Replacing Incandescent with LED in Passenger Cars

It'll be interesting how long the LED's last.  They must be very bright will 18 volts in command operation on them!

Great tutorial just when I need one for my heavyweights.

John what do I need to tone down the ac voltage going to the lights to make them dim?

The rms value of the half wave rectified 18vac is vpeak/2 or about 12.1vrms. Not so far off. I would put a diode in series with the LED string to protect the LEDs from the high reverse voltage.

Right, the reverse voltage of around 14-25 volts exceeds the ratings, even in series for the three diode strings. 

I add four components in a small clump to provide DCS compatibility, rectify and filter the power, and provide current limiting.  I find that 18" passenger cars are plenty bright with 20ma of total power feeding the lights.  I've lit them up with full power to the strips to see what it looked like, and to me it looked like the car was on fire!

I made a clump of a diode in series, 100uF/35v in parallel to com, and a CL2. How many sets of 3Leds do you string in a car after the clump.

Be sure to use full wave rectification or you will create a DC offset on the track. 

I also add a 22uh choke for DCS compatibility.

Jon or someone, can you please explain the meaning of this statement. I'm still trying to learn the electrical aspects within a project like this and any further explanation would be helpful. I've used gunrunnerjohn's methodology in lighting my passenger cars and thus far am very happy with the results.

Another question, what is the exact purpose of the choke? I've seen the reference to the DCS signal but still don't quite understand the effects. Would I need the choke if I only ran TMCC, Legacy and conventional?  Thanks!

Actually many people sense 60 cycle flicker but do not perceive what it is.

That is why many are so tired after a day in the office with old CRT computer monitors.

The newer ones refresh at higher rates, old ones caused eyestrain.

And some of us can actually catch it as a flicker in the peripheral vision...

But, this is a second argument for the Bridge, as it doubles the flicker frequency.

It also ensures that you load the sine wave evenly on both poles.

Also, adding a cap in line in either case reduces the flicker a lot, or completely.

Now, in a Model Train, does it matter? Not so much. We aren't sitting in the train for hours.

OGD;     I don't know about rectifiers, etc. but I do know that your cars look great.

This is a very IMPRESSIVE lighting conversion. If one operates numerous cars with LED lights (or multiple DCS or Legacy trains on a single track) with this type of lighting upgrade, will there be a power drain? Do the LEDs operate under 18 volts of constant track voltage?

Excellent job! You have inspired me to tackle this project in the near future. Thanks for sharing.

My eyes don't see the 60hz flicker, probably a benefit of getting older. .  You can also alternate the polarity of the connections to the LED lighting DC for half the cars in a set, which would accomplish the same thing as the full wave bridge.  Since my cars only use 20ma of current, and I'm strictly running command, I've never noticed an issue, but it is a valid point.  It's no big deal to put in a bridge if desired.

Thank you for your replies. Anybody care to comment on the purpose of the choke?

I assume the dcs signal travels on the center rail and the choke blocks the lighting circuit from loading down the signal.

Here is my "clump", designed with the help of our electronics gurus:

And here is a MTH Superliner sightseer lounge with the LED's powered off the above circuit. Both levels are lit, three segments on the lower and eight on the upper. I tapped the power rails for the lower level light bar to supply current to the circuit. That way I could reuse the stock light bars, and the LED's shine through yet remain recessed into the ceiling. 

---PCJ

There was a lengthy thread on this a few months ago.  The subject was something like "Lights sucked all the voltage out of my transformer" or something like that.  Railride's unit is similar to what I came up with, except I mount it on perforated board.  The earlier thread contains links to some invaluable reference sources on building voltage regulator (which I use) or current regulator (which gunrunnerjohn uses) circuits.  They are cheap.  I recommend you check that thread.

Like RJR, I also mounted RailRide's circuit on perfboard. I didn't even attempt to use that surface mount resistor, instead I opted for a standard one of the same value.



This is a fun and rewarding project, but can be challenging at times, depending on the car being retrofitted.

I guess I could have left out that choke since I don't use DCS. Oh well.

Here's how I mount them in many cars, after connection I use some hot glue to stick them to the roof.  I currently use a 220uf capacitor for better flicker resistance, I never see any flicker with that combo.

GRJ, when I posted some pictures on the buy-sell forum today, I noticed in the adjoining station the tape LEDs like you use are sagging.  Apparently I lot a lot with bad adhesive.  It's hard to fit my fingers in there to insert double-sided tape.

So far I haven't had any come loose.  I do make multiple passes over them trying to make sure they're solidly stuck to the roof.  I also clean the area with alcohol before putting the strips on to remove any oils of any kind that would cause an issue.

The diode gives you rectified DC current, that's it's lot in life.

Hi Folks.  Having read some of the highly technical replies to this post, which are beyond my understanding, could one of you have a look below at the specifications of the lights I'm using and tell me if any of this information addresses some of the caveats that have been mentioned?  I'm hoping that the lights I'm using may already have some of the details mentioned as required to avoid problems like a '...DC offset on the track...".  I'd love to know what this means, in plain English.  I would not take offense to dumbing it down to a kindergarten level (or even lower).  

Specifications:

● Color: Warm White
● Protection Rate IP33(Non-waterproof)
● Output power: 20-24W /5 Meter
● Working Voltage: 12VDC
● View angle:120° ● LED Quantity: 300 leds/5 Meters
● Working Current/meter: 0.35-0.4A
● Working Temperature:-20 to 50°
● Luminous Flux: 250-300 Lumens/Meter
● Size: L500cm (5M) x W0.8cm x T0.25cm

One more question to those of you who watched my video:  What bad things are going to happen if I keep using the simple method I've employed thus far of basically powering the lights directly from the pick up rollers?  (no diodes, no rectifiers, no resistors, etc.)?

The lights are rates to last 50,000 hours.  So far they are not too bright from the 16-17 volts my track measures at.  But I don't want to cause any damage to my engines if the method I'm using puts me at risk for that.  

Thanks in advance.  The great thing about this forum is the knowledge transfer that takes place from those of you with expertise in areas like this.  Much appreciated!

The lights are only rated at 50,000 hours if you don't abuse them.

I'd probably add a diode to protect against reverse voltage.  Also, since you're running them at higher current than most folks, you should stagger the connections to the roller and wheels so that you're not creating a large DC offset on the tracks.  This may bite you in other ways.

When I run them at 20ma, there isn't enough DC offset to cause any issues, but you're obviously doing a lot more than that.

Did you try measuring the power draw of one car?

GRJ, I can understand DC offset if you're using half-wave rectification, as with a single diode,* but wouldn 't use of a full wave rectifier eliminate that as an issue, since the in AC circuit the same current is flowing on each leg of a cycle?

* Explanation:  the current downstream of the diode is reflected in the current upstream, more current is flowing upstream during the half-cycle that the diode is conducting than during the other half-cycle, when no current is passing.

The full wave would eliminate the DC offset, but the average value of the DC current would increase.  I suspect he's already pushing those LED's pretty hard, driving them harder is probably not a good idea.

Before I went any farther, I'd be wanting to measure the actual current being currently drawn by the lighting in the car and seeing if I'm in the danger zone.

GRJ, I think that you and I agree that there should be a current- (you) or voltage- (me) limiting circuit.  It's just a matter of time before his components start failing.

I would agree that half-wave rectification is easier on the LEDs, but the peak voltage is still there, probably about 24 volts if fed by a 18 VAC circuit.  If he's using resistors to drop voltage, as components start failing open, the voltage across the remaining components will increase due to lower IR drop across the resistors with the reduced amperage.

Yes, LED's are diodes. Yes, they only allow current to pass in one direction. But they're not as good at it as the non-light-emitting variety. Their life as a light-emitting device is thus shortened if you continue hitting them with  reverse voltage all (well, half) the time they're lit.

The associated electronics packages like the one I and others pictured maximise their lifespan by feeding them straight DC (via rectifier) that is smoothed out (thanks to the capacitor), and limited to the ideal amount of current (the last module in the chain, either the LM317 or the CL2), since an LED's brightness is determined more by the amount of current fed to them rather than plain voltage.

---PCJ

The interesting thing about LED's is that it's all about average current, not peak current.  You can have peak currents many times the full 20ma rating of the LED if the duty cycle is low enough.

A typical data sheet, 25ma forward current, but 140ma peak forward current.  It's very standard to mux LED indicators with much more than the average current at a low duty cycle, this allows you to drive a lot more indicators with a lot less circuitry.

Correct, that's what my digital 'scope is for.

So, GRJ, what is the actual ratio of meter read vs. peak?

I had to STOP reading.  I felt my head beginning to fracture from within.

I suspect that depends on the meter, I'm computing now and the 'scope is in the closet.

I'm still confused.

To recap...

The fellow is using 18vac track power to power up 12vdc strings of LEDs.

Since the LEDs are actually diodes, they only get power on one half of the 18vac waveform.

That means that the effective voltage they are getting is only about 12v (called the RMS value of the half wave waveform). And that's what the LED string likes.

However, LEDs are not tolerant to a high reverse voltage, the other half of the wave when the LEDs are not conducting so it's recommended that he put a rectifier diode in series with the 18vac to withstand that.

Then Jon pointed out that if you load your track voltage with too many half wave loads, that could put DC on the track which you don't want.

Then John pointed out that LEDs are such a small load, that it doesn't matter. But he half wave rectifies one side of the 18vac in one car and the other side in the next car to balance the loads on the track power. You do that by putting the anode of the rectifier toward the track power in one car and the cathode toward the track power in the next car and appropriately connecting the LED string so that it lights up.

Then there was something about the LEDs withstanding the high peak current but this was discounted when the data sheet for LEDs revealed that was not a problem.

Then there was some other chatter about using a rectifier bridge instead of just one diode to rectify both sides of the 18vac thus balancing the load that way. However that raises the effective voltage back to 18v and needs some current limiting and while you're at it, throw in a capacitor to solve the lighting blinking going around the layout.

Also a comment was made about how a volt meter doesn't measure AC volts the same for all kinds of wave forms like it does for a pure sine wave...unless you spend several hundred dollars for it or look at it with a fairly expensive (relatively) oscilloscope.

And to answer the last question about ratio of meter reading to peak...it's around 1 to 2.

So why are you confused?

Is that all there is?  Why is blood now coming out of my eyes, ears and nose after reading this?  DANGER, DANGER WILL ROBINSON!

Cjack, the solution is to use a voltage limiting circuit.  They are cheap and easy to fabricate.  There was a discussion of these and current limitibg circuits in a thread a few months ago.  The title had words like " ...sucked all the voltage out of my transformer."

I made up a bunch at a few minutes each and install one in each passenger car or station.

This might be the solution!

see johns post above.

What color should I use for 1950s streamlined cars? Warm white at 3100k or Neural White at 4000k?   The warm white looks better for me. More subdued. Neutral looks more like a more modern car, but I'm not an expert.