Hello I have been working on this for many months, many parts, circuits, etc. I dont know if its even possible. I have a 24V A/C source lighting fixture that uses two SCRs in anti-parallel for dimming. Im trying to use this phase angle (and power/current) to drive a LED either directly or with PWM with the same amount of brightness, so the LED replaces the incandescent. The original fixture lamp was 24V/250Watt Halogen so there should be plenty of power for LED, however the LED I want is 100W, 30-35V, 3A constant current driven, so i'm concerned how to get 30V minimum if phase angle is very small.

I tried single diode, works but flickers at 60Hz, dimming is nonlinear due to the power requirements of all LED's.

I hired electrical engineer from Fiverr who built this circuit

enter image description here

After many back and forth with parts, It finally arrived and didn't work at all.. Changed Resistor R3 to 1K and it began to kind of work. I believe this circuit is supposed to rectify, smooth, and regulate a DC output and somehow a 555 timer is in there to generate PWM.. But instead U4 heats up to 126C, max output to LED is 27VDC, low dimming flickers (but almost seems to work) as the LED does light up, just with reduced voltage, flickering thats worse with lower duty cycle.

Any ideas how I can fix this circuit, or use the parts to make a simpler one? The end goal being not using an external power supply as im hoping to get enough juice from the lamp output.

For your info I also include the source dimmer circuit from 25 year old fixture.

enter image description here

  • 2
    \$\begingroup\$ There's a sign outside my favorite barbershop. It says "We fix $6 haircuts". I'm thinking that, now. \$\endgroup\$
    – TimWescott
    May 14, 2021 at 20:55
  • \$\begingroup\$ Double check the connection to pin 7 of the 555 timer. The trace join dot is in the wrong place. Could be just badly drawn, or possibly not connected at all. \$\endgroup\$
    – tim
    May 15, 2021 at 8:50

1 Answer 1


What is right with the circuit:

The "power management" section, while it will be wasteful of power because it uses linear regulators, may work if the regulators are adequately heat sunk. U4 may be heating up for other reasons (see "PWM Wave Generation, below).

What is wrong with the circuit:

Rectification and Filtration:

With a 3A average draw, each diode will dissipate up to 1.5W. A 1N4007 is rated at 1W. So -- they're inadequate. You should always go conservative on power supply products, so they're inadequate twice.

24VAC works out to around 33V peak-peak. Going into a bridge rectifier you'll get a voltage that peaks at the p-p voltage minus two diode drops -- so, around 31V maximum. After correct (keep that word in mind) filtration, you should be able to get something that ripples between 31V maximum and maybe 27 or 28V minimum. That's not going to support your 30-35V LED, especially if your 24VAC power dips below 24VAC.

This means that the whole approach is inadequate.

And did someone say "filtration"? In order to ripple between 31V and 28V, you'd need a filter capacitance of \$8333\mu\mathrm F\$ -- \$10\mu\mathrm F\$ is only 833 times short of that. Fortunately caps that big aren't horribly expensive (I just found a \$10000\mu \mathrm F\$ example on Newark for $19 in singles, and didn't look beyond that). Unfortunately, it's not been designed in.

PWM Wave Generation:

Well, OK, a 555 circuit isn't inappropriate. But it's sitting right next to a processor that's doing practically nothing -- so why bother?

(edit): WHAT??? The 1N5348 is an 11V Zener diode -- that's going to be trying to pull the 12V line down to somewhere around 11V. This may account for why U4 is getting hot -- it's probably getting hot, too.

I'm guessing that the original designer put it in as a cheapo regulator using a series resistor from +30V, then changed their minds. You should be able to just take it out.

Constant supply to the LED

Yes, where is this section?

Temperature control of the LED

It's unknown what this is doing. But if the LED is giving a simple voltage, and you're going with a 555 to control the PWM, why not use a comparator?

Output Power Control

Very questionable, but it may work. I would have taken care to drive it from a CMOS version of the 555, and I would have called out one that has considerable output drive. Or I would have used a gate driver. I'm not sure if C2 is there for any purpose other than to fry the output transistors.

In general, it's better to use one suitable transistor than to try to parallel them. Yes, if you just can't find suitable transistors, go ahead -- but 3A at 30V is not a big challenge.

What to do:

If chucking it and trying again is out of the question, the first thing you should do is try about 1000 times more capacitance at C1. This will mean that you're actually supplying the rest of the circuit with sort-of DC. Monitor the LED current carefully -- if it's supposed to be current-controlled, well, that's not what's happening here. While you're at it, monitor the state of the 1N4007's carefully to, or just preemptively replace them with power Schottky diodes that are rated for 3W dissipation (but they only need to be rated for 80V or more -- the 1N4007 is overkill in that regard).

It is possible to do this, and to do it nicely -- an engineer used to working with switching supplies could probably even do it fairly cleanly. But it would involve rectifying the 24VAC into DC, then using a boost converter up to 35 or 40V, then bucking that down again through the LEDs in a way that maintains a controlled current. That's utterly beyond whoever did this for you.

  • \$\begingroup\$ Thanks! The diodes were actually upgraded to 1N5822 before assembly. Im not opposed to chucking the whole thing and starting over or getting a circuit working on a breadboard before assembly (lesson learned!). Is it possible to hire you from SE? The temp control is supposed to kill power when temp reads above 90c (through microcontroller code).. I will remove D8 Zener Diode, should I remove C2 capacitor as well? \$\endgroup\$
    – Tim Davis
    May 14, 2021 at 21:59
  • \$\begingroup\$ Removed diode still works.. down to 25V output U4 not hot to touch anymore but like you said whole thing feels inadequate \$\endgroup\$
    – Tim Davis
    May 14, 2021 at 22:28
  • 1
    \$\begingroup\$ C2 shouldn't hurt. I'm not working as a contractor currently -- I'm too busy at my fun day job. SE frowns on getting commercial, but if you're in the US see if there's an IEEE consultants group in your area. You want to get someone who's got a track record of successful circuit designs, and you want to get an individual you can interview, not some yokel at a "circuits for money" operation. \$\endgroup\$
    – TimWescott
    May 14, 2021 at 23:14
  • \$\begingroup\$ If I were to get a dc-dc constant current regulator from Amazon after a rectifier/smoother (proper) would that work? I'm confused how the phase angle gets converted to pwm from your suggestion above with boost/buck convertor \$\endgroup\$
    – Tim Davis
    May 15, 2021 at 1:45
  • \$\begingroup\$ Also removed diode.. worked the same U4 no longer burns up \$\endgroup\$
    – Tim Davis
    May 15, 2021 at 1:46

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.