Tawe TMD blog is live

This is a blog for Tawe TMD thingummybobbins.  Also, all my SLA 3D Printer Project posts will reside here from now on.

SLA 3D Printer Log 1: Projection, Projection...

In between dreaming of a house where model trains bring me pre-mixed Robinson's squash and keenly observing the cathartic tidal wave of backlash against Dungeon Keeper Mobile I have received a 720p projector from Ebay, costing £250.  Before launching into the specifics of how to use it, let's examine the unit itself.

The book is A4 width (and 2cm shorter lengthways) for comparison, and the unit weighs 1.13kg

The unit is actually 206 x 162 x 75mm (lxwxh), feels like a quality implementation despite it's low weight and is highly intuitive to operate; this isn't surprising when we learn that this model (Nobo X22PW) is simply a rebranded Optoma TW330, which is going for £410 at time of writing.  That is excellent value for money given that most projectors in that range will be 1024x768 native at most, not 1280x800 native (keep an eye out for native resolution, most projector sellers claim their model will "handle" 1080p, but this is done through upscaling and will make your movie look anything but sharp).

*Review Interlude*

The unit is surprisingly quiet in spite of the fan having to keep a 2200 lumen lamp from exploding (not a hint of facetious here), and that lamp gives plenty of brightness even on std-mode with 1700 lumens.  It is audible, but it quickly fades away during gaming since the noise has no irritating frequency (that goes out the window in bright mode, I suggest just turning up the brightness in windows).  Having played Medal of Honor: Airborne on it I can say that it eco-mode might not be to everyone's taste but with lamps costing around £90 I think most can sacrifice 29% brightness for 66% more lamp life.  This is also a Single-Chip DLP model so it suffers from a little colour separation during fast motion (highly noticeable when you shake your head viewing a black background); finally I would like a better contrast ratio than 2000:1 for more defined extremes in the colours, but I'm rapidly disappearing down a rabbit hole here.

*End Review Interlude*

Back to the point at hand, 1280x800 will allow for a detail level of 0.1mm per pixel for a build area of 12.8cm x 8.0cm; I'm looking for a minimum resolution of 0.05mm seeing as I design for FUD on Shapeways (0.1mm Resolution), giving a build area of 6.4cm x 4.0cm.  One of my ideas up to this point was to traverse the projector along a vertical rail, utilizing a redundant RAMPS axis to adjust the build area/resolution balance, but I forgot to account for projector focus; what's worse is that the minimum distance I could get the projector to focus was around 28cm, which gives an image of 197.5mm x 124mm - well above the 64mm x 40mm required.

Despite this, the brightness of the resultant screen area was highly encouraging.

In order to attain this build area, I need to add an adjustable lens with at least 3.0x magnification.  DSLR lenses can be had used for around £50, given that a UV filter is a common accessory it may be that the lens can transmit UV without hindrance, but apparently DSLRs have UV filters of their own and there may be some in the lens too - an optical testbed will be required to fully investigate the matter, as well as some actual 3D resins (no VOCs, please).

Free Ideas Bucket 1.0

I intend this project to be completely open-source, that means any and all STLs and diagrams related to the build will be available royalty-free and not on my Shapeways Shop; if *you* would like to sell it, go ahead but be prepared for a hard time making money given that the models are free and Shapeways can be used by a Bonobo.  Anyhoo, I present my current list of ideas:

• A standalone built-in computer, either a Raspberry Pi, it's more powerful compatibles or even a full-on x86 PC.  This is to allow the printer to run without having to keep either my PC or Laptop running with it all night.
• An ultrasound sensor to detect the resin level in the bath and adjust the build platform starting point to suit, this is only relevant to top-down designs.
• A touchscreen on top of the machine, adjustable to either face outwards (for desk use) or upwards (for floor use).
• Gullwing doors...  just because...

That concludes the first of hopefully many buildlogs to the SLA 3D Printer Project.  The ultimate aim is to largely free me from having to use Shapeways for every.  single.  solitary.  step.  in the R&D Process for Tawe_TMD.  It's always the way when you make something that you trip up on things you couldn't have predicted, but even if I flounder and fail, I can always watch Zero Punctuation across my wall.

Sirens of the 1080p screen, and the prints of reason.

As far as temptation goes, getting a new PC is up there with crisps positioned but a few feet away after 3 hours of CAD.  I say temptation because my current aim (and New Year's Resolution mind you) is to be able to subsist upon my own business alone - not to have to rely on a job to pay my bills.  I have already made inroads towards this with my Shapeways Shop Tawe_TMD but I've hit a snag:  Research and Development.

I am currently developing my first proper product, which is a Metropolitan Ex-Twin Brake Van from the turn of the century in N gauge (1:148); the body is pretty much sorted out, pending any minor alterations for functionality.  The chassis also pretty much looks like the final iteration, so the bulk of the work is now done leaving me with the R&D associated with functionality - and boy was I in for a ride!

Each one of those strips on the side is 0.1mm high by 0.25mm wide.

In the UK it takes around 3 weeks from order to delivery, in the latter stages of R&D (tweaking) this holds true even for minor changes like increasing the distance between the axle holes; this is doubly frustrating when a mistake is made, such as an item being wrongly printed, this being the direct inspiration for this post and the potential project I will get to in a moment.  Shapeways solved this issue with haste and care and I applaud them for such attention, however, I consider it a call to action for a proper solution.

That solution is my very own SLA (Photo Lithography) 3D Printer; this isn't as difficult as it sounds, because according to many Youtube videos and This Instructable, it is entirely possible to print a 3D object using nothing more than a Z-carriage, a beaker of resin and a DLP Projector.  Feature sizes can be minute, too:  a 1280 x 800 DLP projector (£250 from Ebay) can result in objects up to 12.8 x 8 cm in size with a resolution of 0.1mm in the X and Y; you can even get higher resolutions simply by reducing the distance between the Projector and the resin (e.g:  6.4 x 4 cm with and XY res of 0.05mm).

I have made a not-terribly-professional case study into this idea and found some interesting facets to the idea of running a 3D printer like this:

I was never noted for relishing writing novels...

For the record, here's some additional information for those looking to check all this out (I did this on many assumptions):

• 1 Metro Brake Van = 2656 mm³
• I set a budget of £200 for materials, this includes minor electronic items like stepper motors, and any 3D prints for special parts.
• I don't know what I will coat the build platform with, but it will need to hold things during printing whilst peeling off when done.
• I assume Council Tax to be included with the Rent.
• Food + Taxes is broken down thus:

1. Food = £30/week or £120/month
2. Income Tax = £38.12/week or £165.19/month from here (assuming gross earnings of £15,000 a year)
3. The rest is for other things like electricity, gas, water which I can't be stuffed to calculate, but the remaining £164.81 should cover it

• As for printing, I am assuming a perfect run of setting up supports (SLA prints need copious support material, maybe not in the case of the Metro Brake but I'll apportion it anyway), a successful print owing to a reliable non-stick application (that's what the Sylgard 184 is for - It stops the print from adhering to the resin basin in bottom-up designs), and finally a flawless post-processing run (fat chance...).

It is quite clear to me that running this printer to print Metro Brake Vans to sell is not a viable proposition - even with sufficient resolution, the time spent in preparation will ensure absolutely no free time, an unacceptable situation when this business is intended to help me achieve financial freedom.  3 hours per print sounds low compared to 3 weeks, but remember that Shapeways is taking on all the work in those weeks, and I am free to make new designs while my Metro Brake goes on sale.  This printer makes much more sense for Functional Testing, i.e:  making minor changes around the coupling area or testing strengthening ribs.

To any aspiring Shapies, I say this:  If you have passionate pride in your work, be prepared to test and test and test - go so far as to intentionally break your redundant testing models to reveal weaknesses.  On the note of final presentation, don't be like many of the modellers on this virtual superstore of human creativity and only upload the STL with some token renders; this really shouts "That'll do, next!", and is that the attitude you would tolerate from a carpenter?  An engineer?  A NEST team?  One of the best ways of showing that you have done such testing is to upload a Youtube video that will show it running.

I hope that PC temptation doesn't get the better of me...

Canals and Curious Carriages

Progress is slowly being made in Tawe_TMD - Bearing in mind that I am on a Student's income so leaps and bounds are not to be expected, at least not initially.  I have decided after many pilot runs uploading models of bunkers to my Shapeways Page, I have decided to get one step closer to my ultimate aim of locomotive kits and begin work on a Railway Carriage in N gauge:

Creo Elements/Direct is like Pro/ENGINEER without a price tag :)

This is a Metropolitan Line 8 Wheel Non-Bogie Carriage, originally working the aforementioned line on the fledgling London Underground until 1909 when they were sold to the Burry Port and Gwendraeth Valley Railway for it's own passenger service.  These had eight wheels with a total wheelbase of 28 feet or 8.53 metres which given the sharp curves of a line built on the route of an old canal (Wales' oldest one, no less) would have presented a problem if it were not for it's chassis; the inner 2 axles were fixed to the frames, but the outer 2 axles were mounted on radial trucks with their pivots above the inner axles, similar to locomotive practice.  Incidentally, these carriages had brakes on the inner wheels only as a direct result of this.

The model is far from complete, with side details missing and a lack of a chassis (this will be a seperate unit to allow fitment of an interior and to allow flexibility for future conversion to 2mm finescale); also the roof is a bit crooked at present so I will need to address that.  Other than that, progress has been good with me picking up Creo nearly effortlessly due to muscle memory and experience with Pro/E and I think we could have a physical model to show off by mid-June, early-July.

Hope you enjoyed that  :)

Sources:

1. The Burry Port and Gwendreath Valley Railway and its Antecedent Canals Volume Two:  The Railway and Dock
2. Wordpress:  The Basilica Fields blog

Hunting the Zero: 123D

Hello, and welcome to the chronicle of my first attempt to model anything seriously using Autodesk 123D, done while Bioshock Infinite is installing (ty James!).

123D has proven to be a steeper learning curve than expected - The model I chose features several complex prisms in it's construction, something that extrusions cannot effectively do in this package.  Then again, you need to select a reasonably challenging model to really test something like this. 

After what initially felt like coming adrift from that lovely £1m yacht called Solidworks in the Caribbean Sea, I have had some success modelling the bicycle rack outside my block of flats at Swansea University:

This is what the new age of model building looks like :)

This is just the part where the bikes are locked to, the roof is to follow.


I'll try to describe the method I use to get my curves correct:  First, have a picture handy of the side of your subject; mine wasn't straight so some perspective issues did happen, but the important thing is to measure at least part of the subject and use it to draw up a grid like so:

The base was 1000mm in length (thick black line), so I drew up a 3x3 grid using it in the indispensable Paint.NET

Set a zoom level for all the grid to be shown simultaneously (in my case 66%), and measure the grid line on your screen (73.5mm on my 1080p screen).  From this you can obtain measurements relative to the grid or otherwise by measuring the on screen subject, dividing by the measurement per-grid-line (as luck would have it for me, 1 metre irl = 73.5mm on screen); this will give you it's irl length, an approximation perhaps, but suitable for modelling.

If this still confuses you, I will give you a pictorial guide has to how I made a single support for the roof of the bike rack.

Tutorial:  How to obtain measurements using a photo and one measurement

We start by obtaining a length in whichever axis, x or y, is most suitable; for me this is a vertical component so I use the y axis coming from the baseplate up to where it joins the roof strut:

The lime-green box is the measurement tool, and the support in question is bounded by the box

Measuring on my screen, I get 134mm, so 134/73.5 = 1.823m = 1823.13mm irl.  I should mention at this time that I am modelling to O gauge, or 1:43 scale, so dividing by 43 gives a value of 42.40mm for use in my model:

Now we need to measure the distance from the bottom-left of the lime-green box to the centre of the support, since it does not start there.  Again, measurements on the screen give me 2.5mm, so 2.5/73.5 = 0.034m = 34.01mm irl; again this is divided by 43 to give 0.79mm.

Finally, we measure the line from the top-left of the lime-green...  you know what, I think you know the drill by now :)

Ready to receive arc...

Luckily, I know that the arc the support follows begins perpendicular to the support so I need not measure a middle point for a 3-point arc, I simply adjust my aim until the start of the arc is at a right angle to the grid.

Now we delete the right angled Construction lines to leave our arc; since the support has a circular cross-section of 0.93mm radius, we need to draw that and orient it to be perpendicular to the arc, use the "Sweep" Tool under construct to do this.

Final Verdict:  The MS Paint of 3d CAD

I eventually failed at making the roof of the bike shed, it could have been perspective issues in my photography, but I think the major reason is that precision is god-damn hard in this package.  In Solidworks, you have access to 3 planes right off the bat, that is to say you can start working vertically or horizontally both ways; additionally, solidworks creates a flow-chart of all your moves similar to Photoshop or Paint.NET, so error correction is a cinch - neither of these things are present in 123D, you have to model vertically and orient it when you are done, plus it really drags when you have to begin again all because you failed to measure accurately, causing systematic error which is something which should be easily correctable given how it has a cascading and predictable effect on your work.

So I feel my comparison to MS Paint is very apt:  it's fine if you just want to save a screen-cap from a program to post on facebook, just like 123D is fine for making a custom phone casing or a simple functional item you wish existed; but you try to paint "Fireworks at Chilcompton" with MS paint or make a scale model with 123D for that matter and things will get messy very quickly.

In closing, for anyone wanting to 'sperge out on your models, like I do, get Solidworks.  It will not leave you "Hunting the Zero".

Autodesk 123D: First Impressions

I dread using AutoCAD.

A 3d CAD suite I disuse so much I had to look up it's name despite it being installed on the workstations at Swansea University (not to mention stipulated in many engineering assignments); AutoCAD is probably not a bad program, but nothing in it's UI seems intuitive to me and neither did the UI of 3ds max for that matter.  My memory is expectedly hazy, but it seems to centre around this 3d grid as is standard fare; the problem arises when I expect to be able to draw a rectangle or other polygon and simply extrude it and I simply could not find the function which did that (If I remember it right I pretty much flunked that part of one assignment practically relying on tutorials).  Given the damnation I have bestowed upon AutoCAD, imagine my surprise when I discover a simplified CAD suite made by the same people (Autodesk) that not only is intuitive, but seems to be tied in with my darling emerging technology:

The most laid-back, pants-around-the-knees 3d modelling I ever did see...

Meet Autodesk 123D, a barebones 3d CAD program which is designed to give ordinary humans (you) and greenhorn ingrates of engineering (me) access the world of 3d printing by giving a super-streamlined but fully-capable interface.  It functions pretty much how you would expect - you draw a polygon, you extrude it and BAM! you have a prism in the shape of the polygon.  Now, I realise the you probably want to print something less vapid than a prism, once again, it has you covered; simply select the face you wish to extend the additional feature off of, then extrude (that means draw it out, btw).

Okay, that is as far as I got, but I genuinely believe I will grow to love this program almost as much as SketchUp (almost!); maybe after a while I may miss Solidworks features like being able to instantly calculate a part's volume and even mass once the materials are specified, and you can also run a simulation to see how it will fare under load (certain to make my PC explode!).  I am seriously considering inroads into 3d Printing for my projects and even perhaps as a source of income, everything about this technology makes me squeal in delight considering that now I don't even need a lathe to make rudimentary CO2 engines and other super-complex things.

Until next time...