RepRap: Blog

Shaping is more complicated when you try to automate it blind. The probe tip gets shorter as you etch it, and its electrical characteristics change. Eventually though, I managed something replicable:

Etched 0.3mm dia. 316 stainless steel wire points

A spot of corrosion on the lower one. Must've forgotten to wash it after I made it late last night. That's not coming off. But the profile? Pretty darn close, even if not exactly what I was aiming for. Repeatable probes means repeatable experiments, scientific method an' all. Nice to see the machine making its own parts already too.

The breakthrough was when I noticed that using a previous (badly) etched probe gave a better point second time around. By sheer bloody-mindedness and tinkering I have devised the following scheme, all using the new acidified NaCl etch:

Roughing

This roughens the body of the probe and takes the worst of the cut marks off the end. 30 sec etch at 15mm is what I'm using just because it's a known starting point. Roughly 3.8V at 160mA.

Shaping Pass

This starts off with  0.5mm dips, then 0.15mm, then another 0.15 mm but pushing deeper 5μm at a time. As the probe shortens a lot of these miss the surface, but shaping happens (I suspect there's some self-correction going on here). Current wobbles around a lot ~50mA to start ~5mA at the end. The probe end is roughly pointed but may be rounded at this stage, it varies. I have GCODE to do this to ensure repeatability of motion.

Finishing Pass

The probe is zeroed to the surface and the Shaping Pass repeated. Probe characteristics are less variable now, so a more even shape emerges.

Notes:

The wire must be cut as cleanly and squarely as possible. If the cut end is kinked, this indicates your cutting tool is blunt and not cutting cleanly enough. Good, sharp side-cutters please. 

Roughening is just done with an accuracy of "minute of Sharpie". Mark the wire at 15mm and clamp it with that mark on the meniscus. 

By "zeroing" I mean advancing the probe towards the meniscus until it contacts. If you overshoot, you need to back off about 300μm so that the meniscus drops off, and slowly sneak back on your assumed zero. For the Shaping pass I figured 5 microns was close enough, and got to 1 micron for the Finishing Pass. Don't know how critical those measurements are.

I found it important to wrap the anode lead croc clip in aluminium foil to give its grip some compliance, or the wire drops into the pot.

The anode lead also needs to be secured to the Z Axis Driver to reduce unwanted movement of the probe by the lead - I'm just hanging it over a screw on the Axis Driver Frame Trio.

They look much nicer if you wash them in very hot water immediately after etching.

There are probably optimum shaping and finishing passes but for now I'm keeping it simple. Don't know if more passes help as I've not tried it.

There is now an attachment, provisionally called the Probe Dipper, in https://github.com/VikOlliver/RepRapMicron/blob/main/pika/pika_probe.scad that attaches to a Z Axis Driver (the Driver is now very easy to remove and re-attach). The Probe Dipper has a ridged bar to which a length of wire can be held with a croc clip thus:

The electrolyte and so forth have changed slightly. It's sufficiently acidic that you don't want to leave the spoon in overnight:

56ml H20, 2.5g NaCl, 1.5ml fuming HCl. Wire is 0.3mm dia., cut to 80mm for convenience.

Initial depth of etch 15mm, current ~50mA for 30 sec.

Probe then zeroed to point of contact with meniscus to within 10μm.

Etching time purely dependent on Z axis speed of  15mm/min. Probe dipped and immediately retracted to provide agitation of electrolyte:

0.5mm 3x - Reduce roughened surface

100μm x4 - Create annular grove and fine pitting

Contact x4 - Shape and smooth tip

 

The result of the first attempt has the desired features of a roughened shaft, an annular indent, and a vaguely ogival tip:

Just how essential these features are remains to be seen, but I can't explore them until I have a repeatable method of making probes. The above process was used because I estimated that's what I was doing manually to make a probe that worked (minus the nitric acid). How many of these steps are accurate or indeed actually necessary is unknown.

And so, the next task is to repeat the probe electrolysis on another day. If it is repeatable, I can think about optimizing the shape by changing steps/depth/timing, know that I'll be able to replicate it on demand, and produce a method for making probes as good as or better than the ones I make by hand.

I took the probe tip from the previous test and manually dipped it in the usual electrolyte tub rather than in the electrolytic cell on the slide. The result was a cleanly-etched tip, if a little lopsided and irregular. I guess we can start calling it Probe 12:

The bend at the extreme tip is because I was having a heck of a job manipulating the thing into the pot at right angles to the electrolyte surface. I also couldn't see too well in there, so I over-etched the tip slightly. It shows though that there was nothing wrong with the etched wire. They hypochlorite etch (slightly lightened and at 15mm) is not spoiling it, and the slide-mounted cell does not etch a good tip.

The actual tip is taking on that ogival shape though, its tip is shiny and smooth, and the walls are roughened. The current hypothesis is that I can dismount the Z Axis (only 4 easy-access screws now) and place it over a nearly full electrolyte pot. That'll give good access, I'll be able to see what's going on, and use the Z Axis for repeatable dips.

The overall idea is to make a standardised process that people can follow to come up with a good tip. I can make one by hand, but it's a skill, and that's not how science works. 

I've etched another bit if 316 Stainless wire (note to self, needs to be a 70mm length) with an immersion depth of 15mm. I meant to etch in NaCl, but used the hypochlorite pot by mistake. The tip looks like this immediately after etching:

That was only etched for 45s. That's gone in a Probe Arm, with the tip bent to enter the etching well on the Stage at more or less 90 degrees. Etching in the well with NaCl produced a result that was not at all what I had expected. The tip was irregularly corroded:

I don't know if this is a side-effect of etching with hypochlorite, the annular nature of the electrode in the etching well or what. Two -pronged attack: I'll try etching that same wire manually in the big cell, and I'll put a piece of plain, unadulterated wire in a probe holder and etch that.

Oh,  but first I have to make repairs on the Z axis. I'll replace the old flexure coupling with the newer strengthened design. I tried to close the Z Axis on the microscope clamp, it made cracking noises, and now wobbles (which it didn't do in the video in the previous post). Oops. [UPDATE: That was most likely one of the Z Axis Complementary Flexure screws working loose. Fixed.]

This is just a thought. If you've used a Technical Pen you'll get the idea. If I can fit a probe wire inside a blunted hypodermic needle, I might be able to get a capillary gap between the roughened wall of the probe and the needle. This would allow a reservoir of resin in or above the needle to continuously resupply the probe tip with resin, eliminating the need to continuously dip it - but it would still need to seek safety from UV during curing:

In theory surface tension stops the resin from leaking out. Obviously there are issues with potential blockage, though there isn't a single channel to be blocked as the resin flow wraps around the probe. The probe shouldn't be any more unstable than it is at the moment. Dimensions in the drawing are a bit notional and will very probably vary in practice. I might even put a bend in the probe so I can bring it in at an angle, but that's for future experiment.

For now though I'll stick with a plain probe and muck with the Technical Pen idea once I have a repeatable probe tip scheme worked out. I've printed a bunch of spare probe arms, and the glue on the next tip assembly is currently curing.

I put the Probe 11 into a probe holder and etched it in the μRepRap Stage cell. Immediately apparent was that the hypochlorite etch at corroded absolutely huge fissures in the thing:

Way too much. Also, genius here should have bent the probe tip so that it contacted the meniscus at 90 degrees but got carried away. So having a wreck, I experimented on it anyway as you do.

If you look at the very top of the probe in the image you will see a tiny peak. This is where the probe actually contacts the meniscus. I lifted the probe so that the meniscus was pulled up from the surface, and let it etch until the meniscus dropped off. That little peak is the result.

I'm going to etch another probe in NaCl, but immersing to 15mm. This should give a more even cross-section. I'll then try a few etchings on it to attempt to smooth off about 0.5mm, then try to put a point on the very tip of that. Remembering to bend the probe for 90 degree contact this time.

I will also be buying a better microscope for the side view on the μRepRap. Using the good one to take probe photos has made me realise just how held back I have been by the cheap microscope. 

While experimenting with probes, I was thinking "I need some way of precisely dipping the probe in the electrolyte in a very precise and controlled way."

Well, duh, that's what RepRapMicron does, innit?

So I built an electrolysis cell by wiring up an M8 zinc-plated washer to a glass slide with UV Nail Gel thus:

I switched the good microscope into the horizontal position (must buy a better secondary microscope). Having made sure the positive end of the Z Touch was connected to the probe tip (a disposable grotty hypodermic one), I put salt water in the washer, set the Z Touch Retract to 500μm, and told it to probe:

 

Yes, a video on the blog! (if you can't see it, try this: https://youtu.be/GB33HMUqppU) If you look very carefully as the probe touches the electrolyte, you will see a stream of turbulent fluid heading away from the probe tip. This is electrolysis happening. I measured the probe current at ~10mA, at 5V. Interesting how it takes some time after contact with the electrolyte for the Z Touch to detect contact. Current sensing presumably.

So yes, I believe we may have a way to very controllably produce probe tips. 

Probe 9 was given a pitted surface by etching in dilute nitric acid. This allowed more resin to cling to the body of the probe, but nitric acid is restricted in many localities due to its use in illicit explosives manufacture. It gives a particularly pitted surface because it is a powerful oxidizing agent. So, I pondered on what commonly available oxidizers are in use that haven't been restricted yet. Ordinary household bleach sounded like a good starting point, it also being a starting material for a number of energetic compounds. The stuff under my kitchen sink is 42g/litre which works out at about 4%. I'd have preferred something stronger but hey. Etching as before with a 10mm immersed 316 stainless wire gives this at the tip of Probe 11:

Comparing with the wire etched in acidified 5% sodium chloride to make Probe 10, we can see that there is a promising amount of pitting (sorry, couldn't remember the magnification I used and this is about twice the magnification of the previous image):

Worth exploring further. Caution: Do not add hydrochloric acid to the sodium hypochlorite. This will immediately generate chlorine gas. Probably not in lethal amounts in the quantities I'm using, but I recommend against empirical experimentation in that direction.

Next I'll experiment with getting a regular, smooth, pointy tip. 

I need a new probe, Probe 9 having been put through the wringer a bit. So this time I'll try to quantify and document this a bit better. I also need to find an alternative to nitric acid, so I'll wrap that all into one project. The following sequence was taken of a wire (now Probe 10)  as I etched it in 5% salt water for 20 seconds at a time. Between etches the wire was washed with water, sprayed with isopropyl alcohol, then force-dried with cool air:

 I kept the magnification constant, so any apparent thinning or tapering of the wire is real.

The lower nut slot on the PIKA Motor Mounts was a bit tricky to access, so I moved them around the side where you can get at them more easily:

 Should make use of the Nut Tool more convenient too. Already uploaded to github and Printables.

Having watched Jon struggle with inserting M3 nuts into the slots on a PIKA I have decided to take pity on the poor users and create a Nut Tool. Should save makers from the occasional screwdriver stab wound:

You can find it in the library directory on github, next to the M3 parts and Metriccano libraries. I have also uploaded the STL to https://www.printables.com/model/1692745-reprapmicron-pika-micron-resolution-3d-printer

If it breaks, you get to keep both parts. But I've tested it out and it is remarkably sturdy.

I have uploaded the assembly instructions for the PIKA XY Table. Your feedback is appreciated.

https://github.com/VikOlliver/RepRapMicron/wiki/PIKA-XY-Flexure-Table-Construction

On Printables too https://www.printables.com/model/1692745-reprapmicron-pika-micron-resolution-3d-printer 

The levelling arrangement on PIKA has changed significantly. The Stage has two height adjustment screws on the left: one at the top (+y) and one at the bottom. There is a third on the right at the mid point. The right one is only adjusted of either of the left ones bottom out.

Levelling with aluminium foil is having problems. When I start out with a new foil, the error detected when doing several probes at the same point (the "span" here) is relatively small:

 --- Probing corners (1000.0x1000.0)---

BL:  -93.646 µm  (span 1.000)
BR:  -73.375 µm  (span 3.218)
TR:  -76.854 µm  (span 2.000)
TL: -103.823 µm  (span 4.469)

After 17 levelling attempts, it gets pretty bad:

 --- Probing corners (1000.0x1000.0)---

BL:  -74.188 µm  (span 10.875)
BR:  -70.032 µm  (span 22.531)
TR:  -81.271 µm  (span 15.063)
TL:  -72.313 µm  (span 6.437)

The latest foil I've used is far more lumpy than the previous lot, and that may have something to do with it. However, I'm not convinced it'll be useful for anything other than rough alignment.

So I've tried lowering the probe so that it just makes a mark, and drawing 10x10 100μm grids. Looking at this one:

You can see that the left edge is fainter than the right, so I need to raise the two screws on the left slightly. But that's not too bad over 1mm. If I move the probe and re-zero the height, I'll get the same pattern when I try to make a grid somewhere else, so it's a consistent slope.

The glass surface does not seem to be micron-flat though. There are definitely places where the tip will scratch the ink before others, and I moved both the probe and the glass slide around while probing one of these patches to verify that it's not a probe positioning artefact. I'll have to look out for that, but unsure how much difference it'll make in the end.

On the whole though, the PIKA V0.02 is easier to level, and the built-in microscope pole is very much more stable - which makes it easier to track what's going on (or going wrong!) with the microscope. It is now in the "pika" directory on the RepRapMicron github.

I've got the bits together. All the little tweaks make this a much nicer build. Haven't connected up the CNC controller but don't anticipate issues with that. I took many photos for the docs, and here's what she looks like now:

What I do have problems with though is my aluminium-covered touchplate. Will have to make a new one of those before I can test levelling.

I've decided to bite the bullet and update PIKA now. Main differences are a tripod-style Stage levelling mount, and an integral microscope pole socket. These changes alone remove 24 parts from the design, believe it or not. So it's faster to print, quicker to assemble, and needs fewer purchased components:

 

The other big structural change is to the Z Tower. That now has bracing on the vertical Axis Driver supports, and the Driver is now held by two vertical M3 x 50mm screws that double as stiffeners for said supports. Jon can use M3 bolts instead of screws if he wants (if you know, you know).

Another relatively minor change is the elimination of a lot of unused holes. These were rightly criticised as taking up a lot of print time and introducing potential weak spots, but I didn't know which ones were needed. Now I do.

While the Base, Frame, and Stage have all changed, the extant PIKA V0.05 Axis Drivers still fit the same structure.

From an administrative perspective, I'm now going to separate the PIKA and MAUS directories. This means a bit of file duplication, but gives me the freedom to muck around with the designs a lot more and optimise print speed, hole location, etc. 

Personal note: We have a tropical cyclone coming through on Sunday, and I have to get ready for it. I should get this update printed and start testing before then, but expect some disruption to the blog, SuperHouse livestream, and git as I batten down the hatches and deal with the inevitable power cuts &c.

Been away for a bit. Back on bed levelling. I've updated the bed leveller a bit (it doesn't just crash if the probe is already touching the bed), and been playing around some. I can get the bed levelled to 1μm over a mm or so, but trying to stretch that to 4mm doesn't work quite so well.

The problem I think is that there are four adjustable fixing points on the Stage, and there should only be three. The three-legged stool level thing. Now Jon Oxer is going to hate me for this, but I'm going to design a new Stage attachment scheme, which means I'll need to make a new PIKA XY Table design. The change is trivial to make in the model, not that it'll make Jon feel happier :)

To make me unhappy, I want to see how Jon goes with his assembly, so I know what other changes I need to make, and change them all at the same time. So I am very tempted to go do other things with μRepRap and come back to the Stage issue later. Of course, I want it all changed at once so I can go and do the documentation with assembly pictures - which also has to wait if I go down that path. Your thoughts? 

(I didn't want 3-point anchoring because I'm worried that the Y Axis Driver might pull the Stage out of line. However, I can't see a way out of it.)