Incandescent lamp AC vs DC

last.

They were thermistors. Their resistance went down as they heated up, which reduced the turn on current. One brand name was 'Thermodisk'.

--
Lead free solder is Belgium's version of 'Hold my beer and watch this!'

Apparently this was a 110 V lamp ?

The 6.5 V filament is much thicker and would have a very large thermal inertia, so one would expect that the intensity variation to be far less.

Yes. Sorry not to have included that comment in the above.

I didn't try those at the time. Just commenting about what I did try.

Jon

The way I heard it, more notably incandescents have a slight trend of having shorter life expectancy with DC than with AC.

The explanation is that a very small percentage of the tungsten vapor near the filament gets ionized by the small amount of UV produced by the filament, causing a drift of tungsten vapor towards the positive end of the filament.

That causes the negative end of the filament to suffer slightly more from evaporation than the positive end.

Of course, there are other reasons why the ends of the filament may suffer unequally from evaporation:

1. In gas-filled lamps, there is often a convection current. If the ends of the filament are unequally affected, then one end can get hotter than the other, or be surrounded by gas lower in tungsten vapor than the other end, or one end may have more wind from the convection current than the other.
2. An incident of mechanical shock may have stretched the ends of the filament unequally, causing the ends to have unequal temperature.
3. One end may have more filament radiation reflected back to it than the other.

=====================================

On the other hand, if the filament is thin enough and the AC frequency is low enough, then the filament's temperature can vary significantly over each half-cycle of AC. Increase of evaporation during temperature peaks will outweigh decrease of evaporation during temperature dips. That will cause life expectancy to be shorter with AC than with DC.

- Don Klipstein ( snipped-for-privacy@misty.com)

In , Sylvia Else said in part:

For most filaments, the frequency would have to get to the megahertz ballpark, at least hundreds of KHz, maybe a few megahertz, to have skin effect significantly dim the filaments. AWG 30 copper (diameter approx. .01 inch or approx. .25 mm), at room temperature, has resistance not a whole lot more at 100 KHz than at DC. Tungsten has more resistance than copper, and at typical filament operating temperature has roughly 15 times as much resistance as at room temperature.

- Don Klipstein ( snipped-for-privacy@misty.com)

950 nm is just a little on the short wavelength side of peak spectral power distribution per unit bandwidth wavelength for most tungsten incandescent lamps.

Amount of radiation per unit bandwidth in this case should be slightly more than proportional to temperature to the 5th power.

(Total radiation is proportional to temperature to the 4th power, and total bandwidth in wavelength terms is inversely proportional to temperature, and the peak usually gets closer to 950 nm as temperature increases.)

If the amount of radiation in a 120 nm band centered at 950 nm varied by

3% over an AC half cycle for that particular tungsten incandescent lamp, then I expect the filament temperature for that particular lamp varied by slightly less than .6%.

I also expect visible light output varied more. I seem to think that a 120V 60W 1000 hour 845-890 lumen "A19" incandescent lamp has filament temperature around proportional to voltage to the .39 power, maybe .4, and light output proportional to voltage to the 3.4 power, maybe 3.5.

(based on, "extrapolated a bit from my knowledge", from what a 100 watt

120V 750 hour 1670-1750 lumen "A19" does, as best as I know as reported in

That means photometric output being roughly proportional to temperature to the 8.7 power at 2870 K or so, very slightly more at the slightly lower roughly 2800 K of the 60 W lamp that I described above.

Based on this, I would expect the 60W lamp that I described to have photometric output varying roughly 1.5 times as much throughout an AC half-cycle as such lamp's 950 nm output does.

- Don Klipstein ( snipped-for-privacy@misty.com)

I have had experience with those "discs" of two types:

1: Diode: Power consumption reduced to ~58-60% of "normal", light output reduced to ~26-29% of "normal". Life expectancy has good chance of being increased by a factor of 50-90. 2: Thermistor: My experience is that those "soft-start" tungsten incandescent lamps and claim to double their life expectancy.

I did one notable experiment where I found that when the "disc" was fully warmed up, it had enough resistance remaining to dim the lamp to an extent worth a goodly 50% life extension, maybe 55%. That means

10%, maybe 11% less light while combined power consumption of the lamp and the "disc" is at most 2% less than that of the lamp alone.

By any chance is this the thingy that I somewhat remember as an unfiltered 6-diode rectifier for "Y"/"wye" 3-phase AC? (4.5% less average voltage and similar less RMS voltage than with filter capacitor, 6.8-6.9% less power consumption than with filter capacitor, ~13-15% less light without filter capacitor than with, while life expectancy is increased maybe 70-80%?)

- Don Klipstein ( snipped-for-privacy@misty.com)

I thought your plan was to run it on DC anyway.

DC would surely have to be kinder to the filament by removing the cyclic thermal stress.

Sylvia.

No, I was trying to see what the pros and cons (from the lamp's perspective) of AC vs DC drive would be.

E.g., DC is easiest to "get done" -- dig through box of "bricks" looking for 20W @ ~6V, cut off existing "DC" connector, attach appropriate connector, done!

And, DC is easiest to *dim* (since that *should* lengthen life expectancy of the bulb).

But, before investing the time to do this, I wanted to be sure there were no other issues that would make this an unwise approach.

E.g., Spehro's posted URL suggests DC related failures. I, OTOH, had always *assumed* AC presented more mechanical stress to the filament.

The zinger is the "100 hr" life expectancy. Sheesh! Unpopped popcorn lasts longer than that! :-/

last. A

Could you please elaborate on this? ^^^^^^^^^^^

incandescents.

--> DC shorter than AC

Suggesting that the actual orientation of the filament (wrt gravitational field) would exacerbate the problem)?

--> AC shorter than DC

I suspect *the* solution is to just invest in a small sh*tload of spare bulbs and live with the consequences! :-/

Or, find a solid state alternative and some way to position it at the correct focus :-/

(sigh) Nothing is *ever* easy.

Except that my experience suggests that thermal cycling is less damaging than most people think.

I can say why specifically for a couple of issues:

1. It is well-known that incandescents often burn out during cold starts. That leads to reasoning that starting is what does the damage.

However, there is another explanation: The filament evaporates unevenly. A thin spot that runs hot develops. The higher temperature of the thin spot causes it to evaporate faster and become thinner at an accelerating rate.

Such a "fatal hot thin spot" does incur a temperature overshoot during a cold start, due to less mass and tungsten's positive temperature coefficient for resistance. Such a "fatal hot thin spot" becomes easy to melt during a cold start.

However, by the time a filament with such a condition becomes unable to survive a cold start, its hours are numbered. This bad condition accelerates worse than exponentially during steady operation.

1. I did an experiment with a thermistor type soft-starting "stick-on" "button" ("my words") that claimed to double the life of an incandescent lamp. I found that when it was fully warmed up, it dimmed the lamp enough to account for 50-60% increase in life expectancy. (I used to post 50-55% back when I was working with a rule of life expectancy being inverse proportional to voltage to the 11th-to-maybe-12th power, as opposed to most likely the 12th.) Should the device actually meet its claim of doubling life, it only improves life by 25-33% by soft-starting according to the one experiment I did. (In my experiment, current draw was reduced 2%, power consumption of lamp and device combined was 2% less than lamp alone, and light output was reduced almost 12% - based on the life-extension-figure that I remember better. Actual voltage measurements from the experiment I posted in sci.engr.lighting shortly after the experiment several years ago.)

I am now aware of some railroad crossing flashing signals that put a resistor in series with the lamps for the first whatever fraction of a second into a flash, and I have posted before as to my doubts when someone else posted about their existence. However, I still doubt the soft-starting gains a lot.

1. Traffic lights - incandescent lamps used in those go through usually a cold start once every minute. My experience is that incandescents used in those at least usually have risetime fast enough to indicate lack of effort at soft-starting. And traffic signal lamps appear to me to meet their life expectancy in red and green traffic lights, exceeding for yellow, and to not greatly exceed their life expectancy in Philadelphia's alley lights where the duty cycle is about the same as that of red and green traffic signals and the switching frequency is less by a factor of usually 1,440. (And yes, I am counting only on-time for operating hours, so a 130V traffic signal lamp rated for average lifetime of 8,000 hours should have average life expectancy a little over 4.5 years, and a 120V one should have average life expectancy a little over 2 years in red/green traffic signal duty or Philadelphia alley light duty.)

- Don Klipstein ( snipped-for-privacy@misty.com)

I would think that just means that starting is when the lamp is most susceptible to failure due to the accumulated damage.

Sylvia.

So, your point would be that AC drive tends to disturb where this spot would want to otherwise develop? But, can that be relied upon? Or, are manufacturing variations sufficient to render this moot?

But, presumably, "gains *some*" (or the series R practice would be foolish)?

Are you sure they really power off (completely)? E.g., that perhaps they don't keep the filament "warm" though not visible? (I don't know if operating the lamp at such reduced power would make a difference -- I have a friend who does that sort of thing, I should inquire)

Of course, bulb replacement has to be the driving force behind the rise in popularity of LED traffic lights.

But using filaments designed for higher operating voltages at lower voltages increases longevity (at the expense of efficiency) so how do we know they aren't just having "custom bulbs" manufactured for this purpose? I.e., the "customer" is large enough to warrant same...

Hmmm... I wonder if synchronizing the turn on with zero-crossings has any *measurable* effect (or if it is so far down in the noise as to be warrantless)

Actually, I was not considering AC-vs-DC since most filaments running on AC appear to me to have temperature sufficiently steady to not incur significant thermal cycling stress, which I was arguing against degree of existence from cold starting anyway.

However, now that you bring my mind to this, use of AC as opposed to DC (back to what I said a few days ago or whatever) could prevent one of these fatal thin spots from forming at the positive end of the filament (I hope I said positive end before). And likely mainly in vacuum-containing incandescent lamps as opposed to gas-filled ones, but of current rating high enough for the filament temperature to be fairly steady throughout each AC half-cycle. Then again, I expect the difference to be minor, merely measurable.

My impression is that the slower start (dims the filament for several percent of each "flash" from slowing of risetime) and mild dimming for a significant fraction of each "flash" achieves somewhat significant life extension. I do think that doing this instead of merely dimming the lamp slightly (even with a series resistor) is foolish but done due to strong belief in cold starts being more damaging than they usually actually are.

That would make a major difference in later part of the fall time, and I don't see this happening. I also don't see sign of leakage through LED replacements of the incandescents in Philadelphia.

I have heard of this being done in some stage lighting setups. I am aware of halogen lamps sometimes being different - by sometimes having filament notching at the ends of the filament producing thin spots that are not excessively hot in steady operation but that do overshoot in temperature during a cold start.

Although it has to be a major force, especially in older major cities where the lamps may have to be replaced by unionionized municipal employees, I hear LED traffic signals being sold to the taxpayers mainly from energy savings. And the energy savings can be substantial, since LEDs normally specialize in producing light of one color or another, while red and green incandescent traffic signals need something like 70% of their photometric output to be blocked by a colored filter. Another reason for energy savings is that incandescent traffic signal lamps tend to be significantly less efficient than 750 and 1000 hour household lamps due to being designed to last typically 8,000 hours, and also due to having a filament of more-vibration-resistant design that has greater heat consuction loss. A 92 or 116 watt incandescent traffic signal lamp is typically replaced by an LED one that consumes 11-16 watts or so.

Savings of 80 watts at 50% duty cycle, at even 8 cents per KWH, easily amounts to $28-$30 per year per lamp.

I used to trashpick from the dumpsters of the contractor that relamped Philadelphia's streetlights and alley lights. The alley lights used 92 watt traffic signal lamps.

I have peeked into opened-up traffic signals in Philadelphia a few times before, and found the color of the filaments strongly indicating that they are either 120V or 130V traffic signal lamps. The life expectancy I am seeing appears to me more consistent with 130V. And my "day job" is delivery biker in Philadelphia, I went to school in Philadelphia and I spent close to half my life living in Philadelphia.

- Don Klipstein ( snipped-for-privacy@misty.com)