This really follows on from some of my previous mains powered projects but is closely related to the LED Teardown previously documented.
Now we have all gone out and bought them in the summer, those little flower border lights which are solar powered and charge up during the day and once the night kicks in they act as a border garden light.
Of course they have a limited life being cheap imports that suffer in the good old British weather with failed battery packs and sometimes just failed solar panels.
Normally you buy these things in packs of 4 or more and the light source is a single low power led of the cheap variety kind. Once dead we throw them in the bin and off to landfill they go. Well it got me thinking, why not convert it to a mains powered unit with 10W of LED's. It would have to be safe though and protected from the weather and it needs to be cheap. Could it be done I wondered, and would 10W be too much?
From the pictures you can see that the source of the light is a tubular design of about 60mm diameter of stainless steel and a plastic diffuser.plus another tubular lid which fits on top with the solar panel in.
First thing I did was to remove the original small white led and the square solar panel in the roof.
Idea for this is to mount the led's on a plate fixed to a heat sink facing upward through the solar panel aperture.
Step 1: The LED Specification
Having recently purchased some 10W single COB led's I wondered if it would be possible to use a single one and use a switch mode power supply directly from the mains[240V UnIsolated]
Candidate was a buck switch mode power driver chip FL7701 and inductor 1.4mH coilcraft. Unfortunately the conversion from 240v to the FW of the COB block[12V] does not readily work as the current required through the COB is far greater than the driver chip can handle if you want 10W. The chip can handle 0.5A which with a forward voltage of 12v would only get you to 5W or thereabouts. You could use a forward converter switch mode with isolation which would do the job but the cost starts to soar, after all this was supposed to be cheap and cheerful.
So how could I get 10W with only 0.5 A of current. Well given the conservation of energy theory the only way to increase the wattage is to increase the voltage, and the only way I could do that would be to increase the forward voltage of the led's by using more than one of them. If you look at my LED Teardown instructable you can see why they did this in that design. Browsing on EBAY I readily found some 1W led's with a forward voltage 0f 3V@330mA. Now if I used 10 and under ran them @266mA I would end up with 10 x 3 x0.266A=8W...close enough. The underrun has a two pole approach ….keep the heat down and therefore preserve or extend the life.
Lower junction temp means happy lights.
Step 2: The LED Base
Looking at the pictures of the garden light what is needed is a method of mounting of these LED's and of course if they are sinking 266mA we need to get rid of 8W of energy across them which will require a heatsink.
The inner diameter of the stainless tube is a little under 57mm so if I could mount any of the electronics in a sealed plastic tube and install it on the inside the tube.I could then mount the plate of leds facing downwards on top of the enclosure which would then illuminate the diffuser.
So how would we arrange the leds?
First of all I cut a 46.5mm circle of aluminium with a center hole using a hole saw[see pic] and using some double sided heatsink tape covered one side.You can get this tape on ebay and its fairly cheap, normally used for heatsink attachment see picture. The aluminium was an old power supply enclosure but you can probably buy this on ebay . I used a piece 2mm thick.
You need to cover and insulate the metal from the base of the led but still have good thermal conductivity.
Use a double lap of thermal tape laid across orthogonally in two layers. This will change the thermal conductivity
and we lose another 20 degrees c across the junction but that's what it takes. I will revisit this later on and maybe look at a fully sealed aqualusion solution but not for now.
Step 3: BasePlate
Then I used Autocad to lay out where the leds need to go on the base. See the pictures of this attached as pdf's.
I printed the design to scale and used a hole punch to make a mounting template of the layout to act as a rough guide. Laying this over my sticky base plate I drew the outline of the circles on the tape.
Next I laid out the leds so that I could get some positioning of some copper tape which I would use to link the leds on the surface of the insulating thermal tape.
Making sure that no copper tape violated onto the underside of the "slug" I soldered them all together. Of course you need to make sure that cathodes go to anodes. You could just stick them down and use some hookup wire between the pins although using copper tape helps to dissipate some of the heat into the tape.
On the subject of heat , these generate a lot of it so need a fairly large heatsink. I opted for a 40x40x30 H heatsink which keeps the bottom plate at around 58-60 degrees C. It so happens that his size fits neatly into the removed solar chip Allowing for the thermal heat across the junction to case of the led about 4 deg c per watt and say 1 deg C per watt from plate to case this should mean a junction temp of (8x1)+4= approx. 60+12 degrees C =72 degrees C which should be reasonable.
The total voltage across the leds will be 10 x 3v or thereabouts so next stage will be testing the current through them.
The attached PDF has an outline to use as a template but you can always make your own design.
Checkout the easam attachment which you can download the eviewer to peruse
Step 4: Top Assembly
We said earlier that we would use a FL7701 driver chip for this and playing with the xcel spreadsheet designer came up with a set of figures that might work. Key to the buck converter was to get the ripple down to something reasonable given the RMS value we needed. Ripple has a direct bearing on inductor size and frequency of operation an indirect effect. So if we increase the ripple we have to increase the inductor size and the only way then to reduce the required inductance is to up the frequency.
See the attached picture which lists what I was iterating to and was key to the values on the schematic.
Here are the soldered LED's laid over my template prior to sticking them down. Note the use of the heatsink which has the plate stuck to the bottom with the mounted leds.
Increasing the current to 266mA RMS by adjusting the peak current to 500mA set the voltage at just over 30v across the leds which implies that the voltage was actually close to 3v forward if we have 10 leds .
Note that the calculation expected 286mA whereas in reality we only managed 266.
Frequency should have been 101Khz however looking on the scope seemed a little under.
I will discuss the schematic and the driver and waveforms on the next step.
So plugging in lit the baseplate up like a Christmas tree.
Quick note here on safety. This is a non isolated design so everything that could be elevated to mains level needs earthing thoroughly. This will include the heatsink which if you look carefully has a couple of holes that need to be self tapered via an earth tag to the heatsink and the stainless metalwork and the incoming mains earth. Be careful with the wiring of the leds that no shorting takes place between the leds and ground. If it does then greater than designed voltage appears across the leds and will destroy them quickly.
I have a test setup which has a mains isolating transformer but when connected directly to the mains one side of the inductor is at mains potential which if it gets connected to any isolated pieces of metal would be a hazard.
Step 5: Testing and Schematic
So let take a jump back and look at what we need to drive the leds.
We already said we need to support 266mA or there about so we have already done the numbers.
Referring to the schematic note the following:
Incoming through fuse 1 to bridge rectifier then to filter inductor with two c's.
D1 is the recovery diode and the means to ramp the current down on the inductor. Q1 gate is driven by pin 2 of FL7701 via R3 with D2 aiding sweeping the charge out of the gate on the negative stroke of the FL7701.
Frequency of the output is set by R5/R4. Couple of pins have some decoupling and the CS pin..pin1 is the current sense which is monitoring voltage and hence current through R6.
Refer to the peak current in R6 of 0.5A which will cause the IC to reset and ramp down ready for the next on period.
Note whats missing in this circuit. There is no requirement for a big rectifier DC cap for the input. The FL7701 cleverly takes care of the input variations internally. Given this is usually an expensive part it helps save on cost.
Once the PCB was populated I checked the ripple. Using a current probe on the cathode of the led block gave ripple as 150mA and the average current using the meter was measured as approx. 260mA. This is 100mA down on the max for the leds and lets them run cooler therefore extending their life. Frequency was measured as 81Khz and ramp down as 1.71us. This is 13% of the capabilities of the chip/inductor so should be fine.
The starting point for this whole design was in the use of a 1.4mH off the shelf coilcraft inductor
Step 6: PCB Construction
Note that the images are of the prototype board which had some errors on it which I corrected on the new uploaded pcb layouts. Note the jumpers on it to get round some incorrect pinning....doh.
This caused some blowups before I realised the error...must have been tired!
There are a couple of the topside and one of the underside.
Step 7: Putting It All Together
So here it is slotted together.
I will attach a BOM list of all the parts required later.
Some things to watch out for. I earthed the heatsink at the top and fed it through the unit to an earthing point at the bottom.This is then earthed back to the supply. Be wary of this. The cathode of the final LED is 30V or so below the peak mains voltage of 310V. This will hurt if touched so needs to be kept isolated and any metal parts that could come into contact bolted down to earth in order to ensure a clear path for fault current.Note the use of cable glands top and bottom to stop any water finding its way in to the electronics. The earth screw at the bottom acts as a stop for the mains "cannister" and there is a drain hole in case any moisture finds its way in. This is not a waterproof container but the mains is kept out of the way from fingers and the drain hole is well above ground level.
The top heatsink needs some sealing around the top and this is still to be completed.
I intend to put this out in the garden for the summer and probably add some others later.