down dirty inverter advice needed

[jollyrgr]

I'm assuming you are talking about the screw in fluorescent light assemblies that replace incandescent bulbs, right?<p>Jwax beat me to my first suggestion. I've seen "laptop" invertes (to take a 12V car outlet and turn it into 90W 120VAC) for $10 at the local Wal-mart clearance shelves. I've used the Radio Shack inverter for years and it works fine as does others from the local home improvement center. Other options are to convert a UPS into duty for larger jobs.<p>But there is still another option. This can be done with a 120V to 24V CT transformer, a pair of bipolar power transistors (i.e. 2N3055) a couple capacitors, resistors, and diodes. See an example schematic here:<p><p>Full article here:
http://www.aaroncake.net/circuits/inverter.htm<p>The 2N3055 transistors can easily handle two to three hundred watts. A 24V 2Amp CT transformer (Radio Shack's 273-1512 25.2V is close enough) will handle close to 48 watts. This transformer is $10.49. The transistors are a couple bucks at RS but under a buck at other places. Caps, diodes, resistors, etc. another couple bucks. Already you are up to $15 to $20 without a case. Already the premade inverters Jwax referenced are looking quite good. <p>If all you want to do is drive flourescent tubes, this site will be of help:<p>http://www.talkingelectronics.com/Proje ... erter.html<p>This shows a very simple, easy to build transformer (that's right, you can wind your own transformer) and circuit. They suggest winding the transformer on a ferrite rod from an AM radio. You could go smaller and use a toroid form. This would make the overall driver package somewhat smaller but would make putting the 450 windings for the secondary much harder to do.

[Will]

I know this doesn't help the hobby/dabbler - build-it-yourself aspect of this thread but I believe you can buy 6 inch battery driven fluorescent lights in the local $Dollar store for . . . . $1.0 ?

[jollyrgr]

I haven't noticed the dollar flouros before. I have a quite a few of these Dollar stores around me and will stop in this weekend and take a look around. While I think the White LED lights out there are great, I'd still pick up few of these flouros for only a dollar!

[Chris Smith]

TO....2M MACHINING <p>Yes, you can buy new technology fluro drivers, or get into the complex circuitry your self. <p>The whole idea is to “crank up the speed” and deliver short square wave efficient pulses. <p>This way there is no need for ballast and the tube is brighter, lasts five times longer, and uses anywhere from 33% to over 60% less power.<p> And some [purchased] modern circuits even cut down on the RF produced from these High frequencies. <p>This is important if Your running a video or Audio studio, but less important if your camping or just lighting the garage.

I have built many a circuit in the 80, some from scratch, some from tweaking the “dollar fluro”,.... and have come up with several winners.<p> My favorite, you could not stare at, as it burned your eye even at 100 feet. <p>Camping, it would draw from the 12 battery around point six amps, and You could read a book at over 200 feet on a moonless night. <p>
.....Crank it up, square wave, 600 to 900 volts, short duration pulse, [10 to 30 % dwell] 40 k minimum, 75 k max. <p>Any faster and you cant contain the leads, they bleed all over! [Tesla’s HF effect] <p>Not that hard to do if your "READ Up" on the subject.

[jwax]

Interesting, Chris! Looks like an identical signal for ultrasonics generation, just driving a fluoro instead of a piezo. (500 volts, 40 KHz)
I wonder how much of that square wave makes it through the transformer, however. Seems the core would round off the square edges- a good thing for an invertor or a piezo driver!

[dacflyer]

woah , i found a deal and just what i need ,
i got currious and looked on e-bay they have a new coleman lantern that uses the twisted spring looking flourescent lamp , and runs for untold hrs off of 8 D-cells , and even better yet. i got 2 of them , new in box for $22.50
they got more if your interested in one or 2
just type in "coleman flourescent lantern" in the search box.
CHRIS SMITH >> i'd love to get a copy of that schematic ,for that killer bright driver for the flourescent light that you described..

[dacflyer]

wow.. got my lanterns in saturday..they are much brighter than i thought they'd be.
the lamp is a spring type flourescent lamp that boast 15 watts.. you cannot look at it directly its so bright..and the battery last 12 hrs on hi,,,and 24hrs on low..this could be the makings of another solar project...or at least charge it with solar when out camping.. runs off of 8 "D"
batterys (12v)
now to figgure a foto cell for it.

[Chris Smith]

Dac<p>Most of what I did in the 80s was from schematics for laser drivers. The main difference is that lasers used even shorter pulses and had to be clean and “Ring Free”, and no transformer on the final out put connection [transformers for HeNe’s], just Fets, Caps, blocking diodes, and a huge Resistors depending on solid state or gas lasers.<p>Yes the transformer to Fluro tube MAY clip a lot off the edges while increasing the voltage, but the main issue is, is there enough saturation or duration after the transformer to fire the tube with out the need of a heater or starter, and less about the crispness of the final wave output. [Good transformers make a difference] <p>The square wave is essential in the production stage to maximize the power loss in the switching stage, and not so much for the final product to the tube. That final wave front can be dirty or clean, depending on the loss of your transformer. Also bad or mismatched transformers can defeat the low loss gained from a clean square wave. Low frequency saturation in the transformer wont work.<p> Producing crisp Square waves consume or waste less than a bipolar transistor designs and its associated losses. The crisper you make your driver, the overall consumption goes down. <p>You can use almost any flip flop other than a 555, such a JK flip flops to Drive a FET circuit, as long as both parameters are adjustable. You need a adjustable on time, and adjustable off time, and a over all frequency control. <p>Ideal time will vary with your step up transformer and circuit design, while I always played with the transformers, and then tuned the other two to suit the transformers “sweet spot” with the saturation point not interfering, plus or minus a degree or three.<p> Often you can play with and crank the three parameters and hear the sweet spot, or watch the ammeter dive one way or the other and or use the brightness of the tube to decide on where you are going to settle on. Because the parts you play with will always vary, basically create a “frequency generator” approach so that you can play and find the right combination. No slower than 20khz, no faster than 60khz. <p>On times can be short and powerful, while off times can out number the on time by as much as 15: 1.<p>Also “dirty signals” into a tube tend to generate Radio frequencies, which Aid in the start up of the tube, but play havoc with everything else.<p>[ July 11, 2005: Message edited by: Chris Smith ]</p>

[rshayes]

If you want to find out how flourescent lamps are really driven, try downloading ap notes 55 and 65 from the Linear Technology web site (www.linear.com).<p>The basic points are:<p>Sine wave excitation gives the best electrical to light conversion efficiency. High crest factor waveforms are less efficient.<p>Frequencies from 20 to 100 KHz are reasonable choices.<p>Two bipolar transistors can generate the sine wave required with over 90 percent electrical efficiency.<p>Layout is very important.<p>This is the type of circuit used for backlighting the LCDs in laptop computers. Since the LCD panel attenuates the light considerably, a bright source with maximum overall efficiency is desirable.

[dacflyer]

ok,, so if i understand right,, the higher the freq. the brighter the light should be?<p>does this also pertain to EL displays too?
i have some large ones also..but they light up poorly with 60Hz. even tho they light up on 120Vac<p>
also.. i got from a local tool sale a mini-inverter,, it plugs right into the cigarette lighter..but what puzzeles me is.. it makes
120 volts DC..not AC.. wonder why it is made to do this.. i might hack it to see if its got a bridge in it.. nothing runs with this thing unless its a resistive load.. only 60 watts max,,but it will not fully light a 32 watt flourescent lamp...Grrrrr
what did i expect for 12.00 :p

[Chris Smith]

No, the frequency is relevant, but not the whole issue. Faster or slower, depends on your bread board design, your coupling transformer, and other factors. <p>Clean cut square waves on the generator side use less power, dirty or sine waves excite the gas. Both can be had with out much loss.<p> Bipolar is a waste of time [and power] if power consumption is your issue. <p>Bipolar flip flops were used when Dinosaurs roamed the earth, but today we have Fets. <p>Bipolar as a direct driver worked just fine, 20 years back or so because it also was “dirty” and thus covered the ignition issue, but today we have efficient switching, and even better coupling transformers that can couple square wave with out much loss. All the power supplies of today, use this technology, and thus if you need a transformer, just take apart a switching power supply.

Stick to square wave generation and loss less transformers that can handle the square wave. <p>If you have any “ignition” problems, there are many, many ways to achieve a “dirty” wave front with out the loss from the power supply.<p>The main problem with the bipolar flip flop design is that duration, dwell, and frequency are not adjustable to the maximum performance levels that todays flourescent drivers can achieve. <p>All of the losses are before the production of power, and thus they rate the “power factor” after the transformer, and not before. <p>Fluro drivers have come light years, since the dinosaur Bipolar days.<p>[ July 12, 2005: Message edited by: Chris Smith ]</p>

[rshayes]

The Linear Technology application notes that I cited are the result of a 4 year development program at Linear Technology. Together, they total over 150 pages. The author, Jim Williams, is a well known and respected engineer.<p>The basic sine wave converter was not their invention. The basic circuit was described in a paper over 85 years ago. A transistor version was described about 1960 in a British journal. It is reliable and capable of good efficiency up into the low megahertz range with the proper transistors.<p>The bipolar transistors used are not "dinosaurs". They happen to be the best devices fort this particular application.<p>The electrical conversion efficiency is above 90 percent. The conversion efficiency from electric energy to light is about 20 percent. The transmission of the LCD panel is under 10 percent for monochrome, and probably about 3 to 4 percent for color, due to the color filters.<p>If two lamps are used, the emitting area of 1/4 inch lamps will be about 1-1/2 inches (measured around the lamp). The light from this area is used to illuminate a panel about 13 inches high (on a 17 inch display). The display transmits about 4 percent. The result is that the lamp will have to be about 220 times brighter than the display highlights. That is pretty bright.<p>The efficiency of the lamp is given as about 20 percent. This is consistent with an old electrical handbook (1948) that cites the efficiency of a 40 watt flourescent tube as being 20.5 percent. This is for operation on 60 Hz. The efficiency seems to be about the same at 50 KHz.<p>This is collaborated by the figure 7 in Ap Note 65. This clearly shows that light output is independent of frequency in the 20 to 130 KHz range.<p>Figure 1 of Ap Note 55 shows that the light output of the flourescent tube increases at a less that propotional rate with current. This indicates that the ideal condition is steady operation at a lower current, and tht pulsed operation at higher currents is counterproductive.<p>With a capacitive ballast, the optimum waveform is probably a sawtooth, since this would result in a connstant magnitude of curent. The sine wave is a fairly good approximation, and is much easier to generate with high efficiency. A square wave generator at the thousand volt level would be less efficient due to the losses in the stray capacitances of the circuit.<p>This is the type of circuit used in the millions of laptop computers produced every year. It works well and gives lamp life in the thousands of hours.

[Chris Smith]

I used to have thousands of pages on fluro circuits. I did research on the subject for more than six months, a few years back. Bipolar is old and a dinosaur compared to FET Designs. Fets have power curves that surpass Bipolar for resistance and power issues, RF Issues and Dimming issues as well as color curve issues and has moved light years past the old days of Bipolar. The only thing Bipolar design has going for it is simplicity and cost. <p>Today almost all lighting companies use Igfets and other FET designs as well as proprietary designs. This isn’t happening because bipolar is soo great, its because the power hog designs of old are no longer acceptable. You wont be getting 75 watts of equivalent light out put from a 14 watt draw with Bipolar, yet almost any screw in fluro tube today can deliver this and more. <p> All you have to do is dissect any modern Fluro driver today to see the insides, and a FET will be at the heart of that design. Bipolar has nothing but drawbacks, unless power wasting power is your issue. Bipolar has almost no adjustability unless you incorporate more circuitry like some IC chips. <p>The typical on resistance of bipolar compared to FET has no comparison. Fets typically can have a on resistance of one tenth of one ohm. Try to get that with ANY bipolar? Try to come even close?

[rshayes]

"The typical on resistance of bipolar compared to FET has no comparison. Fets typically can have a on resistance of one tenth of one ohm. Try to get that with ANY bipolar? Try to come even close?"<p>No problem at all. For example:<p>2N173_____1.00 Vmax_____.3 Vtyp_____12 Amps
2N278_____1.00 Vmax_____.3 Vtyp_____12 Amps
2N2728____0.10 Vmax_____.075 Vtyp_____50 Amps
2N2912____0.12 Vmax___________________5 Amps<p>These transistor types were all made over 40 years ago, using factories that cost a few percent of what a present fabrication facility costs. Even then, efficiencies in the 90 percent range were posssible in power converters.<p>If you want to get more exotic, the RCA 40079 tunnel diode would conduct 200 amps at a voltage of around 200 millivolts.<p>The General Electric C612L SCR would conduct 2000 amps with a drop of less than 2.21 volts.<p>Power MOSFETS have some advantages compared to bipolar devices, but low on resistance is not one of them. They are inhierantly very poor.<p>One of these is the lack of stored charge. The particular converter circuit used in the laptop computers switches when there is zero voltage on the collector. The slight delay in turning off due to stored charge does not cause a significant problem. In this particular case, the bipolar transistors are a better solution.<p>The screw in flourescent lamps use small surface mount MOSFETS that probably have considerably higher on resistances than .1 ohm. AT 15 watts, the peak current will only be about 200 milliamps, and a part with an ohm or two of on resistance could be used. I have had several of these fail, so I suspect that their design is marginal.<p>Incidently, these lamps are also driven with a sine wave rather than a square wave.<p>[ July 13, 2005: Message edited by: stephen ]</p>

[Chris Smith]

That explains all of todays FET drivers on the Fluro circuits? <p>"Drive requirements are arguably where bipolars and MOSFETs differ most. Care must be taken to check the drive conditions when comparing the two technologies. For example, bipolar transistors require sufficient base current to achieve their lowest RCE(sat) values, and the base drive loss must be taken into account in power dissipation calculations"<p>You keep mentioning 90% as if this compares to .001 ohm on resistance. It doesn’t, .001 ohm is a 99.99% efficient by comparison to bipolar.<p>You failed to mention the on resistance of the Bipolar because they typically have a high current loss, and they even get hot during their performance because of this. And no one uses tunnel diodes, to make fluro lights. <p>Having built many bipolar and FET designs for fluro tubes, only Bipolar designs need huge heat sinks on the transistors unless your trying to make coffee off of them, and they burn your finger if you touch them, while all of the FET designs that I have built run cool to the touch, use less wasted power, deliver more light, have less design problems, and over all preform better which is why there are no bipolar designs used by the top three Fluro people today. Only the Dollar fluro still uses this concept, and they still get hot. They are old and they are ancient. But they are cheap and easy to build, but in todays world of conservation that are unacceptable. <p>Ohms law is still the law of the land.
.001 ohm on resistance beats any Bipolar any day.<p> Forward voltage and on resistance are not the same.<p>The sine wave that is used in the inexpensive fluro today are due to cost. Expensive fluro drivers don’t tap off the 60 Hz sine wave to time their circuits, but rather convert to DC and then back Square wave. This allows total controllability in the circuit design.<p> Cheaper Fluros like the 15 watt through 75 watt tap off the natural sine wave from the wall power, for convenience, but have no control over the circuit, nor do they need any for the tiny tubes.<p> If you want power control, color control, dimming and RF suppression then Square wave is the only way to go.<p>[ July 13, 2005: Message edited by: Chris Smith ]<p>[ July 13, 2005: Message edited by: Chris Smith ]</p>