If the input voltage drops, the behavior is unpredictable but it will be nothing that you want. You have to give the regulator more volts than its output, that is immutable.
Is there some reason why you can’t run your LEDs from AC? Or switch to a half-wave rectifier ahead of the regulator? If your supply transformer simply does not give enough input volts, you can use a small transformer as a “bucking” transformer and boost the AC line input by 6 or 12 volts. At some point you may have to heatsink the 7812 because of how much power it has to get rid of.
If you don’t give your 7812 enough input volts, it will not cooperate in the most irritating ways imaginable.
Making the regulator out of discreet parts may work better, but the 7812 is made up of transistors and zeners anyway, so it might just turn out the same. I’d focus on getting the regulator input volts really high, aim for 18-20 volts and heatsink your regulator. Get it high, move to a 7815, then use some series diodes on the output to drop the overall supply voltage by .6 or .7 volts each.
If indeed you are only powering 3 LEDs at (figure) 20 mils each, you should be able to give the 7812 (in TO-220 format = 1 amp) a solid 20 volts and should not sweat.
Thanks for the reply - the modules each draw about 100 mA*, and anywhere from 1 to 8 might be switched on at any given time. Given that the current drawn increases rather quickly at and above a 14 volt input, I believe each module consists of 3 LEDs and a limiting resistor, all in series, something like the “12 volt” example here:
https://www.waveformlighting.com/pcb-designs/led-strip-light-schematic-and-voltage-information
*Exactly what LED is used I am not sure. It is apparently something a bit “less” than a 2835, as the 2835 is rated at a maximum 150 mA, not 100 mA. (At 150 mA these modules get awfully hot awfully quickly. But they seem fine, just a bit warm, @ 100 mA.)
More, in a bit...
Eh, took me a bit longer than “a bit”. (eye roll)
I did just post #27, paragraph 4 applies in particular. :-)
Indeed, there are cheaper linear regulated supplies on Amazon, eBay, etc., that with good heat sinking of the regulator chip and output transistor could easily crank out a couple amps, especially given that the voltage drop when limiting the output will be modest. BUT... same problem — if the input is 13 volts, what does the output do?
It occurs to me that SOMEWHERE around here I should still have my old audio op amp prototyping breadboard which had either a 7815 and a 7915 in the power supply section, or possibly a 7818 and a 7918, as I sometimes worked with +/- 18v supply rails and NE5532’s and NE5534’s. I could muck around with the 7815, for example, to see how the output behaves as the input decreases. If below a supply voltage of 18 volts or so, the output pretty much follows the input, no problem. I don’t really need true regulation, just voltage limiting. But if the regulator turns off when the input falls under 17v or so, that’s no good.
Then I would go on the assumption a 7812 would behave much the same, just at 3 volts less all around.
I still think the zener-transistor circuit MAY do what I want, so, I have parts ordered and will see what happens... If it doesn’t work I’m out ~$15 and will have several extra 13v zeners and TIP31C’s to add to the parts bins. Plus I got up to $35 for my Amazon order, finally! :-)
A fellow over at All About Circuits is assisting me. He’s come up with a couple pretty interesting, novel (to me) circuits. Essentially, as the input (source) voltage ramps up, the output voltage ramps up too, but levels off above 13v. This is not quite analogous to audio amplifier clipping, but could be crudely thought of in that way.
The parts count is a little higher than I was hoping, but still cheaper than if I had small regulators on each LED module.
I may still purchase a small PWM buck regulator just to see how it behaves as it’s input voltage sags. Or if it requires RF suppression. They can be had for well under $5 on eBay or Amazon.
Thanks!