RepRap: Blog

I levelled the coin 'A' letter scan with a spreadsheet. The upshot is that as the Y movement goes positive, the Z position goes positive by 0.06 times the Y movement. When the X movement goes positive the Z height goes negative by 0.02 times the X movement. These effects are cumulative. Correcting this with the spreadsheet gives the following scan:

This can obviously be corrected preferably in the firmware but for the moment I intend to write something to pre-distort the gcode. But now I know by how much!

The dark purple patch in the top right is a visible contaminant on the coin that required additional force to either move or penetrate and so shows as a deeper contact point. If we correct this, and flip the X axis to correct the fact it is inverted, we get:

There is now a visible defect in the crossbar of the 'A', and indeed if we look closely at a micrograph we can see it:

# posted by vik-olliver @ 10:59 AM 0 comments

I've connected the touch probe up and written a few scripts to generate a grid of probe points and analyse the resulting data by scraping the results from the CNCjs serial console window. This is what I'm getting:

The probed data (a 30x30 grid with 50μm spacing) looks like this, and suggests there is a constant slope to positioning that I'll hopefully determine and compensate for later:

Bear in mid I'm using a crappy but robust probe 'cos I don't want to destroy any nice ones while bashing things about a lot! I've got a better image of  the modified slide holder that shows the setup a bit better:

There's a fairly extensive wiki entry here that details the process and provides the scripts etc. because I bet they'll be useful for a heap of other things than calibrating the probe.

# posted by vik-olliver @ 1:40 AM 0 comments

I've got the GRBL touch probe working. On a RAMPS board I just used the T2 thermistor input as the touch probe (default config on the Mega firmware). Ground contact for the probe is supplied by a piece of copper wire that has been curled to give it a bit of spring. I've wedged a header pin onto the probe and soldered another onto the ground contact so I can take stuff on and off easily. Not able to play with chemicals to make more conductive glass yet, so I have put a New Zealand 10c piece on it and I'm testing the probe on that. This is being done with the stock CNCjs app's probe function.  I just copy the Z touch data out of its serial console - quick hack or wot?

Mapping the data with LibreOffice wasn't possible - apparently 3D Graph means something with 3D eye candy in officespeak. This has meant learning gnuplot, which takes a bit of wrapping your head around but gets there in the end and produces proper 3D maps in perspective. You'll see some later, but a 15x15 grid with 0.1mm divisions takes about an hour to probe ...

# posted by vik-olliver @ 4:37 AM 0 comments

Why do this?

I'm positioning a probe with micron precision for a teeny micron-scale 3D printer called RepRapMicron. Float glass is a pretty darn flat work surface, but I need a Z touch probe and glass microscope slides do not conduct.

Why tin oxide?

I need a thin-film technique that is accessible to as many hackers as possible. Silvering is excellent, but tarnishes. Conductive tape is too thick, so are paint-on solutions, graphite etc. I do not need high conductivity so futzing around with vacuum chambers and wires is not necessary. I hope.

Why not just buy tin chloride?

Historically, New Zealand has had problems with people making illicit substances because it's so isolated. Shipping is a nightmare. So buying chemicals and lab equipment can be problematic. Plus, I don't imagine NZ is unique in that regard. But if you can get it, fine, buy it.

We do these things not because they are easy, but because we thought they would be easy.

# posted by vik-olliver @ 8:44 AM 2 comments

I'm about to go off on a trip to get the other eyeball sorted out. I haven't managed to finish a conductive slide, and won't be able to play with boiling acid, sprays, etc. for a couple of weeks, so expect a break from my usual tedious updates.

I have realised a mistake in my sums, and need 4x the quantity of acid I previously thought. Also, it has become apparent that tin does not dissolve in 24% HCl unless it is heated to above 60C so leaving it to sit overnight doesn't help. When I get back, I'll repeat the experiments with the stronger tin chloride solution.

Meantime, I might be able to work on the Z Touch software one-eyed when I can.

Might try my luck ordering some lab grade tin chloride, but due to "backyard alchemists" having a history of brewing illicit substances in New Zealand chemistry things are difficult to get. Believe it or not separating funnels are restricted items here, for example!

# posted by vik-olliver @ 1:42 AM 0 comments

Rather than blow bits of firebrick and soot onto the slide with a blowtorch, I've got an old carbon steel wok set up over a gas burner - if you do something similar, burn out the steel surface thoroughly before putting the slide on it like I didn't. The FLIR camera that lets me accurately measure the temperature of the slide before spraying on the tin without being distracted by the furnace heat reflecting everywhere.

Short story, I do need 450C to decompose the tin chloride. I did several runs starting at 300C and they didn't start to conduct until I reached 450C (tin oxide melts at around 1,600C). This being a rough test, I reused the slide. The final slide gave a reading of 2.2M from one edge to the other, but nothing across the other way due to a, uh, significant air gap:

This is probably due to repeated heating/cooling, and the sheer volume of spray that suddenly cools it down to about 250C. To fix that, I'll make a new batch of tin chloride and double the concentration. Less squirts, less temperature drop.

Note: Antimony does not dissolve in hydrochloric acid, so I won't be picking much up when dissolving Britannia Metal in HCl.

I'm allowing the finished slide to cool gently by simply sticking the lid on and gradually turning the flame out. As I reused it a few times the finish on that slide is well grainy.

I've made a new slide holder for the Block Stage with a connection probe built-in, and all I need now is an intact conductive slide to put on it.

# posted by vik-olliver @ 1:46 AM 0 comments

Previous experience with tin oxide has been on fired glass, and the decomposition temperature of tin (II) chloride is listed as 632C. But I never tested it. So this time I heated the remains of a shattered slide to 450C at one end with the other just happening to be at 170C. The abused slide looks like this:

There is a sweet spot on the slide for conductivity marked between. Assuming even deposition (which it's not) and an even temperature gradient (which it isn't) the best conducting deposit seems to be around 265C, give or take quite a bit. This is not the clearest part of the slide but the high end has a best conductivity of 4.5M and the best part about 900K. So in the future I'll try not to get so melty. A hotplate or toaster oven might be more suitable than the improv firebrick furnace and gas torch.

The high end does look nicer though.

The slide did of course fragment due to thermal stress as I was probing at it, but I got good readings before it did.

# posted by vik-olliver @ 4:35 AM 0 comments

It's fun playing with fire and acid. Some precautions needed. This attempt used a much smaller firebrick cavity, and the glass was not reheated between the 12 sprays of tin chloride. I tried masking part of a square glass slide with a coverslip, but the first attempt melted it, and the second blew it to bits when I sprayed the tin chloride on. Tested coverslip fragments were non-conducting. Will probably use another piece of slide as a mask next time.

But the square itself had a reasonably clear coating that was definitely conductive, giving under 9M from edge to edge in places. Wrangling multimeter probes into the right position was tricky (better make something more efficient and pretty) and I ended up wedging the black base (a convenient rubberised battery pack with matches, what could go wrong?) at the same angle as the taped-up probes. Here's a sample picture of the conductivity:

And this is the points of contact. Some 6mm lengths of "clear" glass measured 860K. Note that the far edge of the slide, which was in contact with the flame, is slightly crazed. A bit like me...

# posted by vik-olliver @ 1:52 AM 0 comments

This cockamamie contraption has a multimeter directly clamped to the probe, and to a wad of aluminium foil held in contact with the slide's tin oxide coating with a miniature clothes peg. The whole lot held down with various clamps. The probe tip is only just touching the surface - 1/50th of a turn of M3 with a 2:1 reduction will make or break contact. M3 thread is 0.5mm, so that's in the order of some microns (and a lot of undetermined experimental error).

So now all I need is a part-coated miniature slide, a redesigned slide holder with a contact clip (hey, I have a pogo pin collection .... somewhere), wire on a non-solderable probe, more bloody wires, and Z touch sensing on the GRBL board. A piece of cake 😉

# posted by vik-olliver @ 9:01 AM 0 comments

Power is down, so might as well post update.  The tin chloride from yesterday was poured off, the residue rinsed 3x, and the rinsings made up to 100ml. The solution was clear, indicating minimal copper contamination. 1.3g of unreacted dry residue remained from 8g of tin.

 The shattered slide from yesterday was placed in a firebrick cell like this:

and heated until the edges tinged red. 3 x 0.5 ml fine sprays of solution were directed in, the glass reheated, and the process repeated twice more. The cell was then covered with thick blocks of expendable wood and left to cool:

The slide cracked in one location but was annealed enough to survive a short drop. A splattering of tin oxide was clearly visible:

 The smaller splatters were not measurably conductive, but a higher concentration can be observed on the right broken edge:

# posted by vik-olliver @ 12:32 AM 0 comments

To make a conductive spot on the glass bed I thought I might try a tin oxide layer. This is/was used in LCDs etc. to make the glass conductive. First, you need tin (II) chloride. I have tin in the form of "lead-free pewter" or I guess I could buy some lead-free solder.

I partially dissolved 8g of tin (I've left it dissolving) in 10ml of 24% w/w hydrochloric acid in a beaker. The reaction is quite slow even at 65C in my crude water bath, so you want smaller bits of tin than I used. I wore safety glasses, used ventilation, and avoided creating any Vik Chloride.

Part way through the reaction a drop was placed on a microscope slide and heated with a blowtorch until the droplet went black, then white. Predictably, the slide shattered, however the piece with the sample on remained mostly intact.

The conductivity across this ugly sample is approx 800K, and the sample is far too thick. But, tin oxide there you go.

To improve this I'll wait until the tin has dissolved as much as it will, and dilute the solution by 100x. I recall you're supposed to leave a bit of tin in there to discourage hydrolysis. Then I'll spray that onto a pre-heated slide, >800C. This time I'll use a hot firebrick to add thermal stability. Hopefully that will create a finer - prettier - layer of tin oxide and not shatter the glass. I have previously done this on borosilicate to create iridescent art, but used commercial tin chloride.

I won't cover the whole glass bed to preserve its smooth surface. I'll just have a conductive patch for zeroing the Z probe. The tin oxide may block UV, so might also serve the purpose of protecting the resin reservoir.

I go for another eye operation next week, which puts another bump in the developmental road, but after that my eyesight should be fully restored. In the meantime I can't play with fuming acids, dust etc. after the op for 10 days.

# posted by vik-olliver @ 1:01 AM 0 comments

I have updated the documentation on the Wiki here with details on the optics and connection of end stops on the current block stage, following comments on the models on the Printables site. Feedback welcome.

# posted by vik-olliver @ 1:47 AM 0 comments