POC printer

I thought I'd share my attempt at designing and building a printer. It seems to be some sort of convention that these things have to be given a name. I settled on POC printer because it can a have a number of meanings depending on your point of view and how successful or unsuccessful it turns out to be. Many of the parts are from Open Builds, including the Vslot extrusion so one interpretation could be Primarily Open builds Construction. Some may think it's a Pile Of C***, (and may be proved correct). On the other hand, if it works out as intended, it may turn out to be a Piece Of Cake. I'll leave it to each individual to decide what it means for them.

What I'm trying to do is build a printer which has a larger print volume than my ageing RRP Mendel - especially height wise. I've started out as CoreXY but I still have reservations about this approach. So, the main objective is to start with a frame which is rigid and square in all directions which could fairly easily be adapted to suit other axis configurations.

Any way, I've already posted details of the X and Y carriages here [forums.reprap.org] and here [forums.reprap.org] and also the complete XY axes here [forums.reprap.org]. These posts contain explanations of why the X and Y are as they are so I'm not going to repeat all that stuff here.

Here is a pic of the frame assembly


This is how it was put together. This was done primarily to save on the cost of buying brackets but it also makes a very rigid frame and IMO looks neater.




I've found a cross line laser to be very useful in assembling the beast (although I've been flamed by people who have never tried it for daring to suggest such a thing). With it being a metre tall, a 100mm engineers square doesn't tell me if frame members are straight and not bowed, only that they are square in that small section where the square is being used.

Here is a picture of using the laser to get the Z guides straight, true and at right angles. It's a bit blurred because I needed lowish light for the laser to show up well and I had mounted the laser on my camera tripod, so the picture was hand held at around 1 second shutter speed. You'll have to look closely to see the red line running down the vertical Z guide on the right, and the horizontal centre rail. Hopefully, you'll get the idea.



Anyway, it worked for me.

These are the idler blocks I made.
This is how it looked in OpenScad


These are the parts


Essentially, the top and bottom parts had to be printed separately. The idler pulleys are Open Builds, each fitted with twin bearings. The belts run at different heights but are offset as per the CoreXY reference mechanism because I couldn't figure out how else to keep them parallel to the axes . It's just on the ends of the Y axis where the idlers are stacked on top of each other which brings the belts closer to the centre line.

This is the assembly.


The bolts run right through so supporting the "axles" top and bottom. These bolts are not fully threaded so the idler bearings are a good fit on the shaft but they are free to slide up and down a few mm.

That's all for now but if the Trolls come out in force, I'll not bother posting anything else here.

Quick update. These are the XY motor mounts.


I got some Sorbothane sheet and cut gaskets out of out to go between the motors and mounts. It's used in the HiFi world as sound deadening. I used some thicker Sorbothane under the feet of my old Mendel which use to drum on the wooden desktop and the difference was quite remarkable. Not sure if it'll work in this particular application but we'll see.

This is assembly. There is a hexagonal hole for a captive nut and the top mounting hole which will line up with the top of the leg is slotted. The plan is that a bolt will be fitted to the captive nut and the mount will slide forwards to tension the belts the locked in place. The side screw holes don't need to be slotted as the side bolts will go into Tee nuts which can slide in the Vslot frame.

The XY axis is now complete. Belts all fitted and tensioned. Movement is still smooth.



This is how I tensioned the belts. Now that the motor mounts are locked in place, I could remove the thumb screw at the front although that too has a lock nut. I guess I'll leave them in place for now, otherwise I'll lose them.

looks interesting ive never seen anything like this before

Here is the bed. It consists of a 400mm x 400mm x 10mm thick machined aluminium tooling plate into which I have "routed" a slot to take a PT100 temperature probe. The top has been painted with Stove paint and baked in the oven to make it less reflective to IR probes. An 800w 240v silicone heater is stuck to the bottom. There are two 6mm layers of "Thermoboard" insulation and the whole thing is mounted on a square frame made from 2020 Vslot extrusion.



Despite being derided for suggesting it, I am using glass as a build surface for several reasons:

Firstly, the aluminium plate cost me £60 with delivery etc, then there is the cost of the custom made Silicone heater which is stuck to it. On top of that, I have invested quite a lot of time in machining the groove and drilling and tapping various holes. Despite what others say, aluminium is a soft metal and is easily scratched and damaged. Due to the high replacement cost in both time and money, I'm not prepared to risk it being damaged by printing directly on to it, even with Kapton tape. Toughened glass is a lot cheaper and quicker to replace although I have never had the need to do so..

Secondly, because the plate is so thick and also well insulated, it will take a long time to cool down. Having a removable glass surface means that I can replace it and start printing again as soon as one print has finished. I do not have a need to batch print Yoda heads, nor do I have an impatient 4 year old child wanting another tug boat as has been implied, but I do have a need to print several objects, one after another without having to wait for several minutes or even hours - the parts for this printer are a case in point.

Thirdly, the glass I'm using is 4mm thick, sand blasted on one side, toughened, and "Brite guard" coated. I'm hoping that I will be able to print directly to it without the need for tape/glue/exotic materials/snake oil or whatever. It may not work, in which case I'll resort to one of the other methods. However, because I have 3 pieces of glass, I could effectively have up to 6 different print surfaces to switch between for different filaments for example.

To hide the edges of the insulation and also to clamp the glass in place, I have cut some pieces of aluminium angle. These are held in place with thumb bolts. The rear section is fixed slightly above the glass surface and acts solely as a back stop.



The side pieces are fixed through slotted holes and are what hold the glass in place.


The front piece is just a bit of flat trim which sits below the glass. To remove the glass is a simple matter of slackening 4 thumb bolts (2 each side) and lifting the side rails slightly so that the glass can slide out from the front.

This is my Z axis which maybe takes a bit of explaining. I didn't fancy a cantilever design - they just don't "feel" right, especially with my bed weighing around 7kg. So, I elected to have 3 screws to take the weight and provide lift. I also don't believe that lead screws should be use for guidance. Their job in this application is to provide lift only and I have separate linear guides to keep the bed in place and prevent any wobble in the XY plane. I also intend to use the lead screws for bed levelling. Initially, this will be manual but a future upgrade will be via software when DC42 gets around to writing the code. At that point, I'll change the design to have separate motors for each screw but for now, I'm using a single motor with one continuous looped belt. With that in mind, I elected to use 1mm pitch screws. I really don't get this 4mm pitch lead screw thing that people are advocating for the Z axis. In my view, a course pitch is solely to give a greater linear motion for a given angular motion, so effectively speed but at the expense of lower resolution. I can't think why I'd want a fast, low resolution Z axis.

Anyway, the philosophy behind this design is that the screws will provide lift, and the linear guides will do the guiding, so it's important that the screws don't impart any sideways constraints on the guides. That is to say that they are fully "floating".

Here is a bearing block to take the bottom of the screw.


You'll see that it has a blind hole. This is for a thrust bearing to take the weight. Here is the same block with the lower part of the bearing and the ball race in place. The top part of the bearing goes in next and the end of the rod will sit on that.


Next comes an 8mm flange bearing. So I have a thrust bearing taking the vertical load and a separate ball bearing taking the sideways load of the belt tension.


That takes care of the bottom of the screws. This is how I set about fixing the screw nuts to the bed. Firstly, having printed all the nut housings, I fitted and "O" ring to each nut. These are all the parts.


The reason for the "O" ring becomes clear in the next picture. I fixed the brass nut to the housing with 3mm bolts and Nyloc nuts such that the bolts are just starting to compress the "O" ring. This will allow a small amount of flexibility for levelling without impacting on the linear guides.


This next picture shows the screws in place and you can see the linear guides at either side towards the front. These are basically small Open Builds "Gantry plates" fitted with Delrin wheels which ride on 2020 V slot. A "rod end" ball joint is fixed to each gantry plate and this is how they connect to the bed rails.


This next picture shows the connection from the underside. A bolt goes through the rod end ball joint, then a couple of washers, then into a "Tee" nut in the extrusion. Again, this allows some flexibility for bed levelling but constrains the bed from any XY movement or wobble (well almost)


I reality, I found that it was possible to induce some sideways wobble by gripping the rear of the bed and moving it side to side. It was caused by the 2020 vertical guides twisting about their single bolt fixings. If I was starting again, I would use 2040 or even 2080 for these guides and fix them with 2 bolts at each cross member. In this instance, a quicker and cheaper fix was to add a third guide at the rear of the bed and it is now rock solid from a "wobble" point of view.

This next picture shows the single ZX motor mount.


It looks like a forest of bolts. Basically the Nema 17 motor is slung underneath (4 bolts) a separate carriage with slotted holes (2 more bolts) so that it can be moved back and forth to set the belt tension. The belt goes around two OpenBuild idlers which are raised on bosses (2 more bolts) which form parts of the mount which is bolted to the frame (6 bolts).

Finally, here is a picture of the entire Z axis. Note that there is no connection at the top of the screws. The screw nuts are fixed to the bed frame. The bottom of the screws sit on thrust bearings and are constrained from any sideways movement caused by the belt tension. Other than that, the screws are fully floating and the linear guides take care of any "wobble".


Edited 1 time(s). Last edit at 07/21/2016 02:53PM by deckingman.