First, the temp that is being read back to you is from the exit thermocouple and is not at all representative of the bean mass temp.
Second, try using a consistent amount of beans. I use 283 grams (10 ounces, loss easily calculated) and try to stick with that amount.
The higher the altitude at which the bean was grown, the higher the initial temperature setting should be. Accordingly, try these beginning settings on your beans:
* Yemeni, Africans, Central American - around 470 degrees
* Brazilians, Island, Indian - around 460
* Decaf, Konas - around 450
(Some start with a lower temp and hold it when it reaches about 350 for about 3 minutes, to develop the sugars), then raise the temps. Compare the roast of identical beans and see if you discern a difference in taste.
After your roast hits first crack, let the beans move through the crack for 30-60 seconds and then turn the unit down 20 degrees. That profile works for me. I am grinning because I trying someone else's profiles lately whom I respect and the beans ended up without sufficient fruits. Profile was too flat.
My voltage is usually around 119 and drops to about 116V as it cycles through. Not sure why you aren't drawing some oils at 20 minutes but you didn't share the weights with which you are working. Less beans will take longer (if memory serves me) as there is greater flow-thru of heat and less temp build-up in the bean (I would need to check my notes - got into an argument about this and wanted proof).
B|Java