JNSF DIY lighting project (PCB based)

[gavino200]

I received a few 100mF Tantalum capacitors in the mail. They have the same footprint as the 220uF ones I used for my Orient express but the height is quite a bit less. They may be thin enough to fit under the roof of a train car without being visible or obstructing closure of the shell. I'll try them out when I receive my single tear LED strips. 

 

 

 

[gavino200]

There's two interesting things in this interior lighting video. First, he uses an odd shaped capacitor. It's shaped like a squashed cylinder. Like one of those things used in the sport of curling. The second interesting thing is that the guy doesn't solder his cap directly to the board. Rather he has a pair of solder pads and keeps the cap on a pair of wires to be tucked away and hidden. This would certainly be a lot easier than trying to hide a cap that is soldered in series with the rest of the components.

 

 

[gavino200]

On 3/8/2019 at 7:21 PM, cteno4 said:

This is the constant current circuit I’ve been referring to and we’ve discussed a lot here in a number of lighting threads. It’s really a nice approach

 

I've been thinking about this example. His circuit is indeed well thought out. And they guy does some excellent work (both craftwork and presentation). The drawback is the way his has soldered the components together at the end of the car. It's pretty much the way I did it with my Orient Express. The problem is that it's slow and delicate work. The downstream part of the circuit is nice an simple. Upstream is a headache.

 

I'm very inspired by his car interior detailing.

[gavino200]

A very basic setup from Viessman. It's basically an LED strip with a rectifier, a cylindrical cap and a couple of resistors. It looks like their cap is downstream though the circuit could loop around. 

 

Just adding for ideas and inspriration.

 

 

[gavino200]

No circuit details here. But a testimony to successful results with 100uF

 

 

[cteno4]

2 hours ago, gavino200 said:

very basic setup from Viessman. It's basically an LED strip with a rectifier, a cylindrical cap and a couple of resistors. It looks like their cap is downstream though the circuit could loop around. 

 

Cant see it clearly looks like resistor for each led, a rectifier, and I think Zener diode for voltage regulation as it looks like a couple of extra resistors.

 

you want the cap downstream of the rectifier to keep the polarity constant for the cap and led. Also good to be down stream of the voltage regulator to use lower voltage caps.

 

the other board also looks to use Zener diodes.

 

the curling stone is a low profile electrolytic cap. Basically short and squat.

 

The tantalium caps have to increase in size as the capacitance increases to fit in more plates for charge storage so after a point capacitance is proportional to size.

 

the constant current could be done just as well on a pcb like the voltage regulator. Also capacitor is in same location down stream of the rectifier like the voltage regulator. 

 

The idea of putting the cap on wires was my thought with the constant current circuit with tantalium or compact electrolytic to better hide. problem with the constant current method is using 25v caps and more expensive and slightly larger chip.

 

jeff

[gavino200]

36 minutes ago, cteno4 said:

 

you want the cap downstream of the rectifier to keep the polarity constant for the cap and led.

 

I think we have a consensus on this. 

 

36 minutes ago, cteno4 said:

 

The idea of putting the cap on wires was my thought with the constant current circuit with tantalium or compact electrolytic to better hide.

 

Ah, I misunderstood you. I agree with this in principle. In practice It might be difficult to get all the upstream components on the rails without creating a dark zone at the end of the car. 

 

The beauty of the commercial designs and potentially a custom pcb would be in looping the circuit back down the near end of the strip to get an LED in the middle of the dark components. Thereby avoiding a dark zone. It would be possible to jerry rig that with an wired LED but that's the end of it being simple and elegant. This being said all opinions must eventually bow to a good test. I think I'll order some of those 'bars' to experiment with.

 

[Sheffie]

Don’t worry too much about a possible dark spot, until you actually see one. Human perception can mask out quite a variation in light level, and with a little bit of diffusion and internal reflection, the variation may not be too severe. Something as simple as masking a dark circuit board with light paper or tape could make quite a difference 

[gavino200]

44 minutes ago, Sheffie said:

Don’t worry too much about a possible dark spot, until you actually see one. Human perception can mask out quite a variation in light level, and with a little bit of diffusion and internal reflection, the variation may not be too severe. Something as simple as masking a dark circuit board with light paper or tape could make quite a difference 

 

That's an interesting thought. I agree that all this stuff needs to be tested as perception is what matters.

 

I'm not thinking a bout a spot being dark due to lack of refection off components that tend to be dark colored. I did used the term "dark components". That was a vague term that I (mis)coined. I wanted to refer to all the components that were not LEDs without listing them. 

 

What I'm referring to is when a car doesn't have enough LEDs for even light. Like when a 6 LED strip is used to illuminate a car that really needs 7.  I experienced this with my Hello Kitty install. It is in fact quite noticeable.  If possible it would be better to avoid.  The problem can be avoided at the far end by using an LED strip that can be torn by single units. The near end remains problematic.

Edited March 10, 2019 by gavino200

[cteno4]

I think this is why the contsant current guys liked flipping their leds facing up and bouncing them off the white paper in the roof to even the light as the leds facing down will give distincter light and dark areas.

 

jeff

[gavino200]

27 minutes ago, cteno4 said:

I think this is why the contsant current guys liked flipping their leds facing up and bouncing them off the white paper in the roof to even the light as the leds facing down will give distincter light and dark areas.

 

It's worth experimenting with. I'll buy some of the rods specified in the video and run a test. 

 

Another thing about the "parallel rails" method (please give another name if you have a better one) is that it will likely be possible to put this together using a 100uF cap and still be able to close the shell (not possible with a 22uF cap). If that is the case then the voltage regulator may not be necessary. Therefore the space occupied by "dark components" will be smaller. 

 

I ordered a bunch of loose LEDs, voltage regulators, and caps. I'll buy some of the metal bars used in the video, and then make a test. 

 

I can try:

 

1. No Volt reg - 100 cap - LEDs facing up - with painted room.

2. No Volt reg - 100 cap - LEDs facing down

3. Volt reg - various capacitors 

4. Dark components on rails soldered to 'single unit rip' LED strips.

 

I don't think 1206 caps are practical without a pcb but I can make experimental prototypes to experiment with.

 

I could use an old US passenger train to experiment with.

 

Any other ideas for tests?

Edited March 10, 2019 by gavino200

[gavino200]

I ordered some .5mm OD straight brass wire/rods. 

 

As I see it, regarding the "Brass rail method" (I'm naming it that unless anyone objects), the two main issues to investigate are:

 

1. Can all components be placed on the rails while still allowing closure of the roof/shell.

 

2. Can all components be placed on the rails without perceptibly effecting light diffusion. 

 

I guess a third issue is to see how "easy" it really is and how long it actually takes to do.

 

I'll also experiment to see how small a "bundle of components" squished circuit (like the one featured in the video) can be, and how long it takes to make. 

[cteno4]

Without the voltage regulator you will need to increase your main current resistor to drop your rectifier output down to the like 0.5-1ma x number of leds (I think you liked the that ma range for coach lighting)

 

again its better to have the caps before the resistor as once the track contact is lost the resistor is not part of the circuit and cap discharge current won’t be moderated by the resistor and and give you faster discharge at higher amps — not the nice slow fade out with the resistor between the cap and the leds to regulate the current and cover the momentary track power loss more gracefully. You discovered this in one of your earlier tests.

 

Inside the resistor you will still need to use the 25v caps still as the resistor does not effect the charged voltage on the capacitor, just the charge rate. Without the voltage regulator you will be giving it your full 12v (well your dcc output minus the small voltage drop from the rectifier), hence 25v caps needed. With the caps outside the resistor you will of course still need 25v caps.

 

of Course this circuit with just rectifier — cap — resistor — parallel leds won’t give you even lighting if using dc, it will be like old style of going from dim to bright as more voltage is applied.

 

jeff

[cteno4]

The rail bit is just for the leds really. Might look at the tiny dark bundle places at the very end of a car attached to rails (with just leds on them) with a short set of very thin wires (you are only putting like 5ma on the wire). Then you can attach the led rail to the roof to set its bounce distance from the paper for maximum effect.

 

i guess if you put the capacitor soldered across the rectifier output ans then the resistor soldered off that to one contact of the cap and then to one of the rails. then the other rail soldered to the other leg of the rectifier. That like or rectifier-cap-smd resistor would probably be about 12-15mm long at the far end of a car and then have the first led right next to it. Of course this chain could be on a small pcb that the rails solder to or little buss wire leads solder to to go to the rails on the roof.

 

you have fighting deisgn, space, assembly time/cost and engineering variables here. Pcbs make the whole thing simpler to pop in but facing down you need more leds to give even lighting. Bounce lighting means getting rid of pcb to hold the leds... I’m also interested in making mine dc and dcc compatible as it will be a long time before I go dcc so if I do lighting I want it to first work with dc and then handle dcc if and when the conversion happens... have to keep trying to find the best combo, none will be perfect for all variables but something should give the best acceptable.

 

cheers

 

jeff

[gavino200]

9 minutes ago, cteno4 said:

Without the voltage regulator you will need to increase your main current resistor to drop your rectifier output down to the like 0.5-1ma x number of leds (I think you liked the that ma range for coach lighting)

 

Yes, more juice needs more squeeze. Regarding resistance I just start at the safe lower limit and increase until it looks right. I may wire up a car with externalized wires to investigate this with a potentiometer. Then measure and use the nearest available resistor. 

 

I'm looking for a variety pack of resistors in the useful range (100 - 10,000) . All the variety packs I find are from 1 ohm to 1Mega ohm.

 

 

9 minutes ago, cteno4 said:

 

again its better to have the caps before the resistor as once the track contact is lost the resistor is not part of the circuit and cap discharge current won’t be moderated by the resistor and and give you faster discharge at higher amps — not the nice slow fade out with the resistor between the cap and the leds to regulate the current and cover the momentary track power loss more gracefully. You discovered this in one of your earlier tests.

 

Yes. We can now consider this to be JNS dogma. We all agree. This never needs to be discussed again. Anyone violating this principle should be smote down mercilessly.

 

9 minutes ago, cteno4 said:

 

Inside the resistor you will still need to use the 25v caps still as the resistor does not effect the charged voltage on the capacitor, just the charge rate. Without the voltage regulator you will be giving it your full 12v (well your dcc output minus the small voltage drop from the rectifier), hence 25v caps needed. With the caps outside the resistor you will of course still need 25v caps.

 

Yes. The 25V 100uF bricks just may be small enough. If so, they are the easiest solution. If they don't fit and the rolled/compact/squishd circuit isn't practical, then no dice.

 

The Rule of Thumb (Cap voltage = circuit voltage x2) is now also JNS dogma. Same rules as above.

 

9 minutes ago, cteno4 said:

 

of Course this circuit with just rectifier — cap — resistor — parallel leds won’t give you even lighting if using dc, it will be like old style of going from dim to bright as more voltage is applied.

 

 

I have no DC trains except for my little tram line.

[gavino200]

19 minutes ago, cteno4 said:

The rail bit is just for the leds really. Might look at the tiny dark bundle places at the very end of a car attached to rails (with just leds on them) with a short set of very thin wires (you are only putting like 5ma on the wire). Then you can attach the led rail to the roof to set its bounce distance from the paper for maximum effect.

 

I've done this already. It works but it's a PITA. 50 minutes per car. At least. And that was using stick-on, pre-soldered LED strips. This would be the default backstop is something more efficient can't be developed.

 

 

Quote

 

i guess if you put the capacitor soldered across the rectifier output ans then the resistor soldered off that to one contact of the cap and then to one of the rails. then the other rail soldered to the other leg of the rectifier. That like or rectifier-cap-smd resistor would probably be about 12-15mm long at the far end of a car and then have the first led right next to it. Of course this chain could be on a small pcb that the rails solder to or little buss wire leads solder to to go to the rails on the roof.

 

Zackly!

 

Quote

 

you have fighting deisgn, space, assembly time/cost and engineering variables here.

 

Zackly!

 

Quote

 

Pcbs make the whole thing simpler to pop in but facing down you need more leds to give even lighting. Bounce lighting means getting rid of pcb to hold the leds...

 

 

Yes, I'm very interested in trying the "facing up" brass rail method. It may be superior. Just need to see how efficient the process is. On the other hand since I'm now thinking about detailing the car interiors, I'm not sure I can still talk about efficiency without much hypocrisy.

 

Quote

 

I’m also interested in making mine dc and dcc compatible as it will be a long time before I go dcc so if I do lighting I want it to first work with dc and then handle dcc if and when the conversion happens... have to keep trying to find the best combo, none will be perfect for all variables but something should give the best acceptable.

 

 

So for DC you need the constant current chip, like in the video? To avoid the lights brightening with train speed? It looks like all the same issues are relevant, though the circuit would be slightly different. 

 

I may also want to light my trams at some stage, so I guess I'm still interested in this branch of the algorithm.

Edited March 10, 2019 by gavino200

[cteno4]

For dc you need the constant current or voltage regulator to deal with variable voltage and not have the dim to bright issues. Does mean you need at least like 4.5-5v min for the circuit to kick in due to the forward voltage of the led and the voltage drop of the rectifier and regulator.

 

you could do a mini pcb for the dark components or just lay them on tape sticky side up and solder together. Same with the led rails and either solder together or with buss wire separators. 

 

Jeff

[chadbag]

I am still learning this software, so the schematic may look a bit jumbled (until I can figure out how to rotate things, if possible, etc).  I have not annotated or assigned values to anything yet though I will to simulate it -- you will be able to use multiple values at each spot depending on the circuit you want (and this is not the PCB layout, just the electronic schematic).

 

There is a rectifier first then a voltage reguiator with small input and output caps (as listed in the spec sheets --- you will be able to leave this out if you want).  Then a small regulating resistor on the cap input, a series of caps (the first will be either a tantalum or a ceramic while the rest or ceramic if you don't use tantalum and want to stack the ceramic).  Then the main resistor (the 2.2k in Gavins experiment) then up to 10 LEDs.

 

If running off a decoder you don't need the rectifier but otherwise (DC or DCC) i think you need the rectifier.

 

Schematic-v0.1.pdf

[chadbag]

Here is an updated one with annotations and better input

 

Schematic-v0.1.1.pdf

[chadbag]

This one is layed out better.  Functionally identical except deleted a spurious wiring

 

Schematic-v0.1.3.pdf

 

  

 

 

Edited March 11, 2019 by chadbag

[cteno4]

Nice!

 

jeff

[chadbag]

I am playing around with the PCB part.  I don't know what I am doing :).  I have the parts layed out but I am sure there are some issues with it as I am still working through that part of the tutorial.

 

[chadbag]

Reversed the direction of the LEDs (ie, i accidentally counted down instead of up but was laying them physically up on the PCB).  Also got rid of some spurious wiring that you may not see on the schematic.

 

Schematic-v0.1.4.pdf

[maihama eki]

At the company I work for, we are pretty much no longer allowed to use tantalum electrolytic capacitors in new designs because of reliability issues, and we only buy from top-tier suppliers.

 

There are newer ceramic capacitors with nearly as good volume to capacitance+voltage characteristics.

 

The added benefit is that the ceramic capacitors can be connected either way - they don't care.  Tantalum electrolytic capacitors can react very negatively (exploding) to being biased backwards when enough current is available.

[chadbag]

50 minutes ago, maihama eki said:

At the company I work for, we are pretty much no longer allowed to use tantalum electrolytic capacitors in new designs because of reliability issues, and we only buy from top-tier suppliers.

 

There are newer ceramic capacitors with nearly as good volume to capacitance+voltage characteristics.

 

The added benefit is that the ceramic capacitors can be connected either way - they don't care.  Tantalum electrolytic capacitors can react very negatively (exploding) to being biased backwards when enough current is available.

 

I think most users of this won't use tantalum.  It is set up to allow 1 instead of  row of ceramic but most probably won't I am guessing.  (The first capacitor pad will be large enough to handle either).