DIY balcony/terrace/garden lantern

The problem with lanterns and all LED lighting is the balance of illumination strength vs. illumination duration vs. illumination diffusion. The Citizen above has been good for 9 hours at the medium setting so far. Have to try out full setting at 500lm, better for a small terrace or garden setting; maybe even sufficient for a caravan?

A sheet of Covestro Makrofol® LM 309 held at around 50mm distance to the LED. Transmission 81. Half-value angle 28°. Thickness 500µm. This special LED diffusor film is used in the automotive and aircraft industry. Looks ok considering the size of the COB. Have to try 300µm at a slightly larger distance, spec sheet says 85 at that thickness, but only 21° half-value angle. Well, I can’t have everything. The magic full diffusor with 100% transmission has to be invented still ; )

This stuff can be rolled into a tube of 40mm Ø before it cracks; the problem there is the seam, so vacuum-forming would be better.

If one of you fellow lanternists knows a material with higher transmission at around the same half-value angle, even if I have to get it from the U.S., please link to it here; I’d really appreciate that!

Edit 1: Rolled into a tube, quite sweet…

Edit 2: Turned it off after 24 hours at the medium setting with around 300 of the full 500 lumens. Enough for a good lantern, but 2700K version might be better still. On full setting now…

Hi Bra Ljus,

Do you have a source for small amounts of Covestro Makrofol, I can’t locate one.

If money was no object I would enjoy buying this sample kit of different diffuser films from Corbell Plastics. Different thickness, color temp and textures. Looks very fun to play with.

Diffuser Sample Kit

I got many samples from this source (use Bing translator), in the framework of a product development research project.

But I saw Corbell also trades in these wares, so maybe you can obtain single sheets from them?

Evonik USA also sells thicker similar materials and a flavor of Plexiglas that transmits the light at a 90° angle when fed into the edge of a block, rod or sheet, which is great for use in luminaires, quite unlike the classic edge-lighting plastics that primarily transport the light.

Thanks Bra Ljus,

I may contact Corbell directly for sample sheets. I am really glad you mentioned Evonik’s product, I will look into it.

Right, good luck with Corbell!

While waiting for the parts for the 3V Cree variant, and the two 9V versions working very well, particularly with over 12 hours on full setting, I thought why not go one step further and make these solar powered, so that the battery is charged during daytime and the LED comes on at dusk and then illuminates for as long as the battery can power it.

There is a timer breakout-board at Adafruit that lets one add low-power/timed-use to a circuit quite easily.

Existing balcony/terrace/garden solar lanterns look awful, their light output is visually garish and too low. Would be great to see more solar powered lanterns here!

Hey BraLjus , there are some threads on here on lantern talk . One is the BLF ultimate lantern project , by DBSAR . He is pretty knowledgeable about lanterns . It’s a long thread , but you may get a feel for some members or ideas that can help .
That Coventry sheet is pretty cool . I had never seen that before .

BLF Ultimate Lantern Project*~updates:Sofirn working on design, to produce prototype!

Sorry , can’t get the link :person_facepalming:

Yeah, that’s the only high transmission high half-value angle diffusor material on the market that works. I saw it at the Light + Building and Euroluce fair several years ago. Engineering these plastics is an art. Good to see that it is also available in the U.S. I’m always puzzled how some put a high-lumen LED in a luminaire - only to then kill 50% of the photons by using matt opalescent glass or frosted Polycarbonate or worse.

I clicked around in that lantern thread for a bit, it’s uuultra long, but I happened to come across an image and that looked rather as if a Kenwood blender had gone to bed with an SWAT-issue flashlight ; )

It’s a matter of taste and, as Oscar Wilde said, taste is a matter of taste : )

PS: Instead of a flat bottom, bouncing light right back onto the LED and surrounding heat-sink area, I would have made the internal reflector conical/hyperbolic/asymptotic… just like in good omnidirectional loudspeakers.

I’m not really a lantern-tard, but throw in solar [and LiFePo4], and my enthusiasm grows.

I got one of these recently as a lightning deal (sale price) for $9.99 … https://www.amazon.com/gp/product/B013UCC65G/ref=ppx_yo_dt_b_asin_title_o03\_\_o00_s00?ie=UTF8&psc=1

I’d suggest bumping the total circuit up to 12V (if possible), using [4] LiFePo4 in series, and some kind of jack readily comptabile with:

Expounding on my idea a bit, throw one of these into the body with its own on/off switch (to avoid parasitic drain) :

Then you could measure charging / discharging currents and have some idea of the state of the battery charge.

If you were car camping at night, and the battery circuit dipped below 12V, you could plug it into the car’s cigarette lighter, and the car’s battery would run the light and charge its batteries (assuming car battery had voltage >12.0V).

Thanks, Barry, I already bought 6V solar panels, chargers, batteries and 3V LEDs with matching MCPCBs (see first post).

The Cree XPLAWT-00-0000-000UU30E7, 3000K, 320lm, CRI 90, Vtyp: 2,95V DC, Ityp: 1050mA is just at the sweet spot for a single LED 3V lantern.

If USPS weren’t so slow, I would have tested that set-up already and posted some pictures and running times : (

Edit 1: That looks like a practical little meter… The idea expanding the lantern to solar would mean to use the solar panel itself as the darkness trigger, meaning that once the solar panel delivers 0V, a MOSFET switches the charger/battery <-> LED driver path on.

What do you think about, as a next project, a 4S LiFePo4 12V setup with on-board voltmeter/ammeter and 12V jack to connect solar panel or cigarette lighter lead?

And if you want it to be really boss , have it accept:

(Enough juice to start your car with appropriate cabling)

Ha, next project… I have not even finished the first three (9V, 3V and 3V + solar)… I also would need a car to begin with ; )

At last I got the LEDs and MCPCBs from the U.S. and “skilletflow”-soldered them. Works great. First time I attempted it. This is fun!

Yeah , it is fun when you do something new and it works ! Looks as though you project can move forward now

Got a little mass-production going ; )

Once the pan is hot enough, you just drop in one MCPCB in after the other : )

With over 12 hours full lumens from the 11.1V set-up, it’s now going to be interesting to see what the 3.7V setup with the flat 4500mAh Li-Po battery will yield, and how the Cree’s CRI of 95 will make a difference.

Actually, you want to have a much slower ramp-up to temperature to minimise thermal stresses. Otherwise the mcpcb heats up, the solder melts, the LED sticks to it but is still relatively cool while everything else is still hot, then when you remove it the mcpcb shrinks somewhat while the LED is just starting to equalise in temperature.

When everything cools back down to room-temperature, the LED could be “squeezed” under constant compression.

My IKEA two-plate cooker has a great heat-curve profile - settings off to 6, ha!

Maybe I was just lucky with around 175°C ; )

But these Crees do their thing at 1050mA and the heat-sink reads 55°C, which is quite ok, over-dimensioned as it is. I really don’t like that Cree does not provide the usual figures for heat-sink sizing simulation : (

Done with the 2nd set-up:

1. Cree® XLamp® XP-L LED – XPLAWT-00-0000-000UU30E7 – 3000K, 320lm, CRI 90, 1050mA, 2.95V & matching MCPCB & matching Fischer fcool heat-sink
2. OLIMEX Li-Ion rechargeable battery – 1s, 3.7V, 3000mAh
3. TP4056 USB 3.7V Li-Ion single cell charger with micro-USB

Lasted 3 hours and 38 minutes.

So, for a long-lasting balcony/terrace/garden lantern, the 11.1V solution with the CITIZEN LED is the clear winner, also because it outputs around 500lm compared to the 3.7V Cree based variant.

That 11.1v looks really good on run time. 11 hours at 500 lumens , definitely will get the job done. That’d pretty much last me on a long weekend without a charge .

Yeah, I will go for the 11.1V version. It is very large, even with the proper (smaller) heat-sink, but one has to make a compromise somewhere.

Still, for the 3.7V version, I will like to go ahead also, but with a 7000mAh Li-Po pouch cell, which should be less voluminous.