Atlas Amtrak Pass car LEDs

There appears to be a cap already there, but it's small.  It's the yellow component next to the bridge rectifier.  I'd just parallel a larger one one with it, I'm guessing the chip at the right end of the picture is a regulator chip which should control the voltage to the LED's.

If you're having issues with the board then it's likely others are too.

So for the record, could you post a close-up of this region of the board where we can read the markings on the various parts.  This would allow more informed suggestions as to a suitable course of action.

Ok Stan! I just got back home. I tried to get a decent picture at first. The board is supplied it's negative by a long copper strip that goes under the windows on the side down to touch the frame. That strip wouldn't allow me to get the board out more.

 I did note that the yellow square was labeled C1 which I also would guess is the cap? as GRJ pointed out.

 I will go get a better picture for U.

OK, I unsoldered the board and got a decent shot. That cap says 475 35v. I'm surprised at how small it is. Is it possible that it has failed? This car flickers more than the others and the wheel pick-ups seem good? The only failed caps I have seen have exploded!

 I'm not sure what U1 is?

with over +12v DC going in I measured 3 to 3.2v DC at the circuit tabs on the end of the board.

Might be a .05 NP on the regulated side. Not sure. U1 might be a regulator,not sure either. I would solder a electrolytic cap to the + and - of the bridge,in series with an 18 ohm resistor to one of the leads.

Dale H

The cap marked 475 is 4.7uF - looks about the right size for a 35V tantalum type capacitor.  U1 is marked 7544-1 and the mfr symbol is Holtek so that's a 4.4V voltage regulator IC.  So this board appears to support constant-brightness conventional-mode or command-mode operation.

In this circuit, a 4.7uF provides little flicker tolerance.  So per previous suggestions an electrolytic capacitor installed after the bridge should help.  220uF, 330uF, or whatever you can fit.  The cap should be rated 35V or more.

Stan

My take was it was 475 nano farads (nf) or .475 uf, but I could be wrong. 

If it was 475 pf,then it would be .000475

These type capacitors usually are put on the output.

I can't really see the markings from the pic but whatever it is it is not much as far as flicker reduction.

Looking at pics the net 475D would be 4.75 uf

475E would be 47 uf

Dale H

 So I should add it after the bridge into the circuit. Do I remove the factory one?

I think it was mentioned to add it to the circuit as it is?

 Thank you guys for the extra help. I'm trying!

Joe, you can just parallel the larger value cap across the existing 4.75uf Tantalum cap or replace it with the larger cap, your choice.  I'd probably hot-glue the larger cap laying on the board and tap into the pads where the 4.75uf cap is connected.  As Stan stated, that small a value gives you no flicker resistance.  Also, given that apparently the LED's are running in parallel, they will consume more power than a series connection, so I'd lean to a 330uf or 470uf capacitor for flicker control.

Given the 220 ohm resistor, presumably one for each LED, they're running the LED's at around 7ma each.

Mind the polarity of the added capacitor.  The + and - side attach to board as shown.

But one other thing I suggest you do.  Under power, and before you add another capacitor, measure the DC voltage across the capacitor (between the points labeled + and -).  Separately measure the DC voltage between the "-" point and the circled contact on U1.  The latter measures the voltage reaching the input of the regulator.  Not to scare you but this chip will fail for input voltages above 26V DC.  This would be like that DC buck module you tried that failed for high track voltages...requiring changing to a different DC buck module that could handle higher voltages.  In any event, I believe the design on this board uses some combination of R7 thru R10 to drop the voltage to U1...I can't quite make out the board connections but it appears such.  If you see a meaningful voltage difference between the capacitor DC voltage and the U1 input voltage then that's good and go ahead and add the additional capacitor(s).

Separately, note how the design is low-profile.  A capacitor in the hundreds of uF will be the tallest component no matter how you mount it (upright, on side, etc.).  Should space be an issue, note that you can parallel two or more capacitors to increase effective capacitance.  There is some overhead in that, say, two 100uF capacitors takes up more total volume than one 200uF capacitor...but this sometimes gives you options to fit parts.  

I measured 11.8 across the cap. I measured 3.x (3.2?) across to the input at the chip, where your arrows point.

My memory is so bad I forget just climbing the stairs!

What if I add wires to my 220uf cap and move it to the end out of sight?

 OK, I tried again and wrote it down!

With another voltage 14.3v DC going to the car, it reads 13.5 across the cap, and 4v at the U1 position in your arrows.

Note that with AC going to the board, the voltages will be higher, certainly with the large cap.  I'd suspect with 18 volts AC on the track, you'd be getting around 25 volts across the cap, and you will certainly get that with the larger cap as it'll doubtless charge up close to the peak voltages.  With a 28 volt maximum, you won't want more than 18 volts AC track voltage for these cars.

I think measuring the voltage from the negative side of the cap to either end of the resistor closest to the regulator would be a reading I'd like to see.  I'm curious how much those resistors drop the reading, and it seems you're reading the regulated voltage after the regulator.

The input to the regulator is the center pin of the 3...but should also be on the tab for a SOT-89 package which is what I asked Joe to measure.  I figured it would be "safer" to probe the tab.  I can't quite make out where the trace from VIN goes to on the resistors....but if you can talk him thru it, I'll just imagine you doing that job!

OK, with 14.6 volts AC going in, I measure 4.4v DC at the output of the last dropping resistor thing? and 11.9v DC going into that last one. It appears that maybe the four of them, are wired so that two are in each direction? Maybe they are something else than resistors?

 It appears that the circuit goes into R7, out of that to R10? I need to look closer at how R8 and R9 are wired.

I don't understand the circuit. I will try to explain it?

The circuit trace appears to come out of the cap, into the bottom of R7. It goes thru each R and ends up at the U1. It also has a trace that comes out of R7 and jumps over to R10? I would think that skips R8 and R9. Again they appear to have both connected???

 With 17.1 AC going in, I read 16.1 DC at the cap output, and 4.5 at either the output of R10 or the single pin at U1.

Well, they're labeled as resistors, I'm assuming they're just resistors.  It's curious that only 4.4 volts is going in to the regulator, that's why I was curious what was on the resistors.

I can say with some certainty that the circuit is way more complex than it needed to be, and also much less effective than it should have been.

For conventional and command use, I'd be using a small buck/boost power module.  For command only, a few components gives you effective intensity control with the 12V strips, and having series/parallel LED's as opposed to all parallel is much more efficient in terms of power usage.  Less power in heat, more power in light.

Dumb question, how effective is the light output regulation over a wide range of input voltages?  I'm guessing that's what they were trying to achieve with all the circuitry.

OK forget my eyes! I used the meter. The trace appears to go either into R8+ R9 OR R7+R10!

So maybe they are diodes? I don't know this stuff.

I get

131 ohms for R7

131 ohms for R8

311 ohms for R9

and 311 ohms for R10

Odd looking resistors, and the color codes don't make much sense either.

Although I don't know if it's worth the trouble, it might be useful to draw a schematic of the circuit that exists there if you really want to know what those resistors are doing.

I think I'd be tempted to just drop a larger capacitor across the small one (or simply replace it with the larger one), then slowly bring the AC voltage up and measure what the input to the voltage regulator is.  If it stays well below the 24V max rating of the regulator, that's all you probably have to do.

There's also the matter of adding the 22uh choke for DCS if it's not there.

To satisfy my curiosity,could you furnish the markings on the cap? Usually they are picofarads with a multiplier.

Dale H

Trust me Dale, that's a 4.7uf cap, I use those tantalum caps all the time.  That's how they're marked.

475, that would be 4,700,000 picofarads, aka 4.7uf.

Sorry Dale, I just looked and can't read anything more than 475? I took a couple more pictures and they weren't any clearer. I need to find a magnifying glass, or I'll try my camera on a tripod, set to take really close up shots.

 I'll be back!.....................

Your measurements are consistent with the following.  The resistors are Brown-Green-Brown (150) and Green-Brown-Brown (510).

So the net dropping resistor is 330 ohms from C1 to the regulator IC.  For your example of 16.1V at the capacitor and 4.5V at the U1 input, that's a voltage drop of 11.6V thru 330 ohms suggesting an load current (U1 plus the LEDs) of about 35 mA which seems in the ballpark of something like this.

One can speculate all day as to why they use 4 resistors but bottom line is this reduces the power dissipation in U1 as well as drops the voltage that U1 sees.  So adding capacitance is no problem.

OK, This is the best I can do. I just can't see this small. Maybe save the picture to your computer and then magnify it?

 Click on the picture to enlarge it!

Thank you!!!

  this stuff is interesting but my brain only stretches so far!

Please don't make me do calculus.

I'll bet all the resistors were to spread the power dissipation around.  Those look to be about the size of 1/2W resistors, so you have in effect a 330 ohm 1W resistor.  The power dissipation is .4 watts, you probably would want at least a 1W resistor.  The uneven resistors throw me a bit, I'd have used used four 330 ohm resistors to have the power dissipation evenly distributed across all four.

Of course Stan, I suspect that you wouldn't have done it this way at all anyway.

OK, I think this is the simplest way of fixing up these car's lighting. I added a 220uf cap, a 22mh choke, and a poly-fuse between the truck's + pickup wires.

 The board now takes about a one second delay in power to kill the lights. It doesn't flicker anymore. I did notice that it blinks once entering a curve? Maybe a larger cap would be better? I think it's the way the trucks are designed and the wipers quit for just enough to get a blink entering into curves. I think it's funny because the car goes over my unpowered frogs without a blink!

Looks fine, and the larger cap would probably help with the one remaining blink.  I'd try the 470uf cap, that should do it.

If I remove the stock cap and add a second 220uf cap, will that be too much for the circuit?

Nope, no problem at all.

 OK, here we go. I took another car apart and added 2 220uf caps and removed theirs. With 16.78 v AC going in, I measure 6.8 DC at U1 now. The board output stayed near the same at 4.3 DC at the end of the board.

 The problem is the dropping resistors are hot! I know there's a trade off. I don't remember them being that hot with one cap? I'll try it again as the car isn't bad with one extra 220 cap. Maybe they were hot?

 I have a feeling the best thing is to pull these strips out and swap out everything. I could live with a blink here or there. I can't stand constant flickering. I have six of these now and don't need them to look like Christmas lights.

 Edit: pretty hot with one 220 cap too. I guess that's the trade off for how this circuit is? Maybe cheaper this way? Seems like some power is wasted.

This makes no sense unless it is a universal board which could be adjusted to how many LEDs it drives and/or what the input voltage is. A common/conventional switch for instance. The LEDs driven are not shown. If they are individual LEDs,then each one should have its own resistor. As far as heat, as drawn the 150 ohm resistors would use more than 3 times the wattage as the 510 ohm ones. As John says if you want to make a 2 watt resistor 4 equals ones would work a lot better. I think something is missing here. Also the regulator is supposed to have 2 capacitors,one for input and a small NP one for output,not that it wont work without it. 

Dale H

 There are more resistors at the individual led positions.

I did what I think is the next logical test? I put the stock board on 18v AC and the resistors were not hot. So the extra boost from the additional cap(s) causes extra heat at the dropping resistors?

That makes sense Joe, the anemic little cap they added didn't boost the average voltage much above the RMS voltage as the cap didn't get to charge up like the larger ones do.  The higher voltage with the larger capacitors means there's more power to dissipate.

The benefit of the 12V strips are you use the power more efficiently, so you can supply them with a steady 10-11 volts DC at much less current and not be dropping as much power when running on full track voltage, but you still get the light you need.

I light my 18" cars on around 20 milliamps total current through the strips.  The voltage drop from around 24 volts DC on the cap to the 10 volts on output of the regulator is 14 volts.  The power of .28 watts is easily dissipated by the TO220 package of my constant current regulator.  I've run my regulator module at 45ma (max intensity) for an extended period powering a 12V strip, and the regulator reached around 74C.  Since it's in open air, it wasn't a problem.  I stick the regulator down with double-sided sticky foam, and it insulates the board quite nicely from whatever it's stuck to.

I'm with you on the flicker, I can't stand it!

Dale, the output cap on linear regulators is to prevent oscillation, it's only necessary in certain circumstances.

I think you meant to say the 510 dissipates 3x the power as the 150.  Since the same current flows thru each resistor; power is proportional to resistance.  But if the objective was to make a 330 ohm resistor, it is puzzling why not use four 330 ohm resistors to spread out the power.

Joe, could you measure the length and diameter of the resistor (in mm)?  I just want to confirm these are 1/2W devices.

Also, if you still want to pursue this design (with the flicker caps installed), measure the DC voltage across the cap and also at the regulator input.  Your last report showed 6.8V DC at the regulator input with 16.8VAC (or whatever you intend to run this at) on the track, but I didn't see the DC measurement at the cap.

They're smaller than they look in the pictures, I'd say they're more like 1/4W resistors.  I measured a standard thru-hole 1/4w resistor, that's very close to the measurements.  They would have better power dissipation as they have the board to help heatsink with the mounting.  That being the case, you're really dissipating close to the maximum rating in the 510 ohm resistors, a very bad design criteria.  Rule of thumb is no more than half the rated power of a component.

I didn't understand U Stan until I re-read it.

 I failed to report the voltage at the cap?

I just went down and repowered my fourth car and turned up the Z1000 until I got 16.8v AC. I heard a distinct snap. Now either my meter has messed up. or the circuit has failed. It still works though?

 I get 9volts DC at the caps (220uf + stock one) output. I'm not getting a measureable reading at U1?? I do get 6.7 at position R7 output.

 So I put my meter to ohms and I get nothing for positions 8,9, &10?? I get 131 for R7 still?

 The lights are still on and nothing has burned? I will play with it some more until I can see what has happened.

 I deleted my post above about the size of the resistors. They are just over 5mm long and under 2mm thick.

Might have just proved your theory. Something snapped inside one.

I think it's time to replace those strips with the LED rolls.