Re: CFL effciency?

On Fri, 28 Nov 2003 08:50:11 GMT, [EMAIL PROTECTED] (JM)
wrote:


>I just connected a LOA 18w CFL to a diode to see what happenes.  It appears 
>normal light output, but extremely flickery.  Power factor was reduced.  The 
>fluorescent tube itself was still receiving high frequency from the ballast. 
> It's flickery becuase the DC is switching on and off 60 times per second.  
>Since it's the same thing as powering on and off 60 times per second, it 
>can't be good for the ballast and the electrodes will probably wear faster.  
>It is not the same thing as a magnetic ballast operated with a diode, since a 
>magnetic ballast will output true 60 pulses per second, while an electronic 
>ballast will still output HF while the DC current is ON.  However a good 
>electronic ballast wil have a large enough capacitor so that the OFF time 
>of the pulsed DC will be neatly smoothed out, and not flicker.

I do no see any basis for your last comment about "good" electronic
ballasts. Electronic ballasts are not designed to be operated on
half-wave rectified AC power. The qualities that may distinguish a
"good" electronic ballast from others include input power factor,
input current THD, lamp current crest factor, programmed start vs.
instant start, ballast reliability and life, universal input voltage
capability and perhaps a few others. Using a larger than necessary
electrolytic capacitor increases the cost and size of the ballast,
reduces the input power factor of a low power factor ballast and may
make power factor correction more difficult in high power factor
ballasts such as those using a valley-fill power factor correction
front end. I doubt that any ballast designer has intentionally
modified his or her design so the system would operate on half-wave
rectified input power since this is an aberration. 

Perhaps I have misunderstood you. Is there some other characteristic
of "good" electronic ballast design that you think naturally lends
itself to operation on half-wave rectified input power? 

-- 
Vic Roberts
http://www.RobertsResearchInc.com