This instructable will show you som much needed improvements required to get good results out of a cheap 3D printer of delta type.

I had been curious of 3D printing for a while. I did the usual research and found that what most people recommend for beginners seeking to dip their toes into 3D-printing is carthesian style printers. Thinking that this advice can't possibly apply to me, I obviously decided on a delta. Big mistake! But not on all accounts. I have gained a lot of knowledge in my efforts to get the !#%¤&¤ thing working (at least I like to think so because it makes me feel better). I came to realize that I hadn't purchased a gadget. I purchased a hobby fully capable of sucking up most of my free time.

The printer in question is the Geeetech G2S Pro. This is my first and only 3D printer so I can't really say how it stands up to the competition. I will say though, that my instance of this product was of sub standard quality. The few prints I managed to produce in the beginning were of very poor quality and printing dual-color simply didn't work.

Do you have the same printer and problems or maybe a different delta with similar problems? Don't give up! With some alterations you can actually get really good results! Keep reading.

Step 1: So What Was Wrong With It?

In my experience the most important things with delta printers are:

Low tolerances! There must be as little play as possible between all moving parts. Play is you enemy. Get rid of it! Even small play will have a big impact on the result. To simplify. The GCODE is directing the hotend tip to a XYZ coordinate. If you have a low tolerance machine the hotend will go where it is supposed to.If you don't, the hotend will only go to a location close to where you want it to go. If you home your printer and try to wiggle the hotend - the play you feel is the play that will come out in you prints.
Structural stability. The machine must be tight, rigid and not wobble and shake as the print head swings around. If it does you will get round corners and other quality issues.
Rod length. The diagonal rods must all be of equal length. If they are not, the print head will not move parallell to the bed. This is especially bad if you have a dual head configuration since the two tips will not be on the same height compared to eachother across the bed which makes it impossible to calibrate. Those hours of my life are lost forever.
Rod spacing. The distance between the diagonal rods must be the same in the upper and lower ball joints.If they are not, the print head will not move parallell to the bed.
Hotend. A good hotend is imporant to all kinds of 3D printers.

And what did I get?

High tolerances!
1. The linear bearing were loose inside the casing
2. The original diagonal rod ball joints had noticeable play
3. The linear bearing was not tight against the smooth rod. The was wiggle-room.
The machine would wobble noticeably when printing.
The rods lengths varied a lot. The difference in length was a couple of millimeters between the longest and the shortest.
The upper mounts for the diagonal rods on the pulley have play (the screw holes are too large) which makes it hard to get it to match the distance at the lower mount. This is something that took me some time to realize.
Hotends are like most things. You get what you pay for. If you want reliability, go for a brand name.

Step 2: Pulley Play

These two actions reduces the pulley play to a minimum.

Loose linear bearing (picture 1)

The linear bearing is housed inside what I think is a standard part used in multiple applications. To be honest, I don't think it was designed with 3D printing in mind. The bearing is loose inside the casing. You can feel it if you wiggle the pulley. You shouldn't be able to. I solved this by drilling a hole into the side of each casing, threading (M4) the hole and fixating the bearing with a screw and threadlock.

Pulley wiggle (picture 2, 3)

Even after fixating the bearings there was some wiggle left in the pulley. This is due to tolerances in the linear bearing against the smooth rod. To compensate for that I designed a part to hold a third bearing. The offset of the third bearing stops the pulley from twisting in two directions. Please note that in order to fasten screws to the back side you probably need to thread the hole from the fron all the way through.

Step 3: Diagonal Rods

New rods

I have seen people trying to improve the metal kit rods. It is not worth the effort. The original rods were crooked, of unequal length, heavy and loose fitted. Get new carbon fiber rods with Traxxas rod ends (https://traxxas.com/products/parts/5347). There are full kits on ebay. Make sure you are very careful in getting all the rods exactly the same length. They don't have to be exactly the same length as the original rods. It is recommended to use a jig (like here: http://www.thingiverse.com/thing:1463487).

Joint play

Even with the new joints there can be a slight play. This can be reduced to almost nothing by pulling the rods towards eachother with rubber bands or springs. I attached springs with cable ties.

Parallell rods

Grab you caliper. Measure the distance between the rods at the pulley end (top) and the print head end (bottom). If the distance is not the same you should adjust it. Yous adjust it by adding shims and/or loosening the mount screws on the pulley and re-positon the rod mount.

Step 4: Structural Stability

Increasing stability will increase the overall print quality, especially when going higher in print speeds.

Print the displayed part (see last step), drill, screw and attach with cable ties (see picture).

There are many ways to stabilize the smooth rods. This is just one way. It gives a nice improvement but doesn't eliminate vibrations entirely. The drawback is that you loose a little printable space close to each tower.

Step 5: End Stops

Swith mount (picture 1)

The original endstop switch is mounted on a laser cut plastic part. I found that the the part had wiggle room so I designed a improved version and switched to micro switch.

Adjustment screw (picture 2)

The stock adjustment screw goes through a threaded aluminum block. There is a spring to supposedly keep it from moving. That doesn't work. The threads are quite loose and I found out the hard way that the screw slowly unscrewed itself ever so slightly every time it homed. Unadressed, this results in endless re-calibration.

Simply add a lock nut to the top! The screw will stay put.

Step 6: Weak Extrusion (skipping Steps)

I had problems with extrusion. Unless printing at very low speeds the extruder motor would skip steps causing under extrusion or complete print failures.

This was a result of the stepper motor not getting enough current to produce enough torque to push the filament through the nozzle. The printer kit comes with a Geeetech GT2560 which is a pretty decent mainboard. It is essentially a RAMPS combo on a single board. The board comes with pre-installed A4988 stepper drivers. The stepper driver power output can be tuned. My skipping motors was a result of the drivers being set a very low amperage. To increse power output, turn the tiny potentiometer (see picture) clockwise 1/8th turn and try again. Still need more? Repeat the process. There are more scientific ways to calibrate power out using a reference voltage. For me, the empiric approach worked fine.

Step 7: Notes on Firmware

The recommended firmware for the GT2560 board is Marlin. I used it for a while but then I switched to Repetier firmware. In my experience, Repetier firmware is clearly better suited for this printer and probably other similar printers.

Why Repetier? The main reason is that you don't have to go throught the tedious process of recompile/upload firmware for config changes. If you install the Repetier-Host client on your PC you can edit almost all config settings from within the software (Alt+E). This is a time saver. Other reasons for choosing Repetier is that is seems to be better maintained and also there is an excellend on-line configuration tool (https://www.repetier.com/firmware/v092/).

If you want a starting point for Repetier, download Configuration.h. Upload it to the configurator and tweak it to your liking.

Configuration.h
Download

Step 8: Hotend Upgrade

This may be the last thing you choose to do if you are worried about cost and you probably are if you bought this kit to begin with. However, there is no other way around it. Quality costs. I have tried several cheap hotends from China, most of them inspired by a brand name original. In the end I put up the money and bought a E3D original hotend, namely the Chimera. In retrospect I wonder why I waited so long. The Chimera works very reliably.

I have designed a mounting plate and a fan duct for mounting on the stock spindle plate.

Step 9: Notes on Calibration

This instructable is mainly about hardware improvements to the G2S Pro but I will offer some advice on calibration. Full instructions on how to calibrate a delta is an Instructable of it's own and there are several guides on-line. The process varies slightly depending on firmware. For interacting with the mainboard during calibration I use Repetier-Host, i. e. homeing, move to 0-height, eeprom config.

Here is some general advice on delta calibration:

Be patient. Inital calibration of a delta is iterative, boring and time consuming.
1. Height. Start with adjusting the center z-distance/height using the paper test (at 0-height, the nozzle should just allow sliding a piece of paper under it with some drag). How to adjust height varies with firmware.
2. Endstop calibration. Home, then move the nozzle to a position close to the first tower, slowly go down to 0-height. If the nozzle isn't touching the bed at Z=0 screw the endstop screw up (clockwise), if it touches before 0 (paper test), turn it counter clockwise. Move to the second and third tower and do the same. Home the hotend and start over with the first, second, third tower again. Don't adjust z-height (Step 1). After several iterations you should have paper drag at Z=0 in front of each tower.
3. Bed shape calibration. After completing step 2, check Z=0 in the center of the bed. It is very unlikely that the nozzle is on the same level as before (paper test). It is most likely either too low or too high. You need to adjust delta radius to level the movement. After each adjustment, go back to step 2 and adjust endstop screws. If you have Marlin you have to recompile/upload for each change but for Repetier you can change the setting in EEPROM from Repetier-Host. This speeds up the calibration cycle a lot. If the nozzle is:
  1. Above bed (dome shaped movement): Increase delta radius DELTA_SMOOTH_ROD_OFFSET(Marlin) or PRINTER_RADIUS(Repetier)
  2. Too low: Decrease delta radius
Don't start with auto leveling too soon. You must have a very good manual calibration before
activating auto bed leveling. In fact, I currently only use manual calibration. If you go through the motions and get a good basic calibration you don't really have to recalibrate that often.
When you pass the paper test at each tower and in the center you are done.