I went to a hardware store and picked up six inexpensive small plastic sockets (these bulbs generate virtually no significant heat), mounted and wired them side-by-side on a 12-inch strip of flat black-painted plywood, and glued some small neodymium magnets to the bottom (from used electric toothbrush heads, I always salvage those as they are incredibly useful).
I then folded a small sheet of Rubylith cylindrically domed over the top of the strip of bulbs and attached at the ends to curved black vertical cardboard supports. The magnets allow the unit to firmly attach to the top sheet metal center of my Thomas Duplex unit up by the ceiling.
My small darkroom is now relatively bright red when I use the LEDs, and of course insanely bright orange when I use the Duplex.
Interestingly, I also managed to perform the same trick with the Duplex by using an inexpensive sheet of Roscoe Roscolux #19 (Fire) filter material as a cutoff filter.* The emission spectra for a low-pressure sodium vapor lamp includes numerous additional spikes in the blue and green portions of the spectrum. You don't see them with your eyes, but the photo paper sees them. Get rid of them and you can significantly increase the safety of the safelight.
This is why many people complain that the Duplexes are "way too bright and fog my paper" no matter what they do. It's not the intensity of the sodium emission doublet, providing your paper is Kodak OC safe in the first place. It's the additional blue/green emission spikes at that high intensity. Add a cutoff filter of the right transmission wavelength and the fogging all but disappears.
When I was using Kentmere Bromide graded papers I was able to leave the sodium safelight vanes open all the way and preflashed Bromide #2 paper showed no fog out to the 30 minutes I ran the test. I don't know how far it might have gone, but the room was lit up like I was using a standard 100-watt lightbulb.
That much truly "safe" light is a bit disconcerting and makes you feel uneasy, even though you know it has tested safe. I did find that I had to close down the vanes to compose and focus on the easel as the Duplex safelight completely overwhelmed the projected image.
* See the technical data sheet for this filter here. The transmission graph demonstrates why this material is so effective as a cutoff filter for the higher frequency blue and green spikes. I looked at all of the possibilities and this color was the best choice.