Cold Cathode Fluorescent invert as a HV source - Question

[dbrown]

Doug Coulter or any other member,
was wondering about using a CCFL as a HV power source as Doug mentioned in his post to me. I understand how to obtain/disassemble a unit and even how to connect a Cockroft-Walton ladder for the HV but was wondering what CCFL watt value is best - a 40 watt or is that overkill? Also for a 1 ma load @ 15 kev in the CW (roughly), what would be a good value of capacitor (350 pF or should a larger value be better?) Any preference for half or full wave? and if so, why (besides the small improvement in output ripple?)

Strangely, after searching the forums for information on the CCFL topic, I can't find it - I would think that this would be a common talking point due to its many uses ... .

Doug, thanks again for that idea I never would have even thought of the CW for HV (seriously; felt the auto coil would be fine as a Ion Gun source) much less a CCFL (heck, never even knew the first thing about those amazing little devils. That is an incredible amount of high tech in such a tiny package!)

[Doug Coulter]

Well, here is what I know so far about these nice little toys.

The older tech ones are better -- just the two transistors, a transformer and a couple of R's and C's.
Newer ones that have a "dimming" feature have issues when we try to vary output -- they always have the same peak voltage if you use the "dimmer" input, just change pulse width or frequency, and don't respond well to changing the output voltage by changing the input voltage like the simple ones do.

That is a quite valuable feature, as the simple ones are pretty "stiff" and this makes the output volts or current easy to regulate from the LV side of things.

The older tech uses a small series C (about 30pf) as a ballast to control current. You can replace this with a larger one if you control max current in the input to have a "stiffer" output. There is enough space there with little bit of air wiring to make a volt doubler and have it still all on board.

I've not seen simple ones that go to 40w, the best I've found are dual lamp ones that are rated at 12 watts total (and usually those are 12v input, not 5v).

Because the outputs are usually square waves in the 40-50 khz range, you don't need big caps, but do need good fast diodes which can be hard to find with HV specs -- I was lucky to score some 3kv, 20ma ones on ebay from a source in Germany, those work. Normal rectifier diodes don't -- they won't turn off quick enough and everything just gets hot.

The hot tip is to use spark-gap caps that will flash over just before the diodes fry if you have some other issue like an arc at the stack output that doesn't discharge the stack evenly. I was also lucky to score a sack of these, .0027u @ 2.5kv -- perfect. If you can't get those, yes, then keep the cap values smaller -- with a square wave they just don't need to be a large, usually as those for a sine at the same frequency and a bonus of the SQ wave is lower output ripple.

As I've posted before, all this CRT type tech (good small fast diodes and spark gap caps) are becoming harder to find out there with the death of the CRT -- so if you find a good source, stock up!
Sadly, that includes the older tech version of these inverters too. Get them while you can. I buy them in tens.

I have multiplied my "standard" one up to about 30kv, but to get that I had to coat the board and parts, as the corona loss became similar to the power the thing had....not much loss can be tolerated if you want any output current -- like even enough to "drive" a B&K HV probe. I made various PCB layouts for this use, but I primarily use these for small ion sources and phototube/gas tube power supplies for which they excel and only need a stage or two of multiplier to get to those voltages.

The 12 volt input ones I use oscillate all the way down to about 1 volt input. I use an lm317 variable regulator on the input, and control the output that way, which also acts as a current limiter -- I don't use a heatsink on that as it's cheap and easy to replace, and this cuts the current limit down to what the CCFL inverter can handle even in short circuit output with a fatter cap than they use in the first leg of the output. You can devise, and I have, a little opamp circuit to close a loop and regulate this that way -- a volt divider off the HV output, a reference etc and the opamp driving the adjust pin on the lm317 used as a "smart emitter follower" on the opamp output for your low volt DC input. Add a ten turn pot, and you've basically got what they put in a lot of NIM bins for detector power, except this is much lower ripple and generally much better and repeatable.

This is a link to the one I find most useful -- the old, brute force blocking oscillator type.
http://search.digikey.com/scripts/DkSea ... 89-1025-ND

The thing on there that looks like a to-92 transistor is actually a quick blow 1a fuse. If you put in fat output cap(s) and don't pay attention, it will blow....I've replaced it with a "penny" though and done ok as long as I paid attention to heating issues.

The long skinny ones with the higher output and lots of parts seem to universally not as useful for this work. Too hyper-specialized for a bulb load in a computer monitor. I've worked with them a little bit, and didn't find them as generally good for this sort of work, unless a pwm output wave happens to be what you want -- probably not, but who knows? The inductor you'd need to make that into variable DC would be pretty serious stuff....

Pix
First is a detail view of the type I use. TDK also makes a larger (size, not power) version of this that runs on 5v.
Second is some CW stacks I did boards for to play with how high a voltage I could get out of these at any decent current. The one on the right was the attempt at 30kv or so -- not very practical. This one I put a penny in the fusebox and instead used an 8 ohm R in series with the input power for protection.

The other board isn't laid out as smart, but at the lower voltage it makes, works fine. Cutting the first board about in half for about 10-15kv would be the best you can likely do for this if you need those volts.

As I said, I usually use these for detector power, and in that case you can kind of wedge in all the rectification and filter parts right on the board instead of those output coupling caps. I've used them with one or two stages of multiplication for small ion sources with success, too. I'd think you could improve on the SciAm ion source with something more like the penning source used in the borehole tube shown at:

http://www.coultersmithing.com/OldStuff/pdfs/dox.html

under Phillips chapter 8. you wouldn't need such a high voltage with that.

Edit:
I've managed to float the grounded end of the HV winding on these to do other tricks, but I suspect you'd run into troubles with transformer insulation if you pushed that really far.

Since these draw low enough current to run on batteries (which can be floated off ground) you could hook a few in series using separate batteries for more volts and power output (after you rectify and filter each one separately). Might be a little tricky to do the power switching though....I was thinking of using an IR photodiode on each and an IR source to remotely control that, but never got that far with it. Maybe a long insulating rod working a series of microswitches would do that better anyway.
I have sometimes wondered if that wouldn't be the best way to do a decent linac, just use one of these (multiplied a bit) for each stage...I'm not sure what effect (good or bad) the relatively "stiff", ie, low impedance outputs would have there, but you'd sure have all the focus adjustments you could ever want if they all had their own variable regulation on the input.

I have not tried this yet on these, but a small cap between the base of a transistor on one board and that on another one should let you have them in sync for full wave CT outputs at twice the power.
They are just blocking oscillators, and should sync up in or out of phase fairly easily if you put the cap between the right two points on the two boards....A little bit of experimentation should tell.

[Jerry Biehler]

Another option is a electronic ballast for regular fluorescent lights. They are CCFLs on steroids. About 40khz out.

[richnormand]

You might be interested in this. Also available in a tray of 80 units!

http://www.goldmine-elec-products.com/p ... ber=G16565

Edit:
Not the greatest units but the price is right to experiment. Actually fixed two separate computer monitors with burned invertor boards with these over the past year. They run hot at 12V but they are still going!

[dbrown]

Doug,
thank you - that is a ton of great information! I'll try and digest it but a good bit is above my minor electronics knowledge. I ordered one of the CCFL systems - what a great device - not sure if I want to use that and a CW ladder with it (I have a number of 3 keV caps and only some 1000 volt diodes (will look for some 3 KV.) Thanks, my ion gun parts list is just about complete!

Richnormand - great site - I order two of those - they look really useful for my project and are just too nice not to own even if I didn't need it (which I do)!

You guys are beyond nice and helpful to the point where I feel really guilty - all I do is ask for help and you all do that but at 110% level - boy, not use to that.

I have 90% of all my parts and the vacuum system is complete (but untested for high vac since I don't have the active cold trap yet. System has been leak tested at 1 micron. All I have left to get is another 'Van de Graaf' sphere (will try the steel salad bowls)! and machine the Einzel lenses.

Hope to assemble the final vacuum system in its bench, and complete all water lines this weekend. The cold trap (an old ice creme maker) is on order and when that arrives, I'll test the operational system to the max. Will post a photo.

May get the ion gun assembled this weekend and post the basic item less power supply.

Thanks again. I will continue to post and comment but I can't tell all of you how helpful you all have been.

[Doug Coulter]

For what it's worth I have some of those myself, was trying to get things running on 4 aa cells for example. For whatever reason, they are inefficient as it gets, drawing 160ma with no load, so I used the 12 ones at the lower input (works down to 1v) and just added a multiplier stage to the output of that instead to get to phototube/geiger voltages in that project -- and it draws more like 20 ma no load.
I did try fiddling the bias on the TDK one and got it down to 120 ma or so, but that's as low as I could get it to go and still oscillate even at full input volts. I recommended the one I did for good reasons.
For me, twice the price at 6 times better efficiency is a good tradeoff.

I've not fooled with the ones from house lighting, but of course they need high volts in, so batteries are out, and the application here is inside the corona ball of a van de Graff....

I'll have to cut some up and see how they do, but my take (as just about all the lighting here is with those) is that they are marginal in reliability at the power levels they run at anyway. They seem to have been "generous engineered" a term we used to use on crap GE consumer electronics where it looked like they took parts out and made things (like poor quality PCB) cheaper until it failed right away for them, then put one back in and shipped it. Said in a sarcastic tone of voice. I know that here they fail long before the lamp itself goes bad, but those little guys are rarely the cause of failure in computer monitors.

[Richard Hester]

I don't know that I'd use the little CCFL boards for any serious power, but I have used them as HV sources for PMT-based scintillator detectors. I have two detector/preamp/SCA all-in-one neutron scintillatorsetups that use the litttle chewing gum-sized inverters as the HV source. I had to shield the inverter board, as it was a pretty powerful source of interfrerence due to the high-leakage transformer construction (it seriously screwed up my output ripple measurements). I took the brute force approach and used a CCFL inverter into a doubler followed by a shunt regulator in order to get some reasonably clean HV.

For more power, a board from a large screen LCD TV or monitor may something to look at, though there you'll run into the extra complication of the dimming circuits.

[Richard Hester]

BTW, if you're looking for a fairly stable HV supply with low ripple (800V-1kV or so), the circuit for the CCFL inverter/shunt regulator HV supply is lurking somewhere in "files". It's tested, and it works. It's also not a big deal to put one together on a piece of perf board, which is how I prototyped the circuit (long ago, in a far-away galaxy...).

[Jerry Biehler]

I used to do commercial lighting and have seen the insides of quite a few ballasts. GE ones are one of the cheapest made. Not as bad as sylvania though. Advance makes some real good ones though but they are often potted.

A lot of the ballasts use the same board for 2 and 4 lamp ballasts, they just use different and more components to make the switch, hence the extra holes.

[Richard Hull]

Having been here since Hector was a pup, I have all of Richard's circuits that fell like manna from heaven in these forums in a small GBC bound volume entitled "Richard Hester's Rad Circuits". I have built most of them. They work.

Richard Hull