Day 2 - I have the basic frame built, tomorrow on to the y axis slide assembly…

Day 3 - OK first mistake, I picked up the wrong shaft and pushed the XY linear bearings onto the z axis screw shaft and removed a few small balls from one of the bearings. Well one of the upgrades I had planned was to replace the linear ball bearings with Drylin(c) bearings, so I have just bought that forward and they are now due in 48 hours time.

The Z Axis threaded rods seem a tad short, by design – so I have refitted the flexible motor couplings on the Z axis such that they only just overlap the motor shaft and the rods. This is not unique as there seems to be multiple solutions to z axis wobble on Thingiverse. So one of the early prints will be some Z axis supports – and I have ordered a couple of extra 608zz 8mm bearings for them. Update: I subsequently concluded that adding any Z wobble devices would make matters worse so I have never fitted any.

Hmm… I bought a machine advertised as the latest version with auto levelling but there is no sign of a Z sensor, just a limit switch. Time to complain to the vendor in China, worst case I will need to spend another £10 and just wait for the parts to arrive from China. Not a big problem as they machine will work with manual levelling.

Update: The vendor refunded me £10 and I never fitted any auto levelling as the print quality with manual levelling has always been good enough for me.

I have powered up and re-calibrated the power supply using some load resistors. All the motors and drive belts have been fitted leaving the electronics.

Weird… If you move the print head the Y motor acts as a generator and produces enough back voltage to power up the display !

Warning! Whilst it does not appear to be a big problem for the ANet's 8 bit motherboard, there are reported cases of 32 bit motherboards being fried by the voltage generated when quickly moving the X or Y axis by hand.

Just the heated base plate to install and tidy up the wiring before powering it up.

Day 6 - New Years Eve - Fully Assembled and powered up – I have levelled the print bed and calibrated the end stops. The z-axis being really problematic. I decided to bypass the test prints and go straight for a print of a motor support for the Artificial Horizon.

I have downloaded the latest version of Ultimaker CURA software to slice the CAD models and produce a g-code file for the printer to print. Of Course the Anet A8 does not appear to have a preconfigured printer profile for the latest version of CURA, meaning I spent the rest of the day creating one by reversing engineering the sample g-code files provided by ANET.

02/01 To debug the printing issue I take an STL file (3D design) for one of the examples and run it through CURA then compare the manufacturers g-code to CURA generated g-code. Voila! The manufacturer is using Absolute Extrusion and CURA is producing Relative Extrusion values. The modified g-code header I am using assumes the ANET default Absolute (does it actually support Relative ? I need to test this).

I ran a test print through CURA using Absolute and still no filament comes out. In desperation I take the same STL design code run it through Slic3r, an alternative slicer program, and that runs perfectly OK. So CURA is going to be binned for the moment while I get on with first calibrating the printer and then printing the upgrades that I want to apply.

All the forums say that the machine is just good enough out of the box to print the extra pieces that it should have had to make it a really good printer!

Then I think the next step will be to change the firmware in the printer to something more safe and usable. The Anet A8 is a cheap Prusa i3 clone, with a very clunky interface and no programmed safety cut outs. The electronics for which is basically an Arduino with a CNC shield and I have already found 3 alternate open source firmware stacks, though Marlin looks likely to be the OS of choice.

Update: I eventually decided to upgrade the board to a BTT SKR 1.3 and Marlin 2.0 rather than try to squeeze everything into the ANet's limited memory.