EEFL backlight circuits

[klugesmith]

A dumpster-diving friend just gave me the backlight unit from a LCD TV set or monitor.
18 straight, thin lamps which I have learned are not CCFL, they are EEFL
(with external electrodes, and nothing but glass touching the plasma).

The external electrodes are all connected in parallel, which puzzles me.   
Each lamp's intrinsic glass-dielectric capacitors will offer substantial impedance at the operating frequency,
which I guess will serve to balance the currents. But how do they get all lamps to start?
Really high operating voltage on the shared AC bus?

Along with the panel I got one inverter unit, very similar to the one pictured here:

The part number on mine appears to have a Slave suffix, for use with a Master board and possibly other Slaves.
Is there a politically correct word-pair that conveys the relationship so well?

Anybody know if those are easy to operate on the bench?  Maybe just DC power and an Enable signal on the correct pins.
It wouldn't be too hard to trace the power path, which has a HV transformer and I guess an H-bridge of transistors in SO8 packages.
The one in picture (snipped from ebay) seems to have six of those transistors, so maybe a half-bridge.

[klugesmith]

Learned things from Internet, starting with a 2007 paper about work sponsored by Samsung.
Study_on_the_equivalent_model_of_an_External_Elect.pdf

Scope view shows peak voltages around +/- 2500 V, and current on the order of 100 mA for whole array of lamps.


Still haven't found info about how to fire up an inverter module.

[klugesmith]

It was educational to see 18 luminous tubes in parallel all strike and glow with the same brightness, driven by a plain old 60 Hz NST. 
It's the series capacitance at each end of the EEFL's that limits and balances the current, with enough bus voltage to start every tube.   


In this case I stopped turning the variac at current of about 0.5 mA for the whole array.  About 28 microamps per lamp.   Indicated voltage is 5380 volts.  Couldn't go much higher without spurious arcs between lamp holder metal and the sheet metal panel.

Now that voltage comes from a combination of capacitive and resistive impedances.  For simplicity, assume i is sinusoidal and E = i/jwC +  i*R.
If we guess 28 pF per lamp end, whole array has 252 pF, so there's 5263 V across the equivalent capacitance and 1116 V across the barely lit plasma columns.
Guessing 30 pF per lamp end, 270 pF for array, E_C is 4912 V and E_R is 2195 volts.

How would 252 pF of equivalent capacitance figure into normal operation?
Say we want 180 times more current (5 mA per lamp, 90 mA for whole array).
If frequency is higher by factor of 900 (54 kHz)
then the voltage to push current through capacitor is down by factor of 5, to 1053 V.
I bet this is in the right ballpark.



Next exercise could be firing up the inverter which was salvaged with the backlight panel.
I found connector pinout (24 V DC, ground, on/off, and brightness control voltage) for a Master unit.
The one at hand is a Slave unit, I think with the control pins unused.

[Mads Barnkob]

Very long ago, 10 years or so, I experimented with CFL drivers like these.

Back then I just found some enable pins that was pulled high by the power supply in the monitor, not much else to get them running.

One of the funny applications I could see for CFL tubes was that they ignite from the same end of the tube and once you get duty cycle/voltage low enough, you can make it light up partially. I played around a bit with making a VU meter with a row of CFL tubes, but that project died when the tubes one day was broken by a dropped plier

[klugesmith]

Bad pliers!  At least its fall didn't destroy a loaded ICBM, as in the Damascus accident (Arkansas, USA, 1980).

A closer look at my inverter board shows a transformer primary winding that's bundled fine wire, maybe even Litz wire.

.
The four main transistors say RSS100, which has standard pinout (SSSG,DDDD).
Driver says OZ9981GN, and I found an application circuit that looks promising.

.
from https://www.520101.com/html/internationaltv/002018461.html.
Is the text on that page Korean?
Does any reader here have a datasheet for the OZ9981GN?

[klugesmith]

>>Warning: this topic has not been posted in for at least 365 days.
>>Unless you're sure you want to reply, please consider starting a new topic.

Of course I want to reply; there is news about the artifact under discussion.

In previous post, the maximum current was about 0.5 mA for whole array, or 28 uA per lamp.
That's all we could push through tiny capacitors at 60 Hz (NST on variac).  Voltage was limited by spurious arcing from electrode holders to metal panel. 

After more than two years, plasma projects on this forum inspired me to dig out a high frequency neon sign power supply.   Found one connected to a tube shaped like the word BEER, which lit on first try.   When connected to the array of EEFL's in parallel, the 17 unbroken tubes all lit on first try.   For the picture, a regular neon tube element is in series with that panel.

I bet we now have about 30 mA in the circuit; almost 2 mA in each individual EEFL.  That's 60 times more than we got with mains frequency NST.
Next steps are to properly measure the voltage, current, and frequency.   I think my home-made AC kilovoltmeter and floating milliammeter will serve if upgraded with fast rectifier diodes (which don't need to be high voltage). Must consider parasitic capacitance in the voltage divider.

[klugesmith]

Took some steps toward measuring capacitance of glass-insulated plasma electrodes!

Preparing for a HV holiday lighting project, I found a very small and compact neon power supply.   Nominally 4 kV, 30 mA, 25 kHz.    It lights the bundle of EEFL's, but only briefly for 3 tries each time it's plugged in.  I guess the voltage drop due to series capacitance is too high, so power supply thinks no luminous tube is in the circuit.   

The brief on-times should be sufficient for measurement of AC voltage, current, and phase by oscilloscope, when suitable high frequency attenuators are set up.  Then we can put numbers on electrode capacitance and arc voltage, without referring to EEFL literature;   Of course same could be done with the high frequency PS from neon beer sign in previous post.

In the meantime, with the EEFL's supported away from their original metal holder, they were lit with full voltage of a 60 Hz NST.   Rectified panel meters indicated 12.9 kV and 1.3 mA.   They're both calibrated for DC, so multiply by 1.11 to get RMS AC voltage and current.

[edit] In an earlier post here, 60 Hz measurements at lower voltage and current led to SWAG's of electrode capacitance: 28 to 30 pF per tube end, 14 to 15 pF per tube.   I have not tried figuring C from glass area, thickness, and permittivity.

[klugesmith]

Here's estimated capacitance of glass insulation between electrodes and plasma.  To be checked in lab, after setting up to measure magnitude & phase of HV current and voltage at 25 kHz.

EEFL tube diameter is 4 mm outside and 3 mm inside.   Taking dielectric constant to be 4, the cylindrical capacitance is 774 pF/m by proper logarithmic formula, and 779 pF/m using average circumference divided by thickness. Simple formula works OK because diameter ratio is not far from unity.
Outside metal electrodes are about 20 mm long, for 15.5 pF at each lamp end (not figuring the end closure and fringing field).  Equivalent to 7.74 pF in series with plasma column.
These numbers are barely half of what I'd guessed in a previous post, after lighting tubes with microamp currents using 60 Hz NST.

If today's value is accurate, then bundle of 17 lamps comes with 131 pF of series capacitance.
At 25 kHz the impedance magnitude is 48000 ohms. 
30 mA neon power supply would see 1.45 kV across the capacitance, on top of the plasma voltage, and probably lagging it by 90 degrees.
One supply works, and the other gives up after a few tries.

We might make the second neon supply happy by putting an inductor in series, to cancel the capacitive reactance.
I figure that would take 308 millihenries, at 25 kHz.  Only 170 mH if old previous SWAG for tube capacitance is correct.
Might be lucky and find that in a transformer winding that's insulated for 1.5 kV. How about a variac?
Or wind 308 mH worth of turns on bobbin of a big ferrite pot core, or laminated steel EI core?   For resonating with C in the 150 pF ballpark?
I think this would not interfere with operating the many EEFL's in parallel. since there would be plenty of voltage across each lamp-capacitor branch.

[edit]  EEFL drivers have been discussed in the trade for more than 20 years.
https://www.scribd.com/doc/40383101/EEFL-Backlight-Technology shows inverter topologies familiar here, current and voltage waveforms, electrode capacitance options, etc.
Exercises reported in HVF thread here are just playing with toys.   Making things work using parts intended for other uses, like neon PS and maybe transformer windings.   A chance to practice with DIY voltage attenuators and isolating transformers for scoping kilovolt circuits.