It varies depending on the light type. LED HID retrofits usually need bypassing, but ballast compatible lamps exist. In fact, Home depot sells an LED filament lamp that runs straight off an H39 ballast for yard blasters. Bypass lamps are better IMO because of less failure points and components wasting energy. Same goes with fluorescent, although ballast compatible fluorescents are fine from my experience.

Most cases I've seen for HID fixtures, the ballast is left in place and simply bypassed electrically. One example is with this Spartan branded floodlight that came off a Deli during an exterior remodel and LED changeover. The floodlight was originally 250 MH, but had an LED Corncob lamp made by Light Efficient Design( which unsurprisingly was not as bright). Much to my benefit, the 250w metal halide ballast was still in there and functional, I just had to do some rewiring.

With fluorescent, I've retrofitted one of my basement recessed 4X F40T12 fluorescent fixtures with Feit ballast compatible LED tubes because I was fed up with the temperamental Rapid start ballasts. Most plug and play LED retrofit tubes are electronic ballast only, but these Feit lamps are compatible with magnetic (haven't tried with preheat yet)I got them from Home depot.

Idk if this is really off topic but whatever.

Correct me if I am wrong, but I believe that the only thing inside of "normal" glass that makes it block UV is trace iron impurities. I believe those iron impurities are also what gives many glass products their distinctive teal-green tint if you look closely. Therefore (again if I am thinking correctly), covering a harmful UV lamp with glass that looks teal will provide UV shielding while covering it with glass that looks more bright and clear has a risk of not sufficiently shielding against UV.

Lots of people here use drinking glasses as shields against smaller UVC lamps like the quartz cold cathode ones and the smaller hot cathode ones like the G4S11 and GTL3. But not all drinking glasses have higher iron content, many of by (some fairly thick) glasses at home have no visible teal coloration at all, meaning they might not provide protection.

Has anyone ever encountered this?

I have retrofitted two fluorescent fixtures above my workbench to LED. To be clear, I am only talking about ballast-bypass type retrofits here. When I retroftted them, I took out the ballast. Not only because I wanted to mess around with it, but also because it made it lighter and easier for my skinny self to put back up. 

But for things like streetlights that get retrofitted, obviously they aren't taking the whole thing down to retrofit it. So I would think it would be possible to leave the ballast physically in there but electrically disconnected to make the job easier and faster. Is this ever done?

Also with multi-tap ballasts, are they ever left in as voltage conversion autotransformers to power retrofit lamps requiring a different voltage than the supply?

 All we got was a decent amount of wind , a crapload of lightning , and and an unbelievable amount of rain !! Looks like the really bad stuff got the north side of Columbus .

Nice experiments, RRK. A few minutes proably won't hurt the tube. Even crap  chinese flyback power supply fixtures for 4, 6, 8 W tubes need several hours for a significant blackening.
If they had neon in the gas filling it would probably be easy to notice like in Nixie tubes and some CCFLs.

So bottom line is at least my Osram HO tubes are actually pure Ar filled. I do not have any other branded T5 lamps like Philips or GE or Sylvania on hand for now to compare. Longer lengths of T5 HEs really have a bit of Kr added, very likely to keep burning voltage reasonable, and may be to improve efficiency a bit as well. Economy types of T5 can be reasonably expected to have some even more Kr added to drop lamp voltage further.

But the biggest surprise for me was that I was really able to monitor cathode sputtering going on in the tube clearly by detecting barium and strontium lines in the light captured near the electrodes! 

Yep, I use a nice alternative program written in Rust:  https://github.com/DerFetzer/spectro-cam-rs

Spectro-cam offers calibration. *But* I do not believe a webcam spectrometer can be satisfactory calibrated by spectral response because camera definitely demonstrates strong non-linearity. May be if you take care of selecting a fixed gain (exposure) and disable all dynamic range adaptations in control panel. If this is possible at all...

Regarding reddish color with the cathode part of a barrier discharge. Yes I noticed this as well. A probable explanation is that cathode part of the discharge often shows significantly different excitation of different gas wavelengths due to much higher electron temperature. And different balance of UV + visible lines may mean slightly different color balance of triphosphor components radiation (plus visible light from the discharge). Remember classic Xe/Ne filled plasma balls, where cathode parts clearly show red Ne excitation and discharge columns are blue-gray xenon?

From the recent experiments: here you can compare a spectrum captured from TN20 neon lamp (mostly cathode glow, see orange 585nm Ne line is excited preferentially) and a spectrum captured off a discharge column of a neon sign tube, 585nm is weak and ~640nm + a line cluster around is radiated as much redder light. A confounding factor to note here is also TN20 is filled most likely with Penning mix while the sign tube is 100% neon - here i am sure as I filled it myself

To see the neon glow through the phosphor coating is not easy.   you often have to place the lamp in a deep freezer to solidify the mercury and then it becomes much easier.  On the other hand, all of these gas fillings seem to have changed so frequently and between different manufacturers, it’s possible there is a different fill in your Osram lamp.

I am sure the detectability depends on gas fill Ratio as well, but I find with known neon filled lamps, you certainly can see the Neon glow through the Phosphor coating, for example when I tickle my GE F96T12/VHO with my Martindale I can very much see the Neon buffer gas being excited at the barrier discharge point



(as an aside I notice interestingly with Triphosphor lamps even those most certainly not Neon filled, the barrier discharge seems to preferably cause the red phosphor to glow, I wonder why that is!)


BTW @RRK which spectrometer program is that? looks like by the USB Device name your still using the little-garden spectrometer, but a different program? does that program offer an option to calibrate the spectometer against a known CCT Black body radiator?

Nice experiments, RRK. A few minutes proably won't hurt the tube. Even crap  chinese flyback power supply fixtures for 4, 6, 8 W tubes need several hours for a significant blackening.
If they had neon in the gas filling it would probably be easy to notice like in Nixie tubes and some CCFLs.

Next time when I'm kicking it with friends at a bonfire, I'm going to try toasted marshmallows alone without the chocolate and graham crackers.
I have a feeling that might be very good.
But yep, fire burned smores is absolutely insanely delicious to me. Heck, fire burned foods in general seem to be incredible... even the fire burned
hotdogs were an easy 9 out of 10.

Marshmallows don't have a good taste in me. Though yes I agree, they taste good when you burn them a bit in the firecamp 😀