Re: Dissecting CFLs

["Robert Green" <ROBERT_GREEN1963@xxxxxxxxx>]

"Andrew Gabriel" <andrew@xxxxxxxxxxxxxxxxxxxx> wrote in message

<stuff snipped>

> > I was surprised to see no apparent component damage.  There's a nick in
the
> > transformer insulation, but I think it was my doing with the Dremel.
The
> > can-type capacitor had a slight bulge on top, but nothing like what I've
> > seen in dead ones.  While this is not the old lamp from the Taiwanese
> > formula theft era, the capacitor looked like those that I have seen on
> > failed motherboards.
>
> So do all electrolytics.

Exactly my point.  How likely is it that the bad caps *only* found their way
to the PC motherboard makers?  I think it's more reasonable to assume that
they'd also wind up in lots of other electronic devices of that era.
Somewhere, I still have the MB with the bad caps.  They not only bulged, but
leaked yellowish gunk.  That may still be what I find when I open the older,
failed GE's and Lights (out!) of America bulbs.

I recently pressed the last large wattage LoA bulb I had from very old
inventory into service.  It was running in a photographer's 12" aluminum
reflector with a ceramic base and lots of air vents.  After less than a
month in service, it failed, emitting only a faint glow from a single tube.
It was running from an X-10 appliance module and operating at about 45
degrees downward from horizontal.  ]

At first I thought the entire batch of LoA bulbs was bad but out of five,
four have failed and one still runs 6 hours a day over the terrarium, also
controlled by X-10.  It operates base up and is fairly poorly ventilated
compared to the other fixtures where the bulbs failed so "go figure."  I'm
hoping another bulb autopsy will reveal a problem that might have a
solution.  I might try to laughingly return them to Wal-Mart to see if they
really will do anything but laugh at me before I make them unreturnable via
Dremel.

> Capacitor failure is common in CFL's because they are operated
> above their temperature rating. This isn't normally a problem
> because they are only required to work for something like 10,000
> hours, whereas an electrolytic operating within its temperature
> range would last very much longer than required in a CFL.

Hmmm.  That would explain why base up failures are so common.  If the
devices are operating close to their rated tolerances base down, adding the
bulb's heat to the electronics would certainly hasten the doom of any
temperature sensitive components.  I suppose the tradeoff is one of space
vs. thermal "headroom."

The small N-vision bulbs that are almost exactly the size of a standard
incandescent work in standard reflectors quite nicely, but the CFLs that
stay within the design space of a standard tungsten bulb (at least in the
US) are too dim for my taste.  It's clear that the electronics are getting
smaller but it seems that there's a direct correlation between light output
and component bay volume and there may be a physical limit to how small they
can shrink that base.

The problem seems to be that to run at the tungsten equivalent of 100W, the
electronics bay gets large enough to interfere with proper air circulation
at the base.  I suppose that could be tested by running two bulbs, base up,
side by side and cooling one with a fan.  Another science project.  Anybody
with any teenagers in need of science fair ideas?  (-:

> I've seen a couple of capacitor failures in CFL's, and in neither
> case did it cause the lamp to stop working. It caused the lamp
> to take longer starting and to run dimmer. It is possible that
> some other control gear circuits might cease working if the
> capacitor dies.

Whilst there is the strong stench of electronic death inside of the bulb,
aside from some discoloration on the inside of the bay, there's nothing that
indicates a failure mode in a visually obvious way.  I have at least 10 more
dead bulbs to dissect.  I've been taking photographs along the way that I'll
post to my site when completed.

> > I think this tube is still good.  It flickers when I hit it with a
static
> > shock.
>
> The common failure mode for a fluorescent lamp is that the
> emission material is all worn off the tube electrones (forming
> the dark marks at tube ends).

I assume that's "electrodes" although "Electrones" would make a great name
for an electronic music group!  <g>

> This will not have any impact on a tube flickering when exposed to static.
> A while back, I added detailed description of the common tube failure
> modes to the Wikipedia fluorescent lamp page:
>
http://en.wikipedia.org/wiki/Fluorescent_lamp#Mechanisms_of_lamp_failure_at_
end_of_life

Good job!   It seems quite fair and balanced to me, and I see something I
can add:  lots of migraine sufferers are bothered intensely by the light of
fluorescent lamps of all types.  Long before a dying fluorescent's flicker
is noticed by me, it's giving my wife migraines.

I don't know why wiki's got such a bad rep - it's really remarkable.  It's
better to light a candle than to curse the darkness.  I thought the failure
mode would involve loss of seal integrity but now I understand the complex
processes involved in CFL failures.  The only way to reliably test the tube
on the duff bulb, it seems, would be to hook it up to known good
electronics.

> > Do you have any thoughts on whether the trickle current from an X-10
> > appliance module could cause premature failures in CFLs?  I've been
>
> Yes, if the tube is flickering at all, even dimly.
> That can wear it faster than when it's fully on.

I just bought new 48" fluorescent worklamps for the basement.  The first
thing I noticed was how much brighter they were than the two year old lamps
they were running next to.  They really do dim pretty seriously over time
and it's not until you operate a new one side by side with an old one that
it's obvious.  But the real problem is that for the first time, I can't seem
to "damp" the flicker that results from using the lamp on an X-10 module.

I've put the new lamps behind two appliances modules (one with the local
sense diode snipped) AND a filter, and they still flash intermittently when
off.  As I added each additional component, the frequency of the flashing
dropped to about 1 flash every 10 seconds.  I wonder if the diode snip mod I
used for local control neutralization still allows trickle current to flow?
It does neutralize the local sense ability, but I never checked to see if
trickle current flowed.  That should be easy to detect - I have a space
heater whose neon lamp lights dimly when trickle current flows through it.

There appears to be more than one "local sense" mod and more than one type
of appliance module circuit.  I hope I can find some cure for it other than
building a 110VAC relay box that's run from an appliance module.  That
should cut any trickle current to the device but I'd hate to have to add
even more wall warts to the house.

Since I bought two identical lamps and they are both on at the same time, it
would be a good way to test such an isolated, zero trickle current relay for
bulb life.  First thing to do is to check the existing setup for trickle
current via a meter and then see if one of the other appliance module mods
actually does eliminate any current leaked through the lamp.  Since these
are ceiling lamps with their own pull chains, I could even test them
manually if I turned one off by that chain switch each time I used the pair.
Well, that's today's science project.  I'd really like to know what's
causing the high failure rate of CFL's in my house.

Thanks for the information!

--
Bobby G.