Hmmm... well, here is a two-day last chance to tell me how I am going to blow my old PSU to pieces. I am thinking of skipping testing monday to allow for discussion.
I have wound several hundred wraps of my super-fine wire on the core and replaced it. God knows how many, I didn't even bother trying to count. If the thing reads continuity tomorrow I will see about powering it up. On a local surge protector.
For clarity: This is the last big magnetic part before the HF filtering, a biggish toroid that I am fairly sure gives my old PSU its 12 and 5 volts. On this are now several hundred turns of very fine wire that I hope will yield low-current high-voltage in the several hundred volts range.
I am pretty sure my multimeter won't be able to test the voltage, so I am thinking of making a power-on check to see if the power supply explodes with nothing attached, and then going straight to a known-good lamp with a series capacitor to limit current. If it strikes a light I'm good. If not it's possibly another couple hours' winding, but the hole in the center of the torroid is getting awfully small. I will wrap the lamp or put it under some sort of a shield in case of explosion.
Go on, please. Tell me what may go wrong and how to possibly avoid it.
Thanks all, but especially to sch3matic, for humoring me.
I have wound several hundred wraps of my super-fine wire on the core and replaced it. God knows how many, I didn't even bother trying to count. If the thing reads continuity tomorrow I will see about powering it up. On a local surge protector.
For clarity: This is the last big magnetic part before the HF filtering, a biggish toroid that I am fairly sure gives my old PSU its 12 and 5 volts. On this are now several hundred turns of very fine wire that I hope will yield low-current high-voltage in the several hundred volts range.
I am pretty sure my multimeter won't be able to test the voltage, so I am thinking of making a power-on check to see if the power supply explodes with nothing attached, and then going straight to a known-good lamp with a series capacitor to limit current. If it strikes a light I'm good. If not it's possibly another couple hours' winding, but the hole in the center of the torroid is getting awfully small. I will wrap the lamp or put it under some sort of a shield in case of explosion.
Go on, please. Tell me what may go wrong and how to possibly avoid it.
Thanks all, but especially to sch3matic, for humoring me.
Toroid? Most SMPS use an EI or dual E core for the transformer. Sure it isn't the common mode filter choke? (All voltages have the same ripple, so they choke-input filter them all at the same time. If you have a load on it, it'll put out some voltage through here, but nowhere near as much or as well as from the power transformer.)
Tim
Tim
Stocker said:
The toroid is a part of the HF filter! you couldn't get much power from it.
The main switching transformer is EE / EI type, but it's full,and winding is difficult. (Also, it's too near to the mains.)
It's better that you keep the old AT PSU survive, DIY a inverter and power the inverter with the PSU.
BTW: the switching waveform in the AT PSU is square wave, so a series capacitor won't be a good thing to limit the current. It's only suitable for sine wave source (e.g. 50Hz mains or resonant inverter)
Well, that's all some good information to know, for the next try!
Moot for this try however because the wire is broken somewhere. It's looking more and more like I'm going to have to make a proper inverter. I would like something considerably smaller than a computer power supply, too. I can use the mains filtering side of this one to get me 1/3 of the way there, I suppose.
On a side note, the one I ended up using was an ATX supply. I suppose if I had to, I could still use the AT supply to power my own inverter (which I hadn't thought of, thanks! ) More research will be required however since I would like to do this with parts I have on hand. I have some oddball stuff, but I'll also see about reusing the controller for the ATX supply...if I can convert it to a more CCFL-friendly sine wave output. Otherwise, it's a LOT more research!
Moot for this try however because the wire is broken somewhere. It's looking more and more like I'm going to have to make a proper inverter. I would like something considerably smaller than a computer power supply, too. I can use the mains filtering side of this one to get me 1/3 of the way there, I suppose.
On a side note, the one I ended up using was an ATX supply. I suppose if I had to, I could still use the AT supply to power my own inverter (which I hadn't thought of, thanks! ) More research will be required however since I would like to do this with parts I have on hand. I have some oddball stuff, but I'll also see about reusing the controller for the ATX supply...if I can convert it to a more CCFL-friendly sine wave output. Otherwise, it's a LOT more research!
You could reuse the magnetics to build a half bridge series resonance inverter.
Perhaps 555 oscillator + diode/resistor gate drive + a pair of N channel & P channel MOSFETs + resonance inductor (about 20uH) + resonance capacitor(about 1uF). The CCFL output coil is winded on the inductor.
power it with 12V and adjust the frequency to modify the resonance voltage output.
The resonance voltage at the primary ("low voltage") side may be up to 150 Volts, take care.
I have once get 150V peak voltage out of 24V by series resonance. The capacitor is a 63V cerametics SMD, 0805 size. It becone nonlinear at high voltage and limited the resonance amplitude. However, using this as a voltage limiting method may break the capacitor. You'd better use a 400V metal one. It's linear, so resonance amplitude at no load may go up too high -- the big current may destroy the MOSFETs. So apply some over voltage / over current sensing and shutdown protection.
Perhaps 555 oscillator + diode/resistor gate drive + a pair of N channel & P channel MOSFETs + resonance inductor (about 20uH) + resonance capacitor(about 1uF). The CCFL output coil is winded on the inductor.
power it with 12V and adjust the frequency to modify the resonance voltage output.
The resonance voltage at the primary ("low voltage") side may be up to 150 Volts, take care.
I have once get 150V peak voltage out of 24V by series resonance. The capacitor is a 63V cerametics SMD, 0805 size. It becone nonlinear at high voltage and limited the resonance amplitude. However, using this as a voltage limiting method may break the capacitor. You'd better use a 400V metal one. It's linear, so resonance amplitude at no load may go up too high -- the big current may destroy the MOSFETs. So apply some over voltage / over current sensing and shutdown protection.
Ok, here's another idea for you fellows to shoot full of holes!
How about a regular amplifier-type topology, but freaky-high voltage, and VERY fast?
That is, a "simple" transformer-bridge-filter that gives (+/-? ) 600VDC, with a *serious* step-down supply or maybe a secondary winding supply for +/-15V to power a simple opamp sinewave oscillator (or high voltage discrete off the main supply? ) running about 65-65kHz. Output about +/-10V if using the opamp, followed by a voltage gain of about 60, to some high-voltage transistors (or FETs? ), with again the current-limiting capacitor on each tube.
Any takers? Any comments?
How about a regular amplifier-type topology, but freaky-high voltage, and VERY fast?
That is, a "simple" transformer-bridge-filter that gives (+/-? ) 600VDC, with a *serious* step-down supply or maybe a secondary winding supply for +/-15V to power a simple opamp sinewave oscillator (or high voltage discrete off the main supply? ) running about 65-65kHz. Output about +/-10V if using the opamp, followed by a voltage gain of about 60, to some high-voltage transistors (or FETs? ), with again the current-limiting capacitor on each tube.
Any takers? Any comments?
Problem with that is, as I said, I am a low-voltage solid-state guy. I have -->.<-- that much experience *making* inverters, or anything else for that matter involving over 120-ish VRMS. Similar level of experience with anything over 20kHz, especially in selecting and designing the magnetics.
That said, the goal here IS to design a (multiple) CCFL power source, from scratch, be it an inverter or no. The secondary goal (becoming less important the more I learn about the primary goal) is to spend zero dollars on the project. If by a roundabout way I end up with a textbook inverter scaled up for multiple lamps, I will have learned a great deal more than just going out and buying a pile of inverters (and likely saved a great deal of money, if not time).
Right now, I have as a general tertiary goal of ending up with a power source that gives my lamps a life-extending sine wave of AC voltage with a frequency around 60 or 65kHz, and if it's dimmed, then dimmed with a PWM scheme in the 204-280'ish Hz range
Of course I am likely missing a great deal of information that leads to nOob questions, like this:
Is it easier/better to generate high voltage high frequency sine waves, or to amplify low voltage high frequency?
That said, the goal here IS to design a (multiple) CCFL power source, from scratch, be it an inverter or no. The secondary goal (becoming less important the more I learn about the primary goal) is to spend zero dollars on the project. If by a roundabout way I end up with a textbook inverter scaled up for multiple lamps, I will have learned a great deal more than just going out and buying a pile of inverters (and likely saved a great deal of money, if not time).
Right now, I have as a general tertiary goal of ending up with a power source that gives my lamps a life-extending sine wave of AC voltage with a frequency around 60 or 65kHz, and if it's dimmed, then dimmed with a PWM scheme in the 204-280'ish Hz range
Of course I am likely missing a great deal of information that leads to nOob questions, like this:
Is it easier/better to generate high voltage high frequency sine waves, or to amplify low voltage high frequency?
650V is not enough for a CCFL tube, so you shuld still use a high frequency transformer.Stocker said:
Stocker said:
Build a circuit running directly off the mains and debug it is either dangerous or difficult. At least, standard oscilloscope can't be used to view the waveform safely -- the shell of it is connected with the test cable!
You don't have any high voltage experience, so read the safty operation instructions in the Tube forum.
Keep the semiconductors running at low voltage and use a transformer to get the high voltage for CCFL.
Regards.
Kenshin
That is not what I have seen, from the specs on various commercial inverters. I also (above) proposed as a quick example perhaps 1200V (+/-600 ), which is enough.
edit: oh yes, if bridged this could be up to 2400 which is usually going to be more than enough.
I have no intention of doing any such thing (currently). The above idea involves the use of a step-up transformer on the mains, which would also provide isolation.
I don't have high voltage *design* experience, which is my problem. I have been working with and around high voltage safely for years. When I was young the safety came by blessing and chance. Since my Navy training, it's by blessing and on purpose. Thanks for the pointer though; it's been a while since I studied such things. I'll read that very soon.
The transformer would be a high power HF item, which I don't have. That goes against one of my design goals but can be dealt with. I also don't know enough (yet) about HV HF magnetics to even select the core properly! Like I said, more study is in my future and I do not shrink from it. My wife has different ideas and hassled me off the computer at 2:20 this morning.
This is turning out to be quite a project!
Stocker said:
So didn't I. But I've got a small working induction heater circuit now.
Work up the theory on flyback inverters, then build one for 12 > 60V stepup. Then add a 1:20 winding and the tube.
A very crude inverter can be had with a 555 or two transistor mutivibrator clock and highspeed HV darlington, or MOSFET. A slightly more elegant inverter can be had with any number of SMPS chips (SG3524 et al, TL494/KA7500, etc.), with op-amp controlled dimming and light regulation.
Start collecting dead computer PSUs!
I'm no lighting scientist, but I'd say a current-limited square wave would work best. For sure you want variable current, since any gas discharge has a zener-diode-like response.
Yes and no and no and yes!
HF sinewaves are easy because you can use a resonant circuit to couple it together. RF principles are black magic mind you, so that may be a ways off in experience.
Multiplying (amplifying isn't right here) low voltage is even easier because you just let an inductor do all the work. However, a single-ended arrangement leaves you with an assymetrical output, which means the tube ends up being turned on by a tall positive-going pulse, so it sees DC and likewise, uneven wear.
The ultimate would be a resonant tank at 1-3kV with series ballast inductors feeding individual tubes. The tank is fed by a half or full ("H") bridge MOSFET inverter, also with a series matching inductor to protect the squarewave-switching FETs from the capacitance of the tank. It also adds "springiness", so you can operate the bridge at a lower voltage (say, 200-600V) and use Q multiplication to get the voltage to the tubes. Power output would be controlled by moving the clock frequency above resonance. Such a circuit could handle a kilowatt or so with $25 MOSFETs.
Tim
Sch3mat1c said:
I ah... I don't think I want to make something capable of a kilowatt as my first project involving HV!
From what I have read, a sinewave is easier on the tube than square, but you guys keep mentioning square waves - is it ease of implementation or something more?
Why restrict myself to single-ended voltage multiplication? Makes more sense to me to do AC with +and -
At any rate, simple is good and crude may be acceptable. All of these ideas mean more reading for me. If anybody wants to do the digging for me I'd appreciate it, but I think I have at least a few weeks' reading before I can make much of a meaningful next step on this project.
It's all about switching and switching topologies.
When something goes quickly between only off (near zero current = near zero power) and on (near zero voltage = near zero power), it dissipates very little heat - 90% efficiency is remarkably easy to obtain with modern transistors. The hard switching requirement means it has to operate as a square wave, though. This is where the magical inductor comes in, smoothing out the hard corners and easing current transitions.
Single ended is easiest because you only need one transistor, it can be a small one, but it also means (usually) asymmetrical drive to the tube, as I mentioned. You can do a push-pull setup, but that takes twice as many transistors, kind of useless when a mere quarter of a transistor will suffice.
Anything like half or full bridge is encountered in higher power units; I bet a 300B the computer you are using right now has a half bridge topology in the PSU. (Either BJT or MOSFET, 2SC2625s and C3066's (I think?) were popular in AT size units, I don't know exactly what they use today.) It's not worth digging up four (or 8, 12, 16...) transistors to do a puny load, so full bridge (aka H-bridge) is used in the kilowatt+ range.
You could make a circuit with a cute little driver transformer and everything, use maybe eight 2N3904's in H-bridge to drive maybe 10-20W of CCFL, but that's just not the right tool for the job.
Again, I'm no lighting expert, but I doubt a tube cares about square or sine. After current limiting, it probably looks square anyway, assuming the gas can switch at that speed. (If it can't, it'll look somewhat more resistive, at least at lower-than-average forward voltages.)
Asymmetrical voltage output from an SE output is easiest used directly, but you can add a resonant matching circuit or such to trim it down to something more siney.
Tim
When something goes quickly between only off (near zero current = near zero power) and on (near zero voltage = near zero power), it dissipates very little heat - 90% efficiency is remarkably easy to obtain with modern transistors. The hard switching requirement means it has to operate as a square wave, though. This is where the magical inductor comes in, smoothing out the hard corners and easing current transitions.
Single ended is easiest because you only need one transistor, it can be a small one, but it also means (usually) asymmetrical drive to the tube, as I mentioned. You can do a push-pull setup, but that takes twice as many transistors, kind of useless when a mere quarter of a transistor will suffice.
Anything like half or full bridge is encountered in higher power units; I bet a 300B the computer you are using right now has a half bridge topology in the PSU. (Either BJT or MOSFET, 2SC2625s and C3066's (I think?) were popular in AT size units, I don't know exactly what they use today.) It's not worth digging up four (or 8, 12, 16...) transistors to do a puny load, so full bridge (aka H-bridge) is used in the kilowatt+ range.
You could make a circuit with a cute little driver transformer and everything, use maybe eight 2N3904's in H-bridge to drive maybe 10-20W of CCFL, but that's just not the right tool for the job.
Again, I'm no lighting expert, but I doubt a tube cares about square or sine. After current limiting, it probably looks square anyway, assuming the gas can switch at that speed. (If it can't, it'll look somewhat more resistive, at least at lower-than-average forward voltages.)
Asymmetrical voltage output from an SE output is easiest used directly, but you can add a resonant matching circuit or such to trim it down to something more siney.
Tim
Ah, but I have bags of 2n3904 (and 3906, and 2222/2907 etc. )
If it's not the right tool for the job but it works I may use it if it is an easier implementation. My US-bought screwdrivers all have warnings about how they are not prybars, punches or chisels, but I have successfully used them for all of the above, and more!
How about this (just thought of it)
little transistors in a bridge, with a little tiny transformer for each lamp?
It'd be cute
It'd be tidy
It'd eliminate long wire runs to the tubes from one central power supply
If you guys keep giving me things to look up, I shall be much older before I have a power supply built!
/edit: I keep forgetting to mention:
I have lots of SUPER-WHITE plastic to reflect the light from the inside of the box. I have read that the (in this case, metal) reflector around a tube can sap some of the current because of the capacitive coupling to the lamp.
Would it be a good idea to suspend the lamps by each end and perhaps a few isolated standoffs along the length for additional support, instead of using the reflectors they came to me with?
If it's not the right tool for the job but it works I may use it if it is an easier implementation. My US-bought screwdrivers all have warnings about how they are not prybars, punches or chisels, but I have successfully used them for all of the above, and more!
How about this (just thought of it)
little transistors in a bridge, with a little tiny transformer for each lamp?
It'd be cute
It'd be tidy
It'd eliminate long wire runs to the tubes from one central power supply
If you guys keep giving me things to look up, I shall be much older before I have a power supply built!
/edit: I keep forgetting to mention:
I have lots of SUPER-WHITE plastic to reflect the light from the inside of the box. I have read that the (in this case, metal) reflector around a tube can sap some of the current because of the capacitive coupling to the lamp.
Would it be a good idea to suspend the lamps by each end and perhaps a few isolated standoffs along the length for additional support, instead of using the reflectors they came to me with?
I have just read reports of three commercial CCFL inverters exploding and/or burning inside peoples' computer cases. I know commercially-bought is no guarantee of quality but it occurs to me to wonder:
Is a CCFL a safe lighting option? I can house this project in an all glass and metal case with heatsinking as required, but
Will a CCFL backlight ever be long-term reliable in use? Is there a "safest" topology? I suppose that is a good question and becomes now a primary goal. The cost of implementation may just kill this project for good in favor of bright white LED illumination. I don't really want to burn my house down because I was too cheap to dish out $100 for safer parts eh!
So the question becomes: What is the "safest", or is there a "safe" circuit that can be made by the average DIY'er?
Is a CCFL a safe lighting option? I can house this project in an all glass and metal case with heatsinking as required, but
Will a CCFL backlight ever be long-term reliable in use? Is there a "safest" topology? I suppose that is a good question and becomes now a primary goal. The cost of implementation may just kill this project for good in favor of bright white LED illumination. I don't really want to burn my house down because I was too cheap to dish out $100 for safer parts eh!
So the question becomes: What is the "safest", or is there a "safe" circuit that can be made by the average DIY'er?
It's safe if properly manufactured. Since all lap-top computer has one inside. And there's seldom any report of laptop CCFL accident.
Safe topology for average DIYers? It depends on...what does "average DIYers" mean?
Safe topology for average DIYers? It depends on...what does "average DIYers" mean?
Stocker said:I have just read reports of three commercial CCFL inverters exploding and/or burning inside peoples' computer cases. I know commercially-bought is no guarantee of quality but it occurs to me to wonder:
Is a CCFL a safe lighting option? I can house this project in an all glass and metal case with heatsinking as required, but
Will a CCFL backlight ever be long-term reliable in use? Is there a "safest" topology? I suppose that is a good question and becomes now a primary goal. The cost of implementation may just kill this project for good in favor of bright white LED illumination. I don't really want to burn my house down because I was too cheap to dish out $100 for safer parts eh!
So the question becomes: What is the "safest", or is there a "safe" circuit that can be made by the average DIY'er?
alright then, I was fudging. I was, perhaps unsurprisinly, referring to me.
I have more than 1/2 my life been playing in electronics. Currently I have the better part of a decade in working professionally with, especially troubleshooting and maintaining, electronics. My home equipment includes a nice soldering station and DVM, and all the hand tools I think I would need*. I can make PCBs but prefer point-to-point wiring. I can do my own metalwork and glasswork as required. I can think logically and rationally, figure things out for myself and follow instructions. I have a healthy respect for all things that I know to be dangerous or lethal, but will employ (or play) with them when I am pretty sure I will not get blown up.
The answer to the question should have been obvious to me, as your answer shows, but I was freaked out by the reports and photos I had just seen.
I suppose the worst-case power supply failure could be well contained in a small, sealed steel box with a fuse outside of it. Okay, the primary goal is back to what it was: getting the lamps glowing. I will be extra safety conscious once I have the design worked out and ready for permanent implementation.
*My oscilloscope just showed me it doesn't work anymore. It hasn't been used for years and is sufficiently ancient as to use a 5.25" floppy drive for its memory storage. Something in the measurement or display has given up and I don't even feel like fooling with it. It is on the short list of things to be disassembled and thrown away when my curiosity is satisfied. I am mostly keeping it intact because my toddler son likes all the knobs and buttons. It has a +/- 2000V power supply for the CRT and I wondered briefly if I should try to utilise it for this project but quickly decided a hack job on 4000V is a good way to get blown up.
I have more than 1/2 my life been playing in electronics. Currently I have the better part of a decade in working professionally with, especially troubleshooting and maintaining, electronics. My home equipment includes a nice soldering station and DVM, and all the hand tools I think I would need*. I can make PCBs but prefer point-to-point wiring. I can do my own metalwork and glasswork as required. I can think logically and rationally, figure things out for myself and follow instructions. I have a healthy respect for all things that I know to be dangerous or lethal, but will employ (or play) with them when I am pretty sure I will not get blown up.
The answer to the question should have been obvious to me, as your answer shows, but I was freaked out by the reports and photos I had just seen.
I suppose the worst-case power supply failure could be well contained in a small, sealed steel box with a fuse outside of it. Okay, the primary goal is back to what it was: getting the lamps glowing. I will be extra safety conscious once I have the design worked out and ready for permanent implementation.
*My oscilloscope just showed me it doesn't work anymore. It hasn't been used for years and is sufficiently ancient as to use a 5.25" floppy drive for its memory storage. Something in the measurement or display has given up and I don't even feel like fooling with it. It is on the short list of things to be disassembled and thrown away when my curiosity is satisfied. I am mostly keeping it intact because my toddler son likes all the knobs and buttons. It has a +/- 2000V power supply for the CRT and I wondered briefly if I should try to utilise it for this project but quickly decided a hack job on 4000V is a good way to get blown up.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.