Fixing a chinese 3D Printer: Part 1

TL;DR: You have a lot of work to do until this is usable

IMG_20170521_191100-1

photo of a random print at 120mm/s

Introduction

I’ve bought an Anet A8 für 143€, everything included. It’s a clone of the the Prusa i3 and available under all kinds of brandings and with every changing components. Here are the things I’ve learned on my way.

 

Common Errors which have to be fixed

 

Not adjusting the elastic Couplers on the Z-Axis

The aluminium couplers which couple the stepper motors to the threaded rod have to be lifted a bit. The stepper axis should only enter ~ 5mm into the coupler. Same for the Z-Axis rod.

IMG_20170521_185901

right hand side Z-axis coupler

IMG_20170521_185904-1

right hand side Z-axis coupler

This is something which you probably won’t catch yourself. You’ll have more Z-wobble and may loose steps.

The Z-axis threaded rod should be flush with the acrylic frame if you did it right:

IMG_20170521_185930_HHT-1

right hand side Z-axis top

Right Side X-Axis Idler

This is just faulty design. The right side’s idler, which holds the linear bearings, bends the Z-axis rods inwards. The more you fasten the X-pulley, the more it bends.

IMG_20170521_185854

right hand side X-axis idler

IMG_20170521_185844_HHT-1

right hand side X-axis idler

Symptoms are grinding noises towards the lower end, uneven movement of the two Z-axis and of course missed steps. Also you shouldn’t be able to see the X-Axis rods’s end on the right side of the idler.

 

You can only fix this by printing a new part. You should do this as soon as possible. Otherwise you can only print very slowly and you will damage the Z-axis. Try to keep the X-axis belt as loose as possible.

It’s available at

X Motor mount for Anet A8

 

Y-Axis Carrier upside down

Even the vendor’s instructions are wrong on this one. The Y-axis belt should go in straight lines to and from the stepper and the idler.

IMG_20170521_190052

Y-axis pulley attachment installed wrong and pulling upwards

Otherwise you will have uneven pressure on the belt depending on the position because the carriage pulls the belt upwards.

You can fix this by just turning it around. Yes, that means taking everything apart. Again.

Bed Cable Chain

This is looks really cool! Turn your bed by 90° and you move all the cables out of the way.  This is necessary because the white heatbed plug takes all the mechanical load. This will lead to failure!

The contact will be intermittent. If the software doesn’t catch it this connector will catch fire. Find it on thingiverse at

Anet A8 Y-Axis Cable Chain – Right Side

You also want to hold the chain down. Mirror and print this:

Anet A8 Chain Bed Protector

Optional upgrades

These are nice to have, but not mandatory.

Y-Axis pulley fastener

This helps you fastening the Y-axis belt. Makes your and your stepper motor’s bearings life much easier.

Remixed Y axis belt tightener for A8, A6

Z-Axis Top Caps

This should be unnecessary as all the forces should be taken by the hardened rods. Should. You might reduce a bit of Z-axis wobble.

You can find it on thingiverse as

Anet A8 Z-Axis cap

The Mosfet Upgrade

It’s not strictly necessary because the Mosfet on the Mainboard is quite capable and as long as the connections are good, you will only loose a bit of power. But the heated bed is already at it’s limit with the 12V and the voltage drop over the long wire doesn’t help.

Just connect the heated bed directly to the power supply and put a mosfet in series. You can switch the mosfet with the original wire or any signal wire.

I’ve gone on a spending spree and invested in a optocoupler because I didn’t want any ground loops. Making and testing the PCB took me ~1 hour. Next time, I will buy the ready made PCB für 5 USD.

New Firmware

You should use Marlin or the Anet A8 fork called ‘skynet3d’. Flashing it is really difficult. Luckily there are two alternatives to the ancient arduino version you have to use. That is

marlintool is a shell script which downloads and flashed the software. It’s available for Linux x86 and raspberry. This is what I used; works surprisingly well and can also do backups.

Platformio is the new defacto standard in the embedded world and work in progress for skynet3d. You can find the instructions here. And I copy & pasted the relevant part:

[code] cd <your skynet github installation>
# backup our config
cp Marlin/Configuration.h /tmp
cp Marlin/Configuration_adv.h /tmp
# update from Github
git fetch
git checkout SkyNet3D-Devel # or the branch you use (like SkyNet3D-V2.4.x)
git reset origin/`git rev-parse –abbrev-ref HEAD` –hard
# restore the config
cp /tmp/Configuration.h Marlin/
cp /tmp/Configuration_adv.h Marlin/
# build and upload the new skynet
platformio run -d Marlin -t upload [/code]

Afterwards you have to measure your zero position again. I’ve used this pattern on thingiverse:

First Layer Target

Frame Braces

The acrylic frame is optimized for easy shipping and fast development. You want it stiffer to print faster. I’ve printed some upgrades which help a lot.

IMG_20170506_193724

X-axis end frame brace

It’s on thingiverse at

Frame brace for Anet A8

New Power Supply

The 12V 20A Power Supply is one of the cheapest you can get. Mine exploded.

IMG_20170525_175702

12V 20A Power supply. The transistor at top right blew and shorts to the case

Luckily the house installation fuse blew. I’ve tested it at 45V. It just pulls ~3A and smolders. So this could really burn if the failure mode is slightly different.

 

Auto bed leveling

Get yourself an inductive distance sensor. 8mm sensing distance is good. If you take a npn, normally open inductive sensor you can just parallel it to your  Z-endstop so you have a backup. One possible name is “LJ18a3-8-z/bx”. The inductive sensor’s internal pullup pulls it up to 12V with 1mA and the AVR is powerful enough to sink this through it’s protection Diodes.

There are a lot of guides for auto bed leveling. Beware that Marlin always uses the Probe offset from the EEPROM. Your firmware settings will have no effect. Do a M502 M500 to load defaults from firmware and save to EEPROM after flashing.

Conclusion

Well, this thing is a fire hazard and a time sink. After the first few weeks I had enough parts to build a second printer.

The frame’s stability will seriously inhibit the printing speed.

But it’s fun! You learn how the stuff works, you learn the limits, you learn the workarounds and you learn patience on your 96 hour print.