Profile

Cover photo
Jeremy Nelson
Works at Google
Attended University of Western Australia
Lives in Sydney, NSW, Australia
172 followers|411,952 views
AboutPostsPhotosYouTubeReviews

Stream

Jeremy Nelson

Shared publicly  - 
 
The world's cutest 4th axis! The head is 64mm. ENORMOUS!
1
Gill Nelson's profile photo
2 comments
 
And I guess I look at all those wires and think to myself..."it's such a man's room!" I bet Kris never goes in there!
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
Dishwasher magnets. We have Fisher and Paykel Dishdrawers, which work very well for us, but it does mean there are a combination of possible states the draws can be in. We have been using magnets that Kris and I 'knew' what they meant. These are... more explicit.

Colour laser printer, laminated and glued to some flexible flat fridge magnets we had kicking around with shoe goo glue to keep them flexible. So far they seem to work pretty well and I quite like how they look.
8
Mike Nelson's profile photoMichael Flanagan's profile photo
2 comments
 
Ha! We have a similar thing at home; just a single magnet with "Fill Me" and "Empty Me" that can be rotated to have the appropriate message at the top.
We started with "Clean" and "Dirty" but ran into the inevitable philosophical conundrum of which to chose when there was nothing in the dishwasher.

Yours, by the way, are much more stylish.
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
It's Alive! Actually milling things!

I had a restlessness night so I eventually got up and did some nice soothing CAM. Between CamBam for linux at 3am and a late start for work I spent some time cutting wood chips.

This is a wooden holder for Pixel Qi ( wireless inductive) charger electronics. The central circle is <1mm thick, the little "legs" on the turtle shape are for magnets to hold the phone and the tail is for the USB lead. This version of the design is designed to have a C shaped piece of work fit over the back, but I think a veneer would probably work better.

Overall, it worked very well. It took ages (43 minutes) but there are a lot of things I can do to make that faster- I'd guess I can get it down to 15 minutes with two simple changes and probably under 10 minutes. I have a list of things to tweak but it was a very plausible first attempt- and the mill worked perfectly.
9
Mike Nelson's profile photo
 
Nice to know that there are things that one can do on those restless nights.
Good to see the progress to the stage of actual production.
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
Z-Axis head mount part 2. In which Delrin is shaped and the head mount is completed.

After much cogitation I enlarged the Delrin nut recess and added M3 screws through the base which hold the nut tight. I had spent some time squaring out the nut recess (by manually filling out the corners) so that when I cut the nut I could make it square before realizing, and this will come as a shock to you, I have a CNC mill. So shaping a block with 10mm rounded corner is just code. GCode to be precise.

I started by squaring of the Delrin in my 80mm chuck- this is milling in reverse or lathe like operations on a mill. I now have some lathe tools so rather than holding a 3mm square hardened tool steel bit in a vice (pfff! so amatuer! so dangerous!) I am holding the a 10mm square tool in a vice. Lathing stuff is amazingly satisfying- the finish is beautiful and the there results are satisfyingly precise.

Cutting the 22mm section length of the rod was... harder. Another entry for the "professionals would shake their head in sadness". It mostly definitely didn't involve holding a hand saw against the spinning mass then once I had a decent groove sawing it by hand. It was probably something like using a cut off tool or something.

I'm getting much better at handcoding GCode. My first attempt was just about correct except for one minus sign. A GCode simulator is an absolute must- I've been using OpenSCAM which has been perfectly adequate. (It could do with a "reload gcode" action. Which I should probably just add because its open source, but I haven't got to doing that yet.)

After that I clamped the nut into the recess and tightened the screws, mounted the head on the table and drilled out the Delrin to 8.5mm then move the table to the side far enough that I could tap the 10mm thread. And finally the head was finished. Like so many projects the last 10% took about half the time.

Next- the column mount. 
1
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
Another achievement from the weekend- Y min endstop. More elaborate than I had initially intended I misjudged how far forward the switch needed to sit so after carving the upright (two shoulder cuts and some screw holes- easy!) I need to make a holder for the switch which could hold sit forward of the upright. The screws holding the second plate are neatly recessed so the rubber guard (which sits above this switch) doesn't catch or bunch) and the switch itself is held in place by grub screws which thread into the plate and the servo holes.

The cable is shielded to prevent interference, the edges of the mount are all neatly smoothed and overall its reasonably neat if slightly overengineered.

Its also buried under the rubber guard covering the dovetails for the Y-axis so no one will ever see it.

But this (in combination with the X min endstop) will let me home the mill to a known starting position- invaluable for picking up again after work interruption. I also have a Y max endstops (although no X max endstop yet)- the endstops also act as limit switches, which seems like a good idea. We'll see. The smoothiefirmware has no elegant approach for dealing with the situation where a limit switch has been hit- the machine stops but then you are stuck since you can't drive off the limit switch. When this happened to me I physically swapped the triggered limit switch with a different switch so I could drive off. I may need to make a daughter board which allows me to "untrigger" the endstops.

Yet another little project.

#mill
1
Mick Hollins's profile photoMichael Flanagan's profile photoAaron Drew's profile photoJeremy Nelson's profile photo
4 comments
 
The 3D printer is currently out of action (mostly because I built a new frame for it but haven't had a block of time to move all the componentry across, but partly because I don't have anywhere to leave it set up at the moment). Also I'm vaguely nervous about using a 3D printer part for a component where I care about repeatable precision- this switch is used for registering the location of the bed and I'd like (ideally!) micron precision... but certainly in the 10s of microns. I don't have a good model for plastic deformation or change over time- but I'm pretty confident this mount won't change much.
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
Rejected Princesses "Women to Awesome, Awful or Offbeat for Kids' Movies".

Fantastic.

By a gent called Jason Porath "I used to work at DreamWorks Animation (as an effects animator), but not anymore." Some of these characters came up during lunch time conversation and after he left DreamWorks the site was born. Its rather good and the historical figures are both amazing and outrageous. Here's one of my favourites.
This week we turn our attention to La Maupin, Julie d’Aubigny: sword-slinger, opera singer, and larger-than-life bisexual celebrity of 17th century France. Her life was a whirlwind of duels,...
1
Add a comment...
Have him in circles
172 people
Paul Cowan's profile photo
Dale Quinn's profile photo
Aleta Keating's profile photo
Chris Chambers's profile photo
James Bromberger's profile photo
Peter Cooper's profile photo
Kylie Ding's profile photo
Siobhan Capell's profile photo
Chris Nelson's profile photo

Jeremy Nelson

Shared publicly  - 
 
I made a drag knife for my mill. Drag knives can be used to cut paper, carboard, vinyl- anything you'd cut out with a paper knife by hand. But this is CNC and capable of more intricate and precise cutting... The knife just needs to swivel smoothly in the head, and the cutting point needs to lag by a distance so it can be turned by moving the head in an arc.

It has a couple of 11mm bearings press fit into a 20mm diameter cylinder of alumnium. The knife is a milled down Aluminium pen knife. Its a nice simple design- see my one pager that design. However, I did a truly awful job.

Because of the lathing technique (or should that be 'technique') I was using the centering of the holes in the Aluminium block is terrible- so the pen knife is actually tilted. The bearings are not quite parallel so knife hitches as it rotates. The knife is was not lathed to the precise 5mm diameter shaft I intended- I was holding the lathe tool in a vice and the tool slipped, so instead it ramps from 6mm down to 3.5mm, so one end wobbles.

Its so awful I can't actually quite bring myself to try it. It'll probably work, at least roughly. I'll try remaking it at a later date since its a fairly quick process. But it was a learning experience and while the implementation was awful, the design was actually quite nice and spare. My original idea was complicated- the final design very simple. So, silver linings of a sort!
2
Rico Site's profile photo
 
cool
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
One of the weekends efforts: a sign for my workshop-in-a-bathroom. This was produced from an image developed in Inkscape, rendered in CamBam using an image map that maps whiter pixels higher and darker pixels lower and then milled with a 3mm bit... for about 6 hours.

This was a test cut, which is just as well. The result looks amazing in the flesh, the letters in 'Workshop' are very fine (about 1mm thick) and about 10mm raised from the background. But late in the milling process portions of the wood gave way (see the middle of the W, the base of the R and several other letters). And Connor, who had been quite interested and asked lots of questions earlier in the day, came to investigate progress and carefully turned the light off as he left.

Which was actually the power point running everything.

So the cut didn't finish and it wasn't really worth trying to pick it up again. But it certainly proved that this could work. I intend to mill the sign in Aluminium, so collapsing shouldn't be a problem, although since the aluminium is larger its going to be an epic cut. The mill is, in truth, very slow. On the other hand I'm growing increasingly confident of setting it going and walking away so speed is less of an issue than it might otherwise seem.
4
Aaron Drew's profile photoAdam Haberlach's profile photoJeremy Nelson's profile photo
3 comments
 
So far CamBam seems pretty easy to use, and it actually works on Linux- so ++! The interface also seems good- it makes a lot of sense to me. I was impressed enough to actually buy a license. On the other hand I own very little licensed software because I'd prefer to use OpenSource/Free and as it happens there's nothing I've found in OpenSource land that's even vaguely decent CAM.

However I can't get CamBam to do waterline paths, its CAM actually seems... a bit dumb (it will "recut" areas its already cut if you overly shapes- it doesn't union the shapes then cut the union). Its fairly easy to use but I don't think the CAM is very sophisticated. But its certainly good enough for my use and its at a hobbyist price point not the insane-SolidWorks-call-for-a-quote type price point.
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
A working Z-Axis!

It took most of the long weekend for lots of silly reasons, but the z-axis is finally in place and it works. I have a bit of squeal from the delrin- I don't have the lubrication right yet- but otherwise it seems to be working OK. I need to play a little bit more to get confidence that it will work reliably (I think I lose steps at higher speeds and the acceleration needs to be conservative), but during calibration it was pretty good.

There were a few things I got right- I made some good choices about the bearing recesses, the belt is at a good tension (which is lucky because its not designed to be adjustable) and the overall design works. Locating the column assembly on the column was fiddly and ultimately I got it wrong. I should have got the alignment right then drilled and tapped one hole, then confirmed that hole was good and drilled/tapped the next hole, then done the final holes. Doing all four at once... didn't work. And its a mistake I've made before. Another five or six times and I may well learn something from it.

Anyway, a little bit more calibration and tuning and I should, finally, be ready to CNC Mill something! Woot!
3
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
The Z-Axis head mount. Hand coded G-code, didn't quite get it right on my trial mounts with wood. After fine tuning the G-Code I started cutting the final aluminium version out of 25mm thick aluminium. Sadly my Y-Min endstop was dodgy which meant re-registering after losing position was a semi-manual process (I eventually fixed the issues and could accurately re-register my position but there was much swearing involved in the meantime).

I stopped half way through to make a decision about whether I was going to pocket the material and use a delrin nut or simply thread the aluminium. (I'm using a steel M10 threaded rod for my z-axis and delrin has high lubricity which should free up more power for actually moving the head compared to a steel nut.) I eventually decided I'd make a delrin nut and pocketed the material before I had to take the block off the mill. Sadly I was over ambitious in my cutting process, lost registration and as a result the pocket isn't as neat as it should be. But its good enough for my purposes.

Final shots are setting up for the vertical drilling paths that go through the delrin nut pocket for the Z-Axis rod. I'm just taking my time rechecking the measurements since it would be annoying to have to redo everything because I bungled this particular hole...
3
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
I've added a light switch and a spindle direction switch to the mill's control box. The light switch had previously been cable tied to the side of the mill which was perfectly functional but not very neat. The spindle direction switch was relatively trivial- the headers were already broken out on the PCB inside the case- they just needed wiring up to a DPST switch.

Of course, this was all a MASSSIVE pain to actually do. The Power light needed to be de-soldered and re-soldered to get the decal off. The decal was painful to get right (and I still didn't get the surrounds on the knob right, but I did get the holes aligned correctly). I sweated cutting holes in the laminated paper (ultimately a hole punch, some manual cutting and a bit of sand paper) and then at the last moment I thought I'd got it all wrong when I tried to put the adjustment knob where the power light goes and it wouldn't fit next to the switch (luckily the next morning I was less stupid and realized the layout was just fine... phew!).

I had ordered in a little 5V->9V DC converter to power the light from the 5V supply inside the box, neatly soldered it up and more or less installed it when I finally got around to testing the 5V supply inside the box... only to discover there was only 5V when the spindle was running. Which is not ideal- the light has made a real difference and I use it when the spindle isn't running. Ultimately I accepted defeat and plumbed in the existing external power supply I was using.

All this waffling aside its a gratifyingly neat outcome but a surprising amount of effort to get there. This seems to be my everyday experience of achievement: there are many failures or seeming failures along the way, the outcome never seems to show the sheer labour involved but, nonetheless, its very rewarding to achieve something- no matter how trivial. I suspect this is the Silver lining of the Sunk Cost Fallacy: making us value the things we work hard for, even if they aren't much of an outcome.

#mill
1
Michael Flanagan's profile photoJeremy Nelson's profile photo
2 comments
 
Oddly enough my 3D printer lighting solution is likewise more primitive. The big spinning sharp bits of metal make me desirous of being a bit more neat and tidy.
Add a comment...

Jeremy Nelson

Shared publicly  - 
 
2 Axis! Ha ha ha! 2!

So I've been waiting for the videos to become available in G+ (they are slow to process which, as it happens, is a work issue that I'm tracking, but I'm on holiday this week so I haven't tried to fix it...) but I can wait no longer: the first video shows 2 axis movement. It's alive!

The movement demonstrated is a G2 move which is a circular arc- in this case a full circle. Crank up the sound- its really quite musical! I have another video (not processed yet- boo!) which shows a variety of circles being drawn with a pen held in the spindle. At this stage I have a very expensive etch-a-sketch. Which can cut you.

The other video shows some a "making of video" which is an X-axis CNC finishing run (look ma, no hands!). The finish with CNC is significantly better than by hand: the constant motion leads to a much more regular cut.

The final picture is leveling out the walls of the Y-Axis mill servo mount walls, a process made slower by the fact that my vice is not very high quality and I can't clamp items in it and get them to be reliably level and even. To be fair, I have a mill so I could probably do something about that, but I have other priorities...
1
Michael O's profile photo
 
Are you climb milling on the finishing pass? Does it finish better than conventional milling? I see the chips are recirculating a little. Need air blast:)
Add a comment...
People
Have him in circles
172 people
Paul Cowan's profile photo
Dale Quinn's profile photo
Aleta Keating's profile photo
Chris Chambers's profile photo
James Bromberger's profile photo
Peter Cooper's profile photo
Kylie Ding's profile photo
Siobhan Capell's profile photo
Chris Nelson's profile photo
Education
  • University of Western Australia
    Physics
Basic Information
Gender
Male
Work
Employment
  • Google
    SRE, 2008 - present
Places
Map of the places this user has livedMap of the places this user has livedMap of the places this user has lived
Currently
Sydney, NSW, Australia
Links
Some good food and interesting places. Very convenient to where I live!
Public - 4 years ago
reviewed 4 years ago
Expensive and doesn't really go very far. Airconditioning is frequently broken. If you are a local there are occasionally "cheap" days where you can stock up a card with $2 rides which can make this a useful service for the occasional bad weather.
Public - 4 years ago
reviewed 4 years ago
5 reviews
Map
Map
Map
Really good for my 4 year old. Good value if you get a year round pass. Expensive for a one-off visit.
Public - 4 years ago
reviewed 4 years ago
Public - 4 years ago
reviewed 4 years ago