LEDs in Showcase Miniatures Signals

[peteski]

Hmm. If it matters, the 'G' in the 0404 RGB chip I'm using (Rohm) is already 527nm. That's signal green, no blue necessary. That's why I thought the back-to-back R/G 0402 chips instead would be a reasonable solution. :shouldershrug:

The R/G solution (assuming you can get a good yellow hue) will minimize the wiring, but the driving circuitry will be more complex than one needed for R/Y/G LED.

[C855B]

Well, yes and no. The back-to-back R/G pair is still supported by some off-the-shelf signal drivers such as the Digitrax SE8C. And it's easy enough to do with a PIC if you were needing something fed by discrete one-per-color logic.

I'm thinking that some of the magic of yellow is more in strobing the primary colors to fool the eye rather than hoping to get a solid color mix in something translucent. The trick may be in frequency and duration. Time to do a bit of research.

EDIT: Quick surfing resulted in discussions of "convergence". A RGB source like a monitor produces a perceived wide gamut because the individual color pixels are too small to be seen as separate elements. The RGB chips or separate R/G pairs are large enough for each color source to be perceived separately. We can try to blur the sources with translucent conduits and artificially inducing convergence, but at some angles the separate colors remain perceptible. This was always my problem with the old-school R/G bi-polar LEDs.

If I hit a wall on the R-Y-G thing I'm working on right now, maybe I can resurrect the signal driver test rig I built around a BASIC II. It had a nice PWM function, so hooking that up to the original test signal might be good for tweaking into a yellow.

[peteski]

Well, yes and no. The back-to-back R/G pair is still supported by some off-the-shelf signal drivers such as the Digitrax SE8C. And it's easy enough to do with a PIC if you were needing something fed by discrete one-per-color logic.

I'm thinking that some of the magic of yellow is more in strobing the primary colors to fool the eye rather than hoping to get a solid color mix in something translucent. The trick may be in frequency and duration. Time to do a bit of research.

EDIT: Quick surfing resulted in discussions of "convergence". A RGB source like a monitor produces a perceived wide gamut because the individual color pixels are too small to be seen as separate elements. The RGB chips or separate R/G pairs are large enough for each color source to be perceived separately. We can try to blur the sources with translucent conduits and artificially inducing convergence, but at some angles the separate colors remain perceptible. This was always my problem with the old-school R/G bi-polar LEDs.

If I hit a wall on the R-Y-G thing I'm working on right now, maybe I can resurrect the signal driver test rig I built around a BASIC II. It had a nice PWM function, so hooking that up to the original test signal might be good for tweaking into a yellow.

You got it. If the eye can see each color die as a separate object, you will see green and red glow, not yellow.  But that is why a translucent diffuser lens is important (to blend the green/red light together).  Strobing will not make any difference if you can still see the individual LEDs.  What is important is the proper balance of brightness between the red and green LED to obtain the proper shade of yellow.  The brightness can be controlled by limiting current to a specific value (using resistors) or even by applying PWM voltage to the LEDs.  If using the resistors, the problem is that the values of red/green current-limiting resistors which will result in a proper hue of yellow might not be optimal for straight red or green light.  So you might need additional resistors being switched into or out of the red and green driver circuit.

Either way (resistors or PWM), the driver circuitry will be more complex than when using a RYG LED.

Oh BTW, Kingbright just came out with a line of LEDs in 0201 SMD packaging!  Assuming that you can solder leads to those fleas!
http://www.kingbright-europe.de/kingbright-develops-the-smallest-smd-led-package-2/

[C855B]

0201? Do I perceive a challenge?   My main concern would not be the soldering, but that the mere act of breathing would send them skittering off the microscope stage into oblivion. I already am fighting the 0402s wanting to stick to a couple of tweezers as I maneuver them into the CA, and a mis-timed exhale blew away a couple of 0404s. I'm using the Kingbright Minikin-Air 0402s. They're something like half the thickness of regular 0402s, so there's not much mass holding them down.

No white yet with the 0201s, however. White was also a year behind on the Kingbright 0402s, so I'm not surprised. Anyway, I am looking at - seriously - lighted switchstand lanterns, using the smallest Ngineering tubing for the post and sending wires or a fiber optic filament up to a Shapeways-fabricated lantern. Something in 0201 would sway the decision towards a point source at the top rather than 1,000,000 candlepower under the benchwork pushing its way up an .010" fiber.

[C855B]

Studying the specs on those 0201s, we can't quite dismiss the possibility of hand-soldering. At least not yet. Dimensionally they're not really 0201 - more like 0301.5. So, yes, they're smaller than 0402, but not 25% the size, more like 50%. Basically the 0201s are the same chips as the Kingbright Minikins, just without the wings for the solder pads.

So after messing with the 0404 RGBs and finding it easier to deal with the 0402s because of their larger lands, I am wanting to try the new 0201s.

[Sokramiketes]

Are you actually planning to use the blue LED in the mix?  Or just green and red?  The only possible use I can see for blue is to get a better green hue.

Yes, with green and lunar indications.

[C855B]

A program note - the 0404 RGB chip I used is now "end-of-life", that is, the manufacturer (Rohm) replaced it with an "improved" model, SMLP34RGB2W3. The new chip's ratings are about twice as bright (yikes!), with each color rated for 20mA as opposed to 5mA. I'm sure that running the new chip at 5mA would result in the equivalent brightness. Incidentally, 20mA is right at the limit of current capacity for #38 wire - light more than one element at 20mA and you'll fry the common lead.

I've momentarily back-burnered this project. Milling the head out for three 0402 chips was not working (read: "excruciating"). At this point anticipating samples of the 0102 chips and also waiting for Showcase's developments is the agenda.

[ryan_wilkerson]

Thanks for the update Mike. The info on the max for the wire is helpful.

[sp org div]

Cool thread
Nice to see someone else working on a similar project
On the layout here, we re-cast the old Sunrise signal parts in pewter (thanks Paul!), before Showcase came out with their masterpieces, but have continued installing bi-color 0603's in our cast heads.  I know all of us out here are waiting for the prime 3 color SMD sporting the correct colors, but the layout here needs to keep moving forward at some point.  The solution for us in the interim has been partly in both finding LEDs that have the correct red and green, but also that have the internal dies as close together as possible, as this aids in mitigating the "angle of view aspect scenario", and significantly aids in creating the correct color mix for the Approach indication.  Resistors are used to get the proper mix, and all LEDs are driven pretty hard near their .020ma max for needed brightness.  Drilling through the bottom of the head accommodates an entrance for the wires and LED installation.  The other part of making the color mix work fairly well is accomplished by drilling a scale lens size hole from the front, to meet with the hole that was drilled from the bottom for the LED.... then installing weed trimmer line.  The trimmer line is a great diffuser to finish blending the color mix. 
Good times attaching the wires to the 0603 SMDs, and I cant imagine the 0201's, but fortunately the 0603's fit well.  I can't solder the wires without laying them on masking tape, and pre-tinning with liquid flux.  Coats of ACC strengthens and insulates for placement into the head, held by Epoxy.  Twist the enamel wires together and coat with liquid tape to simulate the flex conduit drip-loop out the bottom and into the signal pole... this also helps protect the wires where they enter the sharp edges of the hole(s) drilled in the pole, as well as mitigate sharp bends where coming out the bottom of the head.  .010" black styrene seals the bottom entry hole.
We also horded a bunch of red / lunars for the long term project, as we all know this stuff is limited duration runs, so get it now, or miss it altogether!
Been posting less than regular over on the SP Org Div blog about the CTC / Signaling project...  I think March 2014 was when we started to address our latest target signal adventure, nearing the bottom of page at: 
http://espeeoregondivision.blogspot.com/p/ctc-signaling-project-archive.html
The layout is almost too large for me with limited help, so hard to follow through on many a project, but fortunately this one is still alive.... at a severely reduced rate for now.
Volunteers welcome.   
Keep the ideas coming, as it has been a collection of ideas that got the signaling project here as far as it has...
Thanks guys!

Jeff
http://espeeoregondivision.blogspot.com/

[Cajonpassfan]

Here are the results, a finished signal:

[img]http://www.everywherewest.com/SM_SA_red.jpg[/img

Yummy!
Following with great interest
Otto K.

[C855B]

An update of sorts, Otto.

Kingbright has just released a new micro LED, http://www.kingbright.com/content/detail/product/28/12. This is slightly different because it is a "right angle" version, the advantage to us being that it is very, very thin (0.2mm, or about 0.008"). This is thin enough that a three-LED "sandwich" might be possible with the Showcase head. Once it becomes available at Mouser I will order a batch and give 'er a go.

EDIT: Never mind. It turns out these right-angle chips have been available for about a year, but aren't stocked by anybody in the U.S. I could go to the U.K. to get some, but it turns out to be pointless - they don't have the correct green (525nm) in this chip configuration. Back to the drawing board.