1

Since last week, every LED bulb and fixture in my household is flickering at a similar rate. Incandescent and CFL are not affected.

Here's what I've had done to try and diagnose the issue:

  • Fuse panel inspected by certified electrician, no issues found and everything tightened
  • City electric provider came on site and tested the meter box including neutral line. No issues found, everything tightened in box
  • Asked neighbors for similar issue. No direct neighbor experienced it, however other people in the same neighborhood on a different street did experience it at the same time but theirs resolved overnight, and mine did not

There are no dimmers on any of the LED switches.

Does anyone have any idea what could cause this or what I could do to fix this (aside from replacing every LED with older incandescent / CFL)? Or any suggestions on how I can diagnose the issue.

I'm in Canada, if that matters.

automaton
  • 119
  • 2
  • 2
    I would want to put a power monitor on your line or use an oscilloscope and make sure the transformer that feeds your home is ok. If something caused a problem in the windings some strange waveforms can be the result causing problems and harmonics this is rare and a hand held meter that only measures RMS won’t see it. – Ed Beal Apr 09 '22 at 17:05
  • 2
    LEDs seem to be more sensitive to problem power than other types of lights. Seeing your neighbours also had a problem would say it is coming from power company equipment. Was there any weather(wind,lighting) storms near you just before it happen? – crip659 Apr 09 '22 at 17:06
  • LEDs are diodes - in order to handle AC power, whose direction fluctuates, mains LED lights have a capacitor in them in order to ride out the "wrong direction" part. Not an expert on AC electricals though, but maybe you've had just enough of a local frequency drop for the dimmer capacitor-driven part to be more noticeable. – Luke Briggs Apr 09 '22 at 17:12
  • Side note: the build quality of the LED bulb matters. Cheaper ones simply have smaller capacitors and are basically much more likely to flicker. – Luke Briggs Apr 09 '22 at 17:14
  • 1
    First place I look is a "Lost Neutral", which is a weird type of power outage that won't kill power entirely, but will cause voltage to be highly variable. Check voltages on plug-in circuits all over the house. If some are below 120V while others are above 130V, and turning on a hair dryer or microwave makes the voltage move by 4 volts or more, that's what you have. Good news is, the power company will fix that for free. – Harper - Reinstate Monica Apr 09 '22 at 20:01
  • Are all bulbs on the same phase? Or are the bulbs fed by different phases (2 or 3)? Is it a positive flickering or a negative? Sometimes hard to tell, but if the LED light is brighter for a short period of time, or if it is dimmer, is a very helpful information. Do other people also see the flickering? Some people are more sensitive. Maybe some neighbors do not recognize the flickering, but a short check would reveal it. Sometimes a medication or physical condition can modulate the eyes'/brain's sensitivity for flickering. – xeeka Apr 09 '22 at 23:29
  • It's not psychological. My spouse and kid noticed it before I did. As mentioned others in the neighborhood also noticed it, but not direct neighbors. Also both the electrician and city power workers could see it. – automaton Apr 10 '22 at 17:05

1 Answers1

2

In North America, most name brand light bulbs are powered by a bridge rectifier, a smoothing capacitor and then a linear regulator in series with ~ 120-150V worth of diodes. Or at least every one I have bought at a Home Depot worked like this. The bridge rectifier and smoothing capacitor output a nominal ~165V peak with some additional volts of ripple. This ripple is dropped across the linear regulator resulting in uniform, ideally flicker free LED power.

This setup is not very efficient, but it is very cheap. The inefficiency comes because the voltage dropped across the linear regulator is directly converted to heat. The larger the voltage across the regulator the more resistant the lights to flicker but the more heat and lower lm/watt. If the ripple on the lines (due to both the 120 Hz from the bridge rectifier, any external noise on the lines, and any voltage droop on the mains) causes the linear regulator voltage to drop to zero, the regulator will enter drop out, and the LED brightness will flicker.

For example, I have a pretty good, flicker resistant LED bulb here that is composed of 7 SMD packages each with an 18V forward voltage (3v*6 junctions). That is 126V. The rectifier outputs 165V with maybe 10V of ripple, so the lowest voltage for nominal 120V input is 155V, while the LEDs will stay lit for any voltage above ~126-127V. That means I have 27V of head room, and any noise/droop/etc on the lines less than this will have no effect on the diode brightness. However, you can also find bulbs with smaller capacitors or less power wasted on the linear regulator that only have 10 or 15V of margin. These will be much easier to make flicker.

My guess is that if you probed your mains power, you would find that it is a few volts below nominal 120Vrms and there is probably some additional noise from a large load somewhere else in your area. The combination of these effects causes occasional cycles in which the RMS voltage dips by more than the voltage drop across the regulator, causing periodic variations in LED power which you see as flicker. This is only a guess, but I have seen it happen in practice.

If this is the problem, your best bet is to simply try a few other brands of light bulbs. A slightly higher voltage drop across the linear regulator or a slightly larger smoothing capacitor will make a large difference if you are riding right up against the regulator dropout voltage.

user1850479
  • 121
  • 2
  • It sounds like the OPs problem just started with LEDs that were working right with no flicker. Unlikely that all of OPs LEDs would start failing at the same time. – crip659 Apr 09 '22 at 18:38
  • 3
    @crip659 They're not failing, they're probably working as (cheaply) designed. Problems like this start when someone else on his local transformer adds a large or noisy load. – user1850479 Apr 09 '22 at 19:57
  • As others have mentioned, LEDs react almost instantly to voltage variations, larger capacitors may help but won't completely cure the problem. Remember that incandescent bulbs basically emit via a very hot piece of wire (filament) ...not very sensitive to minor voltage variations. Next: There may be a new load installed in your area that causes these symptoms. To diagnose, buy, rent, borrow or steal an oscilloscope, preferably the recording kind and monitor your voltage. – George Anderson Apr 09 '22 at 21:58
  • @GeorgeAnderson The capacitor cannot remove flicker completely, but the regulator after the capacitor can. If it isn't working, either it's missing (uncommon) in NA or you are driving the regulator into drop out. Does that make sense to you? – user1850479 Apr 09 '22 at 22:05
  • Thanks for the reply. I have a rack mounted UPS that tracks voltage and it's been consistently 120 - 122v, but it's been like this even before the issues started. The LED lights were working fine for years until this week. – automaton Apr 10 '22 at 00:41
  • How would such a bulb dim past the halfway mark (90-180deg) where the chopped sine wave never peaks above the LED's Vf? Most non-dimmable bulbs I've seen use capacitive droppers (transformer-less PSUs) instead of the linear current regulators found on so-called "driverless" LEDs. Dimmables (the kind that work with regular triac dimmers) use SMPSs with large-ish output caps and monitor the incoming sine wave to determine output power (current). They fade out over ~500ms, so a missing phase or two shouldn't produce notable flicker, but a noisy sine wave can "look" like a dimmer output to them... – dandavis Apr 10 '22 at 08:52
  • @dandavis Modern dimmable bulbs sell for too little to have an SMPS, or at least I haven't seen one in years for sale in NA. They're (nearly?) all a rectifier, a linear regulator (which are now available with the ability to measure TRIAC duty cycle) and a series string of diodes. Similarly I haven't a bulb with a capacitive dropper for sale in an actual American store. The nondimmable have cheap linear regulators while the dimmable have slightly more sophisticated linear regulators. Not sure I understand your question? – user1850479 Apr 10 '22 at 14:29
  • I shuck a lot of bulbs, a wanna be bigclive. If i see a tiny transformer, I assume SPMS. Bulbs from, eg. dollartree and the grocery store have no IC and have a cap in series between the fusible resistor connected to mains and the FBR. W/o an SMPS, i don't see how a bulb lights when the phase angle cut of the triac causes the P2P of the AC to to be say, 75v, when the forward voltage of the LEDs is 126v; how does a linear IC do that? I can see that working a bit past half-way, maybe 0-110 deg, but my bulbs dim all the way down to about 10% power/ 160deg... – dandavis Apr 10 '22 at 16:58
  • @dandavis LED dimmers are trailing edge, meaning the lowest voltages are truncated first. Since the peak is the last part truncated (at 0% brightness), the peak voltage out of the rectifier is constant until the bulb switches off. This is why old fashioned incandescent dimmers usually won't work with LED bulbs and why you can see bulbs start to get 120 Hz flicker if you dim them too much (past the peak of the waveform). – user1850479 Apr 10 '22 at 19:24
  • @dandavis It's easy to find SMPS if you open old light bulbs, but once high voltage diodes became widely available a few years back using a switched power supply is essentially pointless. I'd be (pleasantly) surprised if you could find one still being made in 2022. If you've got one still made in a 120v country, please let me (and the OP) know as it might help him with his problem. – user1850479 Apr 10 '22 at 19:26
  • The cheap ecosmart bulbs I use don't need trailing edge dimmers, junk drawer rotatory ones work fine; even yielding about the same range as an incandescent. They have static resistance that plays nice with neutral-less "fancy" switches like motion/dusk2dawn/etc. They fade to black in ~400ms upon switch off. They can't be linear CC because they operate on low voltages, needing ~30v to turn on, and ~13v to stay lit; real AC, from a variac. Here's one running (albeit dimly) at 15v: https://i.stack.imgur.com/Aob1V.jpg To me, that implies the chain's Vf is either <16v(haha), or it's an SMPS, right? – dandavis Apr 12 '22 at 03:29
  • That ecosmart bulb is still for sale, i've blocked my whole house with them, about 30 bulbs, and have yet to have one fail after a couple years. The new glass ones aren't nearly as good, so maybe I should stock up ;) If you want a capacitive dropper LED bulb, dollar tree; they don't dim, they weigh virtually nothing, and they last about 6 months in use in anything besides a bare hanging bulb. Lastly, an aside fun fact: they've had 1000v diodes for decades, [the 1n4007](https://en.wikipedia.org/wiki/1N400x_general-purpose_diode) for example. – dandavis Apr 12 '22 at 03:50
  • @dandavis Common Buck topology requires that the input voltage be larger than the forward voltage, same as the linear regulator. If it works at low voltage, it could be something more complex (expensive), but for a cheap bulb it might be a capacitive dropper and low Vf diodes, especially if its a few years old. That would explain why it works with an old fashioned dimmer. Could measure the power factor or just open it and see. – user1850479 Apr 12 '22 at 03:51
  • Ok, shucked it for you. I use these but wth, i've got time and money to waste. You'll be happy. There's 5 LED chips in the chain, running at ~8v each, ~40v total. That's even when input is at 16v AC. It's an SMPS, but i didn't trace out the topology. Once opened I was shocked to find 2 screws and a completely modular design. The driver PCB is 1-sided, and the THT side is shown. The SMT side is a bunch of passives plus 1 FBR, 1 diode, and 1 driver IC. I think this is what you're looking for, right? pic: https://i.stack.imgur.com/ur5hE.jpg – dandavis Apr 12 '22 at 04:19
  • @dandavis Really neat. I tried opening one of my Ecosmart bulbs, but only got a linear supply. I'll have to look at some more. Perhaps that would be a good option for the OP. – user1850479 Apr 13 '22 at 04:22