On Thu, 11 Oct 2012 05:55:15 -0700 (PDT), George Herold
wrote:

On Oct 10, 11:50*pm, Jeff Liebermann wrote:
On Mon, 8 Oct 2012 06:27:05 -0700 (PDT), George Herold

wrote:
Hi SEB. *Well I sent an email to Don klipstein on this topic. *And
have permission to copy his reply.
from Don K. below
2: *In incandescent traffic signals, the bulbs for yellow last
longer than for red and green. *So even after being switched on
and off about a million times, on-time is still a significant
factor in life expectancy.

That means for my proposed test comparing a 50% duty cycle flashing
light bulb, with one that is on continuously, the continuous light
bulb will burn out first. *That's the opposite of what I saw with the
theater marquee bulbs. *Now, I'm really tempted to run the experiment.

Experiments can be very useful.

Yep. However, it's more fun to predict, speculate, guess, reverse
engineer, and maybe calculate.

I'd worry most about how you turn on the bulbs.
Maybe just some simple relays?

No. I didn't want to life test the relay contacts, just the light
bulbs. I have plenty of solid state switches that will suffice. The
reason I wanted two was to make sure the voltage drop across the
switch was the same for both the flashing and continuous bulbs.

1,000 hours isn't all that long. (or are you going to over-voltage
the bulbs?)
I guess I'd want at least 10 bulbs in each group. Say 60 watts.....
1200 kW-hrs.

Much as I would like to use a rack of bulbs,
http://pinterest.com/pin/172122016978363241/
http://pinterest.com/pin/172122016978761590/
I think two bulbs will suffice for a start. The plan of the moment is
to use a variac to boost the voltage from 120VAC to about 135VAC,
which should reduce the 1000 hr life to a more tolerable 112 hrs.
Cut-n-paste from a previous posting:

Instead, an accelerated life test can be done with higher than normal
voltages.
http://www.welchallyn.com/documents/Lighting/OEM_Halogen_Lighting/MC3544HPX_Catalog_2_11_09.pdf
For halogen bulbs, they use:
Life = (Vdesign / Vapplied)^12.0 * Life at design voltage
For a 1000 hr lamp running at 120% of the rated voltage, the life
might be:
life = (1/1.2)^12 * 1000 = 112 hrs
which is more reasonable.

That's looking like more money than I'd want to spend on the
electricity.

I'm trying to determine where to run the test. I don't want a
flashing bulb in my bedroom. I also don't want to run an unattended
rack of bulbs in the office which could become a fire hazard. Methinks
just 2 bulbs and a 3-4 day accelerated test will be sufficient.

George H.

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558