Lighting for a MTH SD70 Dummy

If it's just powering a few lights, that sounds like too much power is being used!

All that is on now is the headlight, cab lights and 1 ditch light- I am going to check for a short, then try another converter. I laid a piece of shrink tubing on the heat sink and it did not affect it-still seems too hot.

Can you put an amp meter inline with the input and see the current draw?  Each bulb is about 60ma so your only talking about .24 amps right now.   G

Well, it depends on what kind of converter you have.  I can't quite tell from the photo but:

(1) what is your track voltage (presumably AC since I see 3-rail)

(2) is this an AC-DC converter module? Or a DC-DC module with an external rectifier(s) that I can't see in the photo?

(3) does the module use the LM317 regulator chip (read numbers off the device screwed to the heatsink)

(4) how many 6V bulbs will you ultimately want to power (I understand 3 for now)?

I have those exact same modules, with 16 volts in (1033 transformer), and a 250ma load (24 ohm resistor), the regulator is at 60C, so the 180ma of three bulbs and 18 volts in would probably result in a pretty warm regulator.  Probably not a problem, but I doubt this module will handle the 1A the specify it at.

Update, after I went to lunch and came back, my regulator stabilized at 75C in a 20C ambient.  Apparently, the temperature you're seeing with that regulated supply is not abnormal.  I expected it to handle quite a bit more than 1/4 amp!

If you're running from a constant 18 volts, I'd consider a series diode for the input power of the supply, that'll drop the power dissipation way down for the module and still let it have enough voltage differential to operate the lights.  The module has a full wave rectifier on it, so the polarity of the diode won't be a factor, it'll just chop the input voltage way down.

Well, it won't be 20C ambient once he puts the shell on!  And doesn't an SD70 have more than 4 bulbs(?)

If adding more external components anyway, an alternative to the series diode is use a LM2596 DC-DC module (now going for the astonishing Buy-it-Now price of 99 cents on eBay with free shipping).  There would be no heat issue at all.  You also might be able to use a series diode (instead of a bridge rectifier) to convert the track AC to DC.  Here's a recycled photo:

And no matter which way you go, don't forget the 22uH inductor for DCS compatiblity even if you aren't using DCS now.

Stan,

What is the 22uh inductor?  We ME's were taught about bridge rectifiers and diodes, maybe I fell asleep during the inductor part!!

Thanks

Scott

I'm with Stan here, I'm disappointed in the LM117 module, it really doesn't do that well, even under light loads!  I was stunned to come back and fine a 250ma load had shot the temperature up as far as it did.

Since he already has the power module, the series diode may be the cheap answer.  However, the 22uh choke is still recommended, I put them into all my lighting projects nowadays.

Learn something new every day- typed in 22uh choke in ebay and viola! 

the bridge rectifiers and Chokes all have an amperage rating- what do you guys think would work ?

I am going to build a couple of these and test them- will keep everyone posted

Thanks

Scott

1A diodes, like the 1N4003 will work fine.  For lighting only, a 22uh 1/2 amp choke should be plenty.

Try Digikey for the parts, good prices, and very cheap shipping on small orders.

Since you have the module, did you try this or run the numbers to insure the input capacitor and regulator headroom is adequate to prevent dropping out during the half-cycle?  Some SD70ace's have illuminated number boards, marker lights, and who knows what else in addition to headlight, ditch lamps, and interior lamps.   Maybe some of those are LEDs?

Anyway, if the Number Board and Marker lamps are incandescents, now we're creeping up on 1/2 Amp.  While it's not the end of the world if the regulator drops below 6V during each half-cycle, it is a consideration.  And if there are more than 4 lamps, even with a reduced average input voltage using a series diode, that heatsink might still be quite the hot potato?

And yes, Nothing is so easy as the job you imagine someone else doing!  I now owe you $10.

It didn't drop out at 16 volts from the 1033 with the 1A diode in series with it, it still managed to put out 6 volts at the same 1/4 amp.  I didn't measure the heat, but it was cool enough to easily hold the heatsink, much better.

There's always LED lighting, not that hard and power considerations simply melt away.

I think i'm following this...16vac, full wave rectifier, gives you 22.4 vdc at .25 adc. So the series drop the reg is being asked to dissapate is about 4 watts. Imagine a 4 watt nite light bulb and try to touch it after it's been on awhile.

Right.  Another perspective: 22.4V @ 0.25A = 5.6 Watts going in, 6V @ 0.25A = 1.5 Watts going out.  That's a conversion efficiency of 27% (i.e., 1.5W / 5.6W) with 73% of that input power going straight to heating the cab.

The LM2596 module uses a switching regulator which would run with an efficiency of maybe 85%.  Same 1.5 Watts going out, but only 1.75 Watts going in.  So only 1/4 Watt goes into heating the cab.  Since the input voltage is the same 22.4V, the big win is it only requires 0.08 Amps going in.  No problem touching a 1/4 Watt nite light bulb if such a creature exists.

That would be an LED Stan.

You're right, the conversion efficiency of a switcher will blow away a linear regulator, which is why they're taken the electronics world by storm.

I have a bunch of the little DC-DC switching converters, for the price they were hard to resist.

Stan, I am confused about this.  I think I understand the principle, but is the application correct?  A standard Constant Voltage regulator has a very small ground current.  While the input wattage to the Full bridge is as stated, the Regulator shouldn't be seeing that power, nor that poor of an efficiency?

When I look at the simple CV regulators MTH used in some of the PS-1 engines to drive 6V Bulbs, it was a simple 4 diode set up for Full rectification, a 7806 Regulator, and a filtering capacitor.  Granted the regulator heat sink is the chassis.

Just seems to me the current through the regulator with a .25amp load should be more around .26 to .275 amps, and heat dissipation at 1/4 of the rating should be no where as high as being seen or described?  Am I missing something?  G

What you're missing is a switching regulator does not dissipate all the voltage difference as heat.  You can think of it as a DC transformer with 85% efficiency.  So, if I wanted to get a much lower voltage, the input current to a switcher would be far less than the output current.  Going from 24 volts to 6 volts, for instance, you'd only have about 30% of the output current being consumed at 24 volts.

A linear regulator is like a big resistor, if you want 6volts at .25 amps, you'll be using .25 amps from whatever input voltage you supply, and the excess power is all turned into heat in the regulator.  The "efficiency" of this combo is about 30%.

John, I am not comparing the switching regulators to the linear regulator.  I am asking about the power calculations of the Linear as stated.

It is not making application sense to me.  There are up to 4 Linear regulators inside a PS-1 engine.  If the efficiency is as stated the heat would be significant.  Only 2 of the 4 are on heat sinks.

You really shouldn't have to go to a Switching converter for a few 6V 60ma bulbs.   G

Well, if you're rectifying 18 volts with a bridge rectifier and a capacitor, I can tell you that you're getting around 24 volts out.  To get that down to 6 volts, you have to drop 18 volts.  All of that current goes directly into heating up the linear regulator, that's 4.5 watts.  That math is certainly not suspect.

I have the exact same linear supply that was first mentioned here, and my test confirmed that it does indeed get plenty hot with a 1/4 amp load powered by 16 volts AC from a 1033 transformer.  All of that heat was in the regulator, and yes it has a reasonable heatsink on it. 

I'm not sure what you're looking for here, but the computations and empirical tests agree on the heat issue.  I've computed it and I've tested it, and the results are the same.

You said:

A standard Constant Voltage regulator has a very small ground current.  While the input wattage to the Full bridge is as stated, the Regulator shouldn't be seeing that power, nor that poor of an efficiency?

This is where you're going wrong, linear regulators have very poor efficiency when the input/output voltage ratio is significant.  If the input voltage is closer to the output voltage, you dissipate far less power, that's not the case here.

Perhaps it's the "ground current" of a 3-terminal regulator that is confusing.  Indeed the ground current of 3-terminal regulators (whether the fixed LM7806 in a CV board or the LMx17 in the above module) is small as you say...usually a few mA.

But that ground current is effectively "stolen" from the current that the regulator delivers to the load.  So if you look at the current flows in the circuit, it boils down to the following.  Pick any point in this circuit and the sum of currents flowing in and out of that point must be 0...otherwise electrons would be vanishing from the universe.  So to be precise, the load (output) current in a linear regulator circuit is not exactly the same as the input current.

But circuits are typically evaluated from a "black-box" power flow perspective or output power vs. input power with heat being one output.

As an aside, the lights themselves have an electrical to optical conversion efficiency.  So in this case there's 1.5 Watts going in and, for incandescents, most of that goes out as wasted heat rather than optical "Watts".  LED's have much higher conversion efficiency which is why LEDs are supplanting filament bulbs in the same way that switching regulators are supplanting linear regulators.

HTH.

John, So what are the formula used?  I read 2 technical papers this morning (one from TI and the other from a university and while they did not specifically address the input current calculation they addressed the current flow and the input current was basically the output current and the current returned to ground.  The implication that the ground current was relatively small for Standard regulators, and slightly higher for LDOs.

The small 5V regulator on the MTH smoke unit sees about 18-20VDC input from the FWBR and capacitor for .1amps draw at 5V.  So by the calculations above it shouldn't last very long with that kind of heat stress, but it does.   G

Stan,  Thanks for the diagram, and that is what I assumed was used in the discussion, but for a transformer, and I understand the physic is different the current in at higher voltages is not the same as the current out.  Energy is conserved, but less the transformer efficiency if input is 20V and output is 5 volts at 1 amp, the input current would only be .25 amp.  So I guess I was thinking the bridge was expending some of this Heat energy.

The technical documents didn't really get into details.  Guess I will do some more research.  From the practical side, it just doesn't seem like that much energy was being wasted.

So you simple increase the efficiency of the regulator by lowering input voltage.  G

The regulator in question is feeding the fan and nothing else.  Actual measured current for the fan motor in the PS/1 smoke unit is 45MA at exactly 5.0VDC, at least the one I have on my bench right now.

If we again assume the 24 volts DC available at the input and 5V at the output, we get a power dissipation of .85 watts in the TO-92 package.  That's right at it's limits.  Of course, you have to remember that the PS/1 smoke unit rarely sees 18V track voltage during operation, so the power would be less.  FWiW, I have measured the temperature of the regulator on a Lionel unit running off track power, and it was running at 68C, so it is running pretty warm.  Some Lionel smoke units actually have a TO-92 heatsink on them, likely for that very reason.

If we expand the perimeter of the black-box to include the entire module from AC in to DC out then the bridge most definitely is contributing to cabin warmth. 

But this discussion started with the OP talking about touching the heatsink.  I don't have this module in question to look at but from what I can tell it's just the LMx17 linear regulator IC that's bolted to the heatsink.  The 22V "input" voltage was what I think cjack or GRJ suggested/measured as the average DC voltage AFTER the bridge.

When the bridge is conducting, two of its diodes drop, say, 1.25V on average.  Using an average current of 0.25A, you can black-box the bridge itself as a power converter/transformer.  That's another 1/3 Watt (1.25V x 0.25A) going out as heat.  Of course the DC-DC switching-regulator module also needs a bridge, but the input current is only about 0.08 Amp (vs. 0.25 Amp for the linear-regulator).  So the current flowing in the bridge is smaller and it would dissipate only 1/10 Watt (1.25V x 0.08A) as heat.

So tossing the bridge rectifier into the mix and for the same 1.5 Watts going out to the lamps, the linear regulator module (bridge included) requires about 5.9 Watts and the switching regulator (bridge included) requires 1.85 Watts.  Electrical efficiency: 25% vs. 81%.  And over 10x wasted heat: 4.4 W vs. 0.35 W.

Thanks Stan.  I don't dispute the difference between Switching converters and Linear, I just didn't think the Linear regulators were that inefficient.  I assumed it had a basic efficiency and therefore the input power and output power would be similar (like a transformer).  Found some good documents discussing it and I guess there is a sweet spot with input voltage closer to the needed voltage, but above the drop out for best efficiency.  I guess I need to measure some of the temperatures on the regulators next time I am working on an engine.  G

I would guess those linear regulators with no heatsinks on the PS/1 boards have pre-regulation so they're not receiving a large input/output voltage differential.  Many times a linear is selected to develop a secondary voltage from a slightly higher voltage, then there is minimal power dissipation.