Further Filastruder Fun

I've been doing some empirical tests with my Filastruder in an attempt to optimise the Filament produced from the ABS pellets I bought. I decided to extrude 1m of Filament at a range of temperatures, and measure the diameter at 6 positions along each length.
This allows me to calculate an average diameter and a max-min deviation.
I extruded at temps. from 190C to 205C in 5C steps. With the following results:

I'm aiming for 0.1cm deviation with an average diameter of just under 3mm. The inlet manifold of my reprap extruder is 3.2mm and so I want to stay comfortably far away from this with a minimum deviation.

Looking at this rather small data set, it seems that 200C is the best temperature for the ABS pellets I have. Temperature less than this start to make the die swell a significant factor, increasing the filament diameter to over 3.0mm.

I also noticed that as the temperature increases, the surface of the produced filament become rougher.
I took some x10 magnification shots of filament with my trusty old stereo microscope:

First, 3mm ABS filament purchased from Reprapsource

Nice and smooth.

Second, Filatruder made 3mm ABS filament extruded at 190C:

Not bad, considering it is home made. Next, 3mm ABS filament extruded at 210C:

Noticably rougher. In the prints I have done, I do not see this as being an issue as the extruder hobbed bolt will average out the 'roughness'. So long as the filament is a reasonably consistent diameter, all should be well.

However, I found an unusual problem when using Filastruder filament, but in a wierd Z axis kinda way....

I've had great results with a couple of my Milestag Lasertag gun handgrips.

I decided to try a more complex print, this one is of my infra red focus tubes:

The print failed around Z=30mm, but seemed to recover at about Z=35mm. A second print of the same gcode failed in the same way. This is gcode I have been printing with reprapsource filament without issue for 2 years.

This image shows the failure on one of the tubes at approx. Z=30mm

A print of another of my Milestag gun parts failed twice, also around the 30mm mark:

I decided to try a simple 20mm square tube print to a height of 40mm, this also failed around the Z=30mm mark:

The only common factor here is Z. All the prints have massively varying X and Y deviations. Z is the only common factor.
Which is when I realised that some months ago I was having a problem with skipping on the X axis. Some prints would fail due to a loss of x motor steps as the layer changed to the new layer. It turned out to be a lack of current on the X axis stepper, but prior to this I changed the software to raise the extruder by 10mm at the start of a new layer.
The thought behind this was that the extruder was jammming on the previous layer as it homed to the start of the next one.
My 'gunbot' is in a heated chamber, with the ABS filament entering from the top:

Through a small hole:

My theory was that the roughness of the 210C extruded filament was catching on the walls of the heated chamber when the extruder raised itself by 10mm at the start of each new layer. This caused it to bend inside the chamber, and exert pressure on the extruder. This slight increase in pressure was enough to clog up the extruder after about 30mm of Z printing.

Here is the same breech part successfully printed on the left hand side with Z retraction removed, compared with its failed peer, which has the 10mm Z retraction: