Re: Fluorescent Lamp Heater Filament Current

In <47896707$0$5200$afc38c87@xxxxxxxxxxxxxxxxxxxx>, Steve Carroll wrote:
<EDIT FOR SPACE>

I'm designing a couple of inverters to run a 4W and a 20W tube from 12VDC,
ideally in modified 'rapid-start' mode. (The heater current is reduced or
removed after the tube ignites.)

It appears that most tubes use 3.4 to 4.5V, while some need about 8V, as you
mentioned. I have two tubes in front of me, a 20W coil type, taken from a
compact fluoro and a 20W T10 black light tube.
The compact fluoro coil tube filament has a cold resistance of 4 ohms and
lights up dimly with 3.5 to 4V (DC) applied.
The black light tube filament has a cold resistance of about 8 ohms and
lights up dimly with 8V (DC) applied.

I'll do some hot filament current measurements next.
First, though, a couple more questions, if you don't mind.

1. Do you know if it's better to run the filaments so that thay're visibly
glowing, or just short of that point?

They should visibly glow.

2. Most small inverters built for this purpose, especially 4W to 8W tubes,
use a single-transistor oscillator with positive feedback derived from a
feedback winding on the inverter transformer. This results in a pulsed
output. Does this mean reduced brightness and if so, can it be compensated
for by adding extra turns to the HV secondary, or is a 50% duty-cycle
complementary drive better?

There are a few factors here:

1. There is "current crest factor". That is ratio of peak current to RMS
current. The less that is, the longer the lamp life (whether fluorescent
or HID or low pressure sodium) and the better the efficiency (for
fluorescent and low pressure sodium). I somewhat remember some sort of
industry standard that this figure should not exceed 1.7 for most
discharge lamps.

2. The current through a fluorescent lamp should have little or no net DC
component. The voltage waveform across the lamp should have little or no
net DC component. Voltage waveform, after weighting by magnitude of
current waveform, should also have little or no net DC component.
What a net DC component here does is cause positive mercury ions to
drift towards the more-negative electrode. This can cause a mercury
shortage at the positive end of the lamp, maybe even a majority of the
lamp, after several hours or a couple weeks of operation. This gets worse
with lamps that are longer or have greater ratio of length to cross
sectional area or greater product of length and current density (amps per
square centimeter of cross section of the discharge). The 4 watt
F4T5/G4T5 is more tolerant, but that is a low efficiency lamp (due to high
percentage of lamp voltage being in voltage drops at the electrode
processes).
If the current waveform has negative and positive half cycles roughly
equal in duration and there is a capacitor in series with the lamp, I
suspect you should be good even if the waveform lacks "half cycle
symmetry". You should also be good if you have "half cycle symmetry".

My experience so far is that cheap solid state ballasts with a single
transistor but also with a capacitor in series with the lamp tend to lack
mercury drift problems, at least in shorter lamps. However, I expect
improved lamp efficiency and improved lamp life with a complimentary drive
scheme.

Another item from my experience: Most (but not all) fluorescent lamp
products that run from batteries have the lamps being underpowered.

- Don Klipstein (don@xxxxxxxxx)
.

Relevant Pages

  • Re: Flouro brightness
    ... One night we stood a failed 5 foot flourescent tube with the bottom ... I've no idea how much of the 1154's 50 watts of output power was ... designed to run from 80W mercury vapour lamp ballasts. ... Well, as others have said, there are electrodeless fluorescent lamps excited ...
    (sci.engr.lighting)
  • Re: LED light results -- short
    ... It's a spiral tube lamp. ... I don't know where one could get this exact bulb but there is a flood ... quartz halogen with only 13 watts' draw (this includes the 10 watts to ...
    (rec.outdoors.rv-travel)
  • Re: move where
    ... I got a fluorescent desk ... Unfortanately it uses a pricey tube that would cost ... almost as much as the lamp to replace. ... cool white tubes ... ...
    (misc.rural)
  • Re: Metal Halide part broken off inside the lamp?
    ... there are two types of getters: ... element getters. ... vapors which may escape from the quartz tube seals, ... and is the one shown in the lamp picture. ...
    (sci.engr.lighting)
  • Re: Dissecting CFLs
    ... or cut one of the wires to the tube. ... can-type capacitor had a slight bulge on top, ... While this is not the old lamp from the Taiwanese ... Capacitor failure is common in CFL's because they are operated ...
    (comp.home.automation)