Dimming car lights from block control

You should look at a buck/boost power supply for that application.  It will tolerate voltages above or below the output voltage.  For most of them, you'll have to add a bridge rectifier and perhaps some additional input capacitance, depending on how much current your load draws.

eBay: 172357829700 for $1.64 shipped free.

Technical Parameters:
Model Specification: DSN6000AUD Automatic Buck Module
Module Properties: Non- isolated Boost (BOOST)
Rectification: Non- Synchronous Rectification
Input Range: 3.8V - 32V
Output Range: 1.25V - 35V
Input Current: 3A ( max ), no-load 18mA (5V input, 8V output, no-load is less than 18mA. Higher the voltage, the greater the load current.)
Conversion Efficiency: < 94% ( greater the pressure, the lower the efficiency )
Switching Frequency: 400KHz
Output Ripple: 50mV ( the higher the voltage, the greater the current, the greater the ripple )
Load Regulation: ± 0.5%
Voltage Regulation: ± 0.5%
Operating Temperature: - 40 ℃ ~ +85 ℃
Dimensions: 48mm x 25mm x 14mm ( L x W x H )/1.89'' x 0.98'' x 0.55'' ( L x W x H ) (Approx.)

I had looked at that one, but the specs on the XL6009 say 5.0V minimum.  Pursuing that further, reports show its output can go to 50V unregulated when the voltage drops below the minimum spec.  I suspect the module originally used the LM2577 because its specs do say 3.5V min.  There have also been reports of people ordering a module supposedly using the LM2577, but receiving one with the XL6009.  China probably just substituted the cheaper XL6009 without understanding it was not an exact replacement.

I'm looking at using a 5V regulator into a 5.5V supercapacitor, followed by an adjustable boost regulator.

Jeff

That's what I did to create my YLB - RailSounds Battery Replacement.  Of course, I needed it to fit in a small space, so it was a custom board.

I didn't notice the other one was a different chip, here's eBay: 143186914163 that uses the LM2577 regulator.

I just ran a quick test on the buck/boost module and it does not exhibit any bad behavior. I see no spikes in voltage anywhere in it's range.

It's set at 12 volts and has a 48 ohm load, so it is supplying 250ma.  At around 3.4VDC, it starts regulating, though it's slightly low in output voltage to start.  By around 3.6VDC on the input, it's up to the 12VDC output and that holds all the way to 24VDC on the input.  Coming back down, at around 3.4VDCinput the output voltage starts falling. 

Obviously, with a 250ma load on the output at 12V, I'm seeing around 1.4A at 3.5VDC input, reducing to around .14 amps at 24VDC input.

I picked that up and will give it a test.  It is certainly cheaper than anything I could put together myself.  I can't even mail one across town for that price.

It's hard to beat some of the prices on these modules, that's for sure.

SGJeff posted:

.... I'm looking at designing a regulator module using a super capacitor to maintain constant voltage (or current) to lights for several seconds with up to a 50mA 12V LED load.  That will also eliminate any flickering as cars travel around the loop.

So from a design requirements perspective, you need energy storage of, say, 2 Joules.  That is, if you want to run 50 mA at 12V (0.6 Watts) for several seconds (let's call it 3 seconds), that 0.6W x 3 sec = 1.8 Joules.

If energy storage is done with a 5V supercap (as done with YLB that GRJ mentions), an economically priced 0.47 Farad supercap would be more than enough.  That is, a 0.47F supercap charged to 5V stores about 6 Joules.  You can "read all about it" in the OGR design thread for the YLB here.  This would be if you go the route of a 5V regulator module, a 5V supercap, and a boost regulator module.  As discussed in the thread, consider some method (could be as simple as a resistor) to limit the supercap charging current when you initially turn on track power.  That is , if you have a handful of passenger cars with this "topology", you might get a whopping initial charging current (many Amps) if the 5V regulator attempts to directly drive a depleted supercap.

SGJeff posted:

I picked that up and will give it a test.  It is certainly cheaper than anything I could put together myself.  I can't even mail one across town for that price.

But if you're going with the buck-boost module, where is your 2 Joules of energy storage?  This would likely be done on the input side of the module since presumably you want constant brightness for the "several seconds" at 0.6 Watts.  But here's the rub.  That input-side capacitor must be able to handle full track voltage.  As the photo shows, they use a 35V capacitor (220uf) which is minimum rating for O-gauge AC operation.  Yet, when entering the 6V AC block, you might be  coming from, say, only 12V AC so the the 220uF cap might only be charged up to 15V DC or so.  15V and 220uF is a paltry 0.025 Joules...so ~100 times smaller than what you need!  So you'd need to increase that 220uF to something like a ginormous 22,000 uF capacitor (rated at 35V).

I vote for the YLB method.  I'd think the 3 major components, 1) AC-to-DC stepdown 5V module, 2) 5V supercap, 3) DC-to-DC boost regulator, should be in the ~$5 range if you're willing to do eBay from Asia suppliers.

Here's one solution, just stick your supercap between the two.  Should be as simple as sticking the supercap across the output of the first one, and I'd probably add an inrush limiter to slow the surge when the cap first charges.  Tie the supercap directly across the input of the second one, and feed the lights with the output.

eBay: 122880595090  AC/DC Buck Step down Adjustable Converter 3.3V 5V 9V 12V Rectifier Filter Module

eBay: 142976851135 DC-DC Boost Converter 3.3v 5v 9v 12v 2A Adjustable Step Up Power Supply Module

I agree with you.  That's pretty much the design route I went down. I was going to diode-OR the super cap output with the 5V regulator output so the lights would come on as soon as track power is supplied, and the super cap would charge slowly through a resistor.  I already ordered some 1.5F super capacitors due to their lower ESR.

But it becomes a cost tradeoff.  I figure I could roll my own for under $10 each, including PCB, super cap, regulators, and passive components.  And that would keep a car illuminated for up to a minute with power removed.

But since my block controller will keep some power on the track, I think it may work OK with just a buck/boost regulator.  An electrolytic such as the 1000uF 35V I use in the electronic E-Unit only has to supply enough energy to prevent the flickering as the cars move down the track.  That's why I ordered one to test.

Jeff

Thanks for the suggestion, but if the simple buck boost option doesn't work, I'd build my own PCB rather than use multiple modules.  If the buck/boost does work, I'll probably still do my own PCB for the rectifier, and attach the regulator module as a daughter board.

While what we discussed previously was an elegant solution, I opted for a simpler (and less expensive) solution.  I had purchased a roll of 5V warm white LED strip.  A length long enough to fill a 15" passenger car pulls about 100mA at a reasonable brightness.  With that load a 1F super cap discharges about .1V per second, so the dimming is barely noticeable even after several seconds.  A resistor limits the charging surge, and a diode OR bypasses the super cap so the LEDs illuminate as soon as power is applied.  A cheap 7805 series pass provides the 5V, and brightness is adjusted with a trimpot.

WOW, 100ma would look like sunlight in a 15" car!  I run the LED's in 18" cars at around 25-30ma for the entire car!  Most other folks run at similar intensity.

My office/lab is pretty bright, so that was near the max brightness setting.

That's way more than I ever put out with my LED lighting modules, and I've yet to have someone come back and say the lights aren't bright enough.

A factor is that these are 5V LEDs.  The string I used has 20 of them with individual dropping resistors, so that is only 5mA each.

OK, that makes more sense.  With the 12V strings, there are three in series.  So, if you have 30ma, and six segments, each segment would have 5ma current. 

So I’m doing leds in some mpc passenger cars. I got some xl6009’s and on a test with 12v dc in and my led light roll strip connected I could not lower brightness. I also got som lm2596 units and these drop the brightness nicely but provide no boost if the track power is low. I’m using a diode, cap and choke on the input to the board to get dc and some flicker control what’s the difference in these modules?

There are tons of ways to handle LED's in cars, my modules are just one way.  These are purpose built to use in passenger cars with the 12V LED strips.  Yes, if the track power is lower than around 9 VAC, you will start to lose brightness with my lighting modules.  You can, however, use 5V LED strips if you run conventional, then you'll be able to run at any voltage above about 5 volts on the transformer and still have full illumination.

NYCBuffalo posted:

I’m using a diode, cap and choke on the input to the board to get dc and some flicker control what’s the difference in these modules?

 You answered your own question.  I put the choke, bridge rectifier, capacitor, regulator, and intensity adjustment all in a module that measures 1/2" x 1" x 5/8" high.  Also, my module is a current mode regulator, so the adjustment is much less sensitive, with a voltage mode regulator, all the intensity adjustment is in around 1/10 of the adjustment range.  You go from very dim to full brightness in a couple of volts.  With my module, you go from around 3 ma to 45 ma over the full range of the adjustment pot.

I fold the tab on the regulator over the top of the cap when I install them.

NYCBuffalo posted:

So I’m doing leds in some mpc passenger cars. I got some xl6009’s and on a test with 12v dc in and my led light roll strip connected I could not lower brightness. I also got som lm2596 units and these drop the brightness nicely but provide no boost if the track power is low. I’m using a diode, cap and choke on the input to the board to get dc and some flicker control what’s the difference in these modules?

If you want to both raise and lower the input voltage to some regulated-adjustable output then you need a buck-boost or step down up converter.  Think of buck as lowering or stepping down the voltage, boost as raising or stepping up the voltage.  For anything you'll find on eBay, the LM2596 is used in a buck-only configuration so it can only lower the output.  OTOH you'll find XL6009 modules that are boost-only or buck-boost.  And you'll find a good percentage of them mis-labeled on eBay.  For example here's two eBay listings that both say buck boost implying they can do what you're after:

The "true" buck boost has 2 black square components (inductors/chokes).  The 2nd board only has 1 and hence this is really a boost-only converter.

Correct Stan, the very first buck/boost unit I bought years ago was not really a buck/boost.  It wasn't clearly visible what regulator they used, and at the time I didn't tumble to the fact that the buck/boost design needed two inductors.

That’s the explanation I needed thanks! And yeah ill probably get some of Johns boards for simplicity.

Here is a photo of the prototype. After power has been on long enough to charge up the super capacitor (less than 20 seconds), it maintains essentially constant brightness during a dropout lasting up to 10 seconds, and then very gradual dimming after that.

You may notice it uses an inexpensive commercial step-down regulator.

Jeff

WOW, that's a pretty big module.  What is that, about 5" long end to end?

What exactly is that giant 3300uf capacitor doing?  Given the fact that you only have 5.4 volts coming into the board, why is it rated for 25 volts?  It could be a lot smaller if it were a 6v or so rated part.  For instance, the Panasonic EEV-FK0J332Q, 3300uf at 6.3V is half an inch in diameter and slightly over half an inch tall.

A thru-hole part that would be more compact is the Panasonic EEU-FC0J332S, 0.709" Dia (18.00mm), height 0.591" (15.00mm) 

I'd have probably used my little power module, it's .6" x .9" and about 5/8" high.  It uses a cheap eBay switcher and I stack the components on top for DCS compatibility, filtering, and of course rectification of the track power.  It's good for at least several hundred milliamps at any voltage.

This is the eBay switching module module I use.

I don't get it.   What is this a prototype of?  It appears the DC-output of the DC-to-DC commercial stepdown regulator on the left is feeding your "custom" PCB on the right?  The custom PCB has a bridge rectifier?  But if you have DC coming from the commercial regulator module then why the need for a bridge rectifier?

Not quite.  It the bridge rectifier and electrolytic feed the input to the step-down regulator.  The output is set to 5.4V, which feeds a diode OR and then an emitter follower to provide the regulated voltage to the LED strip.  It was more cost effective to use a commercial step down regulator (75 cents delivered to my door) than adding my own to the PCB.

It is 4.5" long.  The large electrolytic prevents any flickering before the super capacitor charges.  It could be smaller making the board shorter, but this was the prototype that fits fine in my passenger cars.  If I make anymore I'll probably use a smaller 1000uF.

Can you post a schematic of what you have there?

Since you have the supercap to support the long power outage, I'm at a loss as to why you need anything close to a 3300uf input capacitor.  Am I missing something really basic here?  That whole input power section could be cut way down in size, and you should be able to use an LM317T for the regulation.

I do a very similar thing with my TMCC Battery Replacement product, I have an input regulator that steps track power down to 5V to charge a supercap, then that feeds a switching boost regulator circuit to provide 8 volts to the RailSounds battery input.  The actual PCB is .8" x 1.2".  The whole package with a battery clip is smaller than a 9V battery.  The other side has the three terminal regulator, the input filter caps, and of course, the supercap.  My filter caps for the input state are two 150uf caps, plenty sufficient to charge the supercap.

Yes, a simple series pass would work, but that would dissipate several watts while the supercap charges.  The 3300uF is left over from eliminating the flickering without the supercap.  As I said, if I make any more I'd probably use a smaller 1000uF 35V that I have a bunch of.  Bit physical size is not important for my application.

I guess if it's a one-off application and it fits, that's all that counts.   I'm always consumed with making stuff as small as possible since I don't know where it'll be used.

@gunrunnerjohn posted:

.... since I don't know where it'll be used.

As I understand it, this is for conventional block-control.  I see the 22uH DCS signaling inductor installed.  That said, if this might be used downstream in a DCS command-control environment I'd swap in the 35V cap (v. the 25V cap in the picture) now while it's on the bench.  I realize you show a 35V cap in your schematic.

Even with the bridge drop, tolerances in 18V AC bricks... or transformers that put out >20V AC ... push the 25V DC rating (i.e., 25 VDCpk / 1.414 = 17.7 VACpk).  And even if in conventional, if using a chopped-sine controller if memory serves me there were some MTH Z-controllers that used 21V AC bricks; so you could see the excessive peak voltages even operating at lower average conventional voltages.

Yes, all my engines are conventional.  I included the 22uH thinking about its possible use in the future.

Flat out my ZW produces about 21V, which would deliver about 28VDC max after the bridge.  I never run that high or the engines would fly off the tracks on curves.  I used the 3300uF as a "flicker eliminator" before adding the super capacitor.  That alone carries the LED strip load for about 1/4 second.

But if I build more I would use a 35V capacitor, which is what is now in the schematic.  A 2200uF is the same physical size, and the 1000uF 35V that I have a bunch of are smaller.

I thought a bit more about the trade off between the switching regulator versus a simple series pass.  One significant advantage of the switching regulator is to reduce current draw.  While a series pass would only dissipate about a watt under normal conditions (12VAC and 100mA to the LEDs), its current draw and dissipation when initially charging the super capacitor can be significant.  Of course that could be reduced by increasing the series resistance, but that also extends the recovery time after passing through a dead block.

Thermal resistance of a LM317 TO220 is 50C per watt to ambient, so a heat sink would be required to handle the dissipation at higher transformer voltages.  Since the emitter follower only drops about a volt, its dissipation is around 100mW.

Just another follow-up...

I looked at the alternative 317 series pass regulator.  Assuming max dissipation of 2 watts (max transformer voltage and 100mA LED load), a low-profile heat sink with about 15C per watt temperature rise is about the same size as the regulator board.  And a heat sink and LM317T from Mouser actually cost more than the switching regulator at this point in time.

Without a heat sink, the LM317T temperature rise is 50C per watt.  So the junction would be 125C at two watts dissipation without a heat sink - the max rating on the device.  A 15C per watt heat sink would bring the case down to a more conservative 55C (about 130F), and the junction to 65C at 2 watts.

Of course a smaller heat sink could be used with more temperature rise, or the board copper used as a heat spreader for a surface mount LM317.  But at this point I believe the switching regulator is the right choice.  There is even a smaller one available on eBay for the same price, but it uses a regulator that is not recommended for new designs.

Since everything works for your intended purpose, you're good to go.  It just seemed awfully large to me for the intended purpose, I'm just used to thinking small.

I agree the prototype was large, so I revised it with a smaller DC/DC converter.  This one is 2 3/4" long, and about half the weight.  It now uses a precision reference so the output voltage is adjustable.  It maintains regulation for over 10 seconds driving a 12" 5V LED strip at 3.2V.