CFL Driver Circuit TESTING STAGE

[s4w2099]

This driver circuit was made from random stuff that I found rolling on the floor. I only bought the case in Rat-a-Shak for 3 BUCKS! (expensive for the size.

it uses the following components:

555 Timer

5 Resistors

2 ceramic disk capacitors

a power MOSFET IRFP150N (you can change this one with a more common one like the IRF540)

a transformer that I striped from wife's clock/radio/alarm   :-P

Check the pictures out:

This reflector I made it with "Chuck Tender Steak" packaging. It might not have the perfect shape but it was easy to make with since I could do all the cuts with a kitchen knife. I used a little bit of aluminum foil with the shiniest side pointing out.

 

It runs at 12V and its sucking 135mA form the battery. I know that it is only powering the 8Watt CFL at about 1Watt (very underpowered) but I didnt needed more light since it was only going to be in the hallway turned on all night so I could find my way to the bathroom in the darkness. This allows a very long run time even with old, cheap, damaged batteries.

Last night I left it on and it was running from a semi-charged 12V worn out laptop battery that could not hold a good charge for the laptop. The next morning it was still as bright. The laptop battery I believe it is 2.6 AH.

The MOSFET that I used is well... VERY overpowered for this circuit but it was the only thing that I had in hand. This same circuit could easily power up to 120Watts of CFLs with only this transistor a bigger transformer and a VERY few modifications to the circuitry.

I will post the schematics as soon as I am finished with the tests. So far so good. I will include a version that delivers more power to the CFLs for those of you has can generate more power than me :-(

[coldspot]

Nice use of stuff thats handy !!!



Reflectors-

I've been using veggy steamers from a dollar store

they open and close for fitting different size pans and are very shinny for great effect.

CFL Driver Circuit-

I've had a bad run of busting 120 VAC CFL's lately

(building shed/shop and have them hanging around off nails n screws and I've knocked them off a few to many times and lost about 4, glad they only cost a dollar each but wish they didn't bust)  

 and have saved the screw in bases wondering about reuse in some form.

Anyway

great work !!!!!!!!

[s4w2099]

Thanks man. I wish to know where you buy them from because at home depot they can knock you out almost 10 bucks per tube. I had these around but if I had to buy them it would be a pain.

[coldspot]

I've scored a few from dollar store but not for a few years now.

Last winter, Walmart had some for $0.96 each and I wasn't smart enough to buy more that 5, (Because I had the ones from the dollar store in storage, {girlfriend won't let me use them at her house, she pays the bills}).

Only one time about 4 years back I found a bunch at the dollar store, I bought ALL they had,(30-40) but, didn't get to my truck before some guy bought them ALL off me and in a few moments I'd made about $25 !!!

Had extras in storage back then but now with shed/shop I'm using them up.

Lucky to still have a stockpile of 6 VDC battery powered ones that I'll make into 12 VDC for shed/shop after I finish a few more things and actually hook up battery banks and un-plug the extension cord thats been powering me so far.

Sucks to have mills n panels just sitting around waiting for use!!! sucks to only have 24 hours in each day and not be able to use them all with work n sleep needing to use some of them !!!!!!  

LOL!!!!

 

[terry5732]

I got more light for the same amperage with a similar set-up using 48 inch 40 watt tubes than I could get from any of the various compact / low watt tubes I have. I would get visibly nearly as much light pushing 1 watt through them as I got at 10 watts. Every different tube I tried had it's own harmonic frequency(s) that it preferred, so I used variables for R1 and R2 to tune the cycles and pulse width.

[s4w2099]

Wow thats something to take note of. I wasnt aware of that. I will keep experimenting at 1Watt then and try different bulbs at different frequencies/duty cycles.

Thanx

[s4w2099]

I tried a different bulb. This time a 36Watt. The amount of light that it emits is much greater. The circuit was not modified at all. Then I tested using 24Volts and it still works with even greater amount of light. There is no heat coming from the unit neither from the bulb or from the transistor or transformer.

I am not looking for great a mount of light just something to be able to find my way to the bathroom. I will still use it as planned with the 12V battery and the 8Watt bulb. The light that goes into the bedroom through the bottom of the door bothers me a bit (I like to sleep very very dark).

Its been up for 3 nights and I have not seen significant decrease in the battery voltage and there is no burning smells or fire from the unit so I will call it a success. Apparently the efficiency is very high on this one.

One thing that I did thought about was to add a motion sensor to the light so it only turns on when I open my bedroom's door. That way I will save even more watts instead of just leaving it on all night.

Ive seen a few on ebay for 4 bucks and even for a dollar 6USD S&H to my place.

Thanx

[JW]

s4w2099,

 I would like to make one of these, though I am looking for brighter output. Where can I find the schmatic for this circuit? The one you used, I can work it from there. Any links or pictures would be appreciated.

JW

[JW]

Is this it?

JW

[commanda]

At what frequency are you driving it?

Fluoro tubes are far more efficient at higher frequencies. Most (commercial) units run at 10-20 KHz. There are some graphs published on the web somewhere which show typical light output vs frequency. Google is your friend.

I presume that transformer you're using is a mains (60Hz) transformer. Usually won't work much above a few hundred hertz. Need a ferrite core for higher frequencies.

Also, the 555 will need to be a ttl 555, not a cmos 7555. If you get the frequency up where it should be for proper operation, you will almost certainly need a driver stage for the gate of the fet. A bipolar (NPN/PNP) pair of transistors is usually sufficient.

In my diary somewhere is one I made using a ferrite core salvaged from a pc power supply and a fet. This is about the minimum parts count you can get away with.

Amanda

[JW]

Hi

"Part Total Qty. Description Substitutions

C1 1 100uf 25V Electrolytic Capacitor  

C2,C3 2 0.01uf 25V Ceramic Disc Capacitor  

C4 1 0.01uf 1KV Ceramic Disc Capacitor  

R1 1 1K 1/4W Resistor  

R2 1 2.7K 1/4W Resistor  

Q1 1 IRF510 MOSFET  

U1 1 TLC555 Timer IC  

T1 1 6V 300mA Transformer  

LAMP 1 4W Fluorescent Lamp  

MISC 1 Board, Wire, Heatsink For Q1  

http://www.aaroncake.net/circuits/flampdrv.asp"

Thats were I found this schematic. I guess the circuit could be built and hz measured with various capacitor values, with matching resistors. I think this was first published by Electronics Now magazine. Looks like a fun circuit to play with.

This story has inspired me to build one. I want to run some 48in bulbs.  

JW  

[s4w2099]

Hello, sorry for the delay. Here is the schematic of mine. Its very very similar since there are not many ways that you can play with the 555 for this kind of task.

As commanda says it is not the best practice to use a mains transformer since they are not designed to run over 60Hz. If you have at hand a ferrite core transformer it will beat the crap out of mine. The only reason that I made it like that is because I only had that regular transformer.

If you make this circuit with another transformer it might not work because of the different characteristics that our transformers might have. Regardless you can make it work by playing with the frequency and the pulse width.

you can use the exact same circuit with a better transformer and your life would be more simple. I dont remember the value of C1 that I used for mine but it is a ceramic disk capacitor. I didnt read the value because I was testing a value if it didnt worked just grab another one (I have a big pile of them) until I hit the right frequency for the bulb/transformer. Once again get a ferrite transformer and you wont have to do that :-D

[s4w2099]

Hello, according to my meter it measures 12khz. I dont know if its my meter thats giving me a wrong measure. I wish my oscope would be working to actually see whats happening. It is a miracle that that little transformer is actually working at that frequency. I know for sure that the frequency is not 60Hz because I can hear a sound coming from the transformer very very much faster than the sound that power invertes make at 60Hz. Regardless it is being very efficient.

I know that ferrite core is the real thing to be doing this kind of project but I didnt happened to have one so I used a regular one. Its not the first time that Ive done this and all of the times have worked. Choosing a transformer that has about 7 ohms resistance in the primary side would make it work nicely. If its lower the spikes will go back to the chip and fry it.

I understand the driver transistor part because it is commonly used in inverters but I dont know why you would want to use a TTL chip instead of CMOS. I used CMOS because it is harder to fry because the operating voltage is way higher. The voltage spikes from the transformer will not hurt it if you use the suggested resistance for the primary. For sure I can make a more complex model and make it highly reliable but it would defeat the purpose in the sense that you cant just find the parts rolling in your kitchen floor. But I do agree with your concerns. Please let me know why a TTL and not a CMOS.

Thanx

[s4w2099]

Wait a second Commanda, I remember that wife's clock radio had a ferrite core coil antenna for tunning AM maybe I can make a more decent transformer out of it. Is there anything I need to know about winding a ferrite core transformer? Ive never made one before.

[commanda]

TTL 555 vs cmos 7555; Output drive capability. You need huge amounts of current to switch that gate as quickly as possible. You need to get it from off state to on state as near instantaneous as possible. If it's spending time getting from one state to the other it's dissipating heat = wasted energy = possibly cooked fet. Typical gate driver devices can switch an amp or more.

AM radio ferrite cores are ideal.

bthumble has a very good writeup

http://www.smallsolar.org/hardware/fluoro/circuit.htm

I wound one by hand, nearly killed me. Build a coil winder; you've been warned.

Amanda

[RP]

Another good place to look for ferrite is the flyback transformer in anything with a CRT.  These run at ~15Khz and are actually a square "loop" of ferrite which should be even more efficient.

Take a look inside any old computer monitors or televisions.  BE CAREFUL in disconnecting it from the CRT as high voltage can be present for a long time after the unit was last operated.

[yuandrew]

I built the circuit on Aaroncake's website that JW posted. It didn't seem to work good at first with the 6 volt transformer so I tried a 3 volt transformer instead and got more light from the 4 watt tube it was intended to run. It still didn't light the tube fully but it was enough for a nightlight in a dark room. I've discussed that circuit with someone a flashlight electronics forum I also hang out on and he told me that the circuit is mostly DC biased and may blacken those tubes (it eventually did).

Anyway, I also tried operating a 4 inch white cold cathode tube from JKL components and this circuit did a better job with it. You should have no problem getting enough light to  read from a few feet away. I might even try ordering a 6 inch or an 8 inch tube and see if those will work as well. It draws 280 ma at 12 volts with the 4 inch tube.

Currently, I'm messing around with a circuit board from a closet light I brought from Lowes which takes 8AAs or a 12 volt adaptor. If you take it apart, there's a potentiometer on the circuit board that you can crank up to drive the lamp harder. I think it was only drawing about 5 watts when I first bought it but I have it running at 7 watts now. I can still turn it up more but it gets very warm cranked all the way up. You can view the schematic for it here.

http://members.misty.com/don/amtac1.gif

I'm going to be testing it with some other fluorescent lamps I have sitting around.

[drdongle]

What we need is a good 48 V CFL driver

[fungus]

"If you take it apart, there's a potentiometer on the circuit board that you can crank up to drive the lamp harder. I think it was only drawing about 5 watts when I first bought it but I have it running at 7 watts now. I can still turn it up more but it gets very warm cranked all the way up. You can view the schematic for it here."

Apparently(according to the manual of my fl driver) you should turn the pot so it will just light if disconnected/reconnected again. I'm not sure if you'll get much more light output by turning it up higher.

[s4w2099]

I do understand your point and it makes sense.

Today the light finally stopped working after it drained the battery.

[s4w2099]

Be the way, the version of the 555 that I am using is the NTE955M which has a fall rise time of 100ns that does not sound that slow to me.

[s4w2099]

That should be a piece of cake. This time I will make another one with two transistors one of each side of the wave so it does not turn the bulb black, and a ferrite core transformer. I can make one but it will not be made with scrap parts. I will need more sophisticated ICs or more components.

It can either be made from the same 555 and a flipflop with two sets of NPN&PNP driver transistors for the FETS, transient voltage suppressor diodes to protect the FETs and a larger ferrite transformer.

The other option is using SG3525 which is a PWM circuit which can be used to directly drive teh FETs since it has the transistor driver circuit built in. The PWM can be used to dim the lights when not much light is needed like in my case (saving lots of power).

[Slingshot]

How important is the turns ratio of the transformer?  Are you trying to hit a particular open-circuit voltage on the output side?

[s4w2099]

Its important if you want to do it the right way. You need higher voltage than 120Vac in the CFL side. Mine gives me an output of 400Vac in the CFL side. As I said before. I dont have an oscope to really see the voltage and that might be mistaken. Depending on the lamps it should take thousands of volts to get it going (large ones). So it basically depends on the lamp that you are trying to power.

I would say that about 400 volts for small lamps like mine would be nice. Much current is not needed at that voltage. So fine wire can be used.

As I said Ive never made a ferrite transformer before my first try will be as soon as is stops raining and everything dries up around here.

[JW]

Just out of curiosity,

 Could I use a Briggs and Stratton ignition coil for the transformer? I believe I can separate the common ground to the primary and secondary, is that even necessary. Although ive never seen a grounded floresent bulb before. So offhand I think it is. seen grounded ballasts but not bulbs.

 Id like to see a fullwave schematic(with two mosfets). Im guessing from the comments, this cfl driver circuit is halfwave. I have a scope, this would be a perfect project for me. My goal would be for full brightness in a 48in tube. Using a 12vdc source. Not that worried about 20watts draw. Neat stuff.

JW

[s4w2099]

be very careful. If you connect that ignition coil to your scope you will most likely destroy it. I will be working on a fullwave version as soon as I get a decent ferrite core transformer made or bought.

[commanda]

The other option is using SG3525

Discussion here:

http://www.fieldlines.com/story/2006/9/7/21944/73176

direct link to circuit.

http://www.otherpower.com/images/scimages/2006/fluoro2.pdf

My breadboarded prototype worked beautifully. However the pcb version has problems. I suspect I've messed up some capacitor values in the feedback and current limit circuitry. Haven't had time to get back to it.

Amanda

[JW]

Thanks s4w2099,

 for the heads up on that, ive got a dual channel to 20,000volts per channel.I will be sure to look into a probe for that.

 I think these ignition coils do 18,000v on the secondary with 4 ohm coil on primary. They use a magneto (rotating magnet and fixed coil) to fire. Im only looking to make 600vac on the secondary. I guess for starting- flashing, the bulb for like a second with 1000v would make things easier. That could be staged easy enough. But maybe switching the 600v circuit on and off a couple of times would do the trick also.

Jeremy

[JW]

Whoa,

 I just pulled out the book on my scope and it can only handle 1250v per channel and thats  at 20mhz per channel.

JW

[s4w2099]

Hahah thanx, this IC is the best. That is the exact same oscullator circuit used in the inverters but the pulse width is controlled by the output voltage. One thing that I would change about it is the 7812 and instead of that make a transistor series regulator with a TIP31C or somethign that would stand higer voltages, a zener diode and a resistor. I think that way would be stronger agains the spikes. The 7812 would give up eventually depending on how the transformer was made.

Then a pair of voltage suppressor diodes to the gate of the FETS will make it 100% secure. But the basic circuit is great and will function regardless. I will have my schematics availabe so you guys can see what I am talking about.

Thanx for the schamatics.

:-D

[s4w2099]

I knew there was something smelly there.

:-D

[s4w2099]

Here it goes. I made a ferrite core transformer and tried with the SG3525. I failed miserably. Both ways with the transient voltage suppressors and without them. The MOSFETS were giving out a lot of heat and a pair fried along with the PWM chip.

The transformer vibrates like crazy it keeps making this buzzing sound at about 100Khz. I did not used a feedback coil. I think that just by elevating the voltage high enough I would get light from the CFL.

Any clues? I just dont know what info to give you because I am totally lost about this one since I am doing it without the oscope.

:-(

[commanda]

Close to impossible without the sillyscope.

I've got 2 x 2 channel analog's (one of which is now single channel cause a switch broke), a pc based bitscope, and a portable lcd thingy by Fluke, and still can't see what's going on sometimes.

Suggest your first step should be to fix the cro.

Amanda