Monday, 2 May 2016

Smooth PLA Prints

I've been experimenting with methods to smooth away the print lines.

There are some chemical methods available for ABS but the equivalent for PLA uses toxic materials so I'm not keen on that.

I've tried Hi Build car primer but it took so many coats and so much sanding that I lost more detail than I was happy with. This time I have tried two different solutions. Resin and normal nail varnish.

I tried two resins. Zap 30 and a resin specifically marketed for the job, XTC-3D. The Zap 30 is a bit too thick and cannot be applied so easily so that leaves the XTC-3D.


If you are going to use XTC-3D I strongly recommend watching the following video first. It's a bit long but has some useful tips.

The most important take out from the video is to spread the mixed resin thinly over a dish. I found that the working time was inadequate at less than 5 minutes from the mixing pot but using the thin layer in a dish it is much longer.

Where I applied the resin while it was still nice and runny it worked well. Even when it is thin it is difficult to get the right thickness. Some areas in my test sample had insufficient resin but I can work on my technique to improve on that.

Resin was already starting to cure!
It is important to stop using the resin the moment it starts to go off. The thickening resin ruins the areas it is applied to!

For my purposes the resin needs to set horizontally to self level. The vertical areas pooled and had less satisfactory results.

Carve out the lost detail

Resin took some sanding

The disadvantage of the resin is that it is still too think and of no use where I have fine detail on the model cars. It flowed in to small holes and shut lines where I did not want it! When set it is hard and difficult to sand. I had to use the Dremel tool which makes it impossible to sand some tight areas without damaging some of the detail! Large flat areas are OK to work on.

Resin filled the steps on the bonnet

If I was making an organic shape or there was no detail, the resin would be ideal as it is just one step to a smooth surface. Unfortunately for more man made objects, like cars, it has limited uses.

I will use the resin for the bonnets and any other parts that have open stepped horizontal layers on shallow sloped surfaces but will avoid it elsewhere.

Resin filled the detail on the side of the wing

The loss of detail is why I moved on to an alternative.

Nail Varnish

I tried two different types of nail varnish. A clear top coat and a coloured layer.

The coloured nail varnish worked well. It had the advantage of showing where I had used it and the thicker coating filled the thin layers exactly as I had hoped. Even with the brush that came with the nail varnish, I had enough control to avoid getting it in any of the small details.

Nail varnish on the door

Nail varnish dries very quickly so I was able to do three coats in only a few hours and could sand it before the end of the day. It was easy to sand so I was able to avoid too much loss of detail.

The clear top coat was not as good on the layers but I could use it to fill the fine pattern on perfectly horizontal surfaces where the coloured varnish would be unnecessarily thick.

Nail varnish did not fill the steps

The nail varnish was not thick enough for stepped areas, perhaps 10 or more coats would do it but I did not try that. Back to the resin for those parts of the model.

Acrylic Primer

One last test was to check how they took paint. I was sure the resin would be OK but I had my doubts about the cellulose based nail varnish. I let both dry over night before spraying on the acrylic primer.

3 coats of  nail varnish on the bonnet is inadequate

Nail varnish on the front half of the door works well

Resin on the far half of the bonnet fills the step lines

I needn't have worried about the primer. Both the resin and the nail varnish worked well with no undesirable reactions to the paint.

I had another idea which I didn't try until a bit later. That is to use normal household undercoat. That looks to be a bit thicker than the nail varnish. The undercoat has not dried yet so I'll have to report on that another day!


ColorFabb PLA/PHA vs Faberdashery PLA

I've now used both the Faberdashery PLA filament and the ColorFabb PLA/PHA for a while now. Each has their own properties and I use whichever is the most appropriate for the task.

The following is my experience using the two, I have not done any scientific tests, these are just my opinions based on the many successful and unsuccessful 3D prints I have done over the last year on my Ultimaker 2.

ColorFabb PLA/PHA
+ Stronger layer bond
+ Less prone to warping
- More strings
- Less detail

Faberdashery PLA
+ Less strings
+ Finer detail
- More prone to warping
- Lower strength

I can't show strength in a photograph but I have some examples of the warping and the higher detail of the Faberdashery PLA.

The same print came out perfectly, several times, using ColorFabb PLA/PHA but the one time I used Faberdashery PLA it warped. This is easy to fix by adding a brim but it is just a good example of the difference in properties of the filaments.

White = Faberdashery, Black = ColorFabb
It's much harder to see the difference in quality in photos but if you look at the surface of the flat areas on the corners closest to each other you can see how the black ColorFabb is almost bubbly. By eye it is more noticeable that the Faberdashery is crisper.

Although I can't show strength I can give an example of the layer adhesion.


Again in the photos, the white is Faberdashery and you can clearly see where the print on the A-pillar has several deep notches where the layer has not formed properly. If you look closely on the black ColorFabb version, the layer imperfections are still present on the door pillar but far reduced.

I use both. Fairly obviously, I use the ColorFabb PLA/PHA where strength is more important and Faberdashery PLA where the detail is more important.

The other essential tip that applies to both filaments is how it gets in to the printer.

I think that the spools, as used by ColorFabb and Ultimaker themselves, are too tightly wound. The weight of the full spool is often too heavy for the feeder to pull and when the filament has been pulled out and is close to the core it is too curved and gets caught going through the Bowden tube!

I separate lengths of filament in to coils up to about 12m long and use a simple hook on the top of the printer to hang a coil on. I have found this is much more reliable. I removed the original spool holder so the coils don't catch on that as they are pulled taught before the friction gives way and they rotate round the hook.

Air Brush with Workshop Compressor

This weekend was the first time I have used my air brush with my workshop compressor.

It worked well but needed some initial fiddling to get the pressures right.

I ended up with the compressor set to a pressure of 70psi and the regulator at the water separator set to 25psi.

There are several adapters from the larger workshop connector down to the small air brush hose plus neither gauge is likely to be accurate so the above settings are just what I need to set with my equipment.

Wednesday, 27 April 2016

Lawnmower starting handle

When the pull cord handle on Shelley's lawnmower broke I made this as a replacement:

The pull cord passes up the handle, round and out in the bone under the skull. To be knotted then pulled back and hidden in the cross bone.

Here is the handle in design and production:

Shelley found it very comfortable to use.

Pimp My Lawnmowers

Once I'd started I had to go a bit further. I made a stencil and added some skulls to the little lawnmower and the bigger sit on mower :-)

WARNING: Pull starting engines can be dangerous. There is a risk that the handle can be pulled out of your hand so fast that it breaks bones. Always read the instructions. Do not use this replacement unless you are confident that you know how to use the equipment you are attaching it to. You use this entirely at your own risk.
Skull handle Blend file
Skull handle STL file

Printer settings:
Ultimaker 2
ColorFabb PLA/PHA = 12m
Layer thickness = 0.2mm
In fill = 100% solid (for strength but there are modeled in hollow sections where acceptable)
Speed = 60mm/s
Support = everywhere (there is also an additional support modeled in, that supports the supports!)
Adhesion = Raft

Parts of the model have been derived from others. With kind thanks to the authors of the following:
Skull -
Other Bones -
The stencil was derived from:

My designs on

Test drive the Defender

I have assembled the model I am making of a Defender 90 and run it round the track to check clearances and the fit.

It is still in the raw, off the 3D printer, state and there is a lot of cleaning up and painting to do before it is finished.

It ran surprisingly well, as you can see in the video.

The roof, rear door bonnet and grill are separate parts superglued to the main body.

The chassis is attached with three screws and can be seen separately in an earlier post.

Tiger is not entirely sure about slot cars.

Saturday, 16 April 2016

Leaking screen

The new screen in the Discovery has not gone to plan.

Today we had to have a man come out to fix a leak.
Auto Windscreens customer service has been very good and they were a little too prompt sending someone out the rectify the problem.

I called them late Thursday evening and they sent someone out the day after. Unfortunately, they had not booked that with me and I was at work. They did better when they called back and booked someone in for the next day, Saturday morning.

A very nice man, Eugene, with a great beard came out. After his own investigation, he very carefully filled the gap, I had already found, with their sealant and explained that if that did not do it then they would replace the screen again.

Back a few days. We noticed the first time we drove the car with the new screen, how noisy it was.  It creaked loudly, especially when we went over bumps and their was a lot of wind noise.

I could not track down the source of the noises so resorted to the Internet. According to the information I found, it was likely to be caused by a seal that perishes over time and tends not to be replaced with the screen, however, I now know, in our case that was not it.

The wind noise came from the big gap where the sealant did not touch the glass.  I could put my fingers above the roof trim and feel the smooth uncompressed sealant more than 5mm away from the glass for most of the length across the top of the screen. As far as I could tell the bottom edge was sealed.

I only found this gap after it rained and I had a small puddle of water form in the centre of the dash, under the interior mirror!

Before the leak appeared the screen had stopped creaking, I guess it just needed to settle and cure fully. Now it has been sealed, along the top edge, the wind noise has gone. My temporary solution of putting duct tape across the top on the outside, stopped further water dripping under the mirror but not the noise!

It did not start there and the duct tape did not fix the more worrying problem. Two days before the obvious leak the key fob would not unlock the doors.

The behaviour was odd, as the fob worked to disable the alarm but I had to unlock the door with the key! The LCD odometer display flashed a large crossed out B. I believe this means a fault with the Body Control Unit (BCU.)

After some fiddling about with the Nanocom on the Tuesday evening the central locking started working and the flashing B disappeared. Unfortunately by Thursday afternoon the central locking had stopped again and by Friday the flashing B was back.

This time the electrical problems were worse, the central locking switch inside the car also failed, making getting in or out of any door other than the driver's, very awkward! I told Auto Windscreens this on the phone.

By Friday I had noticed more problems and they continued on Saturday even after the screen had been re-sealed. The electrical problems included that the indicators and hazard warning lights only worked with the headlights on. Sidelights were not enough it had to be headlights. This extended to the other switches in the cluster with the hazard switch in the middle of the car!

I am pretty sure that this information might be enough to narrow down where the water was affecting the electrics. Again, I think the information I found on the Internet is misleading. Most reported the symptoms I had as a BCU failure but I don't think it is the BCU getting wet. The BCU sits slightly inset from the body and is not where I would expect water to congregate. I'm guessing but I think it is more likely, in my case, that water was running down the wiring loom and collecting at the first connector it gets to. If I could narrow it down, I might be able to seal it to avoid future similar problems.

The good news is that now the screen is sealed I only had to drive the car with the heater at 22C for a 30 minute round trip with a stop for an all day breakfast and when I got out back at home the central locking worked. I'll be checking the workings properly tomorrow.

I think this is fairly conclusive that it was water causing the electrical faults. Probably draining down the sides and getting in somehow.

Wednesday, 13 April 2016

Tips for printing slot car bodies

I'm working on my next model and a previous post shows the chassis. This time I am creating the body completely from scratch and building in everything I have leaned from my first model. There are a few tips on that post.

I've called these tips but they are mainly reminders for me.

I'm using an Ultimaker 2 which uses Fused Filament Fabrication (FFF.) I use ColorFabb PLA/PHA or Faberdashery PLA filament. Some of the following tips may not be as applicable for fluid/resin based stereolithography (SLA) or selective laser sintering (SLS) processes that many of the professional bureaus offer.


All objects need to be closed, that is solid, all structures have to have a thickness. Sometimes it works with hollow objects that touch other objects but best to complete each part to a solid block.


I have found that for thin panels using solid (100% fill) works best. Some panels are easier to model and support if they are deeper so I use hollow panels in those cases. This has the advantage of being stronger.

I put hollow tubes inside which reduces the plastic used and I have found often prints better than solid or filled spaces.
By trial and error I found that hollow tubes print cleaner than solid blocks. Not sure why.

I have found the exterior needs to be about 1mm thick which results in a fairly rigid panel.
The interior, roof and bonnet can be as thin as 0.7mm and will flex.

The pillars between windows need to be about  1.5mm square. Anything less needs to be handled with extreme care when cleaning up the finished print.

Shut lines, where doors open need to be at least 0.7mm deep. Over 1mm deep is preferable. The more depth the better otherwise the oozing filament fills any smaller gap. In most cases 0.5mm is wide enough. If they are shallow they need to be wider.

For horizontal shut lines which would have an impossible overhang, I have had good results by inserting a long thin sloped support just at the outer edge of the slot. This will leave a hollow void behind. I also make the support barely touch the edges. The void and the weakness of the joint make it fairly easy to push a knife in to clean out the slot.

Details can be as small as 0.7mm but anything less gets lost in the molten filament.
Any edge less than 0.4mm thick has a good chance that it will not form at all. I assume this is because the nozzle on my printer is 0.4mm. Watch out for upwards or downwards facing points or knife edges. They are likely to get flattened at the height they narrow to 0.4mm

I exaggerate protruding details by at least 0.2mm to allow for finishing.

Supports and Printing

I've tried a couple of the free slicers and they rarely get supports in all the right places. Some bits work and others are just ignored!  The automatically generated supports are thinner than the software allows me to add myself. These are easier to remove than my own supports so use them where possible. I only manually design in supports within the model where the slicer fails.

When it comes to slicers I always end up back with the Cura 15.04 series. The others are different but no one is perfect. One day I might try a paid for one.

I have also tried to include my own structures to make it stay on the platform. That way I can position them where they are easier to clean up. Oddly any thin sheets less than 0.3mm thick do not print at all for me, even though they are clearly shown in Cura! Looks like the first two layers are not printed but it might be something to do with allowing for the thickness of the first layer, I don't know.

I've also tried to make the thin rafts easier to break away from the model. They need to be at least 0.2mm away horizontally otherwise it oozes together and forms a solid join.

For upright supports I use 0.4mm walled hollow sections, tubes or sometimes cones, all with chamfered or flattened pointed tops. I have not been able to replicate the supports generated by Cura because, as mentioned above, Cure tends to ignore any wall that is less than 0.3mm!

The printer gives better results with long runs so where possible I like the supports to be long and thin.

Break away or cut away supports need to be at least 0.3mm away vertically from other layers otherwise the layers fuse together. To make them easier to break away I also use a 0.2mm air gap.

To support a raised hanging edge I form very low castellations where the tiny taller bit is 0.2mm away from the lowest row forming a bridge above. Keeping the height of castellations to a minimum is more to do with the tool path up to that height rather than the support required. As stated a straight run prints better.

The filament will only bridge a 3mm gap and only just. At 4mm wide there will be a slight downwards bow and a rough edge. OK for hidden structures but not for surfaces which I need to finish.

A hanging step never works unsupported. It always ends in an oozy mess.
The top edges of windows are the hardest to get right. The best results I've had with windows is to solid fill them and cut out the support after. That also avoids the stringy bits that I find will often form at the edges of pillars.

I keep the connection between the infill for the windows and the body to just touching on a knife edge. At the top edge I extend it to form a tunnel to bridge the step. If I am lucky the infill pushes out by hand because it is so lightly touching.

I like to keep the support structures as separate objects within the source file. This includes the infills to bridge steps and overhangs. This makes it easier to create and adjust the model shape without having to deal with keeping the supports in line with those changes. They can be adjusted separately at the end.

I colour sections to make it easy to differentiate between model and the support structures. It can sometimes take several (lots of) attempts to get the right shape support for any given area.

I have found that adding superfluous supports to change the tool path can avoid those strings of molten plastic to get a better finish in some parts of the models.

Where possible I print large curved panels, like the roof, vertically. I've found this a common recommendation on slot car forums.  The thin layers in sheets have a step pattern that is hard to get rid of when finishing but the layers, end on, have an incremental transition that are easier to clean up. I usually split the model at the roof line to minimise the support structure but this will depend on the design.

Layer thickness or thinness does not necessarily make for a better result. I mainly use a setting of 0.1mm.

Print speed = 30mm/s.
I've tried faster but even 40mm/s has noticeably more problems. Typically poorly fused layers or more stringy bits.