Not sure if you have some ceramic caps for C3 & C4 though; they should be rated for 1KV or higher.

I'm not sure if size is important, I've got 5, 100nF, 2000V capacitors. If they are no good I'll have to order some. Rapid don't have any that small in ceramic 1000V but they have these in 500V so presumably I could use 2 in series.
http://www.rapidonline.com/Electronic-Components/Silvered-mica-capacitors-62503

I don't know if you saw my questions in post 349, presumably they are not options.

 

One of the requirements for good quality light, and high efficiency from CCFLs, is as clean as possible sine output. Does this circuit do that? (I can't tell from the picture.)

Yes, Jim Williams went into that in his papers. Seemed like it was a pretty thorough intro to what was going on some 20 years ago, and still what remains today.

That's the beauty of the Royer oscillator; the primary side is basically an LC tank circuit, and it's really quite efficient - even when driven by BJT's.

I added the model for the ZTX1149A to the simulation. Try a FFT on "Out" from 1mS to the end. I'm getting ~48dB for the primary, no side lobes to speak of, and the harmonics are nearly -40dB down worst case. That's a pretty clean sine wave.

 

I'm not sure if size is important, I've got 5, 100nF, 2000V capacitors. If they are no good I'll have to order some. Rapid don't have any that small in ceramic 1000V but they have these in 500V so presumably I could use 2 in series.
http://www.rapidonline.com/Electronic-Components/Silvered-mica-capacitors-62503

I suppose you could use a couple in series. You'd want to use some very large resistors (say, 10megs) across them so the node between the caps isn't floating; otherwise you can wind up blowing both caps.

I don't know if you saw my questions in post 349, presumably they are not options.

I just got home and don't recall that question. I'll take a look in a few.

 

Just a crazy thought, there's probably good reasons not to do it:
Could we put 49 turns on the toroid and then have a second toroid with a 1:4 turns ratio?

Ahh. Actually, I tried simulating that, and the results were not terribly promising. I think I was tired when I was trying it tho, so that's why I never got back to you on it. I kinda hate thinking of using three toroids for this project.

<ed>Another idea, if it simplifies anything I've got loads of PIC10F202s in 6 pin SOT-23 which could be easily used to create a 100kHz oscillator and could also do the flashing. They cost about the same as 555 timers.</ed>

That's kind of interesting. However, the Royer, once started, is self-resonant as a function of the inductance of the primary and what I've labeled as C2. It's going to be mighty doggone particular about the frequency, too. If you try to drive it at anything but the resonant frequency, efficiency will go right out the window, and you'll start seeing smoke coming from the transistors.

As far as using the PIC10F202's for sequencing ... well, a minimum of 4 outputs are needed, and they can't be like the MCLR input on a PIC12F625. I think that pin 6 being GP3/MCLR/VPP will kill the deal; unless you want to use a couple of PICs and then have to deal with serial communication between them, or some type of interface to a shift register/Johnson counter - but then we're almost back to where we started with the 555 and 4017. If I'm wrong here, let me know - I've never fiddled with the PIC10x series.

 

and theres me thinking at the start that these lights would be the easiest!

Appreciate all this chat about them, dont understand it, but will no doubt end in something great!!!!!

 

OK, no worries, I don't understand these inductors and transformers, I just had the idea that maybe the little PIC could just drive a couple of transistors and the transformer and somehow that could help. It wasn't intended to control the LEDs as well, just the flashing of the tubes, or it could take an input from the LED board.

@Craig:
Would you like me to solder up a version of what you have on breadboard?
I'd have to get you to check a few things first, those wires have been moving around a bit.

 

and theres me thinking at the start that these lights would be the easiest!

Well, had you simply sent away for a supply, that would have been easy, but we wouldn't have gone through this learning experience. The Jim Williams article was particularly illuminating (pun intended. ) I had never dissected a CCFL curly-bulb before ~February of this year. I started taking one apart, but never documented it. Not too late, as I had just cut the base & bulb from it and left it sitting.

Appreciate all this chat about them, dont understand it, but will no doubt end in something great!!!!!

Right now, I'm trying for more or less a "quick and dirty" type of thing. I'm fairly sure it'll work; looks OK in simulation - but there are some things that just don't simulate very accurately. I'm no SPICE wizard, although I'm reasonably handy with it. Never taken a course on it or anything; it's all been learn as I go.

I'd actually played around a bit with a version of SPICE many years ago in the mid-1980's, it was on a Dartmouth time sharing system that our company had at the HQ building. I think it was on a large Honeywell computer. We'd log on to it using a dial-up modem (110 baud - who hoo!) using a Decwriter LA-120; it resembled a large electric typwriter and took the 14-7/8" wide "green bar" computer paper. Those were the days.... When we finally got a 9600 baud connection, we'd marvel at how the paper would fly out of that printer.

 

Butting in again, will this link be of any help? Its a ballast design powered by either 2.4 or 4 volts DC.

http://www.diodes.com/zetex/_pdfs/3.0/appnotes/apps/an17.pdf

 

Butting in again, will this link be of any help? Its a ballast design powered by either 2.4 or 4 volts DC.

http://www.diodes.com/zetex/_pdfs/3.0/appnotes/apps/an17.pdf

Did you look at the circuit that I posted?

 

Not all of them, did I screw up again?

 

Not all of them, did I screw up again?

LOL!

In my reply #360, the circuit is extremely to that in the application note; Except my version is adapted for PNP transistors and salvaged toroids that Markd77 already has.

 

I've got a big pile of ZTX1053A NPN transistors if you want to use them instead. I heard somewhere that NPNs are usually slightly better than PNP. Looking at the datasheets:
1053 has a VCE(sat) of 120-150mV at Ic=1A Ib=10mA
1149 has a VCE(sat) of 140-240mV at Ic=1A Ib=7mA
The 1149 has a slightly higher hFE

 

Ahh, OK - well, I went to extra trouble to flip the schematic upside-down so you could use the 1149's, but I don't mind flipping it back at all. I had no clue you had some 1153's.

 

apologies for my absence, I have been gutting out and redecoarting a room which i am now going to use as my display room. Pics to be posted once complete.
@Mark, if you feel ready to biuld the soldered up version , go for it, As for the wires and what goes where, there was a post back a bit between me and wookie detailing what wires go where. He also documented it aswell! Or if you rather wait until this other circuit is comlete thats cool with me aswell, just let me know

 

@SgtWookie
No need to change the schematic if it doesen't make much difference.

@Craig, I see it back in post 291, could you do a continuity check from the wires to the pins on the connectors, just in case I messed up with the labelling of the wires?
I think you mentioned that you would want the switch on a long wire, if so, how long and would you want the battery pack to be near the switch or near the board. Would you also want a long wire for the speed control knob, or a short wire, or for it to be just on the board, exactly the same as it is on the breadboard?
Remember the resistor that you swapped to make the speed slower, (was it 82k), does that give a good range of adjustment?

 

Mark, I will do the continuity test tomorrow eve if thats cool with you?

As for the speed control, just a wee knob o the board will do, as it is at the moment!

Also, I didnt change the resistor as I didnt have one the correct size! At the moment, i have the knob turned right down, and its very close to the speed it should be. Could do with being a tiny bit slower!

As for the power swtich and batter. Can the battery an board be together, and the switch seperate. I will be hiding all the control gubbins, but it would be good to have th switch close to the arm so i can turn it on and off etc. As for the length of wire for the switch, 300mm should be plenty enough!

 

For the board, I can make it rectangular or square just as easily. I'm guessing about 4X8cm, 7X5cm or 6X6cm.
Also you should have a little bag marked something like assorted resistors. You could swap any of them with a resistance between 10K and 100K for R2 and see which one you prefer.