Hi

I need a sanity check from the electonic experts...

Topic 1.

I have long wished for LED bulbs that can be screwed into the standard Märklin light bulb sockets and now Adarsh is making them.
See http://www.trainaidsa.com
That is great. The problem I foresee, however, is that while each light has its own resistor, there is no room for a diode to prevent damage due to the reverse current that is present if the light is supplied with AC.

In a digital loco it is not a problem because lights can be supplied with DC current from the decoder, one just needs to ensure that the polarity of the light is correct.

In non digital situations however, (such as street lamps or ananlog locos) there will be 16V running both ways through the LED. I believe that typical maximum reverse voltage for LEDs is 5 Volts.

The LED can probably withstand 16V reverse current for short periods so my question is this:

Are the periods when the 5V reverse current is being exceeded short enough to not destroy such LEDs?

I also suspect that even if the LEDs manage to survive the 50 or 60 Hz reverse current, this environment will drastically shorten their life span.

The solution of course is to insert a diode into the circuit to prevent the reverse current but that is not always possible, nor obvious to people if they think they can simply replace the old light bulb with the new one.

Tim Eckert, who is selling them on eBay claims that since an LED is a type of diode, no diode is needed but my understanding is that LEDs have a very much lower reverse current rating than sat a 1N4148 or 1N4001.

Opinions ?


Topic 2

Adarsh is also selling some very nice looking LED strips for interior lighting. I have tested them and they look very promising. The strips have multiple sets of 3 LEDs, each set containing a 130 Ohm series resistor. Using my LED circuit formuli, this tells me that such strips should not be supplied with a voltage greated than 13V as that would put more than 30mA through the LEDs. Current measurements confirm that at about 12V about 23mA is passing through the LEDs and I therefore belive that these strips have been designed for 12V usage.

Tim claims to have tested the strips at 20 Volts. I calculate that at 20V the LEDs will have 84mA passing through them. LEDs typically have a maximum forward current of 30mA.

Am I crazy or will these things not last long with supply voltages above 13V ? (OK I probably am crazy but still...)

Thanks.

the LEDs on the strips are surface mounted but I don't know the manufacturer so don't know what data sheet to look for. I'll ask Adarsh is he can get that info.

The resistors are also surface mounted so difficult to replace.

For station lighting they are probably ideal as I have 12V Dc available, but I think for interior lighting of wagons a 12V regulator may be in order...

If two or three LEDs can be used, they can simply be mounted parallell in different directions.
/Lars

Please note that with 16 V AC the diode will get a maximum voltage of 16 * sqrt(2) = 22.6 V (can easily reach 24 V if the transformer load is low).

I'd use them with AC only if the producer specifies them for AC operation.

EDIT: looked at the homepage - the resistor was calculated for 12 V DC, but they claim they work with 12 through 18 V DC ...

Using a rectifier as recommended will secure the LED from reverse voltage, but will also supply twice as much forward current.

Silicon semiconductors get damaged if the silicon gets too hot.
I think I'd use 7812 voltage regulators to be on the safe side.

Wouldn't it be the easiest solution to use a DC power source for all lighting? Even if not all conventional bulbs are yet exchanged with the new LEDs, they won't mind running on DC instead of AC, would they? Using a DC control transformer even delivers the additional charm of a dimmer...

Regards,
Arthur

To reduce heat you could use 7815 or even 7818 and put more LEDs in series (you couldn't use their strips for that configuration).

Will there be one 7812 per car (no heat problem) or per train (potential heat problem)?

There's an 0.1 A version of the 78xx series with a TO-92 casing (transistor size) - not sure if you can hide it in the toilet ...

Excellent suggestion Hans Martin!

Dale:
Note that you'll still get "flicker", but it won't be as noticeable - especially if the LED's share the same light diffuser.

The TO-92 versions of the regulators are usually called 78Lxx.

Also note that there are plenty of linear regulators in tiny surface-mount packages (eg. SOT-23, or the even smaller SC-70) but they need careful mounting. Flying leads, a blob of non-acidic silicone glue and some heatshrink is very effective.

About input rectifiers:
Indeed most linear regulators don't like negative inputs, but since the Märklin "AC" signal is chopped ±18Vdc, a simple bridge rectifier with a very small capacitor would provide a good enough source for most regulators.
Actually, I just re-read the data sheet for the LT1121 series, and noticed it is available in TO-92. This regulator tolerates negative input voltages (but obviously won't regulate them). You could use a LT1121CZ-5 with no input bridge rectifier or capacitor at all, and get a tolerable 5Vdc output.
You'll still need at least a 330nF output capacitor, but they're tiny now-days. I'd try and fit at least 1µF. Just solder one (or more) across the legs as you have done before.
Linear Technologies have an on-line store (using credit cards), so you could order LT1121CZ-5 (5V, commercial temperature range) or LT1121IZ-5 (5V, industrial temperature range), or LT1121CZ-3.3/LT1121IZ-3.3 for the 3.3V versions.

Good as I already suggested it. But your figure makes it clearer I admit.

It's 22 V, not 18 V. That's what you get if you rectify 16 V AC.
Digital current is rectangular AC, no sine wave. Voltage will not increase if rectified again.

If you see flicker doesn't that mean that the frequency is below 25 Hz? I thought the human eye was incapable of detecting frequencies as high as 50 Hz...?

It's more than 22 V with a 16 V AC transformer and "old" equipment like CU6021 or Delta Control (4f).
It might be 18 V with a MS and an 18 V DC transformer, but should be more with an 16 V AC transformer.
I wouldn't design anything for 18 V (no matter wot your 'scope 'll show) - imagine you'll upgrade to an CS or an 1.9 A MS with a 60 VA transformer and all your "maintenance free" LEDs burn out ...

Correct.

He uses a 63 V capacitor - that'll do even with blue transformers in analog operation!
A 25 V capacitor should be sufficient for digital operation only (but could explode in analog operation) - but it's smaller (easier to hide).

Edit: if you're using the 78L12, a smaller capacitor will do.
If you're making a "digital only" version, a much smaller capacitor will do (rectangular voltage, no sine wave) - if you have clean tracks!
AFAIK there's a rule of thumb: 1 mF for 1 A (this rule is for 50 Hz and sine wave).
Capacity C = (Current * Period) / voltage drop [As/V]
I think the 7812 needs at least 15 V input. With 22 V track voltage there are 7 V "reserve".
With 0.1 seconds, you get: 0.06 A * 0.1 s / 7 V = 860 µF
With 0.01 seconds, you get: 0.06 A * 0.01 s / 7 V = 86 µF

470 µF should hold the voltage less than 1/100th second.
470 µF should hold the voltage for 5/100th seconds.
Not sure how much you'll need to prevent flickering with dirt on the track ...

Too big can be harmful, too: when you turn on the power, all capacitors will be charged at the same time (this can give your booster a hard time).
Some folks use 10,000 or even 22,000 µF per loco - a handful of 470 µF shouldn't do much harm.

Doh! Dale, if I realised earlier you were looking for a power supply circuit of this nature I could have posted it. I'm using a 12v version of this circuit to power some 12v bud lights on my station platform. I think you can pretty much replace the 7805 with a 7812, or 7815 to get 12v, 15v, etc. The other components can stay the same - I've got 470uf caps in my circuit.

I'm running my circuit off a standard 40va Marklin lighting transformer.

Remember to heatsink the 7805, otherwise current draw is very limited due to thermal shutdown - although I managed to run 4 incandescent bulbs before mine shut down (I didn't measure the current draw though) - but I would guess that for small current draw of 60ma you won't need a heatsink. Try it and see, if the current draw gets too large the regulator will just shutdown. Twas what happened to me...

I think the regulators are rated for 1 amp, correct me if I'm wrong.

Don't know if you can get pigtail capacitors in this size but that may be worth a try - smaller profile.

this is a fantastic group effort !

I think the next step would be for him to look for the smallest sized components and do some burn tests... and if the 7812 gets too hot then add a heat sink. Hopefully he can use a small heat sink and small caps - some coaches don't have much space to hide things.

I will suggest that he runs a number of his light strips at a time so he can check the temperature build up at higher loads.

I just realised that his power strips have one 20mA circuit for every 5cm of strip. This means one may end up with 5 x 20mA = 100mA in a long passenger coach. This suggests to me that heat may be an issue.

An alternative for powering the LED strips is a switch-mode module (aka DC-DC convertor)
This company makes some very small units Recom. Other companies do too, but these are about smallest I've seen.
They may not be small enough to fit in a passenger wagon, though.

Regardless, the advantages are that this unit will provide 500mA at 12V from 22V input, with no heatsink. Go the the Products->Innoline page, and look at the R-78xx-0.5 range.

I just found an even better one!
They used to be Datel, then Newport, then became C&D Technologies, and now have been bought by Murata. Is that clear? Good...
7812SR-C is smaller than the Recom unit.
The "H" package is particularly useful for inside a passenger wagon.

Buying small quantities may be a problem...I'll hunt around.

Hi Damon,

that 7812SR-C only replaces the 78L12 right ? It can't take negative input so the rectifier and smoothing cap would still be needed. I understand it solves any heatsink issue though.

I suspect that the size it starting to approach a small loco function decoder ! (which produces DC output)

Yes, if done as intended, but I guess we can do what we want with them, right?
If a coach is in need of 4 strips, I would chain 2 together in one end, and the other two in the other end. And then I would connect those 2x2 modules in series at the middle.
It may take a bit of soldering, but surely not as much as connecting a regulator.

Dale, yes I did measure the MS output -- ±18.3V or so.
However, after doing some research I realised I was guilty of overestimating the sophistication of Märklin's controllers and boosters. [:I]
I assumed the output was regulated -- instead, the controllers/boosters just chop the unregulated supply, so indeed with a 16Vac supply, the track voltage would be around ±22.5V.
These are the perils of being an electronic engineer, and assuming the same things that matter to you matter to someone else. In my design work, every mA costs, and I routinely specify voltage references to the mv, and power supply voltages to ±1%...
We're not in Kansas anymore...

Anyway, back to the question at hand...

I had a thought on multiple LED strips. You only need one 12V regulator per strip (with small output capacitor). Each strip needs 23mA, so even at 23V input (i.e. 16Vac, rectified), total power dissipation in each linear regulator would be 0.25W
There is no shortage of surface mount or through-hole linear regulators in this range.
You only need one bridge rectifier and filter capacitor per coach, which would be easier to hide than several. You then connect each LED strip with integral linear regulator in parallel.

eg. Maxim MAX1616 or Texas Instruments TPS71501
These are both "adjustable" types, and hence need 2 resistors, and an output capacitor, but may be a possible solution.

To make surface mount manageable for non-mass production, you could use the small adapter boards like Aries, that take an SOT23-5 (or similar) and provide a small pinned package.

update, I measured rectified track voltage on my layout and got 27 Volts !