3D Printing

 

I've been following the progress of the 'hobby' 3D printers for some time now but have always thought that the results from them had not yet reached a suitable standard to be of much use for anything although than toys and gimmicks. I was interested in producing printed plastic patterns for locomotive castings and the surface finish straight off the printers just didn't seem good enough without doing a lot of cleaning up and finishing which to me defeated the object somewhat.

I changed my mind when a friend brought some sample patterns that he had printed on an Ultimaker 2 machine to one of our 2½" meetings. The finish was excellent and more than good enough for casting. I did ask if the finish was straight from the printer and I think he said yes, but having had some wheel patterns printed for the 2½" gauge Association, I now suspect that some sanding had been done on them to improve the surface finish.

Anyway, seeing these patterns set me going and I spent weeks umming and ahing whether to get a printer myself. I couldn't really afford one but couldn't resist either!

I spent a lot of time looking at all the models available but wasn't impressed by most of the specifications and I didn't really fancy one built of wood or plastic as I couldn't see them being stable enough for reliable results. The majority of the current models have a minimum layer height of 100 microns (0.01mm) but the better ones such as the Ultimakers go down to 20 microns (0.02mm) although I think it is very difficult to achieve this in practice. The general opinion is that 100 micron layers are perfectly adequate but on curved surfaces the 'steps' produced by the layer height are very visible, even with 100 micron layers. I wanted something better. Eventually I came across the latest model by Wanhao, the D5S Mini, and that seemed to fit the bill. It had a good specification (layer height down to 20 microns same as the Ultimaker), was built with a steel frame, had a decent size build volume, and the price seemed very reasonable for what you got. I bit the bullet and bought one.

I will say before going any further that I wasn't expecting the printer to be perfect straight out of the box and I was prepared to do work on it to get it up to a standard that I wanted. These hobby machines are all built to a price and some of the components used are not of a particularly high specification because of this. The bearings and guide rods can leave a lot to be desired (as I found out soon after buying it) but can be replaced quite cheaply with better quality ones. Some of the toothed belt pulleys used on some printers seem very substandard and don't run truly on the shafts due to oversized bores etc. All these defects cause the quality of the prints to be reduced.

This particular printer is only really capable of printing in PLA type plastic as it does not have a heated bed but to me that isn't a problem as PLA seems ideal for patterns. It would be possible to add a heated bed later if I decide to try other plastics such as ABS but this would have to be a DIY job as there are no plans at the moment to make an upgrade kit.

The printer arrived the day after ordering (from the UK, I wouldn't get one direct from China) and seemed very well packaged. I was dreading it getting damaged in transit but it arrived ok.

There was virtually no assembly to do, just the extruder filament guide tube to fit and the spindle for the filament spool. I probably had it printing within an hour of unpacking it!

First print

The printer has a plastic build plate held onto the metal Z platform by magnets and this has to be covered with masking tape to give a surface that the first layer of plastic can adhere to. I find this a pain to do! It's very difficult to get the edges of the tape (in this case 50mm wide) to touch without getting gaps or overlaps which can affect the print. I'm also finding that the prints can stick too well to the tape making them very difficult to remove without destroying the tape every time. One tip I came across on Youtube is to spray the print with freezer spray and this does help it come off the tape easier. People have experimented with using a glass plate covered with types of hairspray and this seems a much easier approach, one I will try in the future.

The first prints I did were simple test pieces consisting of hemispheres to see what the surface finsh was going to be like and these looked promising. A few blobs but generally pretty good straight out of the box.

Print at 100 microns layer height

I played around for days printing test cubes and various other objects, trying different settings for the layer heights, printing speeds and temperatures etc. with varying results. This is a whole new ball game for me and a very steep learning curve. It wasn't helped by the fact that documentation for the printer and the software used to slice the 3D CAD stl files (Wanhao Maker) is virtually non existent. There is a forum for the Wanhao printers which is helpful but there is not a lot of information there. The advice I got was to stick to the stock profiles in Wanhao Maker (the settings for adjusting the printing) as they produced the best results. Eventually though, I sussed out what all the different settings did and managed to produce my own printing profiles that I think give better results. I found the instruction manual for the Ultimaker software (Cura) to be very helpful as most of the settings are the same and the Ultimaker forum is worth it's weight in gold for someone like me who doesn't really know what they are doing! (The printers are very similar)

After a bit I got around to printing some of the wheel patterns that I had drawn up and had printed by the friend with the Ultimaker. At 100 micron layer height these take 3 hours or so to print and going down to lower layer heights means a correspondingly longer print time. Fortunately, the printer seems very reliable so far so I just set it going and do something else.

The results weren't too bad with a very good surface finish on the rims but nothing like as good on the spokes. I had noticed this on the prints that my friend had done as well and did wonder if it was something wrong with the drawing for the pattern when it was converted to the stl file.

Printed Wheel Pattern

It was as if the layers were not lining up properly on just the spokes. I asked the question on the forum and was told that the problem was due to backlash in the x and y axes drives. I tightened up one of the belts that was a bit loose which did improve things slightly but the problem was still there. I drew up a simple spoked wheel that would be quicker to print and tried that.

Simple Spoked Wheel

You can see that the rim is perfect but the spokes are quite rough.

I decided to check out all the rods and bearings and it was then that some problems showed themselves. The block that the heated nozzle mounts on runs on linear ball bearings on two x and y axis rods and I found that there was play between the bearings and the rod that allowed the nozzle to tilt from side to side. One bearing was a lot worse than the other. The problem is that the tip of the nozzle is quite far away from the bearings so any play in them is greatly exagerated.

One of the 'loose' linear bearings

The ends of the two rods that run through the nozzle block are mounted in other blocks that have sintered bronze bearings running on rods along the top edges of the printer. One of those was also a sloppy fit.

Plain bush

The bearings in the blocks running on the side rods have to be plain bushes as these rods rotate to drive the head and linear ball bearings are not suitable for both sliding and rotating applications.

It was obvious to me that any play in these bearings will add up and allow the extruder nozzle to 'flop' about all over the place especially when the extruder is rapidly changing direction.I think that is why the finish on the spokes is poor compared to the rim. When the rim is printed the nozzle is moving in quite a gentle circular motion so the play is not apparent. When printing the spokes the nozzle is moving rapidly from side to side and that's probably when it flops about and over runs it's correct position.

When moving the head about I also noticed one of the support blocks was moving up and down slightly and it turns out that is because the rod that it runs on is bent.

I got a micrometer out and checked the actual diameter of all the rods. One of them that the extruder head runs on (the one with the most play) was 0.02mm undersize and the others were 0.015mm undersize, except for one which was 0.04mm undersize. No wonder the bearings had play in them.

When I mentioned this on the forum, one of the guys with the taller version of my printer said he had had the same problem and had replaced the rods that the head runs on with better quality ones. This had cured the problem on his printer. Looks like it's a common fault on these machines. He had mentioned it to the manufacturers (Wanhao) who said they were limited by what they got from their suppliers. Maybe they ought to change suppliers!

I've decided to sort this myself rather than try to get it done under warranty. It's not an expensive job and if I got replacements from where I bought the printer there's a good chance the new parts will be just as bad and I'll be back to square one. So I've ordered some new lengths of rod from elsewhere and some new linear bearings as well. When I tried the original bearings on some silver steel I have, one was a good fit but the other was still sloppy so I'm going to change both of them.

In preparation I've stripped down the top end of the machine and removed all the rods. Out of interest I rolled the rods on a surface plate and nearly all of them are bent to a certain degree. Whether they were like then when the machine was assembled or they got bent in transit I don't know but I'm going to replace all of them. At least I know then that everything will be right. I'm just waiting for the new parts to arrive now.

07/12/2014

The 2 lengths of 8mm linear shafting arrived, and when I measured them, these are 0.02mm undersize. Just as bad as the original ones! These weren't particularly cheap but I don't know the country of origin so they may be from a similar supplier to the originals. Out of interest I did try and find out the normal specifications for linear shafting to see what the tolerance is on the diameters but many suppliers don't seem to state what that is. One supplier did though and quoted -0.004mm to -0.014mm on 8mm shafts so maybe the originals are not as far out of spec as I thought?

I removed the linear bearings from the extruder mounting block quite easily even though they appeared to be glued in with something as well as being held by two grub screws. Fortunately, heating the block in boiling water softened the 'glue' and the bearings pushed out quite easily. They are not exactly a tight fit which is maybe why they were glued as well. I ordered four new bearings to be on the safe side and when I tried them on some silver steel, all four gave a different fit! One was quite loose, one tight, and the others somewhere inbetween. These were only cheap ones off Ebay so are probably of chinese or similar origin. It seems that to get better quality ones means paying a lot more money so for the time being I'll use those giving the best fit and see what improvements this makes. I will probably look towards getting some top quality shafts and bearings in the New Year when I've got a bit more cash available.

I also ordered some metre lengths of 8mm silver steel so I may use that for at least the side rods. The surface finish is not quite so smooth as the proper linear shafting which is polished but that probably won't matter for the plain bronze bushings. I may use the new linear shafting for the linear bearings if I can get a good fit there as I think the surface finish is more critical for the linear bearings to run on.

10/12/2014

I rebuilt the carriage assembly yesterday and used the silver steel for all but one of the rods. I had to redo part of the assembly when I realised I'd got one of the pulleys the wrong way round and a collar in the wrong place. I should have taken more photos when I took it to bits! By mixing and matching the new linear bearings to the rods I got good fits on them and there is no play at all in the extruder head. The sign of a good fit of the linear bearings is that they should slide smoothly on the rods but be difficult to rotate on the rods. If the bearings can rotate easily on the rods then the fit is too loose. During assembly I also checked that all the rods were aligned correctly and the bearings slid smoothly.

I'm not sure how long the silver steel rods will last before wear sets in as in the unhardened state silver steel is nowhere near as hard as the proper linear rods but it will at least prove that the rods and bearings were the problem causing the poor finish on the spokes of the wheels.

During the rebuild I also replaced the plastic tube that the filament runs in between the feeder (extruder) and the nozzle. I replaced it with PTFE tube but I'm not sure what the original tube is made from. It feels like PTFE but there is a lot less friction with the new tube when you push the filament through it by hand. The problem with PLA is that it is very stiff and rough and comes off the spool with a permanent curve so there is a lot of friction as it is pushed through the tube by the extruder. This gets worse as the filament gets towards the end of the spool where it is more tightly coiled.

This Bowden type of extruder has both advantages and disadvantages. The advantage is that the moving nozzle assembly is much lighter as the extruder mechanism is remotely mounted instead of on the head itself but there is friction between the filament and the Bowden tube and backlash in the filament feed which causes problems with both the feed and retraction (used to pull the filament back to try and stop/reducing stringing when the nozzle moves over empty spaces).

Once the printer was working again I did a test print of the spoked wheel and .................. it was just the same as before! There is perhaps a slight improvement but it looks like the problem may lie somewhere else after all. The rebuild hasn't been a waste of time though as the nozzle assembly is now rock solid instead of rocking about which I'm sure will improve the prints.

Back to the drawing board!

20/12/2014

Not played with the printer much over the last week or so so haven't got to the bottom of the problem of the unequal surface finishes yet. I had a play today and noticed that the nozzle temperature was dropping by a large amount (20 to 30 degrees) when the print cooling fan came on after the first couple of layers. This cooling is necessary with PLA as it takes a long time to solidy and without the cooling the print will sag as it builds up. I began to wonder if this large variation in nozzle temperature was causing some of the problem with the surface finish. As the nozzle temperature falls it becomes more difficult to extrude the plastic and you get under extrusion which affects the quality of the print.

I noticed that the heating cartridge for the nozzle block wasn't a particularly good fit in the hole in the block and as it's only clamped in place with a grubscrew I wondered if it wasn't making very good thermal contact with the block. I'd read on the Ultimaker forum that people had had similar problems with their Ultimakers and had solved it by using heat sink compound on the heater cartridge to improve the heat transfer to the block. I'd got some of this so refitted the heater cartridge with a liberal dose of this stuff (very messy!) and did another print. There was no significant improvement so I asked the question on the Wanhao forum. The answer came back to insulate the heater block. Why didn't I think of that! It was also suggested to move the thermocouple that senses the temperature to the other side of the block to the fan so I did both. This cured the problem completely and now the nozzle temperature stays bang on the setting even when the fan is on.

I also noticed today that the silver steel rod that I used to replace one of those that the nozzle block runs on had developed significant grooves in it where the balls of the linear bearing run. It was quite rough in places so it looks like unhardened silver steel is definitely not suitable for linear bearings. I put the old hardened rod back with a new bearing and we'll see how that goes. The other silver steel rods that the bronze bushes run on are fine so far.

26/12/2014

I spent the last couple of days printing out some test objects produced by Make magazine for doing comparison tests between different printers. One of them was basically an open square box and out of interest I measured the corners for squareness with a try square. Only one of the corners was an accurate 90°, the other three were either more or less, so it looks as though the x and y axes of the printing head are maybe not square to each other. I also noticed that the finish on two sides was not as good as the other two.

I printed out a simple thin 3 inch sided square and, again, the corners were not square.

Test square

I checked the alignment of the two rods that the head travels on with the rods at the side and they were indeed not parallel. A bit of adjustment using a digital caliper to check that the distances at each end of the rods were the same and another printed square showed the corners to be all pretty much spot on. Obviously, I hadn't got the rods set up quite right when I rebuilt the gantry assembly with the new rods etc. I'm not sure what effect the misalignment would have had on the surface finish though, unless it was causing extra friction when the head moves?

I then printed out another spoked wheel and the finish on the spokes was definitely better.

Maybe the misalignment of the axis was at least partially to blame for the poor finish on the spokes?

Notice that the parts have changed colour! I bought another roll of filament recently as the yellow was getting a bit low. The new orange stuff seems a lot more accurate in diameter than the yellow that came with the printer and is also more flexible. It doesn't snap so easily either. It's probably better quality.

30/12/14

I printed out another of the proper wheel patterns yesterday and was very disappointed to see that the problem with the spokes was still there. This is all getting very frustrating!

The only difference between the spokes on this pattern and those on the test wheels is the width. The spokes on the proper wheel pattern are narrower so I'm now figuring that this may be the problem. I don't know why yet but I'm hell bent on finding out! I tried printing the wheel pattern with 100% fill and then no fill at all to see if that may affect things but there was no difference.

At the moment I'm printing a pattern at twice normal size. This will make the spokes twice the width so we'll see how they turn out this time. If my theory about the width of the spokes making a difference is correct then these should turn out ok. The print won't finish until the early hours of the morning as it's over 12 hours long at 0.1mm layer height!

12/01/2015

Well. another year gone by! Still no solution to the spoke problem. I have determined that it is down to the way the spokes are printing. I did a test the other day where I printed a test wheel and then printed the same wheel minus the outer rim. Without the rim the spokes printed with a perfect finish.

 

It's perhaps not too clear from the photos but the finish on the spokes without the rim is as good as the finish on the hub and rim on the first photo.

The difference is that without the rim the spokes are all printed with one steady continuous outline and the nozzle does not have to make sudden jumps between them. With the rim the spokes are printed by printing the islands inbetween them separately and the nozzle has to jump from one to the other. In the screenshots below, the red and green lines are the outer shells and the blue lines the jumps between each spoke. You can see how the outlines are different in each case.

 

 

It's been suggested that the rapid movement of the nozzle when doing these jumps may be causing the printer to shake and possibly causing the uneven finish on the spokes. I had already slowed down the nozzle speed when it makes these jumps in case that was a factor but I'll try slowing it down even more. It's also been suggested to try printing the outline of the object using more than one wall. I already use two walls for the outline anyway making it 0.8mm thick but I'll try increasing that to three and making it 1.2mm thick to see if it helps. Anything is worth a try at the moment!

20/01/2015

I'm still struggling to find the problem with the spokes! Slowing down the travel speed further made no difference at all. I also tried printing with thicker walls as suggested and that did help but it didn't cure the problem. I've also tried different slicing software (IdeaMaker and Cura) but no difference. I was beginning to suspect a problem with the stock slicer (Wanhao Maker) which was why I tried different ones but it looks as though the problem is not with the slicer.

I did notice when I tried moving the carriage by hand or by using the 'jog' function in the Y direction that it tended to spring back as it approached the end of it's travel. It sort of did two steps forward and one step back! I thought it might be a problem with the belts at first but it turned out to be the 'spring' in the plastic flexible conduit that the cables to the head run in. The plastic conduit is quite stiff and was pushing the carriage back as it got more tightly bent. I wasn't sure that this was an actual problem but took all the clips off the conduit that held it to the back of the printer so it bends a lot more easily. This has removed the tendency for the carriage to be pushed back but it hasn't improved the printing by any noticeable amount. Actually, this plastic conduit is too big and far too stiff for the job it's supposed to do so I may replace it with something smaller and more flexible at some point.

Out of interest I downloaded a design for a spoked wheel from Thingiverse and tried printing that. This had 18 rectangular spokes, as opposed to the tapered ones I've been printing, so I was interested to see how they turned out. The wheel was about 11 inches in diameter so I had to scale it down to 4 inches. The spokes on this printed perfectly! I then drew up the same wheel in Cubify and printed that using the same print settings. It was definitely not as good and had several shifted layers causing lines on the spokes. I then downloaded another wheel from Thingiverse and that printed perfectly as well.

I wondered if there was a problem with the Gcode for 'my' wheels causing the layers to not print on top of each other accurately so I downloaded a bit of software that gives a visual representation of the position of the print head compared to the position given by the Gcode. The visual display is not good enough to see if the layers align correctly but you can step the Gcode through each line and using the visual display step through the code and stop it at a particular point e.g. at the end of a spoke. You can then check the Gcode for this point on all the print layers. If the spokes are vertical sided then the X and Y co-ordinates given in the Gcode for this point should be exactly the same for each layer. I couldn't find any problem with the wheels that I had drawn up so I'm back to square one.

It doesn't look like a problem with the printer as the downloaded wheels print perfectly (although I'm not ruling that out yet). It doesn't look like a problem with the slicer as, again, the downloaded wheels print perfectly. The only thing left is the 3D design software, Cubify Design. Could it be a problem with that? Is there a problem with the way it exports the models as an stl file? It seems strange that the downloaded wheels are fine but my own designs aren't!

24/01/2015

I saw a friend at the club on Thursday and he offered to try his software to convert the model to an stl. He's got the top of the range version of Cubify Design and also Solidworks. I emailed him the file for the 3D model and he converted it to an stl and emailed it back. Unfortunately, both the Cubify and the Solidworks stls were no different to mine and the problem was still there.

I checked the printer over again today just to see if I could find anything obviously wrong but the carriage moves very freely with no backlash. The next thing is to try a different CAD program if I can find a good free one. This problem is causing a lot of wasted time and filament!

I did another print this afternoon and noticed that the pulleys for the short belt on the X axis were about 1/8" out of line causing the belt to twist as the pulleys rotated. I think the belt was trying to ride up on the flanges of the pulley. I took the stepper motor off and shifted its pulley to correct the misalignment. I'm surprised that I hadn't noticed this before but I have my doubts whether it will make any difference.

02/02/2015

Realigning the X axis pulleys made no difference so still no solution to the spoke finish problem. I downloaded the trial version of Cubify Invent and made a wheel using that but it made no difference. I wasn't surprised really as Cubify Invent seems to be a cut down version of Cubify Design with just the part modelling section without the facility to make assemblies. At £32 it seems quite a bargain if you are only interested in modelling individual parts.

The more I think about it, the more I think it can't be a problem with the 3D software as the prints of the wheels without the rims are perfect and the spokes are identical. It definitely seems to be down to the different way that the spokes are printed with and without the rim.

I've downloaded another couple of free slicers (Slic3r and Kisslicer) but haven't so far managed to get them to work with the Wanhao printer. The Gcode they produce probably isn't compatable. They are also painfully slow compared to Cura, WanhaoMaker, and IdeaMaker!

Return to ME home page