Hacker News Reader

Comment by mikkupikku | Hacker News Reader

I have a theory that cheap LED lights with low quality drivers are bad for dogs. LEDs with low quality drivers very often have a high amplitude flicker at 50/60Hz, which is about the flicker fusion rate for humans so we don't perceive it (at least usually), but dogs are known to have a substantially higher flicker fusion rate and probably perceive the flicker. Probably worth considering, especially if you have a dog with epilepsy.

(Incandescents also flicker at 50/60Hz of course, but the thermal inertia of the filament makes this a lower amplitude flicker.)

View on HN

If you are sensitive you can get really, really annoying blips of flicker even above 1khz - https://journals.sagepub.com/doi/10.1177/1477153512436367

They also flicker really badly if your power is not perfect, like you have a decent sized training rig on a different circut.

Incandescents are basically little light inductors and I would imagine the luminosity curve would be sinusoidal vs whatever hell a LED driver chip puts out.

I have a theory that they're bad for humans too. I think the same is true for computer screens too. Something about the flickering and the "unnatural" color spectrum messes with people's heads (anecdotal/subjective). Maybe our brains do extra processing work that detracts from other systems.

The newer phosphor-coated "filament" LED's also have this "inertia" which is nice. And no other electronics to fail also since they are just series-connected LED's.

The phosphors do have some inertia, but I'm not sure if it's really enough, since fluorescents have phosphor coatings too and those are notorious for noticeable flicker. Different kinds of coatings though, different thickness and quantities of phosphors, maybe different composition too. I don't know how comparable they really are.

Besides the epilepsy, what do you mean with "bad for dogs?"

Sidenote: My partner and me both "feel" the difference of cheap LEDs. Its not something we can pinpoint down, but it got way better with hue lights.

We live with multiple dogs and iam really curious. One of the dogs that we often had around had severe epilepsy that very strong medication was needed and the dog died anyways way to early around the age of 2. She had less sizures than in her original home when she was with us which ofcourse might be unrelated to the lighting. But your thought is interesting.

> Besides the epilepsy, what do you mean with "bad for dogs?"

Psychological or physiological unease at least, I assume this from the way rapidly flickering dying florescent lights make me feel.

When you notice flicker, your iris is trying to expand and contract at that rate to compensate. It was one of the things people were pointing fingers at for sick office syndrome, back when fluorescent tube lighting was popular. People, and I assume dogs, have different thresholds when this problem kicks in and flicker becomes noticeable.

I tried Hue lights last year but 50% of them had problems. Did you run into that too?

No issues but I bought my last one like 2022 and didn't need to replace any of them.

Bad LEDs flicker at twice the power supply frequency, so 120 hz which is very perceptible by most humans.

But also note that not all LED lights flicker.

Why don't LED drivers just use plain continuous DC with no PWM?

Dimming. And even without that, the "continuous DC" is usually provided by switch-mode power supplies that themselves have a PWM ripple. How much depends on the quality of the supply.

Cost.

Since AC is pulsating you need to store some energy to get continuous DC, usually in a smoothing capacitor. And that capacitor is relatively big and when durable, then not cheap. And it requires some further complications (like avoid inrush current).

Sibling comments have no clue what they're talking about. I actually engineer these things.

The short answer is cost and heat. What you're describing is a linear constant current driver, which are the gold standard for flicker-free operation. Drawback is a much more complex circuit and a transistor burning off a ton of waste heat.

A constant current driver requires sensing the current across the LED, which involves sensitive analog circuitry. A PWM drive requires essentially nothing more than a FET tied directly to your microcontroller.

There are switch-mode constant current drivers, best of both worlds. Essentially a buck-boost converter in current mode instead of voltage mode. These are slightly more expensive, so they don't appear in consumer lighting products.

All that aside, the reality is that if your PWM frequency is high enough, it doesn't matter. Above several KHz it's imperceptible. The reason that this still isn't done universally is that it's a third of a cent more expensive to use a controller that can switch above 1KHz. All hail the glorious race to the capitalistic bottom.

At work I just finished up a constant current driver circuit. At home I'm building a custom lighting system with bespoke driver circuits. Despite having a CC driver I can pull off the shelf, I still chose 10KHz PWM. It's easier and more efficient, and neither you or I could tell the difference in the quality of the output light.

What about the IKEA light drivers? They claim to be flicker free.

Another optical engineer here.

I used to wonder WTF was so hard about LED lighting. One R, one G, one B, a shunting resistor and... and then I took a glance under the hood. Oh. Yeah, much dicier than heating up some tungsten and frosting the inside of the bulb.

(Yeah, inventing cheap blue LEDs was some serious heavy-lifting in itself, but that hurdle has been jumped.)

> (at least usually).

Nope. Lots of people see the strobing. Its causes headaches if you focus on the lights.

Reply on news.ycombinator.com