View Single Post
Old 07-17-2015, 05:03 AM   #15
Scotophor
Waiting Room Visitor
 
Scotophor's Avatar
 
Join Date: Aug 2014
Location: La Puente, CA U.S.A.
Posts: 44
Default

Quote:
Originally Posted by feldon30 View Post
I'd say annealing is something to avoid if at all possible. It requires precise temperature control and meticulous cooling while producing flammable gases. It also does not help with clouding, only with crazing which is stress fractures caused by laser-cutting acrylic. Acrylic cut by cnc or a triple diamond blade does not have these problems.
Well, I can only use what I can afford and have access to, and laser cutting is available at two nearby maker spaces. Internal clouding will not be a problem, but surface crazing might because laser cutting introduces stresses which become small cracks (crazing) over time, or when the stressed plastic is weakened such as by a solvent when gluing or vapor-polishing. I've done some research on annealing and there won't be a serious outgassing issue other than perhaps some odor, from a handful of such small parts. Remember these are only the master patterns I'm talking about here, not the final parts. The patterns get used once then discarded (or archived) and shouldn't be needed again, unless the product is so popular that the silicone mold wears out and needs to be re-made. I estimate I should be able to make over 50 handlinks from LEDs cast from one silicone mold. I wasn't expecting to make more than about 10.


Quote:
Originally Posted by feldon30 View Post
I showed him the High Definition screenshot of the Handlink and he pointed out areas that have some clouding and small bubbles due to the use of adhesive. He did not notice any crazing which means that the pieces were not laser-cut. My friend said if you have years of practice and use extreme care, you can minimize clouding and bubbles but there will always be some as the CA, MEK, or other adhesive evaporates and leaves voids behind. He also warned that the clouding/bubbling gets worse over time. If we were to look at the original Handlink prop today (25 years later) the clouding and bubbles would be much more prominent.
I didn't expect that the prop was laser-cut, since it first appeared in 1990, before such technology was widely available. I was planning to minimize bubbling by minimizing gluing acrylic to acrylic, except for the power switch and edge-to-edge between the tiles.


Quote:
Originally Posted by feldon30 View Post
It sounds like you've done all the research and making wise choices between all the various troublesome choices presented to us. My money is on using cnc or similar to cut the handful of clear acrylic sheets needed for the non-LED surfaces, plus the framework, plus the back pieces. Then with a milling machine, make a smooth metal mold to produce two-bay rectangular LEDs. Such a mold would take care of 95% of the LEDs on the face of the Handlink. I'm not sure it's worth the time and expense to make another mold just for the one or two triple-bay LEDs.
If you have access to CNC, then go for it! I was going to make the tiles pretty much by hand using such tools as a bandsaw, bench grinder, and files. I don't have a clue how to make a sharp-edged tiny cavity mold on a milling machine other than by making it in multiple parts. I was planning to make the main frame mostly by hand and eyeball on the mill. There are at least five of the 3-chip LEDs in the props (possibly 7 or 9, though I have already acquired modern 6-chip LEDs to use in those additional positions), so for a run of 10 props I think it would be marginally worthwhile to set up the mold with something like ~15 of the 2-chip LEDs and ~5 of the 3-chip. If I end up with too many of one size toward the end of the run I can just not fill every cavity each time I do a casting.


Quote:
Originally Posted by feldon30 View Post
Let's remember we still have to drill holes to add momentary touch switches (the beads of metal), plus the wiring, plus hiding a circuit board in there somewhere. I'd be tempted to use the low-profile Raspberry Pi model A+. It will run for several hours on four AA batteries or the power cell the original Handlink used, as long as it outputs 5V. Then we'd be able to do sound and lighting patterns without fighting with a PIC or a bunch of 555 timer chips.
The original props didn't fully hide the circuitry. A PCB and something that looks like an EPROM, along with some other unidentifiable components, are visible in at least some of them, especially from the back. Power is likely to be a problem, as I expect the original props probably had rather short battery life. There are approximately 110 LED dice in the props (varies slightly by version). Each prop had some that didn't work either because they were duds to start with, burned out during use, were never connected, or became disconnected. Of those that remained, it seems that on average about half of them were lit at any given time. So figure about 50 LEDs lit at a time on average and probably run at their rated maximum of 30 mA gives an average of 1.5 amps, assuming none of the LEDs are in series. If something like a switching regulator is used to provide a stable supply regardless of battery level, it may be practical to wire the LEDs in series pairs, which would reduce the average LED demand to a more reasonable 750 mA. Then you have to add the load from the control circuitry and sound effects... still quite a lot to pull from a 500 mAh alkaline "J" battery. One factor that may help is brightness matching. Run all the LEDs at their rated maximum current, then decide which ones appear dimmest, and increase the series resistance of the others until they all look the same brightness. This may save some 10-20% or so. Anyway, I expect that I'd be lucky to get 30 minutes of run time from a battery, assuming I don't do something exotic like hide a LiPo cell in the J case (or elsewhere).

Although I'm fairly well-versed in basic analog and digital electronics (self-taught), I never got into any kind of programmable microcontrollers or similar. I was thinking of using a handful of CMOS ICs to set up an 8-bit linear feedback shift register outputting into a 16-bit serial-in, parallel out latching register then into an 8 x 8 matrix, for 64 LED channels with 255 states. Stepping the LFSR by 17 steps then latching for display would give no discernible correlation from one display period to the next. At a fast rate of 30 cycles per second this would give 8.5 seconds before the pattern repeats, but the handlink usually flashed much slower than that. At a sedate 2 Hz rate it would be over 2 minutes between repetitions. I haven't finished studying the episodes for all the handlink light patterns yet but so far it seems like the props used much fewer than 64 channels, so eventually I'll probably be able to simplify this further.
__________________

Last edited by Scotophor; 07-21-2015 at 08:15 PM.
Scotophor is offline   Reply With Quote