Hacker News new | past | comments | ask | show | jobs | submit login

The efficiency claims of these are very nice:

Sulfur Lamp: 100 lumens per watt. [0]

Random LED Light bulb: 88.888 lumens per watt. [1]

~20% improvement.

[0] https://web.archive.org/web/20030818061414/http://195.178.16...

[1] https://www.amazon.com/TCP-Equivalent-Light-Bulbs-Non-Dimmab...

That's very unfair in that there's LEDs on experimental lab benches running 200+ lumens per watt (very handwavy). Something tells me you won't be getting amazon prime shipping on a 100 L/W sulfur bulb anytime before you're already getting amazon prime shipping for 200 L/W LED bulbs... Theoretically LEDs top out just under 300 L/W under ideal lab conditions, so a theoretical max of 100 L/W for sulfur is not so good.

The real competitor of these "ten kilowatt" class bulbs is theater-type arc lamp type devices where you can't deal with the weird spectral effects of LEDs and other quantum mechanical goofiness... Today you can buy COTS xenon lamps that run around 40 or so watts/lumen, so 100 L/W is a nice upgrade for stage and film theaters.

I don't know what sportsball stadiums use for night games. For raw bulk ugly illumination, sodium discharge lamps are COTS around 150 or so L/W, far surpassing what the sulfur bulbs can do in the lab. Interesting thought experiment is immense capex was dropped to illuminate sportsball fields and replacing the lights makes half the demand go away... so are sportsball fields going to be twice as bright at night or recycle all that power company infrastructure or is legacy pro sportsball going away with the boomers before sulfur bulbs could arrive commercially or ?

I don't see an easy way to decouple the magnetron and stuff from the lamp for sale purposes. A kilowatt class xenon lamp runs over a hundred dollars and figure a kilowatt class microwave oven and its innards could double the cost of a sulfur lamp while halving the cost of the electricity... this is really bad news for theater goers... it'll make theaters even more capital intensive which means even more risk adverse. If you're tired of formulaic remakes now, imagine when financial risk is twice as high due to double the bulb cost, LOL.

This got me wondering what's the theoretical limit to lumens per watt and I found this:


My theater knowledge is dated, but there’s a major potential benefit for LED in theater: they can modulate quickly.

I’m familiar with two kinds of high-power theatrical lighting: tungsten and discharge. Tungsten is inefficient and has poor power factor when dimmed. Discharge lamps can’t cycle quickly, so, as a practical matter, they are turned in well before a show starts (to make sure they all work!) and stay on, at full power, until after the show. A motorized shutter modulates the light output. This wasted tons of power when the shutter is closed.

LEDs can cycle essentially arbitrarily quickly, and a good driver gets a power factor near 1.

I think it would be interesting to design a theater lighting system with a DC bus. Tungsten lights (where needed) would be driven by PWM, and LEDs would be driven directly by PWM if the voltages matched or with a DC-DC converter otherwise. There would be a battery to even out the load to minimize demand charges.

I don’t know what a three-phase AC-to-DC converter rated for, say, 100kW would cost. You’d want one that can be controlled such that it can share load in a controlled manner with a battery on the DC side.

What benefit does a DC bus have over decentralized AC-DC converters? You'd still need fairly high voltages to keep current manageable, and high-voltage DC has its own set of "fun" challenges.

I haven’t made any effort to analyze this for real, but:

A circuit to drive a tungsten lamp at variable brightness off AC is cheap (a TRIAC and its driver) but has crap power factor. I imagine a MOSFET to drive the exact same tungsten lamp from DC at a few hundred Hz PWM would be almost as cheap and could be filtered to have a clean input waveform. (Or it could be driven above 20kHz to eliminate the annoying audible buzz that tungsten lamps make.)

A big LED driver (AC to DC, constant current or constant voltage plus some dimming mechanism) is not particularly cheap, and the nice ones are largeish. A pure 24V “dimmer” (12-48V in, same voltage out, adjustable PWM frequency, and DMX controls) can be had quite cheaply with excellent performance. A constant current driver would be more expensive, but a DC-input driver ought to be smaller and less expensive than a comparable AC-input driver, especially if you care about power factor and inrush current.

(Common commercial AC-input LED drivers often have godawful inrush current, because they have big filter caps to get good performance, and they don’t spend the extra money on fancy circuitry to limit the inrush current. This means that you can’t actually put 20A of driver on a 20A breaker because the inrush current will trip the breaker. With DC input, a much smaller filter cap should give comparable performance.)

Anyway, this is all mostly speculation, and economies of scale matter. But those DC DMX dimmers really do exist.)

If by theater you mean cinema, I'm pretty sure state of the art lighting is using lasers. Centralized laser in fact, where a single laser drives many projection rooms through fiber optics.

By theater I mean people doing things on a stage with stage lighting.

> The real competitor of these "ten kilowatt" class bulbs is theater-type arc lamp type devices where you can't deal with the weird spectral effects of LEDs and other quantum mechanical goofiness...

Sulfur lamps have CRI < 80. High CRI (> 90) LEDs are real competitors to HID lamps here.

Ultrabright water cooled LEDs have swallowed whole all remaining market niches for this technology.

LEDs are cheaper, and, more importantly, already there.

This is some kind of commodity now?

Link where I can buy one?

Similarly, sodium lamps get 100-150 lumens per watt. https://en.wikipedia.org/wiki/Sodium-vapor_lamp

Good for low light pollution too.

Popular in Australia; are they popular elsewhere?

What fraction of the lifetime cost of lighting is the power, vs. things like the manufacture, installation, maintenance, and disposal? Having a magnetron up there seems like a pain. Also, the noise?

The magnetron dies, and the thing doesn't scale down to a keychain flashlight like the 88lumn/W LED. We do have LEDs with >130lumen/W though...

Plus a slightly longer lifetime (60k hours vs 50k for LED)

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact