Fluorescent Lamp Inverter

[Bosium]

I forgot to mention I added the bipolar totem stage, made a HUGE difference as it actually turns the mosfets off hard and so improves efficiency.

I am considering going back to single-ended flyback configuration instead of push-pull.

[Chris Smith]

Depending on which way you go, either a stored cap delivers the current pulse, or with the direct Coil feed, the windings deliver the current needed to excite the tube.

Your second statement shows you that your current is consumed by your circuit, and not the tubes because one or two is the same.

The tubes will light with almost nothing, the hard part is getting the "correct amount of nothing" to the gas with out a great loss or drag.

My Tesla coil would light a four foot tube at over eight feet away, so again, almost nothing is needed.

All your work needs to be focused at getting the small current through all of your circuits with a minimal loss.

My suggestion for efficiency is to go bi polar gating on the FET, but use a stored cap to deliver the punch with high speed switching doing the controls.

Use a smaller passive charging circuit to the cap because you need a stepped up voltage, and the cap will deliver the current while the FET will deliver small short amounts of current to conserve the over all draw.

Think of a small caliber high speed bullet doing 3000 mph, instead of a 44 caliber bullet doing sixty miles per hour. Its the shock that does the job, not the elephant stepping on the mouse.

Work on each part as if they were not a whole, then tune them later for the best response. Use a wall powered Diode /cap voltage Trippler/Quad to get your voltage for your experiments, and refine your gate configuration using a cap that will be in the final design.

Place a resistor on the feed to the cap to simulate a small charging circuit at High Voltage, and try to make the tube light while not draining the cap voltage and not consuming a lot of power from the Gate switching section. This way you can check each of the stages for efficiency.

[Bosium]

Hi Chris

Thanks for the input, I appreciate you taking the time to reply.

What you have suggested will require a complete rethink of what I have developed thus far, I will need to think about it some..

Will post back soon with comments.

Thanks again.
Gareth

[rshayes]

In the 1980's, Heathkit sold a battery fluorescent lantern kit (GD1246A). This used a single bipolar transistor (MJE181) in what appears to be a class B oscillator at around 50 KHz. The secondary voltage was about 500 volts peak with a series capacitor as a ballast capacitor. This drove a F6T5CW fluorescent tube. Starting was done by connecting the cathodes across the battery with a separate push button switch to preheat them.

The transformer had three windings, a primary and two secondary windings. The secondaries are connected in series with the base of the bipolar transistor between them. This uses the circulating current in the secondary circuit as the base drive for the transistor.

At this frequency, power level, and voltage, square wave operation may not be very desirable. Each transition causes loss in the stray capacitance of the circuit. This loss is proportional to frequency and proportional to the square of the capacitance. At moderate power levels, this loss can be a substantial fraction of the output power. Trying to create a square wave into a capacitive load also results in high peak currents which increase the stress and loss in the switching device.

Sine wave operation salvages some of this energy, since it is stored in the stray capacitance and recovered on the next cycle rather than being immediately dissipated. It also reduces the magnitude of the peak currents in the circuit.

Preheating the cathodes is also important. It reduces the voltage required to strike the fluorescent tube and increases the tube life. Operation of the tube with cold cathodes sputters the cathode material onto the side of the tube. This causes a black area on the ends of the tube and eventually makes the tube hard to start. Striking a cold tube may require over a kilovolt. This will require a higher step up ratio in the transformer, which in turn increases the primary current during normal operation.

[Chris Smith]

Good analogy from the 50s?

However, we have moved on from those good ole days.

Current draw can be less than 1/2 amp, the tube can be too bright to stare at, and black tubes and heater elements are a thing of the past.

The 50s to be exact. And you can even dim a tube, something the 50s never had.

[rshayes]

"However, we have moved on from those good ole days."

Actually, no. There is no basic difference in the present day fluorescent lamp and those of the 1950's. Several smaller sizes are now being made and the phosphors have better color characteristics, but the basic operation is the same.

The cold cathode lamp is not new, it was well established by 1950.

The compact fluorescent lamps used to replace incandescent light bulbs are still a hot cathode design. The transformers used in these units are wound without interlayer insulation. The wire insulation is only good for a few hundred volts. Trying to strike a cold cathode tube would probably destroy a transformer of this type of construction. I have had some of these units fail, and the failed units showed black deposits at one end as well as melted plastic. The probable explanation for this is that the cathode opened up and did not preheat. This would cause sputtering of the cathode material on to the tube wall (black deposit). The reduced cathode efficiency would cause the end of the tube to overheat (melted plastic).

Xerox machines (as made by Xerox) used fluorescent tubes as light sources. These tubes were preheated by separate transformer windings. These lamps were cycled on and off for nearly every copy made, and they practically never failed, even after tens of thousands of on-off cycles. Preheating the cathodes is very important for good life from a hot cathode lamp.

[Bosium]

Hi Stephen

I agree that heating the cathodes will help in striking, however until I make a workable solution to heat them from the 12V line I will use cold-starting. I dont really care about some black deposits on the end of the tube and if they fail early then I will replace them.

I have thrown around an idea of a 2-transistor warming circuit which works by tying the one side of the top heater element to 12V and then an RC delay will switch the transistors on for a few seconds at startup, one of which shorts out the secondary winding of the trafo and therefore bridges the top heater to the bottom heater, and the other transistor ties the bottom heater to ground so that current will flow directly from the battery, through both heaters and both transistors down to ground, but bypasses the transformer secondary.

As for sine-wave excitation, I am willing to believe that it might be better for the tube and help in ionising, but from what i've seen it doesnt seem to be as efficient in terms of current draw.

My whole aim is to have some camping lights that are bright and do not chow the power from my battery.

G

[rshayes]

The method that Heathkit used for starting was quite simple. A multipole push button switch connected the cathodes to the battery through a resistor. When the switch was released, the cathodes were reconnected to the normal circuit. About a five second preheat was recommended.

As far as an efficient sine wave converter is concerned, look at Ap Notes 55 and 65 on the Linear Technology site (www.linear.com). These describe circuits for cold cathode lamps in the 1 watt range, but it isn't difficult to scale the power up to the 10 or 15 watt range. The peak output voltage can be substantially reduced if it isn't necessary to strike a cold tube. The push pull circuit using two bipolar transistors is fairly simple to design and can operate at efficiencies in the 90 percent range. The trick is the use of an inductor to feed the center tap of the push-pull oscillator. On alternate half cycles each end of the tank circuit is connected to ground while the other end is allowed to swing positive. The waveform on the center tap is a series of half sine waves. The average value of this waveform equals the supply voltage. If the circulating energy in the tank circuit is about five times the energy delivered to the load, the waveform will be a fairly good sine wave and the output voltage will be proportional to the supply voltage.

The oscillator can be used alone without the switching regulator. This is simpler to build, but it will allow some variation in lamp current as the batteries discharge. I don't think that this would be a real problem for a camping lantern. I would try the oscillator alone as a start.

The oldest reference that I know of to this oscillator circuit was a paper about 1920 using vacuum tubes. A transistor version was described in a British paper about 1960. It isn't really a Royer converter, which produces a square wave using a saturating magnetic core for timing.

With a sine wave in the 50 KHz region, a capacitor can be used as a ballast. This is simpler than winding a third inductor for an inductive ballast.

<small>[ November 23, 2005, 05:21 AM: Message edited by: stephen ]</small>

[ringo47stars]

It sounds like the problem I had with my transformers heating too much. They have a normal operating temperature so I kept them running until I noticed they charged up like a battery. So the lighting tube acts like energy and feedback to the ballast makes it heat up. Maybe use bigger components to handle the heat.

[Chris Smith]

"ACTUALLY, NO. THERE IS NO BASIC DIFFERENCE IN THE PRESENT DAY FLUORESCENT LAMP AND THOSE OF THE 1950'S. SEVERAL SMALLER SIZES ARE NOW BEING MADE AND THE PHOSPHORS HAVE BETTER COLOR CHARACTERISTICS, BUT THE BASIC OPERATION IS THE SAME"

Even the Tubes have changed, single pin, no heaters, etc.

Actually todays Fluro drivers operate in the RF region and or in the HF region and don’t need wasteful starters, ballast, or heaters.

In the 20s when fluro was just getting its legs, wall power was used at 60 Hz and thus starters and a ballast to run the heaters were needed, usually with a buck /boost start circuit to ignite the tube.

Todays Fluro tubes don’t have two pins at the ends with a heater element and don’t incorporate a “ballast” to knock down the power to run the heater or use a buck boost trigger to ignite the gas.

They use sophisticated FET drivers, high frequency high voltage and short pulses to maintain the gas using far less wasted current.

Yes, we still use twin 40s, as wasteful as they are and only because of the cost to produce both the drive circuit and the tube.

However, they will go the way of the dinosaur because of their wasteful energy ways, A.K.A., the heater element at each end which consumes most of the total power used to maintain the tube, and their short life cycle when it should be in the 1000s of hours. Todays twin 40 is the same as the 50s, and soon to be phased out with all the other old and wasteful ideas.

[Will]

there is an article on deign of a 25 W fluorescent ballast using an International Rectifier chip - in Electronic Design Mag of Nov 17 - Goto www.elecdesign.com then look up 'articles ' then recent articles