Bridge Rectifier For LEDs

Estimating "how hot" the LM317 gets is a tedious calculation.  There have been some OGR threads that have drilled deeper into the math.  For your stated application of driving a 12V LED strip in a caboose or passenger car I suggest setting board's output current to 15 mA using an 82 ohm resistor.  The LM317 case will be the hottest spot on the board but you should be able to touch it without burning your finger.  Mounting the LM317 to double-sided foam-like tape will help as the tape will draw heat from the LM317 case much more efficiently that ambient air.  

As an aside, for the same 15 mA output current, if you instead drive a single 3V LED (such as for a headlight) the LM317 will get hotter...and might be pushing the envelope but that's a separate discussion.

As long as the LM317 is of the SOT-223 package style, it will work.  When you say "1 Amp" LM317 note that this board is intended for low-power applications...such as the 0.015 Amps suggested earlier.  You will have serious heat problems  operating this tiny board at 0.1 Amps or whatever.

Note that the 100uF capacitor might be marginal for flicker rejection.  I believe GRJ's assembled board uses a 330uF - in other words 3x larger.  Your mileage may vary.  You might consider buying an assortment of lower cost thru-hole radial capacitors and soldering the leads to the SMD pads.  This might provide additional mounting flexibility.  Note that this bare-board can serve a variety of applications - including non-rolling-stock in which case a very small capacitor could be used so the entire board can fit in the cavity under Realtrax or Fastrack.

The MDB10SV package has 4mm lead spacing.  The other two part number you show are ~5mm and ~2.5mm.

FWIW, my use of the LM317T in my lighting module allows me to light the 12V strips in any size car with up to 45ma without overheating the regulator.  If you crank it up to 45ma, the lights are pretty bright, and the regulator gets up to about 60-65C, but it still works just fine.  I typically run 18" passenger cars at around 25ma for the 12V strip for nice intensity to the lights.

1A and 0.5A LM317 in the same package (both SOT-223) will run the same temperature driving the same load.  The larger package GRJ uses (TO-220) in his lighting module will be at a lower case temperature because it better dissipates or draws heat away from the internal semiconductor circuit.

The current I'm referring to is the overall current driving multiple 3-LED sections.  So in my suggestion of R1=82 ohms, that would put 15 mA total into the strip and if you have 9 LEDs total (3 x 3-LED sections) then that 15 mA total current would be divided 3 ways or 5 mA per section.  To each his own on how bright to run lights.  I see you understand the inverse relationship between R1 value and brightness.  

What can be confusing is the seemingly obvious statement that overall LED brightness is exactly proportional to current.  For example, say you set R1=82 ohms for 15 mA total current.  You try 3 x 3 sections (9 LEDs).  Each section will get 5 mA of current.  Suppose you try 5 x 3 sections (15 LEDs).  The total 15 mA current will now be split 5 ways and each section will get 3 mA.  The total light output or brightness (measured in Lumens) stays the same...it's just spread out over a longer strip.   Likewise, if you use only 1 section (3 LEDs) the single section will get all 15 mA of current.  Again, the total light output stays the same...it's just concentrated in the shorter strip.  In other words, if you stand back a long way from the LED strip, you will see the same amount of light from each 15 mA configuration.

Hi:  I guess I might as well get my two cents into the discussion.  The correct number for the LM317 in the SOT223 package is LM317DCY.   This device is provided with internal short circuit current limit as well as thermal overload protection.  What this means is as the temperature in the package reaches some preset limit, the regulator will begin shutting down.  The power dissipated in the regulator is approximately the input voltage minus the output voltage times the current being delivered.  For example, assume an input voltage of 15 volts and an output voltage of 6 volts giving a difference of 9 volts.  lets say the current output is 0.1 amps then the dissipation in the device is approximately 0.1 x 9 or 0.9 watts.  The junction to ambient thermal resistance of the sot 223 is listed as 49 deg. c/watt so that the internal temperature of the device will rise 0.9 x 49 or 44 degrees above ambient.  You can extend this calculation to any set of conditions.  Another point is that the large tab on the package as well as the 3 leads do provide some heat sinking and the calculations above were for junction to ambient I.E. no heat sinking so there will be some safety factor. 

Best regards,

Phranzdan

DaveJfr0 posted:

gunrunnerjohn posted:

FWIW, my use of the LM317T in my lighting module allows me to light the 12V strips in any size car with up to 45ma without overheating the regulator.  If you crank it up to 45ma, the lights are pretty bright, and the regulator gets up to about 60-65C, but it still works just fine.  I typically run 18" passenger cars at around 25ma for the 12V strip for nice intensity to the lights.

60C...sounds dangerous inside plastic passenger cars.

I'm going to guess that is 25mA total for the whole strip, which sounds perfect. I was figuring Stan meant each parallel node got 15mA for however many LED's are in series per node (3 or 4 at 12V), but yes, I want them realistically lit, I do not need them super bright. 

I haven't seen any issue at all with the regulators running at maximum current (45ma in my case) and 18V track power.  I actually did use them at full power for one application, I put in yellow LED strips for heavyweight cars, the strips required a lot more current for proper lighting.  With the lighting module mounted on the double-sided foam tape as I normally do, I didn't see any noticeable temperature rise in the ceiling of the car over where the module was mounted.  As I mentioned, I was concerned about the power dissipation, so I measured the case temperature after 15 minutes on the bench to see where it landed, it was 65C.

FWIW, I've laid out several little regulator modules with the SOT-223 regulator package, and it was not nearly as good at handling the same power levels.  I had considered it for the lighting module, but in the end I opted for the larger TO-220 package to avoid overheating issues.

gunrunnerjohn posted:

I forgot about that one Stan.

Speaking of which, they seem to be out of stock at Hennings...

Are just the assembled boards available? (already have a roll of the LED's

-Jim

They should have plenty of the lighting regulators, if not we'll make sure they do.

I checked, looks like they have plenty of the regulators, they sell them in pairs.

Stan, Thanks for the comparison recap of the various approaches suggested above. I chose to roll my own early on based on the discussions between you and John using the basic LM317 constant current approach before GRJ's final kits were released. Although not as compact or sophisticated as John's final design, they have worked well in my Lionel baby Madison's.

Thanks to both you and John for your contribution and help..`

DaveJfr0 posted:

... I must have put the wrong part number down, as I did order 330uF caps to experiment with, but the wrong voltage came.

Just to be clear, what do you mean by the wrong voltage?  You really want to use 35 Volts (or greater) capacitors for O-gauge AC track voltage.  You should NOT use 25V, 16V, etc.