N scale searchlight signal build thread

[peteski]

I've been watching this build thread - very interesting and informative.  It also explained why you guys (mike and you) have difficulties stuffing the wires down the mast:  those LED leads are relatively heavy-gauge magnet wire.  While I don't know what gauge this wire is, I have magnet wire which is probably half the diameter of this wire.  At least what I can see from these photos. The LEDs use somewhere between 1-5mA - there is no need for such "heavy gauge" wire.

[GaryHinshaw]

I agree that the wire is heavier than necessary for these LEDs (I believe it's 36 ga.), but that is what RR-CirKits ships.  That said, the thin-walled tubing makes threading a snap, and the leads are quite robust to handling.  If people have suggestions for other 3-colour LEDs (preferably pre-wired) that render a proper signal green and a footprint that is 1.2 mm wide x 1.6 mm tall (or smaller) please post them here so people have options to consider.  I'll stick with these since I have a full stock on hand, and they're working for me.

Now, where do we get to the bouncing roundel part of what makes searchlights have souls?

I forgot to mention this earlier, but it's worth noting that the Signalman board has an "H2 Flicker" effect programmed into the firmware that (somewhat) mimics the bouncing roundel.  I confess that I'm not familiar enough with the prototype effect to comment on how effective the software rendition is, but you might want to look into it.   (I think I remember you commenting that you were using RR-CirKits drivers?)  Here is a screen shot of one of the mast panes in the jmri programming interface:



This example shows that, if the aspect to be displayed is yellow (Lamp A: "H1-Y/Steady") or green (Lamp A: "H1-G/Steady"), set lamp B to red (Lamp B: "H1-R/Steady).  The LED will then briefly transition to the desired aspect via red.  It's a decent effect, but definitely not a mechanical roundel...

[GaryHinshaw]

Step 4: assembling the mast and preparing for paint
With functional tests complete, it's time to assemble the remaining components to the mast: in my case the ladder and the signal heads.  This is a finicky process that requires care to get everything properly positioned.   Just be patient.

I attach the ladder first so it's not dangling around while I'm attaching the heads.  The gluing order I follow is: 1. bottom to signal base, to place the ladder at the right height, 2. top to the mast, 3. brackets to the appropriate ladder rungs, 4. brackets to the mast, making sure they are level.  I use extra thin CA glue for these steps, which makes very strong bonds.  The one exception is that I'll use a dab of gap-filling CA to initially attach the brackets to the ladder rungs, since I can get a bit of working time to make sure they're properly attached and properly aligned.  I'll go back over the joint with thin CA at the end, for extra strength.

For the heads, I first attach the head to the bracket with a gap-filling CA, then when it is properly aligned on the bracket, I'll dab the joint with thin CA for extra strength.  After this step, these joints are so strong that I can't break the bond without breaking a part.  The last step is to secure the bracket to the mast.  The one thing to check here is the clocking angle of the searchlight relative to the base.  On Tehachapi, most of the signals are aimed off axis because of the extensive track curvature.  In this shot, you can see how the heads are aimed ~10 degrees off straight:



Here is what things look like at this stage:



(I'm never 100% satisfied that I have everything properly lined up, and it seems to depend on what angle I look at the signal from.  But these imperfections usually fade once everything is finished and mounted on the layout, and better is the enemy of good enough.)

The only remaining parts to attach are the target plates and brackets, but I paint the signal first before doing that.  So the next step is to clean the signal with alcohol or other solvent, and prime the stainless parts (mast and ladder) with Tamiya metal primer thinned about 30% with lacquer thinner.  Once that is dry, painting can commence.

[parkrrrr]

  If people have suggestions for other 3-colour LEDs (preferably pre-wired) that render a proper signal green and a footprint that is 1.2 mm wide x 1.6 mm tall (or smaller) please post them here so people have options to consider.  I'll stick with these since I have a full stock on hand, and they're working for me.

Not prewired, but these seem to meet the other requirements. The only possible caveat is that these have a lens size of 1.6 x 1.1, but are listed as metric 1616 devices, so the pads might stick out too far to the sides.

[C855B]

Not prewired, but these seem to meet the other requirements. The only possible caveat is that these have a lens size of 1.6 x 1.1, but are listed as metric 1616 devices, so the pads might stick out too far to the sides.

Yeah, these have a 1616 footprint. Could possibly be the chip Showcase Miniatures supplies, which is wired for them by Richmond Controls. If so, it would fit in the H-5 head, but not the H-2 Gary is showing at the moment.

[Sokramiketes]

  (I think I remember you commenting that you were using RR-CirKits drivers?)  Here is a screen shot of one of the mast panes in the jmri programming interface:



This example shows that, if the aspect to be displayed is yellow (Lamp A: "H1-Y/Steady") or green (Lamp A: "H1-G/Steady"), set lamp B to red (Lamp B: "H1-R/Steady).  The LED will then briefly transition to the desired aspect via red.  It's a decent effect, but definitely not a mechanical roundel...

Cool!  I was using RR-Cirkits stuff for the CN training layout at their Homewood facility.  But on Modutrak we have a microprocessor piggybacked onto a TracTronics signal driver circuit.  Bob is our ex-Motorola engineer and I let him loose with the circuit designs.  Here's a video of the effect:

Not a valid vimeo URL

[wazzou]

Your craftsmanship is exquisite Gary.  Every aspect looks very clean and believable.

[railnerd]

just sayin' this should be nominated for a "best of" thread.

[GaryHinshaw]

Thanks for the props Bryan and Dave.  It means a lot coming from guys who have produced models I really admire, e.g., this and this.

Nice video Mike.  I think your effects are better than the RR-CirKits effects.  Their transition from green <-> yellow is pretty effective, but the bounce to red is less so.  But it's possible that I haven't programmed that case properly yet.  I'll explore some more and eventually try posting a video.  Fun stuff!   Just out of curiosity, does anyone have a pointer to a good video of the prototype in operation?

[C855B]

... does anyone have a pointer to a good video of the prototype in operation?

Not mine, but I've used this vid from a signal collector as a reference for several years:

/>
It's a GRS Type SA, essentially the same thing as the US&S H-2.

[GaryHinshaw]

Bingo.  I knew you'd come through for me.  Now we need sound too!   

[Sokramiketes]

Just so everyone understands what is going on, the swinging vane has three lenses that change position in front of a single light. 

The order of the lenses is red in the middle, with yellow and green to either side.  Without power, the mechanism is weighted to default to red for safety reasons.  It takes a powered move to display yellow to one side, and green to the other. 

A move from, say, green to red, is from an energized state to a non-energized state.  Power is pulled off the solenoid, and the roundel recenters.  Since it is a weighted pivot motion, it gives a little flicker as the lenses passes over center a bit, dimming, then brightening again.  Depending on how tuned up the mechanism is, it could be more or less bouncing. 

Same from yellow to red.  It was pulled to the side, then released and flickers to red.

Now, to upgrade from yellow to green, you have to pass through red.  Its an energized to energized move, all the way across, but you get that flash to red in between the yellow and green aspects.

(Yes, using red/yellow/green instead of stop/approach/clear for model railroad clarity here)

This is why I love searchlights so much.  They have a soul, they have motion and life.  And they draw engineers out of the crowd at trainshows when they flicker, with an "Oh damn, was that animated?"

[Sokramiketes]

You can see the weights and pivot on the bottom of this mechanism.

[ Guests cannot view attachments ]

[C855B]

Adding a little to Mike's comments...

The vane holding the roundels (colored lenses) is basically a miniature semaphore moving in front of a bulb. It is moved by coils that work like an old-fashioned voltmeter - centered when off, DC to swing to one side, reverse polarity to swing to the other side. The clunk you hear in the video is the vane hitting the stops. This is precisely why they are being replaced en masse as signal systems are being upgraded, they are electromechanical which means they're not 100.000% reliable, and therefore require periodic inspection and maintenance.

Aside from Y-R-G roundels, a somewhat common configuration is Y-R-lunar, typically in the bottom head of a multi-head installation, especially around interlockings. I have also seen lunar-R-lunar.

Oddest arrangement I've encountered was orange-blue-orange used for tracks and service bays around the diesel shop at SP's Taylor Yard in Los Angeles. Blue obviously for normal servicing "do-not-touch!" restrictions, orange apparently the choice for "not blue" versus running indications like Y or G. Several years ago somebody on eBay had the incredibly rare blue and orange roundels listed, but I was out-sniped in the bidding war.

[GaryHinshaw]

Great info guys.  Thx.

Step 5: painting the signal and preparing the target plate

With the signal cleaned and primed, it's ready for painting.  The bulk of the signal is brush painted with Vallejo Model Air aluminium (71.062): I use a Micro Brush for the stainless steel parts and a small flat brush for the rest of the parts.  This paint goes on very smoothly and covers remarkably well given how thin it is.  I mostly get away with 1 coat coverage, but I'll do a second coat, or a touch up coat, as necessary.  One items that need several coats, to avoid light leaks, is the printed resin head.  I usually end up going with 4-5 coats, checking between every 2nd coat to see how the light leaks are doing.  Here is an example on a dwarf signal that is still in progress:



This head will need at least one more coat on the sides, and probably two on the visor (which hasn't been painted at all yet, as it will be painted black). 

The target plate is the "face" of the signal, so it should be prepared with care.  The parts come in an etched fret:



I only use the top two parts.  Both of these usually need to be reamed out a bit with a file to fit properly on the signal head.  Before removing the top support bracket from the fret, I check that it will clear the stop ring on the head and ream it out a bit, if necessary.  Here is a fit check showing that this one clears the stop ring:



Do the same for the target plate, making sure is just reaches the stop ring.  Here is a before and after comparison.  Before:


After:


I use a round tapered file for this.  It doesn't take much time, just a lot of fit checking:


Next, bend the bracket to shape and glue it to the back of the target plate.  I use a simple jig to get these parts aligned, consisting of a tapered bamboo toothpick to which I have wrapped one layer of Tamiya masking tape (for reasons shown below):


Since the bracket ring has a larger diameter, it can fit around the tape layer, which keeps it concentric with the hole in the target plate (which is fitting snug around the toothpick):

(The purple dot is a Sharpie mark that shows where the sight hole in the target plate it.  The bracket should be perpendicular to the position of the sight hole.)  I glue the bracket with a dab of thin CA, then clean the assembly with alcohol.  Next, I prime the back of the target with Metal Primer and set it aside to dry:


Next up is painting the target and visor, and finish sealing light leaks.
-gfh