A friend and I - both electronics hobbyists - figured replacing the blown LEDs in his home exterior lighting fixtures couldn't be THAT hard... (pause for all to say 'yeah, it's not straightforward you two idiots!'). I'm seeing some behavior that I can't really explain or understand, so thought it was time to give in and seek some help.

Here's the basic setup, what we've tried, and what seems strange. Thank you all in advance for your help and insight.


The lighting fixture gets installed into standard 120V AC and its first component is a variable output power supply with the following specs:

  • Vin: 108-135 VAC 50/60Hz
  • Iin: .10 A Max
  • Vout: 14-20 VDC
  • Iout: 350 mA C.C.
  • Max Pout: 7W
  • PF: >0.92

The output leads from the power supply then feed an LED cluster - all surface mounted LEDs with some circuitry buried under the surface foil which I cannot see nor peel back without destroying the copper layer underneath - so I have no idea how the LEDs are wired. But that doesn't really matter, except I was hoping it would shine some light on why our replacement attempts aren't working...

Strange thing #1

I have two of these blown fixtures from my friend - call them A and B. I've tested the power supply on each, feeding 120V straight from the wall and hooking up a multimeter to the output leads from the power supply.

  • A reads 27V
  • B reads 29V

...both obviously higher than the rated 20V max output. Something seems wrong here... but we can't tell if the power supplies are both malfunctioning, or just rated incorrectly (as the power output is relatively stable, fluctuating +-.2V while read continuously, which seemed reasonable given possible 120V line fluctuations and this being a variable output transformer). We have plans to pull one of the working fixture and test that power supply, but have not yet.

Strange thing #2

My friend bought a 12V LED bank to wire in as a replacement a while back - obviously this will blow given the power output of the supply. But he noted that when he hooked it up to power supply A, it just blinked on and off. When I hook the same 12V LED bank up to power supply B, it lights continuously (although it will blow, and already smoked once, so I disconnect it pretty quickly - I know we're killing it... it's a burner at this point anyway). If I add any resistors in series with the LED bank to try to regulate the power/voltage down, the LEDs just blink, even if the resistor is too small to really have any significant effect on the current.

For my personal tests, I assembled 10 LEDs (2.8-3.2V) in series on a breadboard to handle the 29V output. When I connect my LEDs to power supply B, they just blink as well. It's actually probably more accurate to say they flash bright, then fade down to dark... flash bright, then fade down to dark... repeatedly. Adding any resistors into the series only makes the LED flash pattern dimmer, but the flashing pattern continues.

My research would suggest that the power supply is going into safety mode and cycling between attempting to power, and shutting down. I've measured the Amps draw for my LED cluster on a bench power supply and it appears to be well within spec for the fixture power supply:

  • 10x LED Cluster @ 30V : 5 mA

I've also tried to test the Amps draw on the 12V LED bank when hooked up to the fixture power supply, but can't get a reading. This is what happens:

  • 10A Setting: LEDs do not illuminate. 0.00 reading on meter.
  • 200mA Setting: LEDs illuminate. 0L reading on meter.
  • 20mA Setting: LEDs flash. 0L on meter.
  • 2000µA Setting: LEDs flash. 0L on meter.


Much appreciated if you're still with me! The resulting questions we have are:

When we try to hook up new LEDs to the existing power supply, why do they blink and not illuminate continuously?

Is there a way to utilize this power supply and repair this lighting fixture with new LEDs?

Is there something we should be wary of with the power output on these power supplies being higher than the max spec? (I have to assume this is why the stock LEDs have been burning out)

Thanks again for any insight!


2 Answers 2


Vout: 14-20 VDC

A reads 27V

It's a 350mA constant current power supply, so the load will set the voltage. For example if the load is 5 LEDs with 3.2V Vf at 350mA, so total 16V, then it will output 16V to get the required 350mA.

The supply is specified to work in constant current mode with an output voltage range of 14-20V. Outside of that, it will probably trip a safety and go on an on/off cycle, ie it will blink.

So if you put 10 LEDs, total 30V, that's outside the voltage range where it operates in constant current. Since it tries to regulate current to 350mA anyway, it will increase its output voltage until the overvoltage protection trips, and it will blink.

You don't give any info on the 12V light you used for testing, but if it smoked, that means it's low power, probably a LED strip with resistors. So the power supply tries to push 350mA through it, but the load can't take all the current, output voltage rises, and it shuts down. If you add a resistor in series, output voltage rises more, so it also shuts down.

If you want to test it, you need to use a load that will drop 14-20V at 350mA. For example you could use a 47 ohm resistor (or ten 470 ohm resistors in parallel, or 20 1kOhm resistors in parallel, if you only have 1/4W resistors). Then check with a voltmeter, you should have a stable voltage.

When we try to hook up new LEDs to the existing power supply, why do they blink and not illuminate continuously?

I think that's answered above.

Is there a way to utilize this power supply and repair this lighting fixture with new LEDs?

Yes, you need LEDs that will take 350mA with a voltage in 14-20V range, for example five 1W LEDs in series.

Is there something we should be wary of with the power output on these power supplies being higher than the max spec?

No that's normal, on a constant current driver, voltage will increase if the load doesn't draw enough current.

(I have to assume this is why the stock LEDs have been burning out)

Most likely cause is "cost-optimized" heat sinking causing it to cook to death. So if you replace the LEDs, the new one will also suffer the same fate, unless you also fix the heat sinking issue.

With the original LEDs: if it blinks and all the LEDs light, then the LEDs work, and the driver is toast. If not all the LEDs light, then there is a blown LED and the driver may still work.

Note these power supplies basically cost nothing, and these were also probably cooked to near death, so unless this is a learning experience you're wasting your time.

If you want 12/24V LED strips with excellent color rendition, ask in the comments.

  • \$\begingroup\$ Definitely a learning experience - no better way than to have something real world to solve. So, if I'm understanding this, the circuit load must always be 350mA. The power supply determines what voltage is required to achieve that. If the circuit cannot pull 350mA, the supply will up the voltage to try and achieve it until it blows the circuit. If the circuit wants to pull more than 350mA, the power supply will trigger safety shutdown. \$\endgroup\$
    – b. insler
    Commented May 19, 2021 at 0:34
  • \$\begingroup\$ Yeah, that's a good summary. The voltage at which load draws 350mA should be between 14 and 20V on it. Otherwise the power supply will give up. \$\endgroup\$
    – bobflux
    Commented May 19, 2021 at 7:39
  • \$\begingroup\$ Now I've also learned that my resistors are 1/4W - thank you. Two follow ups: (1) Turns out the LEDs I'm using to test on my breadboard are 20mA max. Are these simply not burning out because the power supply is shutting down first, but would burn if the supplied power was constant? Or would these work with the correct resistor(s) in place? (2) I understand how you got 47Ω=10*470Ω, but how did you get to 20*1kΩ resistors? Looks like 470≈500, and then number of resistors and value are doubled, but I'm unclear on how this math works to balance the wattage difference & keep all components safe. \$\endgroup\$
    – b. insler
    Commented May 20, 2021 at 0:40
  • \$\begingroup\$ Yeah, 20 1k resistor in parallel, divide 1k by 20, 500 ohms, close enough. 20mA LED on 350mA current is fine for a short blink but if applied continuously, it should smoke after a few seconds. \$\endgroup\$
    – bobflux
    Commented May 20, 2021 at 7:03

Leds are usually powered by a constant current source. The circuitry will vary the actual voltage so a given current is achieved. There is a limit to how much it can vary the voltage.

It could be possible both the supplies have failed. They probably have electrolytic caps that have a finite lifetime. For the cheaper units, the lifetime can be very short. This might explain why the leds failed - the failure mode might be loss of regulation that smokes the leds.

  • \$\begingroup\$ Interesting - thanks. Any thoughts on the "flashing" behavior? Why would the power supply measure as a constant voltage, and provide constant power to the 12V LED bank directly, but begin to flash when a resistor is introduced? (more just curious to understand the behavior at this point, even if ultimately the supplies are broken) \$\endgroup\$
    – b. insler
    Commented May 18, 2021 at 6:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.