On Thu, Oct 22, 2015 at 1:37 AM, Mike Ross <tmfdmike at gmail.com> wrote: A few comments on 3D printing for projects like this... About a year ago I created a 'blinkenlights' project for an HP Access system running under SimH (for pic: https://www.flickr.com/photos/mikebikes/22213398140 ). An HP 2000 Access system in real life consists of a CPU and an I/O processor. For this project, I used a 4-core ODROID-C1 running SimH under Linux-- one core dedicated to the main processor, one core for the I/O processor, and one core to manage the lights (and one for Linux to play with). The lights are connected by SPI to Adafruit 24-channel PWM boards daisy chained together. The I/O processor box is empty except for a PWM board and some LEDs. The core running the lights computes the percentage each display bit is '1' for the HP cpu cycles between SPI updates. This becomes a PWM value which is gamma corrected so that the dimming looks correct. I forget the display refresh rate, but with SPI it's in the many hundreds to thousands/sec range. The boxes are printed an a well-tuned and heavily modified Makerbot Replicator 2. Probably each box took 48 hours of printing (that doesn't include equipment problems and design errors!). Because the size of the box exceeds the dimensions of the build plate, each surface needed to be split into (usually) two pieces-- the entire project then screws together using screws and thermoplastic inserts. Not visible in the photo is that there is a black 3D printed part inside where the LEDs mount, creating light tunnels preventing light from escaping to an adjacent LED (and also precisely aligning the LEDs). I was expecting to have to sand and paint the pieces, but while you can see layer lines, it's really not distracting. It turned out well, but only for a 'one-of' project. Doing this on any production scale would not be practical (as others have pointed out). The quality, time to print, and build volume size all cause issues. Since this project I have been moving toward using a combination of my CNC mill and 3D printing. I create positives by 3D modeling (with OnShape or OpenScad) followed by creating a positive using prototyping boards and CNC milling. I cast the positive in silicon and create polyurethane parts. If you haven't seen it yet, an excellent tutorial is the 'Guerrilla guide to CNC machining, mold making, and resin casting' ( http://lcamtuf.coredump.cx/gcnc/ ). You get much better quality with molding and can mix your own colors to get exactly what you want. There are some parts that have very complex geometries where 3D printing is still useful (some of my light tunnels, or switch interfaces match this criteria). For some parts (like PDP-8 and PDP-11 switches) you can get a great result with two part molds-- they look a lot better than a 3D printed switch. I do have a vacuum former and have tried using that for creating things like the front trim for a PDP-8. It's much easier and you get a better result with a mold. If you were trying to get the details of the embossed 'digital' part of a PDP-11/70 trim, that would be really hard (for me anyway) to get a good result from vacuum forming, but pretty straightforward with molding. One more note for original equipment holders-- I don't have, and in most cases have never seen, the device I am creating. It is all done with pictures on the web and whatever technical information I can find. So when you are taking pictures to share, try to take at least some pictures absolutely perpendicular to what you are photographing (including a ruler attached to the object would be an extra bonus!)-- you'll make life easier for us modelers! Thanks.

On Fri, Oct 23, 2015 at 10:26 PM, Alexandre Souza <alexandre.tabajara at gmail.com> wrote: Actually, thinking about it, there are two approaches, and they're both valid: 1. Take a lot of trouble, spend a lot of money, and produce as close a replica of the original as possible. 2. Do what he did with the PiDP-8/I; something 'close enough' which is reasonably accurate and evokes the original very nicely, but doesn't even pretend to be a precise replica. I don't think it even *needs* a bezel IF option 2. is selected. Print a different perspex, add a few more switches, paint them purple instead of brown, add a couple of rotary switches... put it in a wooden box again. Job done very simply. But whether it's 1. or 2. I don't see any strong argument for NOT making it full-size. But we can all talk about it; he's the poor bugger who actually has to do the work :-) Mike http://www.corestore.org 'No greater love hath a man than he lay down his life for his brother. Not for millions, not for glory, not for fame. For one person, in the dark, where no one will ever know or see.'