RepRap: Blog

Hydrochloric acid (Spirit of Salts) is easier to get than nitric acid because people don't tend to make explosives from it, and it is widely used in the building industry for removing concrete splat. So I tried using that to prevent the cloudy precipitate when etching a probe and it worked just fine.

I used 40ml of 5% salt solution with 5ml of 0.5% hydrochloric acid to etch a 0.3mm dia. 316 Stainless wire probe as per earlier. The result came out a bit finer than I intended, but I waited for the end of the wire to fall off and it didn't. So, um, less etch time on the next go perhaps?

Anyway, here's the point compared to a 0.5mm hypodermic needle:

I might manage slightly finer detail with that. But as I need the nail gel parts to be at least 30μm thick, this point may be overkill. Still, I've made it so I'll mount it and we'll smash it up once I've got my ducks in a row for the next attempt at layers.

Oh, I definitely need a better USB microscope over the RepRapMicron. It's nigh on impossible to see the resin dots. I might try shaving the end of the housing off this one so I can get it closer to the probe tip. 

I kinda got some layers in a recognizable form. This attempt was a shakedown of the hardware and software pending the fitting of a finer probe. There were two print attempts of a 400μm "lollipop" sliced with PrusaSlicer and the dipify_gcode.py script with Top Coat nail resin. I changed the dipify script so that the reservoir is now at (0,-1000), and as long as I set (0,0) past the edge of the foil reservoir the probe should get dunked.

The first one was intended to be 35μm tall in 10μm layers, the second 50μm tall in 8μm layers. Neither worked to plan. I tilted the slide under the trinocular microscope so you can get some idea of it in 3D:

Sorry about the fuzzy, but the microscope has a limited depth of field.  The purple blob is a marker so I can find things, made from UV resin and a bit of Sharpie (actually a Bic Vivid). It set solid, and I might try integrating marker ink again. I can't honestly tell if the closest object is actually 30μm tall. I strongly suspect that the probe only contacted for the first layer. So I tried again with a thinner layer. The one at the back is definitely thicker and I'd believe 50μm. I could see contact for the layers so it was layering. but it blobbed out. Why? Well, let's look at the probe:

 

It's covered in gelled resin (which mostly washed off). This caused it to dump a lot more resin than intended on the second object. In the future I'll have to get the probe lower and further back to stop the reflected UV from gelling the resin.

Here's what it looks like from the top. Remember, the circles are 0.2mm across in this one:

The To Do list looks like:

• Label and store this attempt.
• Make a new slide.
• Make a new probe.
• Modify dipify script to lower probe before UV exposure.
• Try again. 

I've sorted out a niggle with the RepRapMicron's ground plane. It was tiresome to get the spring-loaded clip onto the aluminium foil that's used as a reference touch plane and UV shield for the reservoir. I've soldered the ground wire to the M8 washer that magnetically clamps the slide down. You clamp the slide in place, and it's grounded!

I've stuck the wire to the other side of the washer with hot melt glue as a nod to strain relief.

As I've increased the length of the slide covered in foil from 25mm to 30mm, the edge of the foil/reservoir can go right up to the UV LED, with the probe at (0,0). This means I do not have to set the reservoir location in code every single time. I just relocate the part on the slicer's print bed. I get more working volume too.

You do have to make sure the probe goes a fair bit over the reservoir to stop the UV LED underneath setting the probe tip into a solid block!

The "dipify" code had a round of debugging, mostly to do with maintaining safe Z height.

I'll try the new slide and the latest dipify out before making a finer probe.

No image to show, this is all software. The new dipify PrusaSlicer filter is now on github https://github.com/VikOlliver/RepRapMicron/blob/main/gcode_segmentation/dipify_gcode.py

It now handles layers and turns the UV LED on between them. There is also a scaling factor to deal with PrusaSlicer's assumptions about how big extrusion widths and layer heights can be.

All the parameters are internal, but I've started adding an argument parser. It's not actually used or called yet. 

I've run it through the V0.05 hardware dry, and it doesn't seem to do anything particularly stupid. Now to get the probe and slide hardware sorted out as per previous post so that I can do some layers. That's going to take a day or three...

PrusaSlicer, bless it's heart, can't cope with the concept of a 6mm print head. As I'm aiming for a 15-30μm line width, that means I can't run the slicer on a scale of one millimetre to one micron.

Back in the RepRap days all this was easily configurable by the user, because nobody was sure exactly what 3D printers looked like yet. Now things are more specialised and gymnastics are needed to cope with exotic extruder sizes.

I'm changing the "dipify" GCODE converter to rescale slicer output. That way I should be able to use probe deposition size values that give sensible layers. I'll start with a scale of 1mm to 10μm and see how it goes.

Other To Do items:

• Detect new layers and modify the Safe Z height as the object gets taller.
• Automatically stow the probe and expose the part to UV in the sliced GCODE.
• Make a slightly finer probe to get dots in the 15-30μm range.
• Design a slide/ground probe with a more conveniently located Z Touch plate. 
• Make a 3D test object (minimug?).
• Establish a practical layer height.
• Maybe speed things up a bit.

Then I can move on to layered objects.

I tried writing a program that watches the probe and detects the flexing that occurs when the probe contacts the slide. Turns out that while the Mk I Human Eyeball can do this, a program has problems with all the video noise. I learned a lot about USB video devices under python though and may be useful later when controlling printed mechanisms:

Oh, I have to deal with writing the Everything Open presentation, and deal with the "Silly Season" that is looming upon us. Jingle Bells and all that. 

Using OXX Cosmetics UV Gel Top Coat (acrylates copolymer, hydroxypropyl methacrylate, hydroxycyclohexyl phenyl ketone) I wiped off the applicator brush on the brim of the container and touched the slide. With Probe One I printed what I'm starting to think of as the "Test Loop." I printed a few other bits and pieces, interlocking 80μm and 30 μm test squares, blobs etc., and cured the whole thing. First for 30 seconds with the built-in LED, then for 200 seconds on a 4W UV LED lamp. No significant shrinkage was observed.

 

The Test Loop consists of a nominal 100μm diameter circle with a 100μm line sticking out of the side. That's pretty much what I got. Note that this was done with two perimeters and 5μm intervals between dots. I could easily get 30 dots between dipping Probe One in the reservoir.

Under the trinocular microscope at 100x I tried to lift some test pieces of print with a hypodermic needle tip. They were not wet, but they did not stay together when poked. They were stuck flat to the glass slide. Fragments attempted to straighten out, so they were not in gel form. I had been printing for over 3 hours at this point, so the Top Coat definitely cures after prolonged exposure.

I went to the (now hardened) reservoir blob and pried the edge of that up. The thin edges tore, so I started prying at it until I found a piece that could be considered strong enough to peel away from the glass slide in a sheet. The material was quite flexible and could be bent over on itself. It returned to its original shape quite positively if slowly when released.

I propped up a torn segment on a 0.7mm diameter pin to get a look at the thickness:

The minimal durable film thickness appears to be approximately 30μm. I tried to determine the thickness of the printed Test Loop. Fragments edge-on were beyond the resolution of my optics. This would put the printed layer at very roughly 10μm or less.

Slicing Problems

While the probe dipping code works, and the PrusaSlicer integration is fine, the slicer doesn't accept layer and line thickness proportions that can be used on a scale of 1mm = 1 micron. I'm going to have to add some scaling to do layers, or use the PNG to GCODE script to stack my own. Possibly both. Also having to specify the reservoir location in the script each time I slice is a pain in the butt. May need to change the reservoir location to something closer to XY=(0,0).

Conclusions

UV Nail Gel Top Coat is a viable material for resin printing tests in air, though repeatable layering is yet to be accomplished.

With UV resin there seems little point in trying to print details much finer than 40μm at this stage as they are not robust enough to survive being detached from the glass slide with the equipment I have.

With 30μm voxels, it would take approximately 1mm^2 to make a complementary flexure that could move +/- 40μm. That would take several hours to print each layer at current speed, though that could be improved as micron precision is not required.

Probe One is a little too coarse for optimal resolution given 30μm voxels. 

Given a maximal thickness of the printed layer of 10μm and a minimal viable mechanical thickness, any printed object needs to be at least 3-4 layers thick to survive manipulation.

Before starting the "Top Coat" nail resin tests, I thought I'd check the state of Probe One (acid/salt etch on Nichrome) to see if it was still in good shape after being used for all experiments since the 17th October. Well, yes. 

I haven't been treating it particularly gently. I have been cleaning it with an isopropyl alcohol spray. But then I've been gently wiping it dry with the corner of a Sorbent white tissue. It's stood up better than anticipated, and the surface still seems to have pores in it.

That XY Table I said "Don't print this model, it's only a mockup"? Well, I had the model and there was a printer next to me. It kinda happened. In my defence I only printed it at 60% scale.

 

If you're also going to not print it, I warn you that my flexures were just eyeballed for the sketch and are too thick causing unwanted resistance/rotation. Also it needs a base of some kind (preferably with diagonal bracing) to stop the bent flexures digging into the floor. Other than that it moves surprisingly well, no sagging and the centre stays level when you move it about.

As I say, needs connections and mounts for axis drivers. But, uh, wow. Useful exercise. I'll keep it around to model in OpenSCAD if nobody else does it hint.

Now, stop playing with it Vik, and get on with the resin experiments. 

I've had this buzzing around my head for a bit. By clever arrangement of the flexures, I think it might be possible to print-in-place the XY Table. So I roughly (really roughly) drew this up in Tinkercad, which is really quite good for 3D sketches:

The green bit is the stage, yellow bits are flexures, red bits structural beams, and the grey bits will be hollow boxes but I've made them see-through so you can see the guts. The outer hollow box is the frame.

If you fancy tinkering yourself or looking at a 3D view, I've stuck it on the Fab Lab's public area: 

https://www.tinkercad.com/things/dTJWUZJloZ0-reprapmicron-xy-flexure-mockup 

Attaching the X & Y Axis Drivers is left as an interesting exercise to the enthusiast. Again, this is just a rough mockup of an idea. It's not meant to be a real printable thing. Do feel free to fix that... 

Got enough stuff up and running. Bodged a 12V laptop power supply into the microscope. New USB for the USB microscopes. I got some of the "OXX Cosmetics Cherry UV Gel, Nail Polish, Jelly Finish." Label states ingredients as "Acrylates Copolymer, Hydroxypropyl Methacrylate, Hydroxycyclohexyl Phenyl Ketone, Silica, CI 15850, CI 19140"

This appears to have some kind of dispersion in it (the silica?), but the particle size is not significant for the resolution of the μRepRap. It is very thick, and wants to stay in large blobs. Attempts to make small dots resulted in 50μm-80μm features as per lower left corner of the structure. The approach of using "dipify" software to make a pre-determined shape was abandoned, and a 500μm square manually created by dragging out the initial contact point, recharging the tip for the next side, dragging that out etc. This was Probe One, which is rather blunt, so perhaps different results may come from finer probes.

500μm Square, "Jelly" nail resin

Side view using lower quality microscope

 

 

Closeup using trinocular microscope

An attempt was made to create some kind of tail to hang on to the finished item. The resin was first cured using the built in μRepRap UV LED for 30s, transferred to a 4W UV LED for 180s, tested, then exposed for 8 minutes.

Adhesion to the glass surface was very good. The bulk of the large blob had enough strength to stay intact in the centre, but cracked when removal was attempted. The underside appeared to be of a similar consistency to the top surface. There was a notable lack of the clear, uncured fluid seen around attempts at curing 3D printer resins.

The strength of the resin was insufficient for the edges to be cleanly lifted from the slide though did have cohesive properties and hints of a stable layer were visible. The relatively thin single layer of the experimental square did not have enough strength to survive attempts to separate it from the slide resulting in the following mess:

Square destroyed by attempts to lift it

 

The "Jelly" nail resin did cure, and was easy to see. It was not immediately possible to create fine structures, and the results - with single layers anyway - were of low strength. Making this work with a finer probe might be challenging, and require multiple passes. I'll call this as being more suitable than 3D printer resin but not ideal. The "Top Coat" is supposed to be thinner and harder, but with poorer adhesion and harder to see. That'll be next, but that's not the only work on right now.
 

I haven't been able to use the workshop since the storm. It looks like one phase overvolted significantly. The printers still seem to work - they were turned off - but I have lost the microscope power supply, bench USB supplies (several), and all the charger units for my DeWalt power tools which are now useless as they're older XRP models. The RepRapMicron is on a 25 year old "survivor bias" PSU built like a tank, and that still works.  A few other odds and sods don't.

Most annoying. Still, the things that were running on the backup inverter have survived. I still have my laptop and screens. Obviously I have a lot of fixing to do. We'll see what we can get running.