I just finished testing over 90 different light bulbs...

itchaboyagain

Hi guys! My name’s Derek and I’m a light bulb addict. looks around the room nervously

I’ve spent the last few months putting together a large database of light bulbs (93 as of this moment), basically any of them I could find.

I’ve lurked here for a while so I’m excited to share this! Hopefully, some of you find it useful.

As many of you are aware, the lighting industry is currently doing whatever they want with regard to flicker, and worst of all, they can say what they like and we just have to take their word for it.

Even companies like Waveform which perform well with flicker, don't release the actual data on their lights.

With this database, I hope to give you and everyone else, access to the information that these companies don’t seem to want to give us.

This includes the flicker index, depth, and waveform, as well as the light spectrum itself and a few other metrics.

One interesting thing I discovered is that incandescent/halogen lights completely stop flickering after the 72w mark! Very cool. I thought they all flickered, but apparently not.

There are of course lights that I have not yet tested, such as Yuji’s MaxCRI and Sunwave lineup. They're just so damn expensive…

(Yuji if you’re watching send me samples I won’t be mad)

As well as random GE and Phillips bulbs and many other smaller brands.

Hopefully, over time the database will just get larger and larger and maybe we can convince these companies to take their lighting more seriously once their dirty laundry is aired for everyone to see.

Any thoughts, suggestions, or comments are welcome!

Enjoy!

💡 Explore the Light Bulb Database 💡

Sunspark

Nice work, did you measure all these numbers yourself like flicker wave, spectrum, etc. or did you use datasheets?

itchaboyagain

Sunspark I took all of these measurements personally!

seeker_of_no_light

I really like how you can quickly gather the information from the table format. If you measured these all by yourself, which device did you use to do so? From the supplied charts, I assume it's the Opple Light Master 3, since another member here has made a similar website using that device pretty recently.

itchaboyagain

seeker_of_no_light Oh very cool I wasn’t aware of that site! Yes that’s what we’re using, nifty little tool for the price!

Gemsand

This is amazing! I sent a small donation in appreciation for the value of this! I dream to find something like this for LED lights used as backlights.

itchaboyagain

Gemsand Oh that was you! Thank you very much for that 😃

By backlight what do you mean?

ryans

Thank you @itchaboyagain for this. I sent a small donation a few weeks ago for this awesome resource.

I appreciate testing halogen / incandescent lights as a comparison too, but the results were not what I expected.

Some of the halogens like the GE 29W Halogen shows 10% flicker at 120Hz which I expect (60 * 2 = 120Hz, for USA electric):

However, the GE 72W Halogen shows a much higher flicker rate of 10,000Hz and 0.2% flicker:

I wonder if anyone here knows how to explain this? Why one halogen has 120Hz flicker and another 10,000Hz Flicker?

GregAtkinson

This database is so cool!

I think I'm extremely sensitive to small amounts of rapid flicker (I think flicker as low as 1% can affect me if there are abrupt changes in brightness). Though I'm not 100% sure about that.

I didn't want to shell out $64 for a two-pack of Yuji bulbs (but if they eliminate my headaches then it would be totally worth it). So I bought the Philips Ultra Definition, 2700K, 1100 lumens 11w bulb.

I don't have the Opple LightMaster, but here's the screen shot from @itchaboyagain's site.

Image with no description

Here's a screen shot from my setup (photodetector + oscilloscope) on DC mode. It shows pretty much the same thing. Mostly flat with dips at 120hz.

However, on AC mode you can really see how much it flickers. Its range is 25 mv (goes up to almost 10 mv and down to a bit more than -15 mv).

So the flicker percentage is about 4% (since on DC mode you can see that it's generating a bit more than 600 mv). And the drops are pretty sharp/jagged, indicating abrupt changes in light output.

It seems like the main drawback of the Opple LightMaster is that it doesn't have the AC mode (or at least I'm not aware that it does), so you can't zoom in.

Here's the bulbs that I use around my house. They are the Maxlite 60 w dimmable 2700k A19

As a side note, here's a typical incandescent bulb that flickers about 10% but has very gradual changes in brightness.

JTL

GregAtkinson So the flicker percentage is about 4% (since on DC mode you can see that it's generating a bit more than 600 mv)

What formula are you using to calculate flicker percentage exactly?

GregAtkinson

JTL What formula are you using to calculate flicker percentage exactly?

Unfortunately I've been inconsistent in my posts. A few weeks ago (including the post above) I was using (range)/(max) which is equivalent to (max-min)/(max). Over the past week or so I've been using (max-min)/(max+min), which seems to be more commonly-used.

KM

I think we should use (max - min) / max. The reason is that the formula (max - min) / (max + min) is not linear. It produces increasingly less differences the smaller the flicker gets. But we are especially interested in small flicker values, too. So dividing by (max + min) hurts our interest of making even small flicker values reasonably comparable.

The first thing I did after purchasing the Fauser LiFli flicker detection device was to change its flicker percentage calculation from "/ (max + min)" to "/ max" (it offers this as a feature). Because the lowest percentage on its scale is 2% (which actually means 0% - 1.5%), and if I used "/ max + min", too many light sources would fall into this range. Light sources which when using "/ max" stay well above that range and are ones we probably want to stay away from.

To me it seems "/ (max + min)" was an arbitrary choice to somehow account for the form or area of waves, too (in a very rough way).
You can calculate a few test values to realize how (max - min) / (max + min) is producing values even for stronger flicker that are not very helpful for comparison (for example PWM that only goes down half of the amplitude would be equal to 33% flicker, when we'd naturally expect 50%) and how quickly calculation results approximate each other the smaller the flicker gets.

IMO the only reason to use this formula is because as you say, sadly it seems to be widely adopted already and could make it easier to compare to some manufacturer's values. But again, for smaller values the differences become increasingly tiny, and no manufacturer is even bothering to print multiple positions after decimal point. So the smallest official value we could compare to is probably a very rough one like "< 1%" anyway.

GregAtkinson

KM I think we should use (max - min) / max. The reason is that the formula (max - min) / (max + min) is not linear. It produces increasingly less differences the smaller the flicker gets. But we are especially interested in small flicker values, too. So dividing by (max + min) hurts our interest of making even small flicker values reasonably comparable.

I totally agree! I'll use "(max - min) / max" and mention which formula I used going forward.

GregAtkinson

Maxlite changed the manufacturing process and now the new 60w 2700k bulbs flicker pretty badly. There are a few old, non-flickering ones on eBay. Here’s an eBay listing that has them for about $2 per bulb with shipping.
https://www.ebay.com/itm/155900291630

Here's a couple pictures showing the differences. the newer (bad) box is smaller; the older (good) box is larger.

I bought about 150 bulbs from various sellers that will hopefully last until my kids put me in a nursing home. It’s a large box, but still a very poignant reminder of my limited time on earth.

jen

GregAtkinson I tried the Maxlite 2700K 60W bulbs a couple weeks ago, but as @GregAtkinson and I expected they might given the spiky shape of the flicker waveform, they made me sick with concussion-like symptoms, although were definitely better than LEDs with high flicker. They were similar to other low-flicker lights that still injure me (data here). These were the MaxLite 2700K A19, E9A19D27/49/WS1, Code 14099718. So maybe they have the new, worse engineering? These had 0.17% flicker (calculated in the standard, misleading way used by the lighting industry).

0.5 seconds (UPRTek flicker meter) with the x-axis at y=0:

0.05 seconds and zoomed-in y-axis:

GregAtkinson

jen So maybe they have the new, worse engineering?

Unfortunately these are the old/better engineering. The new ones are way worse 🙁

Berezny

Your discovery about incandescent/halogen bulbs after the 72W mark no longer flickering is a fascinating tidbit. I bet that kind of data is exactly what lighting enthusiasts and professionals are looking for but struggle to find, especially those in part time electrical jobs. Will your database be accessible to the public soon? I imagine a lot of folks would love to dive into the data.

async

Any way to make it easier to find the ones available in European versions, or similar decent lists that does?