite thick so I opened the hole out from the back using a conical milling bit that I bought for PCB track isolation milling.
The point is actually about 0.3mm so I was able to countersink the hole from the back until the point came through. That means the rim of the hole is very thin indeed and there is a double taper leading to it. The first is created by the drill that made the thread hole in the acorn nut and then a shallower taper made by my mill bit. It will be interesting to see what flow rate and die swell I achieve with this close to ideal shaped aperture.
I turned down the front of the nozzle to a point to give some clearance to the work piece as suggested by Vik Olliver somewhere I can't find now.
The extra hole on the face next to the thermistor is to allow me to introduce a thermocouple during calibration.