I have had a similar problem with an all-metal hotend and PLA in a bowden setup. Because of the bowden I needed to have a few mm of retraction to counter oozing. With ABS I could get succesful prints most of the time (not always, it would occasionally jam), but with PLA the hotend would always get clogged somewhere during the print. First layers were nice, then the extruder would get a hard time and jam totally. I have a Bondtech setup on that printer and it would put so much pressure on the filament that the 4mm couplings would spring loose, releasing the ptfe bowden tube (and piling the filament next to the printer).

I finally nailed it down to the heatbreak (the threaded part between the cooling body and the heating block). After taking the whole hotend apart (including the stuck PLA), I found that the PLA was stuck somewhere in the heatbreak and could not be removed by any force. I used pliers and a vice, so it was really, really tightly jammed. My concusion was that the PLA had glued itself to the inside wall.

Aparently PLA sticks quite well to metal surfaces. It seems that polishing may solve the problem (http://www.dta-labs.com/products/prometheus-v2), but I took the easy way out and exchanged the heatbreak for a teflon lined one I found in my spare parts bin. Used my Dremel to make it fit the inside of the cooling body as the top needed to be thinner. Result: problem solved. I have been happily printing ever since (4 months or so) without a single jam and at retractions up to 6 mm (!!). I just never run the nozzle higher than 260 C now, but I only need 250 for ABS and can do with 235 for PETG and 220 for PLA at high speeds. As I can buy teflon lined heatbreaks for a dollar or so, it is also cheap to replace the heatbreak if the teflon liner gets over-heated by accident. I now have a couple of the right size as spares.

Some people claim that it is dangerous to run teflon at temperatures near 260 C. However, this is the temperature limit where it still maintains most of its mechanical properties and is the recommended limit for engineering purposes. It actually melts at 327 C, and only starts to decompose at temperature way above 400 C. http://nvlpubs.nist.gov/nistpubs/jre...n6p327_A1b.pdf. Beware however that your themistor is not indivudually calibrated and that your temperature reading can be off by 5-10 C (typical) or even 20 C (extreme).