Project Logs

Collapse

• Repairing. Filling the air bubbles

  mclien • 02/16/2025 at 17:00 • 0 comments

  This was the firs idea for the repair:

  That left me with still more than half of the bubbles still there, even if smaller.

  So I figured, since the holes are 0.6mm and the syringe being 0.3mm in diameter, it might be better to try to hit exactly the "top" of the bubble with only one drill hole. like this:

  And while filling the bubble with resin carefully retract the syringe, while still pushing more resin slowly in and let e bit of it "overflow" on the top.

  It took 2 additional attempts to get nearly all bubbles filled.

  some very small ones are still there, but most of the parts that look wrong in this pic are actually optical distortions from the overflow, which forms little lenses:

  Next planed steps: grinding the both sides flat, testing differed grains of sanding paper to see which give the best diffusion for the LED light.
  

• That was the theroy, this is the outcome

  mclien • 02/10/2025 at 21:10 • 0 comments

  It seems the 2 layers weren't glued properly, so resin moved into that capillary. Leading into 2 problems:

  First, detach was not to easy. Some of the screws didn't come of easy. Most could be detached with a bit of force until the thread moved inside the resin.  A better choice would be stainless steel screws which stick a lot less to the resin.

  I break of one head of a screw (bolt), which I could remove with just more force and pliers.

  But the bigger problem is the resin "going sideways" between the wooden layers generates a cavity in the resin, having only one letter without a small or big air-bubble inside (the "L" at about the 8/9 position):

  I did try to fix it by getting resin into the bubbles from above. So I might know the success of that in 24h (curing time of the resin.

  Ideas to prevent this: 

  better, tighter method to glue the 2 layers together before applying the resin.

  Do not use epoxy with a long curing time, but resin for 3D printers, use acrylic as the bottom layer and cure every cavity right after the injection of the resin by applying UV light from underneath.
  

• The dirt problem

  mclien • 02/09/2025 at 20:30 • 0 comments

  What was to expected, is the fact, that having the letters as holes in the surface is. it will collect dirt/fuzz:

  Especially carrying the watch in the pocket of a black trousers:
  

  It is not too difficult to remove, but hardly the best option. The watchmakers choice would possibly be a a glas/mineral on top of it, but that would require considerable extra designing steps.
  

  After spending some time on the technical support hotline of my favored resin supplier, I was going for "fill it with resin". So I ordered a resin with low viscosity and the hint to make that even more low and get the bubbles out by heating the resin to 60°C.  To apply it in the (very small) cutouts, I was planning to use a syringe. I guessed 2 sizes of inner diameter (0.15mm/0.006"  and 0.25mm/0.010" with blunt tips. You don't want to accidentally poke yourself with a sharp syringe filled with epoxy).

  The burned flanks should help to prevent resin soaking to much into the wood, but also reduces adhesion. As I feared those small pieces would come of easily, I thought to stick the 2 top layers together first and fill the letter cutouts and the LED cutouts underneath, hence hold the resin in place by from fit, too.

  So this is a cut section of the idea to fill the resin from top with the syringe and let the resin rise from button to top through the letter cutouts:

  Here are the 2 top layers glued together:

  (these are the pylea runes, because those are smaller to be easy applicable to the bigger roman letters)
  

  With the 2 top layers on top, some failed laser cuts on the bottom side with the separating file in-between and poked through the film with the bolts:

  And finally screwed it all together:

  And here is a pic of the dry-run, how I planned to apply the resin. As you run out of hand immediately when working with resin anyways, there is now way to take pictures by yourself at this task.

  I did use 2 resin mixes (only the clear on and one with a bit of with transparent color paste) and did 2 ways off applying: "only till it reaches the top edge and "a bit of overflow (to the surface tension bubble). The later one needs to be refilled a drop from time to time, until the resin starts to get less viscous, which I did only start to use as I saw the resin didn't get soaked into the wood too much.

  The smaller syringe (0.15mm) was too hard to get the resin through btw.
  

  It is curing right now, I'll take some pics of the further progress after that. I intend to grind the excessive resin flat to the wood level.
  

• Assembly

  mclien • 02/04/2025 at 20:19 • 0 comments

  So finally putting everything together:

  Bottom part has some small layers of wood to hold the PCB in place, the small ring around the LED pcb needs to be handled carefully (once in place it is stable) and the little screw which becomes the on-switch.

  The brown rectangle is a shallow engraved pocket for the wires soldered at the back of the LED ring
  

  Closeup of the first step of the stack up:

  there is a little groove to apply for the hex nut part
  

  Detail of the push button setup. (it is quite difficult to get the the 2mm hole in the 3mm plywood (this is the best of 4 tries)

  On top of that  the LED frame, covered with a adhesive diffusor film. Also supports the parts a bit.
  

  And the next layer with the second film and cutouts to adjust to the letters on top (those are a bit different depending on which kind of alphabet is used (this one is for the pylea runes):

  Also holds the LDR and battery in place.
  

  And the final cover of wood (1.5mm thin), with a bit nicer wood than the poplar I used for the rest (birch IIRC). I noticed the not very precise cutouts, there was a problem with the x axe of my lasercutter, that wasn't running smoothly.

  Hold together with M1.6 brass screws (or is the correct term nuts and bolts?).

  backside with the magnetic connector and front at 16:44

• Soldering

  mclien • 02/01/2025 at 21:31 • 0 comments

  Soldering smd parts is tricky using a soldering iron from the 90th, no hot air solder and no solder paste. But it is doable, you need to be creative on how to fix the parts, till you make the first connection.

  So here we go:

  I removed the USB socket by disassembling it (cut away the metal hosing with the saw, nibbling away the resin and get rid of all the metal tongues except D+/- (yeah might have been easier to remove it entirely and use the pads directly, but this way, I could bend the tongues a bit out of the way to prevent shorts)
  

  First par was the magconnector, barley doable because is sticks out just some 10th of a mm. I then covered most of the backside of the trinket with capton tape and fixed it with the USB and Bat holes. One capacitor is TH, that is the one that makes a few connections, so the wire is helpful. The sot-23 Mosfet just sits in place (and won't stay there by itself
  

  So, to hold it in position: (plus the resitor parallel to the cap)
  

  Not exactly the next step, but almost all connections on the backside. I used some un-tangled CAT5 cable here, which is obviously not make with insulation suitable for heat.. (have to find better material for the next project..)
  

  Few steps later, I2C wires from the RTC to the trinket:

  And here is the finished population:

  And the backside with the backup battery placed from stacked to side-by side. Involves some cutting and bending of the solder tongues.

  The hole in the matrix board for reaching the pin underneath the trinket.

  Next up: assembling. I re-opened it for the pic and I did notice the amount of dust on the LDR, we'll come to that problem later, which also effects the LEDs/letters.
  

• Cutting the matrix board

  mclien • 01/29/2025 at 18:50 • 0 comments

  (something like Dez 24 and January25)

  First I printed the shape resulting from my CAD work an arranged it n a matrix board:

  PCBs are made of a layer of different materials including fiberglass and resin matrix and some more. I have seen some people using a belt sander and alike to get boards in the desired. I do not recommend that unless you have a really good extraction system at your sander.

  To reduce chips and dust going everywhere I used a fine saw blade on a Goldshmiths' saw. Those are usually work while pulling, I added some tape (sticky side up) right over the handle:

  That way most of the particles do stick to that pieces of tape and can be folded in, once you are done sawing. This only works to some extent as the result show, but I did clean that up by sticking tape to it, too.

  Not a quality cut, but I remember to have been a bit in a hurry (which you should never be for such work). Turns out I needed to adjust things anyway.

  I did the same cutting the corner of the trinket (sorry Lady Ada!).
  

• Parts placement CAD

  mclien • 01/28/2025 at 15:11 • 0 comments

  As I have no idea, if there is any tool to help you to place and connect parts on a matrix board, I'm went with what I used to.

  Having the parts placed in my 3D CAD, I created a top an bottom view with the relevant parts:

  I was planning to cut pieces from the trinket an the PiHat as both the corners are not carrying relevant traces (trinket just a mounting hole, easy to be sure, since layout is present/ PiHat datasheet of the DS3231 shows those pins are some of the many NC pins).

  So I drew the matrix holes and the parts, using a color code to mark all pins connecting each other:

  Pic is the one with the already placed parts with pins-to.connect close to each other.

  Then place the connections with lines (of course the mechanical CAD does not prevent crossing "tracks". I finally found ways for all connections, shifting the RTC one row up, which spared me cutting something of the board. I used other line types for the respective "other side" of the board.
  

  To not get lost in the process of placing parts and wires (tracks) in actual making, I doubles each view and deleted every part/wire I had placed in one view.
  

• Prototyp 2, packaging

  mclien • 01/27/2025 at 20:32 • 0 comments

  (about Oct to Dec 2024)

  Thanks to the incredible adafruit team, there was nearly no need for 3D models to make from scratch.

  Here is the github link: Adafruit CAD parts

  I think only had to make the LED ring from scratch (the 18-LED ring is not an adafruit part). And LiPoly batteries are basically littls bricks. So after "some" back and forth (like is it easier to use a LiPoly which is thick an small or use a very thin with a bigger footprint) I came up with this arrangement:

  View from the bottom.
  

  violet: PushButton,  blue/yellow: DS3231, backup battery moved from back to the side, brown: Trinket with the USB socket removed. Green: stripboard (correct word? hole matrix board?) and top the mag connector.

  And the vie from the top side. Top right corner: that is the piece of wood, which holds the M2 brass bold, the "handle" for the push button. The rectangle hole in the stripboard is for  accessing the IO pad on the bottom side of the trinket. and the ring that holds the LED ring in place.

  Next layer of plywood holds the  components in place and separates the light of the LEDs:

  On top of that the is the layer holding the Lipoly, directs the LED light to match the shape of the letters i the final layer and holding the LDR:

  I thought I could fit a bit bigger LiPoly, but I had no small enough Pushbutton to make the room.

  
  And the top layer with the actual letter cutout:

  A keen watcher will notice the extra "M". As there was a "spare" LED anyways and having in mind, that at midnight the would be no LED on at all, I use the "M" for that.

  There are some other options, like "24" or light up all LEDs as a "0". I might change that for another Prototype, when the 18h LED is used as battery indicator. Or some other variant,  I haven't thought so far.
  

• The PiHat mystery

  mclien • 01/27/2025 at 17:23 • 0 comments

  Well at least it was one for me.

  After the setup on the breadboard, I tested the reading of the time. All worked fine as long as USB was connected. There is an example sketch in the arduino ide, which sets the time of the build. I could read time and display it, but the moment I was running from battery, I git a "not RTC source found".

  As I'm not an electrical engineer at all (I try to expand my elementary school battery-switch-lamp knowledge as I'm going) I figured it our "backwards". I compared the adafruit DS3231 board with the PiHat and noted that the resistors were different, which is easy to understand, once you realized, that the PiHat is made for 5v logic, while the adafruit one is made for the 3.3V logic (at least I hope, I have right in my head now).

  Anyways here is the reworked PiHat, with the correct resistors: (and 2 cables to connect a new coincell, because the soldered in was already drained)

• Prototype 2, changes and breadboard

  mclien • 01/27/2025 at 16:47 • 0 comments

  So after the long dark teatime of the soul (aka pandemic restrictions) I finally came back to this in the 2nd half of 2024.

  And I did a lot of re-thinking the whole project which lead to quite some changes.

  watchface.

  As notable in the project pic, you don't need 3 separate for the last 3 "digits" after the "V". The one LED for II (2) and one for I (1) will give you 0-3 with just 2 LEDs. After a while you might even note that the pair of XX follows the same rule as they are needed only to light up together, which would result in only 16 LEDs need in total. But you won't save much space since unless you invent a new "double-X" Letter.

  Would I have come sooner to the later idea, I might have used the adafruit neopixel16 ring to get an even smaller 2nd prototype. Here is the comparison of the 3 pixelrings (sadly the 18 has a lot more space between the LEDs, which leaves it bigger than necessary:

  smaller RTC modul

  The DS3231 is available in a SO8 package, but the SO16 seems to be much cheaper and I didn't find a module with a SO8 package. I did find a small one meant to be used as aPiHat. I have a preferred IC for the future, the mircocrystal RV-3032 or RV-3028, but there was no complete modul which was smaller than the PiHat with the DS3231, so I went for the PiHat for the prototype.

  Size comparison of the DS3231 modules:

  Power latching

  I chose a circuit with MosFets this and the ability to have the power-on button double as an input while the circuit is powered on.

  MCU/functions

  adafruit trinket for size reasons, also I miss just one more GPIO.

  In addition to the obvious power on, PWM for the pixelring, there is an LDR for sensing the ambient light and adjust the brightness of the LEDs accordingly. 2 pins for the I2C communication.

  connector/charging

  As there isn't much room to place an USB port sideways, I decided to use a magnetic connector on the back of the case. As there is now no risk of an standard USB plug being used, I placed the charging logic outside the watch. So the 5 PIN mag connector from adafruit works fine. GND, USBpower, LIPoly+ and 2 datalines for usb.

  I was thinking about using one of the serial pads on the backside of the trinket as input to measure the voltage of the battery and all parts are placed, but that requires some adjustment in the general board setup, which I was reluctant to use in the first run.

  Here is a pic of the breadboard setup:

  yellow: charging circuit

  blue: RTC incl. backup battery

  red: pin headers (simulate the future mag connector)

  black circle: LDR

  purple: Trinket

  green: power latch

  cyan: LED ring with laser mask

  black: lipoly battery

  That worked fine (I even tested the voltage measure setup by using the pin of the LDR
  
  

View all 13 project logs