Tuesday, May 28, 2013

How to use the GTL3 Bulb: A Simple and Inexpensive UVC Source

WARNING: UVC, or short wavelength ultraviolet light, is dangerous. It causes sunburn, eye damage including cataracts, and skin cancer. Even relatively low power UVC sources, like the GTL3 bulb, can cause injury at close range. DO NOT directly expose your skin or eyes to an operating GTL3 bulb. Window glass and most plastics block UVC fairly effectively, but they may not block longer wavelength UV that the GTL3 also produces.


Recently I needed a low power UVC source, and a quick web search turned up the GTL3 bulb. The GTL3 looks like a standard incandescent bulb but contains a small drop of mercury to create a mercury arc. In the picture above the mercury is the dot at the right end of the bulb. GTL3s are cheap if you look around a bit; I bought three for about $20 including shipping. They were made by Ushio, a Japanese bulb company.

The Ushio spec sheet shows that they operate at 10.5V and 300mA, and produce 160mW of 254nm UVC. Their efficiency is about 5%, which is better than most UVC CCFL bulbs, but lower than UVC fluorescent bulbs. They have an E17 screw base, which is a somewhat unusual size between regular (E26/E27) and candelabra (E12) bulbs. E17 sockets are hard to find, but E17 to E26 adapters are readably available.

UVC bulbs are also called germicidal bulbs because UVC light breaks down DNA, which is why it kills bacteria, viruses, and just about everything else. See warning above. Most UVC bulbs are modified fluorescent bulbs with no phosphor and a quartz envelope to pass UVC. Like other fluorescent bulbs they operate at high voltage and require a transformer or inductive ballast.

The GTL3 seemed like a simple alternative to fluorescent UVC bulbs. The spec sheet implies that it operates like an incandescent, just connect a 10V supply and go. Strangely, when I did a search for GTL3 wiring and circuits I found very little. So I applied 10V to one of my brand new GTL3s and… nothing happened. It drew about 100mA, well below its rated current, and there was no visible filament glow or purple mercury arc. What was going on?

One clue is that, according to the spec sheet, the GTL3 should draw 300mA at 10V when operating normally, meaning with the mercury arc running. My initial tests did not ignite the arc, so I was measuring the filament current alone.

This implies that the mercury arc, not the filament, conducts most of the current in the GTL3. Given this, you would expect the GTL3 to behave electrically more like a low-voltage arc tube than an incandescent bulb, and therefore that it might need a higher voltage to start.

So I added a 10 ohm resistor (for current limiting) in series with my GTL3, and slowly turned up the voltage. At 15V the filament began to glow slightly, and at 16V the arc ignited and the current jumped to 500mA. Lots of pretty blue mercury glow that will fry your eyes in a hurry. I quickly turned the supply down to 14V, the current dropped to 300mA, and the GTL3 was running just like it should.

The picture to the right shows a GTL3 with the arc turned down as low as possible. I was using a DC supply, so the arc forms around just one of the electrodes

Before the arc starts the GTL3 looks electrically like a regular incandescent bulb with a 100 ohm cold resistance. Once the arc ignites the GTL3 has roughly a ten volt drop with a dynamic resistance of about an ohm. In other words, it behaves like back-to-back 10V zener diodes in series with a 1 ohm resistor.

Unfortunately, this means that the GTL3 requires a ballast to run properly. However, because it ignites and operates at such low voltages, it’s still much easier to use than a fluorescent UVC bulb. A 33 ohm 10W series resistor works well as a simple ballast with a 24V AC or DC supply. The GTL3 will operate from supplies as low as 17 or 18V, although under these conditions it’s difficult to properly regulate the operating current with a resistive ballast, so I recommend using an active current limiting circuit if you want to go below 24V. Running on DC, rather than AC, will probably shorten the GTL3’s life, although I don’t know how severe this effect will be.

Resistive ballasts become impractical at higher supply voltages because they dissipate so much power. Capacitive ballasts are a better choice if you’re running from AC. A 6.8uF 400V film capacitor works well for 120VAC 60Hz, and a 3.3UF 600V capacitor should be good for 220VAC 50Hz. Be sure to include a bleeder resistor in parallel with the capacitor to discharge it when the power is turned off. Capacitors this size can be dangerous! A 100K 1W resistor will work for both 120VAC and 220VAC. Below is a picture of a GTL3 sterilizing my workbench. The capacitor in front is the ballast, and you can see the discharge resistor connected across the capacitor leads. There is so little filament glow the the light from the arc completely washes it out.


What can you do with a GTL3? I needed UVC to test a small titanium dioxide photocatalytic oxidation (PCO) reactor designed to remove ethylene from a plant growth chamber. PCO can be used to break down almost any volatile organic compounds, so it eliminates most smells as well. Lots of info on the web. Of course, the classic use for UVC is as a microbial sterilizer. It kills everything! It would be fun to see what kind of effective range a GTL3 bulb has. Easy enough to expose some agar plates at different distances and see how far away you have to get before anything grows. Could be a simple science fair experiment…

70 comments:

  1. Hi! Thanks for this helpful article. We have this same bulb and are trying to make it light but haven't been able to. We called their company but they weren't very helpful.

    How did you set this up to make the bulb light up? Is there a way to make it light without a ballast?

    Thanks!

    ReplyDelete
  2. Very useful info!
    According to Wikipedis, these low voltage lamps work this way:

    They look like an incandescent lamp but with the envelope containing a few droplets of mercury. In this design, the incandescent filament heats the mercury, producing a vapor which eventually allows an arc to be struck, short circuiting the incandescent filament.

    ReplyDelete
  3. Thanks Russell. I have successfully used your AC and DC simple ballasts to drive this lamp (I only blew up two of them!) Cheers.

    ReplyDelete
  4. Hi

    Could you add an schematic of the circuit? I think that the capacitor, the resistor and the bulb are in parallel (the three), or are in parallel just the capacitor and the resistor and then in series with the bulb?

    ty

    ReplyDelete
  5. Thanks Russell, I’ve used your encouragement to light up an older but similar G4S11 ozone lamp on AC. It’s 4 watts instead of 3 so I used an 8.5 microfarad capacitor in series. This ran the lamp right on its specs on the very first try - 380 ma and 10.3 volts. Nice snap when discharging the capacitor with a screwdriver. . .

    ReplyDelete
  6. Hi Russell,
    Thank you so much for that! Is it possible to use in an E17 base light fixture? ( a cord with plug and just a socket type) I would like to make a shoe sanitizer...
    Thank you for sharing your knowlege!!
    Miss Hurdle

    ReplyDelete
  7. Did you ever post a circuit diagram? I too am powering a PCO with this bulb, but haven't figured out how to power the bulb. For an AC power source, would that capacitor ballast be in series with the bulb? And is there any resistor in series with it? Or just the discharging one in parallel?

    ReplyDelete
  8. Well, in spite of absence of a wiring diagram, and trying to work around the many vague aspects in the text description, I gave this a go. Two burned out $20 bulbs later I give up. Guess I'll have to continue searching for someone willing and able to sell me a suitable ballast and for information on the appropriate power supply. A 24V AC adapter with a 33 Ohm 10W resister and the recommended capacitor in parallel, each bulb flared for a few seconds very brightly in blue and orange before filaments broke and they stopped working. I'd shown the article to several electronics nerds (the robot designing, generally tinkering, guys running the best electronics shop in town variety) and they were each left scratching their heads at the often conflicting suggestions in the description. 24V or 120V? AC or DC? So much room for error and no single circuit seemingly being described. A shame, as I really, really need a compact UV source, soon, to polymerize some spirit of gum turpentine so I can cook varnish this summer and out that on a new doublebass...

    ReplyDelete
  9. The ballast is in series with the bulb. The voltage is dropped across the ballast, and a capacitor is highly efficient compared to a resistor so it doesn't get hot. You should put a high value (100K or so) resistor across the capacitor to discharge it in a few seconds when the power is off, otherwise the capacitor could deliver a lethal shock.

    The higher the voltage, the lower the capacitance to give the required ballasting. A ballast feeds a nonlinear device with a constant current and thus prevents runaway. However, the filament of this unique tube shunts current from the arc so that it doesn't form at lower ballast current. It is complicated by the fact that cold filaments are much lower resistance than hot ones.

    So, the tube is designed to heat up nicely at 24 volts, then the arc forms and runs at ten volts. So you need a little ballasting reactance, and run at 24 volts. The reactance can be a capacitor, resistor, or even a special transformer with a high leakage inductance....which I suspect is what this tube was designed for.

    Use AC so that both electrodes get lit and you can use a capacitor ballast. Although maybe a current limited voltage supply like is used for LEDs might be a good DC replacement. Whatever the current rating of the bulb was when the arc runs... 100 ma?? And 24 volts to light the filament.

    ReplyDelete
  10. Cool. It makes ozone too, if the type with the quartz glass.

    ReplyDelete
  11. Thanks a lot for your info on this device. I was wondering how it actually worked and before I got one in my hand had thought it may have a bimetal starter built-in.
    I have looked at many (and many more) circuits to drive various type of fluorescent devices but usually they were aimed at much higher voltage and/or lower current.
    The simplest seems to be a capacitor (+ parallel bleed resistor) in series from my 230V mains. In my case this would be 4uF 450Vac + 100k 3W (+ fuse and switch).
    A more complex approach (not yet tried) would be a cheap boost/buck (SEPIC) module feeding a Royer oscillator. The detail design of this is a little vague due to the voltage/current requirements. I was hoping to use FETs and a simpler transformer with centre tapped primary and single secondary.
    The ultimate aim is to make a cheap eraser for my small pile of EPROMs.

    Phil

    ReplyDelete

  12. i really appreciate this stuff.it is very amazing content.thanks for sharing such a useful stuff.it really helped me very much to solve many proble.

    Ballast for UV Lamp

    ReplyDelete
  13. This comment has been removed by the author.

    ReplyDelete
  14. 24VAC will light it up, but when the arc strikes it will look like a low impedance (runs about 10 volts). So you'll probably still need some ballasting. Got another transformer or something you could run in series with either the 220V or the 24V?

    ReplyDelete
  15. This is an awesome experiment, well documented as a case study :). Thanks a lot for sharing, Russel!

    I was searching for days on manufacturing websites with no luck. Apparently no one cares to provide any details, not even technical specifications, not to mention schematics :)
    This is the first and only place in the net I found valuable information on this topic.

    In my case I bought couple of LB1000 replacement bulbs for GermGuardian GG1000/GGH200 models and there is no spec info whatsoever on the 4 pages label print with the instructions. No voltage, no watts, not even socket type, which ended up being E17. Nothing on the bulb either, except a stamp "ON09", whatever does it mean.
    After digging this product in Amazon and Alibaba and lots of controversial information finally I got an idea that the working voltage is 10.5V and working power 3.5W (commonly presented as 12V, 3W bulb) Now I know what I'm dealing and calculated 6.5uF for 110V at 60Hz, which was confirmed by your "study" :) So, I am going with 110V and 6.8uF capacitor with at least 100kOhm shunt for discharge as you suggested.
    Again, thanks a lot for your great article which is priceless even from 7 years ago :))

    Cheers,

    RJ

    ReplyDelete
  16. I have a LB1000 bulb as well. How do I calculate the capacitor and shunt for the 10.5 Volt 3.5 watt bulb for 120 Volt AC?

    ReplyDelete
  17. Also What would I need to run it on a 18 volt DC power supply? Or a 24 volt DC, or a 20 volt.

    ReplyDelete
  18. ==== WARNING - THIS IS DANGEROUS =====
    If you are inexperienced or not sure what you are doing - get advice locally or someone else qualified to assemble it. There are dangers associated with this project which are not immediately obvious.
    If you are asking this question you fall into this category!!!

    -----------------------------------------------------------------------------------
    Calculate the bulb current:-
    W = V * I
    3.5 = 10.5 * I
    I = 3.5 / 10.5 = 1/3(amps)
    Bulb current = 0.33A
    =====================

    Calculate the capacitor:-
    Z = 1/( 2 * Pi * f * C) (impedance of capacitor)

    Assuming 120V 60Hz supply
    V = (120 - 10.5) = 109.5
    Z = 1/(2 * Pi * 60 * C)

    using V = I * Z
    109.5 = 0.33 /(2 * Pi * 60 * C)

    (2 * Pi * 60 * C) = 0.33 / 109.5

    C = 0.33 / (109.5 * 2 * Pi * 60)

    C = 7.99e-6 = 7.99uF (~8uF)
    ===========================
    This is not exact as the bulb impedance is not constant but should work while the bulb voltage is much less than the supply.
    Capacitors should be AC voltage rated for the supply voltage used (or better).
    DC voltage rating is not a good guide.
    (NB electrolytic capacitors are NOT suitable for this application)

    The resistance across the capacitor is for safety to prevent charge being left on it when switched off. 100k 1W (and rated for at least 1.5 x your supply voltage ) should discharge the capacitor within a few seconds.

    Include a fuse and isolating switch in your circuit and put it in a box to prevent accidental contact.
    Make sure there is no way for the bulb light to escape while operating. (Even by reflection)
    Note that the light from this bulb will degrade many materials quite quickly. (Organics, plastics and YOU!!!)

    For low voltage DC supplies it is much more difficult to provide a general answer due to the variation of the bulb voltage from heating to sriking the arc. A constant current converter of some sort but I have not seen one for this applicaion yet.

    Use this a your own risk - Note the warnings above.

    Phil

    ReplyDelete
  19. I assume no one has tried using something like an MC34063? This would be my "go to" option at such low voltage.

    ReplyDelete
  20. Can I just use a 10.5 volt power supply from my old sony? Do I really need 24v to get it going? I'd really like to avoid going the ballast route. I'm not an electrical engineer.

    Thank you.

    ReplyDelete
  21. If you had a DC constant voltage power supply you could design a current regulator that would work. You need the current regulator or ballast because the tube has a negative resistance region in its voltage vs. current curve. LEDs have the same problem; you can blow them up feeding them with a constant voltage, so they also need a ballast (usually a resistor).

    A constant 10.5 volts probably isn't enough to get the filament hot enough, but if it did, the mercury arc would suck a lot of current. You still need a current regulator. A ballast does it; constant current IC supply works too. They are used with LED lamps on the AC line.

    ReplyDelete
  22. Thank you for the instruction about uvc e17 bulb. My bulb (10vAC 3w) worked with 4.7uf 400v capacitor. I=2Ï€xCxUxf. In my case the voltage and frequency are 220V 50hz. So C=I/(2Ï€xUxf)=0.3/(2Ï€x220x50)=4.34uf. So I chose capacitor 4.7uf 400V

    ReplyDelete
  23. Hi Jim,
    Thank you for the information.
    Will something like this work? https://www.trcelectronics.com/View/Mean-Well/APC-8-350.shtml

    Thank you.

    ReplyDelete
  24. reply to gtl3_hopeful: The led driver that mean output dc voltage. But with gtl3 uvc bulb you should use ac voltage for full operation electrode. The best way to use this bulb is 1 capacitor at least 4.7uf 400v (220v 50hz)

    ReplyDelete
  25. An LED driver might work. But a capacitor sure is simpler. That bulb was designed to work that way.

    If you try the constant current DC driver, it will light up on one electrode which might be OK for a while. But it might build up heat right away, I can't be sure. Neon bulbs run off DC have a similar shortening of life.

    I had a radar transmitter that had a traveling wave tube that was run off a DC power supply. The manufacturer started getting complaints from the field that the heaters shorted. They changed to an AC supply and no problems. Just the way the magnetic field of the heater warped it inside, or something.

    ReplyDelete
  26. Just be careful because a charged capacitor with no bleeder can store a lethal charge. If you put a voltmeter on it, it will drain faster than with nothing across it. always be aware of what has voltage on it, and where the current will go if you touch it. You don't want it to go in one arm and out the other, because your heart is in the middle.

    ReplyDelete
  27. Hello Russell,
    i saw your videos on UV lights on youtube and would much appreciate your advise.
    with so many products available, in regards to killing bacteria and other nasties around a 3 bedroom house, do you recommend any particular lamp UV light to buy?
    thanks very much
    Pat

    ReplyDelete
  28. I am surprised there is still interest in this funky bulb. I designed an inverter about 10 years ago to run the GTL3 from 12Vdc. I "Gitified" and here is the link. I will polish it up overy the coming days.

    https://github.com/Geato/GTL3-Lamp-Driver.git

    ReplyDelete
  29. For people who purchased the "LB1000".

    The link below looks more like the "LB1000" than the GTL3 does. Russell's article still applies; use a capacitive dropper to limit the current.

    http://www.cnlight-lighting.com/6-1e-uv-lamp.html

    "Self-Ballasted Mini UV Lamp (10V/3W)
    Features
    Made of quartz glass, classified to ozone and without ozone.
    Connect into main wire voltage 220V/50Hz or 120V/60Hz with incandescent lamp or special made capacitor, no need of ballast.
    Switch on-off up to 20,000times."

    ReplyDelete
  30. For all, who found this blog in search of a universal low cost driver for this lamp.

    I have found a cheap solution to run it from 6V to 32V DC input.

    The trick is a little DC-DC step-up down, buck-boost, converting PCB, with the ability to limit the current. Like this:
    https://www.ebay.de/itm/Down-Boost-Buck-Module-Lm2577s-Lm2596s-Power-Dc-Dc-Step-Up-Voltage-Converter/272315965237
    To make it work first power the PCB and adjust the output voltage to 16-18V.
    Then place and Amperemeter at the outputs and adjust the current to 333mA or lesser than that.
    Then solder/connect your lamp to the output and voila we have a warming up and very little later igniting UVC-Lamp.

    Have fund and best regards: A.Keibel@systragon.de

    ReplyDelete
  31. @Captayne, that Ebay part is a DC output and unfortunately the bulb will burn out quite quickly. It is designed to run from AC. DC causes metal migration. It is a nice little power supply though ...

    ReplyDelete
  32. Anyone who knows which type has ozone or without ozone. I bought a gtl3 bulb and don't know

    ReplyDelete
  33. If I recall, the GTL3 did produce ozone. Any light source operating @ 254nm will produce it

    ReplyDelete
  34. @Geato: Well electromigration might happen, but I think the current is tooo low for this in this case. In former times tubes were used instead of transistors with much higher currents, and voltage and also on DC. My circuit works well since days. Worth a try, and the wide voltage input allows it to run und many sources like batteries, lonely old power supplies ...
    If one is afraid of DC operation, he can change polarity himself from day to day. ;-)
    If my burns in the next time, I will follow up on it here.

    ReplyDelete
  35. But on the website I bought they said there are 2 type of gtl3 bulb. That are ozone and without ozone. I dont know how to distiguish

    ReplyDelete
  36. @Captayne, Looks like it is not metal migragration at all. Here is a nice link: http://donklipstein.com/uvbulb.html#oz

    And, it looks like the ozone producing line is at 185nm which could be filterd out using specific window materials so it is definately possible to have ozone producing and non-Ozone producing bulbs.

    ReplyDelete
  37. Thank you very much Russel! I thought it was 10vdc and I couldn't get it to light. Later with 12Vac and it did not work either. Up to 24Vac with resistance dissipating too much heat, which was impractical. I found your information and now it works wonderfully. Thank you!

    ReplyDelete
  38. Thanks for this article Russell.
    I'm experimenting with a cd4047 complementary output astable, and driving a L293 Dual full H-Bridge chip! This gives a square wave output, pos/neg P-P till about 30V max.
    Tried from 5Hz up to about 1KHz and still good. Just need to limit current and we're good to go! Using about 17 - 19VDC and current limit.

    ReplyDelete
  39. @Geato

    Thank you for your contributions to GitHub. Would you be interested in selling any pre-built drivers from your schematics?

    ReplyDelete
  40. I guess it would depend on how many. The transformers are hand wound as the number of turns are too few to wind by machine ... at least on my winder. Actually, you gave be a fantastic idea. I could redesign the board to use a planar transformer so the windings are actually pcb traces. Like this: https://www.cettechnology.com/what-is-a-planar-transformer/

    ReplyDelete
  41. @Geato

    I currently require 4. What number makes sense regarding your time commitment and the price? I would greatly appreciate this.

    ReplyDelete
  42. I checked stock and I have inventory on most of the parts. We should probably take this discussion offline. How can I get ahold of you?

    ReplyDelete
    Replies
    1. Would like to get in touch. Pls contact me dadarara@gmail.com

      Delete
  43. If you power this GTL3 bulb properly, it works great, but it does require a bit of work to set up the circuit. I came across Russell Wallace's website above, which recommends limiting current flow with an in-series capacitor for AC supplies. Ensuring current doesn’t exceed 300mA, I built a UV-C box for sterilizing masks. Much appreciated, Russell and included comments – thank-you for your posts.

    I used a 6.5uF motor capacitor in series with the bulb, and that gave me 300mA current flow, right on spec. (6.5uF isn’t a standard value; I used a 3.0uF and 3.5uF in parallel, both rated at 470VAC, along with a 100K parallel resistor to discharge the caps for safety). The USHIO website lists specs for the bulb as 10.5 VAC, 3 Watts, and 300mA, but these are round numbers. If you do the math (Watts = VAC*Current); 10.5*0.300 = 3.15W, a little higher than the listed 3 Watts. However when I assembled the circuit, powered with 120VAC, the bulb only dropped ~ 8.4 VAC when carrying 300mA, for an effective power of 2.6 Watts. Pushing the bulb to 10.5 VAC would exceed this current limitation, and likely shorten the life of the bulb. The GTL3 bulbs require intermediate size E17 sockets, available online.

    Using a capacitor to limit current flow however allows you to place two bulbs in series and still maintain the same current flow while doubling the UV-C light output. I later added a lock-out switch, to protect from direct UV-C exposure, and a timer, to limit exposure times, but the basic components worked fine for me. I calculated that a 10-minute exposure time would be more than enough to fry any bacteria or viruses in the chamber.

    As usual, proceed at your own risk…

    ReplyDelete
    Replies
    1. Interesting. Is capacitor and resistor in parallel? Is powered 120VAC? How about 220 or 230VAC?

      Delete
  44. Using my calculation method from March 20, 2020 at 6:26 AM (Unknown) I built an EPROM eraser box.
    Making everything safe was a bit of an exercise but happily with a timer set to 20mins my first EPROM was totally erased. It probably helped that the EPROM window is under 1cm from the bulb.
    The series capacitor was 4.7uF 400Vac on my 230V 50Hz supply.
    I'm not sure of the UV intensity/exposure but it seems to be enough.
    The various adverts for this type of bulb seem to vary from 3W to 3.5W so assuming 3W may be beneficial for long life (C= 4.0uF) but this was not my primary concern.
    Many thanks to Russell for starting this blog.
    Phil

    ReplyDelete
  45. I found a fairly efficient and stable method of lighting the GTL3. I use a buck boost variable output with current limiting circuit such as "USB buck boost" on Amazon. It should be one that has a 10 watt output. First use a 47 ohm 5 watt resistor as a test. Set the voltage at 17.5 and then lower the current to 0.3 amps. Then GTL3 needs a ballast resistor and 10 ohms is about the lowest value you can use safely. When the arc forms in the bulb it briefly draws more than 0.3 amps and lower ohm value can damage the bulb. A capacitor will provide the short burst of power that the arc needs to form. Once you have the buck boost set at 17.5 volts and 0.3 amps, put a 1000 microfarad 20 volt capacitor in parallel with the bulb. The bulb and capacitor together then go in series with a 12 ohm 2 watt resistor. After the filiment heats the bulb should light up. If not you can raise the voltage to 18 and current to 0.35 but then lower the current to 0.3 when the bulb lights. Once the bulb lights the voltage will automatically drop to about 13.5. The bulb runs on 3.2 watts and this circuit will draw about 4.2 watts. -- Bruce

    ReplyDelete
  46. The person who has written this seems very well skilled in the art of writing also the content seems to be very well researched, if information is needed regarding UF membrane workings then visit….

    ReplyDelete
  47. @Geato any interest in building one more driver? Thank you.

    ReplyDelete
  48. @AndreJ can you give me your contact info if still interested

    ReplyDelete

  49. Recently launched an amazing UV Lamp which is known as UV Lamp,i have been quite fond of germ free freak and then i got to know this UV Lamp

    ReplyDelete
  50. The arrival of theUltraviolet lamp supplier prove really beneficial for the safety during this pandemic.

    ReplyDelete
  51. "Proving" something is effective is a lot harder than you think. Not all purple lights are going to be effective against germs.

    I have an EPROM eraser which I can use to disinfect small items. It's about 20 cm long and has a UV fluorescent tube in it. But how can I see if it's effective?

    ReplyDelete
  52. You provide a great source of UV Light Norcross in your post. Thanks for sharing an inexpensive source and all the information about it.

    ReplyDelete
  53. See my post: Unknown March 20, 2020 at 6:26 AM (and July 17, 2020 at 4:47 AM)
    ==== WARNING - THIS IS DANGEROUS =====
    This goes through the calculation method
    Yes the resistor is in parallel with the capacitor.
    Approx 8uF for 120V 60Hz or 4.7uF for 230V 50Hz.
    Note that the voltage rating of the capacitor should be an AC value (not DC) of at least your supply voltage or preferably greater. In my case 4.7uF 400Vac for 230V.
    Look for motor start/run types if unsure. Also the resistor should have a similar voltage rating - 100k 1W or 2W axial types are likely to meet this.
    Bear in mind that this circuit is dangerous and should be enclosed to prevent being touched.

    ReplyDelete
  54. Is it a good idea of having qty of 10 GTL3 bulbs powering an aquarium water sterilizer ? In principle I need one UVC 25w T5 lamp. So will 10 GTL3 bulbs provide the same effect? Not sure how to efficiently power these lamps Don’t want to use 220vac. 24vac or even DC would be preferable. But read above that maybe the lifespan will be short. How short?

    ReplyDelete
  55. I have no knowledge of aquarium sterilization nor any comparison between GTL3 and T5.

    The complication (and cost) of using 10 GTL3 devices rather than one T5 would probably outweigh any advantages of not using 220Vac mains. It may be better to investigate the DC inverter route to drive the T5 bulb if you really want to use a low voltage supply (there are many commercial and constructor designs available but you will need a fairly hefty supply/battery - as you would for 10xGTL3).

    Lifespan of the GTL3 is probably very dependent on the drive conditions (for each bulb). Under-running will usually extend life but will reduce output and efficacy.
    Difficult/impossible to put figures on that. You will have to make your own assessment. (Under-running also makes the design more complicated as the filament has to be hot before the arc can strike.)

    Note that you cannot just put GTL3s in parallel as they have a negative resistance characteristic. Once one strikes it will lower the voltage available to the others. This will 'hog' the current until it blows after which the next one will do the same. Goodbye all! (Unless they fail short-circuit)
    Each one must have its own regulaton circuit.

    Series connection of 2 or 3 lamps is possible but then the voltage requirements will increase. 10 lamps in series would require 200-250v to strike but probably ~100V to run - current limited of course. Good luck with that design.

    ReplyDelete
  56. Excellent! Mine works great! Used your simple circuit with 6.8uF. Great job, Thanks!

    ReplyDelete

  57. Tech On Electronics is the best Led Bulb Driver Manufacturer in India. These products are suitable to bear the heavy voltage and known for their shock proof nature. Our products are available in various sizes and specifications. We also Provide Led Tube Light Driver, SMPS Adapter 12 volt,Mobile Charger these products at market best price.

    ReplyDelete
  58. @Geato I had a few questions about your driver and would love to order a few units. Are you still making them?

    ReplyDelete
  59. @John Davis I don't really want to make any because winding the transformer is a pain. I open-sourced the board and you can find it here: https://github.com/Geato/GTL3-Lamp-Driver ... for people that want to build their own. You can also message me there through the "issues" tab.

    I did start down the path of designing a "planar" version of the PCB meaning the windings are formed by the copper traces embedded inside the PCB. Net result is no copper windings! I cannot attach a PDF to show you.

    Unfortunately, like a lot of things in life, it got put on the back burner.

    ReplyDelete
  60. WhAt about using a 220/24V transformer first. That would be more safe than running the
    whole thing from 220V, and the ballast capacitor should only be rated ~50VAC.

    ReplyDelete
  61. The point of using a high ac supply through a smallish valued capacitor is that it approximates to a constant current through the device. This is important as the device needs a higher voltage at a certain current to heat the filament to a temperature where the coating starts emitting electrons. Once that point is reached the mercury vapour will conduct and the voltage across the lamp will drop to its (lower) operating level. If the same size capacitor is still in circuit with a low voltage source then the current will be much higher (and possibly burn out the lamp).
    On 24V ac the capacitor will have to be much higher in capacitance although lower in voltage rating.
    There will be two values to calculate (at roughly the same current):
    1. where the discharge has not started and the filament needs 17-20V or so to heat up
    2. when the discharge has started and it needs about 10V to run.

    If you can design a detector/switch to change between the two values as required then that may work although bear in mind that a discharge is very non-linear in its voltage/current characteristic. The disadvantage is increased complexity.

    The high voltage/single capacitor is certainly not the only way to drive these bulbs but is simple. Dangers of high voltage circuitry must of course be considered.

    DC fed oscillator circuits can have the required characteristics (current limited ac output) but also require more complexity.
    Another possibility is to use a lower voltage transformer but current limit it on the high voltage side. I have not tried this so don't know if it wouls work in practice. Again cost and complexity increase.

    ReplyDelete