Introduction: Adding 'Oomph' to the Garden Solar Light

Picture of Adding 'Oomph' to the Garden Solar Light

Tired of the feeble light from your Solar-rechargeable Garden lights? You can more than triple the light output with this new circuit. Works even with red or orange lights - as long as it runs off a single AA rechargeable battery!

Step 1: Remove the Old Circuit Board.

Picture of Remove the Old Circuit Board.

Cut the wires as close to the board as possible. We'll need the wires from the solar cell and the battery holder, so cut them as close to the circuit board as possible. If there are wires from a light sensor (colored Green here) we can cut those too - it's not needed anymore.

The board can go and join the others we removed from other projects.

Step 2: The New Turbo Circuit.

Picture of The New Turbo Circuit.

The new circuit uses fewer parts than the original but can deliver over 80mA, almost 4-times the current to a White or Blue LED while using about 250mA, which is double the efficiency of the original design.

We will need to use 1/2W LED, which are readily available on eBay:
here, or here. They each have slightly different products, but I've used both and they're both reliable.

The real star of this circuit is the driver transistor, FJN965, from Fairchild Semi. Capable of handling up to 5-amps of current in a TO-92 case, it will start at .9v and run the light until the battery drops to under 0.3-volts. You can get them from

Resistors, capacitors and inductors can be obtained from your local supply houses (eg Radio Shack, The Source), Sure Electronics, or at online surplus stores, like

The second image shows the layout using a Perfboard. S+ and S- indicate the wires from the Solar cell.

Step 3: Perfboard? or Try Just Winging It Freehand...

Picture of Perfboard? or Try Just Winging It Freehand...

We will be using a 100mA white LED with a 140-degree spread. Start by slipping the leads through a scrap piece of perfboard to keep it upright.

If you want to follow the perfboard assembly, do that now, then solder the leads to the finished board.

The circuit is simple enough that the whole thing can be assembled freehand, which I will show you here.

Start by flipping the case over and bend the shorter (-) lead of the LED flat against the plastic. Then push it INTO the plastic by holding a hot soldering iron against the end of the lead until it melts it. Now it's SOLID!

Step 4: What IS This Sh*t???

Picture of What IS This Sh*t???

Making our many legged creature.

Start by holding the 'C' lead of the 2N3906 to the 'B' lead of the FPN965 and soldering them together.

When you have attached the 2 capacitors and the diode to the transistors, this section is done.

Let me emphasize the need to COOL the parts after applying solder. The rule-of-thumb is to blow on the solder for TWICE as long as you apply heat. So if the joint took 2 seconds, cool it for 4 seconds. Heat is the second highest cause of failed circuits. (Number one is operator/designer stupidity.)

A few more images of the process of building our amorphous creature. Fine needle nose pliers are a must-have for little circuits like this. As is a GOOD soldering iron with a 1/16" or 1/32" tip.

Step 5: Attaching to the LED

Picture of Attaching to the LED

I've trimmed and cleaned up all the little solder joints and I've bent into small hooks the 3 leads which require attachments.

Step 6: A Quick Test, And...

Picture of A Quick Test, And...


The green wires from the Light Sensor is left unconnected.

A 'feature' of this circuit is that once on, it will not be 'distracted' by flashes of light, from passing cars etc.

Here's another approach to make your garden light more effective - by adding Bling to the Oomph!

If you like this, see other LED ideas on my website:

Step 7: Another Option.

Picture of Another Option.

I have a light which sits on a gate surrounded by tall trees, with perhaps a max of 3 hours of sunlight in the afternoon. I added a small 2v, 30mA solar cell which is glued on facing West. It is simply added in series with the existing one. I also put a 2000mAH Sanyo battery in. Now even on overcast and rainy days, I get a good light showing the gate.


qs (author)2014-07-21

I've been told that Fairchild has obsoleted the FJN965, but the 2SD965 is still available and costs under 20 cents a piece. If you have eBay, you can check this:

jason.roy.37 (author)qs2015-01-12

So do I use the 2SD965 in place of the FJN965? This will actually be the first circuit I'll ever have built from scratch, so I'm very new to all the different parts and what they are called/do.

qs (author)jason.roy.372015-01-12

Yes, they are identical. Fairchild had a licence to make them and decided to call them by a different name.

KimS11 (author)2017-10-05

So how much illumionation time do you get from a half-decent days sun????

oldguywithtechexp (author)2017-05-01

Hello QS; still love yer designs. I am having trouble geting yer OOmph circuit to start reliably. Could you please describe the mechanism that begins the whole converter process? I have read all the comments and replies; but no joy.

I am using 2SC3202 specd at 2A: schottky diodes both places; and a 2N2907 with 22uH. Help

qs (author)oldguywithtechexp2017-05-01

Hi OldGuy,

Very briefly, the circuit starts when the PNP begins conducting, which turns on the NPN, and this starts charging the coil. At the same time, current is drawn through the 150pF capacitor, which further increases the drive to the PNP and, subsequently, the NPN. When the capacitor is charged, the drive is cut off, and the magnetic field built up in the coil collapses and cause a voltage to appear across the ends. This voltage is placed in series with the NiCd battery to turn on the LED. Simultaneously, a small portion of this voltage is diverted through the diode (D2) to charge the 1uF cap. This provides about 3.5v to drive the PNP, which improves the brightness and efficiency. The cycle then restarts.

As far as your circuit, double check the specs for the transistor you're using - the datasheet I have rates the C3202 for 500mA, not 2A. And the gain at that level may be a tad low. This may be offset by reducing the 1M-ohm to 720K, which will likely help with the startup reliability. If the output is low, you can try to increase the drive to the NPN (by going down to 680k), but remember you are already pushing the transistor to its operating limits. Fwiw, the FJN965 is also available as the 2SD965 on Ebay  The 2SC2500 also works here.

Another thing I'd watch for is that the pinouts for the 2SC series is different from the BC series, which is different from the 2N series!

Good luck!

dph987 (author)2012-06-24

Here is a spice simulation of the new turbo circuit. I have substituted an FZT849 for the FJN965 being the closest type available in the simulator (8 amp) and I needed to reduce the 1meg resistor down to 100k before it would operate properly. I also found that increasing the 1 microF cap to 2 microF in the peak detector arm improved its voltage stability but takes slightly longer to startup.

However, I also found that this circuit will not start if the voltage is below 1.1 volts (!). It should work down to at least 0.8 volts to be usable.

ArturoC6 (author)dph9872017-03-26

Can you emulate a circuit to use a lithium instead and i can wire up 6 solar cells in series instead of the normal two? Id like to have the batteries cycle from 3.9v to 3.2v if possible. I just do not know how to use that program.

qs (author)dph9872012-06-25

Even if SPICE is getting close to real-world performance, it's not a valid emulation if you substitute a part that has less than 1/2 the gain of the FJN- / 2SD965. The 2SC2500 is also acceptable.

Your observation about the 10uF cap is a valid one - it does improve the operation somewhat.

This circuit was designed to make use of the power curve of the NiCad cell, so operation below 1-volt was not a consideration. However, for other applications, substituting a Schottky diode for the 1N4148 / 1N914 will allow the circuit to start at 0.7-volt and continue running until 0.4-volt.

TroyJ16 (author)2016-09-21

Are the diodes both the same kind?

qs (author)TroyJ162016-09-22

Any diode that can handle 150mA or more is suitable here. I've used the 1N4148 and they're still running after 5 years. If you can afford Schottky diodes they are even more efficient.

MrDiy88 (author)2016-06-23

Hi, great article. I have this exact problem. The Northern Ireland weather is almost always cloudy and overcast. How many hours will the battery drive the LED for?

Is this my solution? or is there a better way for me to get my solar lights working in the garden right into the next morning but making sure the battery gets a good charge on a cloudy day? Many Thanks

sitnah (author)MrDiy882016-06-25

MrDiy88 - I'm no expert, but I upgraded mine with solar panels that put out more mA than the one that came with it. Mine came with a 40, I upgraded to a 225. They run all nite long now.

See "Solar Walkway Lights Solar Panel Replacement" by Vlorbschnat on this site (Instructibles)

qs (author)MrDiy882016-06-24

Hello there!

It's always questionable whether a battery is being charged on overcast days. In the northern latitudes, even on sunny days, it is necessary to point the solar cell in a southerly direction, angled approximately 35-degrees to get the maximum exposure to sunlight.

A good rule of thumb is that typical solar cell will produce about 50mA to charge the NiCad under full sunlight. So that translates to about the same operating time, since the JT here uses about 50mA while operating. So basically, you'll need 10 hours of sunlight to run the light through the night. Less sunlight? Less time.

That's the reason I developed alternate circuits, notably the 'reverse' Joule Thief and the Blinking Joule Thief. Both are attempts to wring a bit more operating time out of whatever sunlight we see each day.

I'm a fan of the blinking JT circuit since it can multiply run times by a factor of over 3, meaning I get 3 days of "reserve" for each day of sunlight.

Let me know if this answers your q!

alC1 (author)2014-08-20

Can you help me with this?

I have purchased the Moonrays globe type thru Walmart @ $12 to $15 each. After about 8 months some of them are developing the following problem: the Circuit or the Photocells are going bad. What I mean is that the "Moonray" brand 1.2v AA NI-Co batteries will not recharge in the unit. When I swap in a regular AA (for test purposes only) the light will glow. When I insert the suspect battery into another unit, it recharges fine and works well. This is the 4th one to exhibit this problem, and since I am heavily invested into this unit (I have about 40 of them operating around my property) I want to be able to repair them. I can upload a pic of board if that will help.

dawp (author)2013-02-02

This may be a bit off-topic, but I was having a problem with solar powered lights not lasting through the night. I was using the type with one Ni-CAD or NiMH cell. I put an ordinary AA alkaline battery in one and found it burned brighly all night. What is more, the lights have continued to operate this way for over two weeks. I can only conclude one of the following (1) there is enough energy in a non-rechargeable to outlast a rechargeable or (2) As in some of the advertised devices that recharge alkaline batteries, the solar cells in the light provide a sufficient recharge to keep the battery going. (3) The rechargeable batteries that come with solar lights are of inferior quality. I I tend to go with option (2). Has anyone else done this?

netdragon (author)dawp2014-05-20

dawp: This is VERY dangerous. Your solar panel is probably not going to be able to tell it's an alkaline, and your alkaline has been CHARGING! Heck, even most chargers don't have a safety mechanism to prevent charging alkalines. Lucky for you, the solar panels are only drip-charging the alkalines. Eventually, it'll burst, maybe within the year, and cause a nice little mess inside your light's battery chamber. It could potentially cause a fire or explosion as well. That's unlikely due to the speed at which it's charging, but I have to tell you it's possible. There's whole discussions about drip-charging alkalines. It's dangerous for those who don't know what they are doing, and there's a reason why commercial companies say NEVER to do it.

The effect you are seeing is that alkalines have a lot more energy density than rechargeable batteries. It's a bit of a trade-off. Plus, since you've been drip-charging the alkaline, it's lasting even longer than it otherwise would. Considering even efficient circuits are still going to draw 100mAh or more, were it not charging, it'd have died within the week.

Also, please view webgiant's response with extreme skepticism. Amp hours are a measure of energy, not power. And as I mentioned, alkalines are typically higher energy density than rechargeables, with the exception of lithium ion, which I'd highly recommend against using outside. Lithium catches fire with a nice giant green flame when it gets wet, and can explode if in high enough quantities.

webgiant (author)dawp2014-03-15

The option you have not considered, regarding your solar lights not lasting through the night, is that the sun falling where you have your solar lights did not fully recharge the original battery. Based on the solar cell size and composition (generally amorphous, which is less efficient than mono/polycrystalline silicon), most garden lights, even in full sun all day long, will still not charge the battery enough to last all night long. Solar garden lights are intended to provide light for about four to six hours in full sun, less in less sun.

As an alkaline battery is charged in a completely different circuit than one used for NiCd or NiMH batteries, what is more likely happening is that your garden lights are simply running off the battery without the solar cell being involved, other than to shut off the light in the daytime. A full charge battery, such as an alkaline, can keep a single LED light on for 1-2 weeks, especially if it only runs at night, without recharging the battery. The LED "throwies" you may have seen on Instructables run for 1-2 weeks on a similar non-rechargeable battery, and they don't shut off in the daytime.

A fully-charged rechargeable battery has at least the same, if not more, amp-hours of power (amperage) than a regular non-rechargeable battery. With solar garden lights you get what you pay for, so rechargeable batteries in cheap solar lights will probably be cheap as well. Some of them last for years while others do not last very long at all.

JLouisR (author)2012-12-30

Hello; on my lamps I have also glued a new cell at 45 degrees facing south to get more light especially in winter. The problem may come from overcharge in summer if the new 2000mAh battery you put is Ni-Mh.I think It would not be possible to put even a simple TL432 shunt regulator, because end of charge on NiMh is difficult to detect with easy circuit. You can see the lights I fixed on my site, just for images because it is in french:.

s1 (author)2012-01-30

Plenty of 2SD965 on ebay.

Now_Is_The_Time (author)2010-12-09

I would like to first thank the original poster for this circuit.
I have been trying to build one of these myself using this schematic as a guide. Unfortunately, I cannot get an FJN965, and so I have used the NTE11 which is apparently an equivalent component( I think the pin out may be different though). I have used a discrete inductor as well as a wound one that I have made, and I am still having trouble with getting the voltage to boost to 3.3 V , though it does turn on and off depending on whether sun is out or not ( turns on when I cover with my hand and turns off when I let the light hit my solar panel ). My battery voltage is mimiced at the LED.

qs (author)Now_Is_The_Time2010-12-09

If all you is 1.2v at the LEDs, then the transistors are not owrking - check the transistors (polarity, pins, etc.) and the diode, D2 - it should be a low-signal silicon diode, 1N914 or 1N4148 or equivalent. Do not use a power unit. The 1uF capacitor at the anode of D2 is also critical.

jsilvers1 (author)qs2011-12-16

do u have a parts list for this build

Now_Is_The_Time (author)qs2010-12-09

Yes I am using a 1N914 and have the 1 uF, but I may have the pins switched around as again it is a brand-equivalent, yet different transistor. Thanks for that input. I will test it some more soon, and reply back.

emy vesverus (author)2010-08-17

please explain the operational for the circuit..

qs (author)emy vesverus2010-08-17

It's an adaptation of the 'turbo-charged' 2-transistor Joule Thief circuit which was presented on my web-site. Take a look there first and I'll try and answer any specific questions you have.

emy vesverus (author)qs2010-08-18

QS...i just wanna ask...why when i change the battery to (6V,1300mAh) and the solar panel to (7.2v,1W), "i had remove the inductor"...the LED always On at day and night... sorry i'm still new in electronic...

qs (author)emy vesverus2010-08-18

This design is specifically for converting the power of a 1.2v battery to 3.3v needed to operated the LED. By using a much higher supply, you're forcing the circuit into unknown, and possibly unhealthy areas.

If you have a 6v supply already, then you will not need to 'up-vert' the voltage at all. Just connect the LEDs through a suitable resistor and add a light-sensing switch for day/night operation.

Hope this helps - let me know your exact application (# LEDs, etc) and I can provide more help in assembling your light.

rmfungaro (author)2010-04-03

QS can use an inductor 47uF in this project? 22uf not find in my area ... congratulations for the project!

qs (author)rmfungaro2010-04-03

The acceptable range for the coil is 15-30uH, so 47uH is too high.

If you have 15-ft (5m) if a thin insulated wire, you can make your own - just wind the wire around the barrel of a felt marker (anything 1/2", 12mm in diameter will work). You don't need to be neat, just jumble wind into a donut shape and tie or tape together after removing from the form. This will give you a 22uH air-core inductor, and will be able to handle up to 1-A of current.

You can also get an assortment quite cheaply here on eBay.

ravingking2008 (author)2009-11-05

clever stuff . i have much to learn

chadeau (author)2009-09-25

So radically convenient-totally understand-cell repl sensor by input reduction-GREAT ! Thanks...

chadeau (author)2009-09-25

Is sensor connected after testing,or ...

qs (author)chadeau2009-09-25

The sensor is not needed since the circuit actually measures output from the Solar cell, and only turns on when there is less than 0.3v output, which is deep twilight, before turning itself on.

Rusdy (author)2009-07-28

Oops, forgot to add photo of mine, using slightly modified your circuit

qs (author)Rusdy2009-07-28

And thank YOU for sharing Rusdy! Fantastic job! With the added boost to the base drive, it is sometimes tricky to turn off using the solar cell - sometimes their internal leakage is high enough to bias oscillations on. The FPN965 is available on eBay (possibly as 2SD965) for 15 cents apiece. I prefer thru-hole components to SMT anytime.

Rusdy (author)2009-07-28

Thanks for sharing the circuit qs! After experimenting myself with different circuits and component values, yours is definitely the best! I'm using your double coil configuration (from your website for garden light). Unfortunately, the LED doesn't turn off until there is quite a bit of light, so I have to modify it with adding a second transistor. Anyway, getting cheap components is a major problem here down under, so I have to change the transistor (using FMMT617 instead of FJN965). After many (lots) of iteration, at last my lightbox is finished (well, actually still evolving!)

Wesley666 (author)2009-06-05

Cool instructable. I was disappointed though, the name said Oomph and as stupid as I was I thought you were adding music to the solar lights (I am German and there is a band named Oomph!) so naturally I was curious. Nice Instructable though.

qs (author)Wesley6662009-06-07


Lost something in the translation huh? I have a project which involves music here.

Wesley666 (author)qs2009-06-08

Cool music Instructable! I don't have an Arduino = ( I should get one.

digitalenigma (author)2009-03-16

could i use this same design and replace one solar cell with four in parralell and supplement the one aa battery with 4 aa ? or would that through all the voltage off?

d-lite (author)digitalenigma2009-03-16

Explain to me your needs - are you planning to run more LEDs? How bright? If you are not going to use the Garden Light enclosure (and solar cell), then there are different options available.

digitalenigma (author)d-lite2009-03-16

OK I bought 4 working solar garden lights from wal mart @ 4.00 a pop (would have recycled broken ones but couldn't obtain any) So, I want to make a modded light out all the parts I have a set up for a container I am working on from cd spindles and I have perfboard and a separate battery pack thats holds 4 AAs, I basically want to compile one circuit 4 battery's and 4 cells, I also bought some bigger LEDs from radio shack rated @28,500 mcd FW current 20 mA FW supply 3.5v 4.0 max, I would also like to wire in the photocell, but also have a on/off switch I am familiar with basic electronics, but can't really read schematics yet, I'm thinking of trying to integrate your circuitry with either a few standard LEDs or one of the big ones I mentioned earlier with 4 solar cells 4 AAs a ,a on and off switch, a potentiometer in the battery line to control brightness to extend battery life and maybe the photo cell to allow it to double as a outdoor light I'm aiming for a indoor outdoor solar light (with removable extendible solar cells for when indoors, photocell when used as a auto/on porch light) sorry for all the detail but from all the Instructables I keep winding back up on yours thinking the answer I need for the better power regulation/rebuilt circuit is there I just don't know enough about electronics to reverse apply it to my own needs, by myself.... but anyways room isn't much of an issue and I can wing everything besides the circuitry, because I don't want to fry my parts ect..

qs (author)digitalenigma2009-03-16

So, you are basically going to rip the lights apart for parts.

And with thise parts you want to assemble a single light with 4 LEDs. is that right?

Then the simple solution is to take 3 batteries, and charge them with 3 solar cells with a diode in between - like the picture below.

My suggestion is to charge the lights up first and see how well they serve your lighting needs. If mains power is avilable, it may be a better solution to build a LED "bulb" like this one here, which has the brightness of a 25-40watt bulb but only uses 3 watts for under $10.

If you bought the 10mm LEDs from RS, then be warned - they are not "brighter" they just have the beam more focused - so it forms a very tiny spot of light, so they may not suit your needs.

Rusdy (author)qs2009-04-09

I definitely like your 'simple' approach. But, I found using rechargeable batteries in series is very likely to damage one of the battery too quick (uneven charging / discharging cycle). So make sure to use fresh same batteries if you want to do this. All my solar garden light fails the same way, that is, one of the battery fails earlier than the other, causing charge reversal. So, I definitely like your single battery solution instead. Though, your 250mA current consumption will definitely make it last a wee bit too short. I'm trying to use LM2623 at the moment, definitely can't beat the efficiency from the chip manufacturer :) (70% as it claimed by the application note AN1221, April 2002 edition from the manufacturer). The only problem is, this chip only available in surface mount. So far, the pain of getting surface mount components, backyard job on surface mount PCB, and soldering them... still progressing. Painstakingly slow for a newbie in surface mount like myself :( All in all, I like your simple solution :)

digitalenigma (author)qs2009-03-19

no im definatly wanting to experiment wth solar/renewables, i could buy a led bulb but i was just in the middle of the southern us ice storm, and more than ever i realise the value of things that dont plug in! but yhea ive already disasembled the lights minus the circut boards and have been experimenting with them on a breadboard, so im going to save your scematic above and see what comes about, thanks very much for the info , the pic looks much simpler than what i had in mind or what comes in the factory circut!

qs (author)digitalenigma2009-03-20

The simplicity comes from the fact that 3 NiCd batteries just happen to fit the voltage of the white LED almost perfectly. That is why there is a minimum of adjusting and fussing.

peppeska (author)2009-04-05

Hi!!! Thanks!

Great Job and great instruction!

here my job! ->

and here my circuit ->

peppeska (author)peppeska2009-04-06


ALadyDragon (author)2009-02-27

Could i get a break down of the bits used. Im very very new to these kind of map and need a little more help. I want to try and get my old solar garden lights a make over and i feel this might just do the trick.. I just need some help.

About This Instructable




More by qs:Joule Thief LED circuitsA Triple Channel Musicator - the TriM...Musicator Jr - Mk 2
Add instructable to: