Introduction: Extend Defective Fluorescent Tube-Light Life

Picture of Extend Defective Fluorescent Tube-Light Life

I present here a non-standard scheme to extend the life of defective Fluorescent Tube-Lights.

As I will explain in the next steps this method is based on study of the the basic fluorescent tube and its operation using the wire-wound-ballast and electronic-ballast.

The tube I modified six months back is still working indicating that the normal life of the fluorescent tube can be extended by 20 to 30%.

This scheme came to mind based on experiments conducted on the electronic-ballast designed by me:

I found that while fluorescent tubes were originally designed for wire-wound-ballasts which needed active filaments at both ends for their operation, electronic-ballasts are capable of operating with only one filament. In this mode the electronic-ballast is operated outside specified electrical limits but appears to have sufficient design margin for continuous operation.

Warning & Disclaimer:

All content provided here is for informational purposes only. I make no representations as to the accuracy or completeness of any information. I will not be liable for any errors or omissions in this information. I will not be liable for any losses, injuries, or damages from the use of this information.

Step 1: The Modification

Picture of The Modification


1. The burnt-out/fused filament and blackened tube end

2. The filament resistance shows open (OL)

3. The normal filament shows 20-30 Ohms resistance

4. The copper wire to be used for shorting the terminals at the blackened end. One strand from multi-strand household wiring cable.

5. The terminals shorted out

6. The resistance is now Zero

Step 2: Theory 1 : Fluorescent Tubes With Wire-Wound-Ballasts

Picture of Theory 1 : Fluorescent Tubes With Wire-Wound-Ballasts

Fluorescent tube-lights were designed to operate with wire-wound-ballast chokes. A more efficient electronic-ballast is now commonly used with these tubes.

Looking at the figure the following is the sequence of operations leading to the start up of the tube:

1. When the AC-Mains is switched ON, the circuit is completed through the ballast, the starter and the two filaments. The starter which is a mini-glow tube starts burning and heats up a bi-metallic contact within the starter. This causes the contact to close and a starting current now flows through the ballast and filaments.

2. This starting current heats up the filaments and electrons are emitted from them. With the starter contact closed the glow subsides and the bi-metallic strip cools down. After a short delay the starter contact opens.

3. Sudden opening of the contact while a current is passing through the inductive-ballast causes a high voltage to be generated based on the inductance value L of the ballast and the rate of change in current di/dt.

4. This high voltage L x di/dt across the tube along with the electron-emission from the filaments ionizes the gas within the tube and a discharge is setup between both ends.

5. This discharge and the coating on the inner side of the tube creates the fluorescent glow.

6. At this time the inductive ballast limits the current flow through the tube to designed limits and the voltage across the tube is not sufficient to make the starter glow and the starter contact remains open.

With continuous use the filaments deteriorate and usually one filament burns-out/fuses and that end of the tube is blackened. A continuity test on the fused filament would show open-circuit while that on a good filament would show 20 to 30 Ohms.

Step 3: Theory 2 : Fluorescent Tubes With Electronic-Ballasts

Picture of Theory 2 : Fluorescent Tubes With Electronic-Ballasts

The electronic-ballast is designed usually as a half-bridge resonant-converter:

Looking at the figure:

1. The input AC Mains voltage is rectified using a bridge-rectifier

2. This voltage is fed to a half-bridge resonant-converter operating at a typical frequency of 60 kHz.

3. The output of the half-bridge is connected via an inductor (L1), one filament, a capacitor (C1), through the other filament to the center point of C2/C3.

4. When the AC Mains is switched ON a starting circuit immediately energizes the half-bridge converter and a high voltage is developed across C1 which is the same as across the two filaments.

5. The mechanism for Ionization and setting up of the discharge is quite different as it depends on the frequency, the high-voltage and to some extent on the emissions from the filaments.

6. After start up the current is limited by the impedance of the inductor L1 and the resistance of the filaments.

While experimenting with the electronic-ballast I found that the tube-light started up and worked when one filament was by-passed and it worked intermittently when both filaments were by-passed. As I was working with my own design it was possible to monitor the tube current, the input currents and the dissipation of the active components. The conclusion of these experiments was that the electronic-ballast was capable of driving a defective tube-light with one filament by-passed.

I was able to confirm this with a commercial electronic-ballast as presented in the video.

Step 4: Experimental Measurement Data

Picture of Experimental Measurement Data

I made some initial measurements of the lamp current using a digital multi-meter. However, as the frequency of the AC waveforms is ~60 kHz these values are not reliable being outside the frequency-response range of the multi-meter.

After setting up an isolated AC Mains power supply using an isolation transformer I was able to take more accurate measurements using an Oscilloscope with a bandwidth of 15 MHz.

The oscilloscope probe is 10:1 and current measurements are taken across a 10 Ohm resistor.

The following is the summary:

  1. The oscillation frequency is ~ 62 kHz
  2. Under normal conditions the Voltage across the tube is 106 V and the current is 28 mA indicating a power of 29 Watts for this 36 Watt Fluorescent tube.
  3. The voltage across one of the filaments is 2.4 V and the dissipation would be 0.7 Watts at 24 mA.
  4. With one filament shorted the voltage across the tube is 159 V, the frequency is not so steady and the current is 42 mA indicating a power of 66 Watts.
  5. The current increase is 42/24 or 1.5 times
  6. The power increase is 66/29 or 2.2 times


kavish laxkar (author)2017-06-08

will it not short circuit phase and neutral?

ajoyraman (author)kavish laxkar2017-06-08

As we are using an electronic ballast the Phase/Neutral do not come directly to the terminals of the fluorescent tube. In the circuit for the electronic ballast (my earlier instructable) you can see that the phase to neutral connection is rectified and fed to a half-bridge. The half-bridge output after a resonant inductor forms one output of the ballast. The other is the mid point of two capacitors connected in series between the rectifier positive and negative. So, we will not short the phase and neutral.

frisbrob (author)2017-04-10

I posted a day ago on this saying, Oh bother, just go buy a new bulb. I realize that it was probably not the right thing to say but I will explain why I did. I am one of the old guys that tinker. I have tinkered since I was a little kid and always will. I have over the years found that some things are just not worth all the trouble, this is one of them and it is nothing new. I did this years ago on two of my old fish tank lights and this might be the reason you need to do this just like I did. I didn't have the money to buy new ones at the time so I did this to the bulbs and they went on working but only for about a month and then they were finished. One thing about this is that when I did it back then they didn't work when I put the tube back in and turned it on, I found that it had to go in a certain way. Now what I am saying on this subject is if you have the money to buy a new one do so instead of doing this. Bulbs last for years, by the time you go through all the trouble of finding wire, stripping it, uncoiling a strand of the wire, wrapping it around the tube prongs putting it back in and having it not work then having to take it back out and flipping it around, putting it back in and trying it again, you could have went to the store and bought a new one and been finished for another few years. That's all I was saying, if your doing it just because you like to tinker, great go for it, have fun.

ManifoldSky (author)frisbrob2017-05-02

I could have found wire, wrapped the terminals, and plugged a bulb back in un under two minutes. Where is this mythical store you are talking about shopping at?

frisbrob (author)ManifoldSky2017-05-02

Good for you. What do you want, a first place ribbon for the fastest wire finder, wire stripper, terminal wrapper, bulb replacer in the world? I have no idea what you are talking about when you ask what this mythical store I shop at is. You can buy new bulbs at any hardware store, lighting store, pet shop, where ever.

Syx (author)frisbrob2017-05-03

The mythical store you can get to in under two minutes, frisbrob.

JakeS9 (author)Syx2017-05-03

corner store

ajoyraman (author)frisbrob2017-04-10

Appreciate your views, I also wanted to explain the theory behind this hack.

ghostgeek (author)ajoyraman2017-05-03

And MOST of us appreciate your Insight on this. I have seen more than my share of fluorescent tubes fail during my lifetime. Many times I see the black end of the tube like this. I had no idea that in some instances the tubes life can be extended with such a simple modification. Thank you for sharing and I will file this one away and give it a try the next time I see a tube fail in this manner. :)

graydog111 (author)frisbrob2017-04-10

Instructables is not about saving money. It's about doing fun stuff that may or may not work or save any money.

ghostgeek (author)graydog1112017-05-03

That's not necessarily true. Instructables is full of life hacks like this, which frequently can be about saving money.

andrewty (author)graydog1112017-04-11

Instructables is about learning how to do "stuff".

it's the learning that is important.

ghostgeek (author)frisbrob2017-05-03

"some things are just not worth the trouble." What "trouble" is it to wind a strand of copper wire around the terminals? Would take less than a minute to mod, and I don't even have to heat up my soldering iron to do so. Then you admit you did it once yourself when you didn't have the money to replace the bulb. Certainly it was worth the trouble then, wasn't it? Or imagine if somebody needed the light, but couldn't get to a store to get a replacement bulb. Or maybe someone is just environmentally conscious, and want to keep the nasty toxic materials that are within a fluorescent tube out of the landfill as long as possible? My point is, there are many times when this modification certainly might be worth the "trouble."

rvan willigen (author)2017-05-03

First I was attracted to this idea.
Sometimes we had to change the tubes every several months.
Since they are rather expensive, it sounded as a welcome idea.

However, having Plecos in our fish tank, I am afraid of experimenting.
It is an old fishtank and replacing the lightsystem would also be expensive.
So we have to keep on replacing these expensive tubes more often ...

fga1939 (author)2017-04-10

Greetings Ajoy Raman. Thank you for the presentation, well done. In the schematic under step 2, you show a starter. I assume this schematic is for a light fixture that has an independent starter device as opposed to a fixture that uses self starting tubes. Do you have any comments relating to your approach for the self starting tubes?

ajoyraman (author)fga19392017-04-11

I am familiar with only the two types explained in the Instructable. I suggest you check this site for details of other types of fluorescent tubes:

fga1939 (author)ajoyraman2017-04-13

Thanks for the link to A good source of information.

jlsmoothash2o (author)2017-04-10

so I read your post and I am going to basically boil this down to why this works.

1 when you wired your electronic ballast you wired it incorrectly. Meaning that your ballast is only making connection with one of the two pins. The ballasts should always be wired so both pins are active.

2 The reason this works because of the incorrect wiring. And will not work if your lamp is wired correctly. Basically all your doing is making sure both pins are being used. Because of the incorrect wiring that was done. So what is happing is you have burned out one pin. Now you are fixing that to use the second pin.

3 When wired correctly the gas inside the bulb will break down do to the charge that the ballast produces. Some times the mercy inside the bulbs cocks off and causes the bulb to fail. Which will happen more often if the ballast is not wired correctly.

ajoyraman (author)jlsmoothash2o2017-04-11

Your connection scheme is probably for a different type of ballast which does not need the filaments. The one I have used and tested is based on the wiring scheme from the manufacturer. The experimental results also show a large increase in tube current when one filament is shorted. The ballast of this type would probably be damaged if both filaments are shorted.

jlsmoothash2o (author)ajoyraman2017-04-12

yes on an electronic ballast they have changed their schematic. Because they are only using one wire. That still does not change the fact that both pins need to be wired.

In a T12 bulb which is and older fatter bulb that ballast still uses multiple wires coming out of the ballast to wire both pins. This is because the bulb itself uses more power and the wires coming out of the ballast cannot handle the load across one wire.

When you change over to a T8 bulb which is a smaller bulb and requiring less energy to light the bulb only one wire is needed to carry they load. So the compainies have save on their production of the ballasts by reducing one wire. So since there is only one wire coming out of the ballast that is how they drew their schematic. They did not draw in all the pins because that would mess up the general public and people would be calling all the time saying they are missing wires. Electricians here in the USA are taught that a jumper should be installed when wiring this type of lighting.

Please know I am in no way discounting your research. Infact your research shows how important it is to wire your lamps correctly the first time. And when done correctly your bulb life will be extended. Probably longer then what you are experiencing by rewiring them correctly after the fact. Because you will not damage the pins inside the bulbs themselves. Which is the main reason why these bulbs burn out in the first place.

JohnC430 (author)jlsmoothash2o2017-04-11

how did you deduce all this?

Caspar (author)2017-04-09

I made something similar in the antarctic in 1972. A motor-start
capacitor across the tube, value chosen to be series resonant with the
existing choke. It could boost the voltage across the tube to 1000V,
enough to start the fluoro at -30 degrees C, without the starter. The situation was quite tough for the tube and it only lasted a few weeks then the capacitor
blew up.

JohnC430 (author)Caspar2017-04-11

cool idea and also funny that the cap blew up after a few weeks. what was the voltage across the cap?

Erelu (author)2017-04-10

One more thing - Intructables is about what us old heads used to call tinkering. Do it because you love it. Every time I hear grey-hairs talking about "this new generation can't do a thing" I tell them about this site. Not just explore, but write about it clearly afterwards.

Ignore the haters. Save the world one Instructable at a time!!!!

JohnC430 (author)Erelu2017-04-10

I am an power electronics engineer and have observed that actually this new generation can do a whole lot of things better, more accurately and more efficiently that we ever did. Fortunately for me I "keep up with the times" so I am not out of the loop.

andrewty (author)JohnC4302017-04-11

I am completely out of theloop and very definitely grey haired.

fredellarby (author)Erelu2017-04-10

It's a well documented fact proven by several neutral scientific sites that 87% of the information on Instructables is assembled by males, 58-67 years old.

robi_ncc (author)2017-04-10

I have down this but every time my Electronic-Ballast is burned.

dlemke (author)2017-04-10

I found this instructable very informative. I am currently working on commercial coolers which you see every day at convenience stores and they use an F70-T8 bulb, which is hard to find. I believe the industry is in the process of converting to LEDs, so the florescent bulbs could be on their way out. They have yet to commit one way or the other on the 70" bulbs and both styles are very expensive. It could be that this being used mainly in a commercial environment, it is being charged a higher rate. The bulbs are ridiculously priced at over 30.00 per tube and over 100.00 per LED replacement.

I am in the process of making my own LED lights using individual power supplies and just by passing the current ballasts. I'm sure I will be breaking some codes and it won't be UL approved, but LEDs work better in the cold than fluorescents and require less power. They also last much longer and have inexpensive power supplies compared to ballasts.

For those who wish to replace the old tubes with LEDs, I suggest you do not use the old ballast, but get LED bulbs that have their own power supplies. No sense powering both.

This instructable has removed some of my ignorance of how these lights operate and for that I am grateful. Thank You.

Erelu (author)2017-04-10


This explains something I came across while working on converting some old T8 fixtures for full spectrum T5. With certain ballasts the "end" socket was open circuit. With others, a shorted socket was required. One could use a shorted socket, then remove the jumper.

Don't count on my memory on that scenario, It's not what it was. Look it up:-)

BTW, very nice explanation of the theory behind this. I will feel much more comfortable moving forward with my project.

kwhenrykerr (author)2017-04-10

Thanks, well done. I will add this to my tool box of ideas.

mikenaly (author)2017-04-05

Im thinking a complete short of the filament would result in enough of a current increase in the circuit that it would shorten the remaining life in the good filament. Would this work with and possibly be more efficient if a resistor with a value between the 20 - 30 ohms. I dont know whst sort of voltages and currents woild be involved, but it would be interesting to find out if s resistor of a proper power rating could be used and if so how the light would compare with your fix.

ajoyraman (author)mikenaly2017-04-05

Excellent observation!

I measured the currents once again which are as follows: Normal operation 128mA, One filament shorted 202mA, Both filaments shorted 286mA. The tube glows in all conditions.

A rough calculation of filament resistance based on the impedance of the 1.3mH resonant inductor (490 Ohms at 60 kHz) gives a value of approximately 200 Ohms. This would be the hot resistance of the filament, 20-30 Ohms would be the cold resistance.


The current increase with one filament shorted is ~1.5 times which should be tolerable by the electronic-ballast and give an estimated 20-30% increased life for the tube.

The current increase with both filaments shorted is ~2.2 times which is not desirable for the electronic-ballast.

Instead of shorting adding a 200 Ohm / 8 Watt resistor would be difficult.

mikenaly (author)ajoyraman2017-04-05

Thanks for the math on that, and youre right, an 8 watt resistor WOULD be rather unwieldy for use. I fidnt know what the numbers were to work with.

ajoyraman (author)mikenaly2017-04-10

Sorry for the earlier math which was based on measurements using a multi-meter which did not have required frequency response.

I have repeated the experiments with an oscilloscope and added the results as a new step.

The resistance is still ~ 8 Ohms but the current is much less 24 mA so the dissipation would be 0.67 W.

In the setup I used a 10 Ohm resistor to measure the current, so it is possible to have an 8.2 Ohm 1 Watt resistor instead of the filament. However, the experiment shows that the current still goes up. This is probably a characteristic of the fluorescent tube with one end shorted.

Sylphhawkins (author)ajoyraman2017-04-09

I understand that. It is operating as per it's original design, that is, as a resonating cavity for relatively high voltage / mid-range frequency AC (perhaps taking the place of a secondary coil in resonance to the input primary (the filament + inductor))
. Since you have a home setup, have you experimented with a higher output AC frequency? (or have you setup a simple inverter circuit where you can vary the output waveform frequency, such as a 555 timer IC inverter circuit, I can run mine anywhere from 50Hz to 1MHz).
Running the AC at a high frequency, say, 200KHz, would increase the resistance of the resonant inductor (given your calc of 490Ohm @ 60KHz), and would probably reduce the overall "filament" current.

I would hedge my bets that the light and the phosphor glow would be more uniform and stable.

ajoyraman (author)Sylphhawkins2017-04-10

Yes, using a higher frequency than the 50/60 Hz mains frequency gives a 10% increase in energy efficiency.

Using a frequency of 60 kHz is a trade off between the switching losses in the power-devices of the half-bridge and the frequency for best energy effeciency.

JohnC430 (author)ajoyraman2017-04-09

Mind you i have not done these experiments so i am just guessing at the values i am mentioning.

I am confused. Why 8 watts? what is the voltage across the heater pins on the good side? at 8 watts you can apply V^2 = 200x8 = 1600; so V=40 Volts across the filament. i would think the actual applied filament voltage from the electronic ballast is no more than 4-6 Volts so a much lower wattage resistor could be used. and also is the resonant inductor really 1.3mH? with 60KHz operation i would think it is very much lower. how much is the old type inductive ballast?

ajoyraman (author)JohnC4302017-04-10

Yes, your observation that the voltage across the filament should be 4-6 Volts got me thinking and I have now carried out a more accurate measurement of the electrical parameters. I added another step with this data.

The resonant frequency of the half bridge is based on the 1.3 mH inductor and a 5 nF capacitor giving a calculated frequency of 62.4 kHz.

MikeM144 (author)ajoyraman2017-04-09

As you are modding the light anyway the 200ohm/8watt resistor could be mounted inside the fitting connected across the wires that go to the socket for that end of the tube

The casing could then be used as a heatsink for it if a heatsink pad was used or a bolt on style resistor, a switch could even be added to remove the resistor from the circuit when that tube dies and a new one is inserted

Alternatively a similar value inductor to the one in the tube itself could be added so that the heat aspect is partially removed also with a switch to remove it from the circuit if a new tube is later installed

That way the tubes themselves wouldn't need any modification as the mods would be inside the fitting itself

troberts27 (author)MikeM1442017-04-09 you really have nothing better to do than over think things do you? Its just a trick to get your light to work until you can get a new bulb. Calm down with all the reengineering Mr. Science.

MikeM144 (author)troberts272017-04-09

Firstly I feel sorry for you if you think it would take time to have such thoughts. For you it might, but for the rest of us we tend to look at things and ideas instantaneously spring to mind

No time other than looking at the "thing" is required, or any longer than it takes to skim read a description

About the exact time you spent in fact....Hypocrite much?

And why would it be a one off thing? If someone was keen on being green and environmentally responsible they WOULD try to do this with each bulb that broke

So it would be worth spending the time to do it in a way that would last and be easy to reverse once a half dead tube dies completely, especially considering the chemicals and energy required to make them as getting an extra 30% per tube would equate to 30% fewer being made for each person who did this

The technology, as we know, is being phased out. This is way too late to the game, but I think you may have stumbled upon a great idea..

Electronic ballasts could be designed in a way to sense a single failed filament, and then internal circuitry could be used to perform a variety of life-extending functions for a tube. All this could be done without user intervention, but most importantly, it would be safe! No way in hell I would leave the room with a fixture modified to run outside of its' limits.

This would only be useful for scenarios where changing bulbs is more tedious than the cost of bulbs though.

When I say a "variety of life-extending functions", I am referencing the theory sylphhawkins posted about changing the frequency to operate beyond EOL conditions. A number of experiments and tests could be/probably have been done to figure out other ways as well, again, the tech is near obsolete anyways. Which begs the question: Not possible? or killed by manufaturers and lobbyists?

ajoyraman (author)2017-04-10

Thank you all for your comments. To give more clarity on the electrical parameters I have included an additional step giving more accurate measurements of voltage, current and power using an isolated setup and an oscilloscope for mesasurements.

ndpani (author)2017-04-09

Great Idea. This instructable will be extremely simple and useful especially in an emergency situation or a remote location where replacement tubes cannot be procured immediately. Even a non technical person can carryout this.Expect more of such ideas.Thanks.

FlyinngDolphin (author)2017-04-09

Like you state, this is for information purposes.

Here in the U.S. I found there are too many types and variables of solid state ballasts in my collection of fluorescent fixtures. This became apparent when I tried replacing tubes with direct replacement LED tubes.

While I have gotten 10 years life out of my Fluorescent tubes, the
consumer level ballasts do not hold up. Replacement ballasts cost mote than the fixtures. And every batch of fixtures I purchased tends to have lower quality ballasts as time goes on.

I was more than glad to place a
large order of 20 LED integrated fixtures where I could just get rid of
the tubes and fixtures and save a bunch on electric costs. Plus about 4 more inches of headroom in my shop.

A real fix is to upgrade to LED lights. That in itself is another separate instructable. People buy the Direct Replacement LED Tubes in a home center, the ballast burns them out, they are returned and put back on the shelf. One actually caused the ballast and plastic fixture to catch fire.

I would suggest purchasing the LED tubes that require the ballast to be removed or bypassed. Otherwise there are too many variables with the different ballasts, at least here in the U.S.

johnb559 (author)FlyinngDolphin2017-04-09

Thanks for mentioning that, I think I might have repeated your mistakes had you not mentioned it.

ChrisK346 (author)2017-04-09

Excellent and very helpful, thank you!

The Freak (author)2017-04-07

Just to be clear, this only works with electronic ballasts right? If so, how can you tell the difference?

About This Instructable




Bio: I am a retired Electronic Systems Engineer now pursuing my hobbies full time. I share what I do especially with the world wide student community.
More by ajoyraman:Android OTG Development InterfaceArduino Uno DDS Audio Signal GeneratorSimple DIY Electronic Enclosure
Add instructable to: