Indeed you have seen the possibilities :)
To understand the temp profiles, you might want to do some investigating of how the slicer assigns temps. You can study the BFB in an offline database and query the 'patterns' the slicer is designed to generate. They are not critical, but to meet the designer's intent, you want to at least understand it.
A tip from my analysis... the first layer temp is much higher using PLA. 235 where 215 is the norm. There is overshoot control in the switching of nozzles. there are several 'maintenance temp' calls which have lower counts in the table. Buddy is adding up all the temp calls for each temp in the file.
The blob is two fold. I am glad you mentioned the start blob. I take out a line in the BFB that parks the nozzle just outside the wiper. This helps a little. I think there is a slight bug in the start because I notices that is starts out fat and only slowly reduces the trace width. This could be managed with some manipulation in the pressure setting (M108). The other setting you want to know about retraction settings. This is where the blob comes in. Let me explain why we may want to affect these values.
First of all, the original cartridge bowden tubes stretch anywhere from 1/4" to 3/8". This was accounted for in the slicer. Our 4mm bowden tubes barely stretch. So we can back of the retraction settings that makes up for this stretch. I find that setting the 2 first values on M227 and the equivalent value in M228 can be cut to 2/3 their original value. Stringing is an indicator of when you hit the lower boundary. The last 2 values also have an affect but I haven't got to the bottom of it yet. They are unique to Cubify.
There is also a painful housekeeping routine for each layer and within layers This is when the extruder is stopped, the plate drops a bit, and the retraction takes over. This works okay, this is when it is destringing, but it also does this without retraction at times. This is when the blob occurs. Basically pausing while filament is expanding from heat. This could be solved with a slight offset between retract and re-instating the original position. Basically pull back a little more than you than you subsequently forward.
Another method for appropriate single models is to remove the entire M103 series and make it a continuous print. This works well for towers, for instance, when only skin layers are put down.
Also look at those #vector calls. These are slicer modes. Walls, sidewalks, infill, etc. each have their own call so you find certain sections a bit easier.
And if you really want to have some fun, put the program's .XML files through the decoder :)
