Theres 10 years worth of reading on this subject, and at least a 100 different opinions, but to answer your questions as briefly as possible;
1. Am I correct that the ducting goes from machine to cyclone to chip extractor?
Yes, the flow is from the lathe to the side of the cyclone, then out the top of the cyclone to the extractor unit.
2. Do the 'heavier' chips pass through the cyclone?
Yes, and no. Normally, the chips will fall down the cyclone into the collector, but sometimes, if there are a LOT of chips, as in a thicknesser machine, sometimes they will go through the cyclone.
3. The existing ducting on the chip extractor is 100mm, so this obviously needs reducing down to the fitting on the cyclone - if I then use smaller diameter ducting between the cyclone and each machine (lathe, bandsaw, sander) will this reduce the efficiency?
Yes, and no, again. Depends on the bore of the cyclone pipe. If youre thinking of one of those 50 mm dust deputy things, then youre wasting your time and money because the 100 mm pipe will just suck everything straight through the cyclone. Dont ask how I know this, but I do have a 50 mm dust deputy for sale.
I now have a 62 mm cyclone made from an old 3/4 size traffic cone, and all my ducting and blast valves are 62 mm, and its quite acceptable performance when running one machine at a time. But I would like a bigger cyclone in the future.
4. Is there a practical limit to the extent of ducting between each machine and the cyclone/extractor setup - my workshop is approx 5m wide with machines on opposite walls, which also implies running ducting at ceiling level.
Yes. the further the pipework travels, the less flow you get. Every metre, every fitting and every turn slows the flow. Work out the most direct route across the workshop, even if that is a diagonal. Use as few elbows as possible, preferably long sweeping bends. Dont use T joins, they are the worst for restricting flow. If you have to, use Y branches and put a blast gate as close as poss to each machine.