Weekend Update 10/4/26

[Maletrain]

Peteski, I don't have the specs for the bi-color LED that I grabbed from my parts cache.  That is why I first used a very high ohm variable resistor in series to see what value resistor to grab for it when using it across my DCC rails.  I just started with high resistance and dialed it down to get the brightness up.  I didn't try to go to LED burnout to see what the minimum resistance could be.  (And that would only apply to my particular DCC voltage source, anyway.)

If you already think that a resistor with a tenth of that 51K ohm resistance would be the best all around value for using the car on DC as well as DCC, I could try that.  But, it is going to be a while before I get around to making the part to change out with what I already have on the car.  I'll post the result here if I get it done and tested with some DC locos at crawl speed.  It just isn't a high priority for me, considering the other things on my to-do list right now.  And, I am not sure how transferrable the conclusion would be for some other LED on somebody else's DC power pack.

BTW, it is the red LED in this particular bi-color (2-wire) part that dies first as voltage increases.  I have several that are now just green LEDs, due to past experiments.

[peteski]

Steve,
I have been dealing with LEDs since mid-70s I have probably installed well over a thousand of them in various projects.  From the very dim red ones 50 years ago to the today's super-bright LEDs. Though-hole and tiny SMDs.  There is nothing scary, mysterious or complex about them. Powering them properly only requires very basic electrical and math knowledge.

Just looking at your LED, it looks like standard T-1 (3mm) size through-hole LED. It is bi-color with each color LED die wired in reverse parallel.  Just a run-of-the-mill red/green LED. From the physical size I can estimate the LEDs can easily withstand 20mA of current, but typical operating current could be 15mA or less.  The Vf (forward voltage) of the red LED is likely around 1.7V and green is about 2V.

For all intensive purposes we can just use 2V for both of them.  That is because we are not powering them by hooking them up directly to some power supply - using a series resistor we are controlling the current.   The 0.3V difference between the Vf each LED develops will result in insignificant difference in the current passing through the LEDs.  LEDs (especially some of the newer blue and white ones) are sensitive to excessive reverse voltage, but in this case each LED's reverse voltage will be clamped but the other LED's forward voltage. Basically the LED that is on automatically protect the dark LED from excessive reverse voltage. Also remember that these are standard LEDs, not the super bright (efficient) LEDs which can operate on 1mA or less.

To calculate the LEDs (maximum) current we need the highest supply voltage (track voltage)  and the desired current we picked and Ohm's law.  For example 10mA should be enough for bright glow.  Easy math:  (Maximum track voltage minus LED forward voltage) divided by the desired current.
Since you liek to be cautious we can use 50V as the highest track voltage. So for 10mA @50V we come up with (50-2)/0.010=4800 Ohms.  5.1k resistor  is even an overkill and no magic smoke will be released. Of course  at lower voltages the current passed through the LED (and it's brightness) will be diminished.  If you were to assume that the highest track voltage is 30V then we would have (30-2)/0.010=2800 Ohms.

Looking at this another way, even if you used a typical 1k resistor then at 12V the current through the LED would be (12-2)/1000=10mA. With that resistor and 30V track voltage the LED would see (30-2)/1000=28mA (still most likely within safe range for an LED), and since that 30V from a cheap DC throttle would likely also be very short peaks at 60 or 120Hz.

You mentioned that powered with DCC the LED glows mostly green. I wonder what happens if you turn the test car around 180 degrees. Is is still mostly green or now it is more red?   The DCC signal is not symmetrical so there might higher percentage of one polarity voltage than the other.

Also, adding more weight would  likely improve contact with the track.

If you really want to get fancy, adding a second resistor and 2 diodes you could control each LEDs current separately. But that is getting to be a bit much for this simple tester.

We should  really have this discussion in a separate thread, not in the WU.

[Maletrain]

Peteski, I understand the math.  What that discussion did not get into was how bright the LEDs would be at DC loco crawl speeds - which was the only real issue with what I have already built.

I do agree that you have shown that a ~5k ohm resistor should be safe for this LED.  So, using one would certainly make the LED glow brighter at whatever reduced DC voltage was found to make a particular analog loco crawl slowly.

Regarding the potential for a non-symmetrical DCC wave to alter the color of this LED based on which direction the car is facing on the track, I have put it on the same track in both orientations and do not see any difference. 

For this batch of LEDs, the green definitely dominates the red.  And that was true for 3 different LEDs from the same pack. It surprised me that the green was so dominant, so I tested a few to make sure it wasn't one of the LEDs that I had previously damaged.  I do have a few where I have killed the red completely with an over-voltage in some DC applications.  But, I put those in a separate container, marked "green only".  I can definitely see a little red dot in these LEDs when they are getting an AC voltage such as DCC.

[peteski]

True, with the typical LEDs, green ones are more efficient and with both driven with the same current the green will emit more light than red. But for two-lead bi-color LEDs I would assume that the manufacturer would match them a bit better.

And as you mentioned, with a fixed value resistor and reduced supply voltage, the LED brightness will be greatly reduced. Of course if the voltage gets close to 2V or  less, the LEDs will go dark.

As an alternative, instead of a resistor you could utilize a constant current diode (aka. current limiting diode)  to limit the LED current independent of voltage, but the diode itself will l waste about a volt across itself.  But from probably around 4 Volts and higher, the LEDs will receive fixed amount of current.  Kato uses such device in their newest version of the passenger lighting kits.

[Maletrain]

I am not familiar with the behavior of current limiting diodes.  Google tells me that they limit current (not voltage) only in one direction.  In the other direction, they pass current rather than block it.

So, wouldn't a bi-color, 2-wire LED need two of them in series with the LED, limiting the currents in opposite directions, so as to limit current in both colors of the LED?  If they are available in fine enough steps of passing voltage, then maybe a pair could be picked to make the two LED color intensities equal?

I don't understand them well enough to know whether putting 2 of them with opposite polarities in series would create twice the voltage drop.

But, because the only real issue here is making the LED(s) light brightly enough at a DC voltage low enough to make a loco crawl slowly enough, I am thinking that putting even an additional 1 volt drop in the LED circuit would be counter-productive.  A 2 volt drop would be worse, obviously.

I'm in the KISS camp on this.  So Peteski's previous suggestion of just using a lower value fixed resistor seems to be the best path "forward" for folks who want to use one of these cars on a DC layout.  a few K ohms would probably work about as well as anything, and it is easy to put together.

[peteski]

The Wiki link I included earlier, mentions that some CLDs  are bidirectional. If you have ones which are not you could connect 2 CLDs in reverse series and they will become non directional CLD, but the voltage drop of a bidirectional CLD would then probably be like 4V across, so it is not a good design trying to have the LEDs lit at lower voltages.

I agree that for this circuit using the KISS principle is the best choice. A simple resistor will work the best.