A Wooden Gear Clock

This is my take on Brian Law's Clock #1

https://www.woodenclocks.co.uk/clock-1/

I converted the free pdf linked above into 3D files using inkscape and Fusion 360.  For the amount of effort involved it would be sensible to pay the man.  What can I say ?  I'm cheap and enjoy a challenge.  Also I've paid for the clock #22 plans so it's not total freeloading ☺

All gears are birch ply cut with a downcut spiral to reduce tear out.  Maple was used for the escape wheel and pawl resulting in smoother sliding surfaces than possible with ply.  The frame is oak.  The drive weight was a rolling pin in a previous life.

A good running clock doesn't rely on tight tolerances, in fact a bit of slop is necessary.  The gears don't need to be tightly meshed because there is no change in the direction of rotation.  The shafts need a bit of a loose fit in the frame due to the 'living' nature of wood  and frame sag caused by the drive weight.  This is one of the reasons why my clock uses no bearings.  The only place I did consider bearings is on the shaft supporting the drive weight.  One  place precision does matter is the bore of the escapement wheel, this bore must be as close to centre as possible.  A off centre or out-of-round escapement wheel will not provide a regular tick-tock beat and may not run at all.

Friction is the enemy.  Clocks are gear-up drive trains where wheels (the larger gear in a pair) drive pinions (the smaller gear).  Following the gear train from the large wheel at the weight pulley to the escapement wheel speed is gained through each succession of gear pairings.  This increase in speed comes at the expense of torque.  My clock powered by a 5 lb weight  can be stopped by a feather at the escapement wheel ... literally.  Friction must be reduced where ever possible and this is especially true for the top shaft where minimal torque is available.  The shaft are highly polished where free rotation is needed and the non-fixed bores are lubricated by wax crayon shavings.  Gear teeth are sanded smooth.

My clock is slightly different than the original design.  The shaft sleeves were not used.  The gears are either interference fit to the shaft or free rotating.  I'll leave the reader to figure out which fit was used on any particular gear.  The pinion on the lowest shaft has been pinned in place.  Two threaded rods tie the frame together rather than the fussy pins+pegs seen on the original.  One weak part of the original design is the pawl on the winding ratchet.  This pawl needs to be heavier in cross-section or completely redesigned.  It's holding power is marginal at best and if it slips the drive weight will crash to the floor.
                                                                                                                                           

   Links of possible interest:

a good explanation of clockworks
https://www.finewoodworking.com/2009/01/07/designing-wooden-clockworks

Lots of general information
http://garysclocks.sawdustcorner.com/

A wealth of information if you dig deep enough
https://lisaboyer.com/Claytonsite/Claytonsite1.htm

Forgot to mention my clock runs 9 hrs before the weight touches the floor.  Accuracy is reasonable but for time keeping
I recommend  something from Walmart.

Son of Sisyphus

I've been a longtime admirer of Bruce Shapiro's kinetic art and his marble-in-sand machines in particular.  It's now possible to buy his creations from sisyphus-industries but this is a DIY blog.  Plus I'm cheap.

Electronics to drive small stepper motors can be had for very little money thanks in large part to the 3D printer revolution.  This build uses an Arduino Uno with a cnc shield and A4988 drives.  The 3 steppers are Nema 17 and the power supply is an old computer psu.  The software is grbl using bCNC as a sender.  Grbl has the ability to slave 2 motors to 1 axis which allowed a  build with very little mechanical rigidity.

A view from below

2 motors slaved as 1 axis

The construction is very simple. Gt2 belts and pulleys, Oilite bushings on hardware store grade cold rolled steel shafts, a coil spring and a magnet.
The top side of the table has a layer of baking soda and a 3/4" steel ball. A strip of LED lights add colour and shadow effect.

Compared to the commercial offering this one is much noisier and noticeably less smooth in it's motion.  On the other hand it's much less expensive and has all the same capabilities.

first test run

Senseless Fun with Probes

 A probe can be as simple as a nail, a spring and a length of copper wire.  I had some unused items in the shop which led to a bit of overkill.  The large metal thing in the photo is an old PC power supply.  Over to the left is a board to read pulses from a photo interrupter.  In the centre is a dial indicator with a metal tang screwed to the top of the plunger.  This tang breaks the beam of a photo interrupter whenever the plunger is forced upwards.  The dial indicator is handy for quickly finding high/low spots in a sheet of material.  For actual probing the nail and spring mentioned earlier would serve the same function.

 This is the bowed wood that was probed.  The shiny bit is a quarter to give a sense of scale.  A size for the engraving and a centre origin point were decided on at this time,  the origin point was marked with a felt pen.  The  reason for marking the origin will become clear in the next step.  To probe a grid of points smartprobe.ngc was used,  this is included with the Linuxcnc package.  The probe work piece coordinates were zeroed at the the centre of the material.  Z zero is set to the photo interrupter trigger position at this origin which is also the high point of the material. 

The text file created looks like this:
-3.000000 -2.750000 -0.404709
-2.800000 -2.750000 -0.346447
-2.600000 -2.750000 -0.292887
-2.400000 -2.750000 -0.244701
-2.200000 -2.750000 -0.201554
etc etc etc.  It's pretty dull reading but it can be put  to use or at least something resembling a use.  It's time to visit ScorchWorks.
http://www.scorchworks.com/
The 2 Scorch programs used in this post are F-engrave and G-Code-Ripper.

The smiley moon started as an SVG from OpenClipart.  F-engrave needed a DXF file so InkScape was used for the file conversion.  I won't go into how the g-code file was created in F-engrave except to say it's about as simple as this type of software gets.  The critical point is the origin and scale are the same as the grid probe file.

With a probe text file and an engraving g-code finished it's time for G-Code-Ripper.

A: open g-code file
B: open probe text file
C: save the adjusted file

That's it.







Remember the felt pen mark made earlier ?  That's where the cutter was touched off for the engraving work piece coordinates.  It was done this way because the probe wasn't in the spindle or even near it.

Anyone tempted to try this is advised to go to YouTube where there are several examples.  The software has a clever probe function which wasn't used in this post

Easy Examples of Cutter Compensation

A description of my methods:
> G54 work piece offsets are used
> Tool table edits are done manually, don't forget to reload.
> Programmed to part drawing,  full radius compensation
> Linuxcnc with Axis GUI
> No tool changes so the D word will be used to read the tool table
    * using D word won't update the tool displayed in Axis.
       MDI a tool change if this is a distraction.
> No tool length (G43) offsets
> No G41.1 or G42.1, something about old dogs and new tricks   
> Cutting inches on a machine configured in inches.
> Ideal stock, no large amounts of excess stock, no obstacles.
> Running the programs in single-block simulation is the best and safest way to
   get a grasp of the concepts.

Outside cut
(1/2" end mill)
N1 G20 G17 G40 G49 G90
N2 G54
N3 G0 X-.75 Y-.75 Z1
N4 Z-.1
N5 G01 G41 x0 y0 D01 F60
N6 Y3
N7 X3
N8 Y0
N9 X0
N10 G40 X-.75 y-.75
N11 Z1
N12 M02

Outside cut description
N1 setting inches, XY plane, absolute positioning, cancelling tool length and   
      cutter comp
N2 read work piece coordinates
N3 get ready to approach, this needs to be a cutter radius or more away from
      intended start of the cut
N5 approach and enable cutter comp, D01 calls tool table
N6 start profile
N9 finished profile
N10 cancel comp with a lead-out 

Inside cut
 (1/2" end mill)
N1 G20 G40 G17 G49
N2 G54
N3 F30
N4 G00 x2 Y0 Z.5
N5 G41 X2.5 Y1 D01
N6 G1 Y2 Z-.1
N7 G3 X1.5 Y3 I-1
N8 G1 X0
N9 Y0
N10 X3
N11 Y3
N12 X1.5
N13 G3 X.5 Y2 J-1
N14 G1 G40 x1 Y1.5
N15 Z.5
N16 G0 X2 Y0
N17 M02

Inside cut description
N1  A safety block in case something unexpected is active.
N2  G54 is probably already active, better safe than sorry.
N4  Setting up for approach.
N5  Approach and kick-in comp, must travel at least one cutter radius, this is
       the  lead-in move.
       Look at the coordinate readout after the lead-in to see how G41 altered the
       path. Looking isn't necessary but may clarify the concept.
N6  A bit of a down ramp for the cutter.
N7  An arc to ease into the profile, (G3 or G2 can be used now as comp has been applied at N5)
N8  Starting the profile.
N12 Profile complete.
N13 Arcing away from the profile, can not be used as a comp cancel.
N14 Lead-out here while shutting off comp. (G2/G3 can't be a lead-out move)
N15 With lead-out complete the z will retract straight up.
N16 Go back where we started, this is optional.


An example from Linuxcnc Documentation
(1/2" end mill)
(this is crafty , maybe too clever if looking for a basic example)
G20
F30
G10 L1 P1 R.25 Z1  (alters the tool table)
T1 M6  
G00 Z0
G41 (call comp)
X4 Y3 (lead-in from tool change position, comp alters path)
G1 X5 Z-1  (ramping to Z depth)
G3 X6 Y4 J1 (arcing into contour)
G1 Y6  (contour starts)
X2
Y2
X6
Y4     (contour ends)
G3 X5 Y5 I-1   (arcing out of contour)
G0 Z0  (straight up, comp still active)
G40 (cancel comp)
G0 X1 Y1 (the lead-out, comp alters path)
T0 M6 (back to tool change position)
M2

A screen shot of Axis.  White is the outside example,  red is the inside example.
The gap between the red and white lines is 1/2",  the diameter of the cutter.

Pictures or it didn't happen.  The black is the outside example,  the arrows are the inside example.  Between the cuts in the top left area some remaining uncut material can be seen.  This should appear in all 4 corners but this is cheap plywood and the cutter was dull.

That's all for now.  Be aware there are lot's of potential gotcha's with G41/G42 due to 'look ahead' and gouging.  Linuxcnc is really anal on the gouging part,  that's a good thing I suppose.
  

Painting carvings

Finishing carved items with varnish or poly looks good in wood but sometimes you don't want the finished work to look like the base material.  Reading sign makers forums this leads inevitably to vinyl paint mask such as Oracal Oramask.  Unfortunately that's not available locally and it's costly to have it sent to my address.  The good news is a decent substitute is available at Walmart.  If you live in North America one of these stores is probably nearby.  The product used in this post is a removable adhesive covering and the brand name Con-Tact The most common use of this product is probably keeping sticky/stainy items from ruining shelves.

    My usual poor photography on display again.  For example the thin white oval inside the Ford emblem is near razor sharp, not like the jagged thing you see here.
Here's a step-by-step of the globe-like item in the picture:
1) After sanding the plywood 2 coats of white oil-based enamel were applied. Good old Tremclad in the rattle-can.

2) Allow the paint to dry fully before applying the vinyl.  The directions aren't very specific about how to apply.  What worked best for me was lining up one edge and slowly peeling off the backing film while pressing the vinyl against the work.  Hurrying this process resulted in air bubbles, lots of them.  The 'arrow' part wasn't covered with vinyl.

3) Time to carve.  I used a sharp v-bit and cut without using the dust boot.  My thinking was the brush might lift the vinyl and I also didn't want the sticky adhesive on the brush.  The vinyl cut fairly clean and any fuzzies were trimmed with a utility knife blade.

4) Once the cutting/clean-up was complete another coat of white Tremclad was applied.  This way if any paint was to seep beneath the vinyl it wouldn't be noticeable in the final product.

5) After about an hour it was time to spray the blue.  Oil-based enamel,  Ford Blue if you haven't already guessed.

6) Now it's time to carefully remove the mask to expose the white beneath, a razor and a dental pick help here.  The blue paint was allowed to dry to the point where it was unlikely to smear but not nowhere near fully cured.  A few hours ?  It was a pretty thick coat.  If you wait too long before removing the mask it could bond permanently with the top coats of paint and make ragged edges.

7) There were some small traces of glue remaining on the white surface.  Paint thinner cleaned those up easily.  Alcohol would also work.  Probably best  to do this after all the paint is really well cured.

The globe-thingy is a model for a weathervane.  The finished item will be cut from pvc sheet instead of plywood.       

Making a toy truck

The beginnings of this toy can be found posted 5/31/18.  Much drawing and measuring went into this project but most of the time passed in waiting for the right free material and deciding how to make the tires. Now that most of the design details are settled it's time to buy some quality birch ply and build 2 more with decent materials.  I hope to document the process in an upcoming post.

Half tone cnc

Need something to decorate your cnc workspace ?
Here's something simple that only needs a bit of flat material and whatever paint/stain you have on hand.

 On the left is M.C. Escher's "Eye".  The item to the right is a reproduction of the sexiest automobile ever produced (in my biased opinion).  The eye is a piece of varnished ply,  spray painted black before carving.  The Jaguar is a piece of shelving veneered with white melamine, blackboard paint fills in the dots.  Both were cut with a 90 degree v-bit.

The effect created is hard to photograph,  easier to explain.  Up close it's pattern of dots,  at a distance the dots become an indistinguishable part of the image.  These work best outdoors or in a large room.  Inch size dots on a building size canvas would be an awesome display. 

The software used is full of features and at the same  time simple enough to be self-explanatory.  Thank you Jason Dorie,  here's the link:
http://jasondorie.com/page_cnc.html