RepRap: Blog

Capillary action gets a bit too carried away when you dunk the probe in a blob of 3D printer resin. It's designed to flow nicely, and for our purposes we don't want it flowing quite so far and enthusiastically. So to thicken some I gave it a half-second burst of UV, which allowed me to tease out a line less than 0.1mm wide - about 50μm in this photo. Back to micron-scale again!

I think the inclusion might be Sharpie marker off a dirty probe tip. I only noticed it after I washed some dust off with IPA, but this might be a way of making things easier to see. Unfortunately said IPA also washed off some finer detail I had done by bouncing the point on the slide, so I didn't get to measure those dots. It took a lot of other stuff off too. Still more things to learn to handle this material, obviously.

# posted by vik-olliver @ 3:52 AM 0 comments

This was pretty slapdash, but I thought it was time to put some resin on a probe and see what the heck happens. I used a wooden toothpick to put on what I thought was a small blob but was about 100x what I needed. Then I dunked the probe into the very edge of the blob and trailed it around to see how far it would draw out. Then I exposed the whole thing to UV light for a couple of minutes to set it good and solid, and this is what it looked like:

The image on the right is before UV exposure, the one on the left after it. For scale, the white circle is 6mm in diameter, so we're looking at about 0.1-0.2mm line widths for the most part. I was moving the probe around in 0.3mm increments, and as it moved to the left needed to drop the probe by 0.015mm to keep contact.

Note that the resin has contracted and balled up under UV exposure. This is not ideal, because the smallest surviving blobs look to be bigger than 50μm across - sorry, I totally wrecked it trying to remove the resin. The big blob came off easily, but the fine stuff pulled apart and I totally failed to pull a strand off. Multiple layers next time, I guess.

Anyway, a change from the usual scratchings, and more data points.

# posted by vik-olliver @ 8:19 AM 0 comments

I wondered if there might be useful diffraction effects if I illuminated the probe with a laser that might make it easier to see when the probe was near the surface. The result was ... scary. Fortunately it didn't fry the optics.

Note how it lights up all the little bitty dust grains on the slide.

Oh, I've added an LED under the stage, so this is the view I get on my monitor when flying the thing. Having a functional "Home" on the controller is making a noticeable difference in repeatability.

# posted by vik-olliver @ 6:15 AM 0 comments

Yeah. Why am I not scratching the dye with the probe? I'm sure I can even see the probe bending.

Oh. *SMACK* Because I have put the slide in upside down. The dye is on the bottom of the slide and I am trying to engrave glass. This does not work.

Good job it wasn't one of the good probes 'cos it now has a wee bend...

# posted by vik-olliver @ 5:12 AM 0 comments

The blocker on further work has been the shifting axes when making movements of more than 100 microns. This puts the Z height out, which is pretty crucial. In theory I could compensate for this in software, but the drift varies depending on where the stage is zeroed.

So endstops have become of critical importance. I've built some out of 50mm lengths of 0.6mm wire, crossed over under the bottom of each actuator. Hopefully these will position me within 100 microns of a known point, and then I can start pre-distorting gcode to position the Z height correctly.

With all the wiring and dodgy crossed-wire switches it looks more like a sci-fi IED inside than a 3D printer. Wouldn't be the first time that confusion has occurred, and travel with RepRap prototypes has always been exciting because of it!

# posted by vik-olliver @ 11:18 PM 0 comments

I've been trying to put a direct drive M3 threaded rod on a stepper motor to drive the microscope mechanism directly. This is not working out well. The reason being that the axis of the motor and the axis of the screw hardly ever line up, causing the threaded rod to rotate in small circles. With the geared drive in the OpenFlexure microscope, the gears allow the motor shaft and M3 screw to not be exactly parallel, and the flexure mechanism halts the rest of the wobble. There is an incredibly large force in that wobble and it will bend and torque the most enthusiastic of flexures out of shape.

So for the moment the parallelogram mechanism goes on the back burner and I'll just have to compensate for the axis drift in software. That's probably where this will all end up anyway, once I have axis zeroing sorted out.

# posted by vik-olliver @ 4:36 AM 0 comments

I had a mad idea in a hotel room a few nights back about a way of doing a parallel flexure XY axis without having to support the weight of the motors. I just had to get the idea out of my head and ended up going down to reception to beg a wad of paper to scribble on.

The big issue with the OpenFlexure design is that the motion of the axes is not truly linear but circular. This doesn't matter much for small-angle microscopy or lining optical fibres up, but it does matter to me.

In theory you can calibrate it out, but in practice without a lab-full of precision gear that's a bit awkward to do. Also the range of motion is a bit small given I need to stack photosensitive resin, the workpiece, and UV light on the printer and move the stage around between it all while dodging a probe tip.

So I've mostly been fiddling around in an effort to find out what kind of flexures work, how much motion I can get out of them before they fail, whether I can print them flat and fold them up etc. Consequently my bench looks like this:

My waste basket is full of the fails. Note the pantograph. This is a key element in combining the motion of the X and Y linear drivers without one motor having to support another. It also gives me a handy method for scaling down motion. Yes, the centre is unconstrained for rotation, but I plan on having an XY-constrained stage, so that'll eliminate that problem. No idea how much constraint this'll need, or accuracy will get lost in the flexures, but I'm going to suck it and see.

# posted by vik-olliver @ 2:42 AM 0 comments

New microscope exchanged by Jaycar with no problems. Homing the axes however, problems. Notably my brainwave for just running the axes out of the drive nuts produces so much vibration the nuts move, and the drive screws won't reliably go in again. In fact it vibrated a mounting screw loose!

I guess I'll have to do it properly and put switches in the base. *Sigh!*

# posted by vik-olliver @ 8:34 AM 0 comments

 I've uploaded the build files for Titch's main body and base. I have also created a collection on Printables of the parts used in the project so far. Please join in.

# posted by vik-olliver @ 6:40 AM 0 comments

I've published the files and assembly details of the glass slide holder on Printables here.

# posted by vik-olliver @ 11:35 PM 0 comments

No microscope, so I thought I'd noodle around with some useful bits of 3D printed hardware. Looking for a way to hold the glass slides in place.

Now I've settled (for the moment) on squares of glass microscope slide, it has been worth developing a holder that will clip them to the Block Stage platform. I came up with this little clippy gadget which I have given the snazzy name "Glass Slide Holder." 

Once I know how well it works I'll post the files on Printables.  Why is one corner sliced off? Because it would collide with the probe platform.

Be nice to test it out, but without the microscope I can't see a damn thing!

# posted by vik-olliver @ 7:01 AM 0 comments

The illumination in the USB microscope/macro camera that I've been using has failed. Should be under guarantee, but development this weekend is now slightly embuggered. I'm sure Jaycar will get a replacement in fairly promptly, have good working relations with them.

Just before that, I was measuring how much height difference I was getting in a 0.4mm radius. Looks to be about 0.02mm which is not terrific. So we're looking at a 5% slope, roughly. Once the microscope is back, we'll see how easy it is to compensate in software.

Just realised I can use the probe to draw calibration marks for re-zeroing the probe. That might help.

The glass microscope slides turned up too. I've cut one into 1/3rds (basically inch squares in old money) which *just* fit on the stage. Will probably make a holder with clips for them that screws on to the stage, as I'm currently holding them there with blue tape...

# posted by vik-olliver @ 1:34 AM 0 comments

I've made a holder that clips onto a standard 40mm, um, biological sample pot. This suspends the wire in the electrolyte and gives you an easy way to hold the wire at the right level with the positive croc lead. Makes probe tip manufacture a heck of a lot easier. Files and process on Printables here.

# posted by vik-olliver @ 11:42 PM 0 comments

I have *a* way of levelling the axes on the Block Stage: Drive the motors until they back the adjustment screws out of the drive nuts, at which point the motor just spins on top of the drive nut. Then drive them back in half way. Crude, noisy, probably good to only 100μm max, appears to work...

# posted by vik-olliver @ 11:13 AM 0 comments

I've bent a few probes now. They're delicate little snowflakes. So I've created a storage spigot and added it to the probe arm files. This spigot can be glued inside a pill container lid to allow the probe to be protected by the container without rattling around:

# posted by vik-olliver @ 12:49 AM 0 comments

Scrappy innit? I could tell from the microscope that the probe was tracing out a square, however point contact is a bit patchy - don't expect to see individual 20μm squares as this is a rigid chonky tip.

As this was scanned in both X and Y, we can pretty clearly see that the probe is too high in the top right corner of the image. Rough sums as follows:

Height of stage: 75mm

Travel from centre to endstop: 2mm

Therefore Pythagoras tells us 0.025mm height variation on the flexure stage.

We're moving 0.2mm here, so we'd expect a height variation of roughly 2.5μm, which feels about right. A few more test patterns required, and a way of actually making sure I'm properly centred to get some consistency. Perhaps a pattern of groups of concentric squares varying in height?

# posted by vik-olliver @ 10:26 AM 0 comments

So I've tested out the new probe holder fitted with with a fairly broad etched tip. Back to a glass slide, which I have to mount sideways to fit on the stage (more on order).

Over 400μm of logo I'm noticing a roughly 7μm variation in probe height. This might be due to my stage not being perfectly level, or the stage dropping slightly as the X and Y axis move. That should be a very small effect but I'll have to run the sums because 7μm is indeed very small.

If I make smaller things though, the Z height variation will be less in proportion. Still, here's what the test looked like. I'll run up a proper test grid and run some off with the axes positioned at different maxima and minima. See what's actually going on.

Speaking of which, I'm starting to get annoyed about not being able to "home" the axes, so that might happen in the near-ish future. Personal life getting in the way of that.

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

I've added a probe socket to the probe arm files.Partly because I wanted to stiffen up the wire probe tips I'm making, and partly because I realise that hypodermic needles are not universally available. These press-fit onto the end of the probe arm.

The probe wire can either be just wedged in the slot (which conveniently tapers inside), glued into place, or melted into a spludge of plastic with a soldering iron.

# posted by vik-olliver @ 7:51 AM 0 comments

I may be blaming my tools here, but I'm running into a persistent problem with my clever idea of using glass cover slips: They seem to be uneven. Or the cheap batch I have are anyway. This might not normally matter, but when you're controlling the probe height to a micron or two it can be a pain in the butt!

I'll have to switch back to slides. Which means I'll have to cut some down because a whole slide won't fit on the new stage. Or make a new stage, I suppose. I'll have to fork out for more slides eventually as I have like one spare and all the others have really interesting specimens on...

# posted by vik-olliver @ 9:58 AM 0 comments

On this one I ran the 400μm logo etch 10 times, lowering the probe height by 1 micron each time. The photo shows the top of the arms of the 'u' under a higher magnification than I usually use. You can clearly see the separated ridges extending in a Northwest direction as the probe is flexed progressively out of position and in the direction of its inclination.

How high a magnification? I dunno. Too high for me to have useful light levels for general work. The objective lens is bloody close to the slide and I'm not going to risk my expensive calibration slide on it anyway!

# posted by vik-olliver @ 8:43 AM 0 comments

The Mk 1 probe holder was a bit tricky to set up, so I've been writing my own flexure library and made a new adjustable probe arm with it

While fitting it I buggered the probe tip though, so guess what my task is for tomorrow?

# posted by vik-olliver @ 6:37 AM 0 comments