LED lights: new contestants

 There they are

I have bought two new LED lights to complement the one I evaluated here.

One is also an MR16 (12V) socket light with 4x1W warm white LEDs. I bought three of them over eBay from a seller in Hong Kong for €4 each. The output is estimated between 320 and 360 Lumens. It has a rather crude heat sink made of aluminium with sharp edges on the side. I guess, at this price this has to be tolerated.

The other is an E27 socket light with a multi-chip (chip on board) LED and a nominal power consumption of 3.5 W for use with 230 Volts. It has a slightly colder light and I got it at a price of €6 for two at the local supermarket. I found the form and design tempting (E27 LEDs are still rare and expensive) and bought the pack although the luminous flux was given at a meagre 140lm.
A look inside showed a small transformer with additional electronics covered in white glue. The LED multi-chip module consists of four lengthy LED chips, which apparently consist of two LEDs each (full brightness at about 7.4V).

A short test

Firstly, I tested the E27 light next to a 40W incandescent light bulb (all following photos have been taken with camera in manual mode and set to: no flash, exposure time 1/100s, aperture 4.97, F/5.6):

The narrower beam angle and colder light colour (3800K) of the LED light is clearly visible. A comparison of the luminous flux is difficult. While the LED light is clearly darker than the incandescent light bulb, it seemed to be brighter than the estimated 16W "equivalent".

Now, I put all three lights next to each other and powered them up to compare the light output:

From left to right: the E27, the MR16 3.6W from the earlier post and the MR16 4x1W light

The E27 again falls back in light output, but not as far as the numbers (140lm to 320-360lm and 300lm). Maybe the colder white colour creates a more glaring impression. The new 4x1W MR16 light has a narrower beam angle than the older one which makes direct comparison of brightness difficult. It seems to be it brighter an also has a slightly less yellowish hue.


Finally, two shots from the front (naturally, the order of the lights is reversed):

 

So, whereas the new MR16 light is a really good buy at this price, the E27 - while maybe being a bit brighter than the estimate on the package - is a no-go because of its low efficiency (<50lm/W), which lies below that of a typical CFL. The low price does not make up for that.

Solar lamp: Improvements

Can it be done?

The solar lamp from my earlier post was nice, but not very efficient. After all, it powered LEDs with a combined power of less than 1.5 W from a solar panel with ca. 3 W and a battery with ca. 28 Wh. Also, with a charging current limited to approximately 120 mA, the battery only got fully charged after a series of long sunny summer days. Using a venerable IC to regulate charging like the LM317(T) is nice and easy, but its voltage drop reduces the efficiency of the circuit dramatically.
I decided to improve the charging of the battery and add a MR11 socket LED lamp (12 V, 1 W, 80 lm) I bought earlier this year to the already present LEDs (six SMD parts with max 84 lm combined).

Ideas that were left out

• I intended to create a voltage limiting circuit made of two transistors with a very low voltage drop for battery charging. In my simulations and the first tests after implementation it seemed to work, but after the second day in the full sun one of the transistors got fried. So I removed this part of the whole circuit and opted for a simple diode which connects the solar panel to the battery.
• Constant current regulators would have been nice because the continually discharging battery leads to gradually darkening lights at night. Also, one of the lights (consisting of three LEDs in series) got badly damaged (though not destroyed) because a resistor of 30 Ohms was not enough with the battery fully charged at 15.5 V and a max. LED current of 80 mA. After a few experiments, I decided that I could not fit them into the cramped space and used variable resistors in series to regulate current.

Simulations

I used QUCS again to simulate the circuit before and during implementation. This time, I created a whole project instead of a single circuit for better modularity. The files can be found here.
Again, as QUCS has no models for LEDs, I have simulated the white LEDs by a series of four normal diodes.

central circuit with LED sub-circuits

three LED sub-circuit

simulation dependent on PV voltage

and with rising battery voltage simulating charging

As you can see from the simulation charts, the LED and charging currents are higher than in the first version of the solar lamp. The inner resistances of both PV panel and battery are included in the model. Also, the point where the LED current drops to zero has moved to right on the voltage axis, because the lower resistor in the voltage divider has been reduced from 20 k to 10 k. The LEDs are only approximated in the simulations and value of the series resistors might be a little different in reality.

The result

new circuit in the test environment

whole ensemble (circuit still with broken overcharge limit parts)

Some other LED lamps I purchased and installed

Apart from the MR16 LED lamps described in my first post, I bought some other lamps for different sockets. One of them is a replacement for a MR11 (aka GU4) 12V/20W halogen bulb. (about €6 at the time of purchase in February 2011). As with the MR16 LED lamps, it works with DC and AC voltage, although the AC frequency (50 Hz) can be noticed as very slight flickering when the lamp is the only light source and something moves fast under its light.

the MR11 LED lamp installed

switched on (low exposure time)

The new lamp has about 130lm and is a bit darker than the halogen bulb that it replaced. It is still OK for reading if it is close (< 30cm) to the book/newspaper/magazine.

Apart from the MR11 lamp, I ordered two E14 230V LED lamps (then about €8, same German online shop). Other than the other LED lamps I decribed so far, they do not have the three-chip LEDs, but rather one-chip versions. However, they have a CRI of 70 and emit a greenish yellow light like the others. With their luminous flux of 250lm (@ 3.5W), they are a bit darker than the 40W incandescent lamps they replaced (and the 17W CFL that was installed there at one time), but their light is sufficient for the small corridor they illuminate.

two E14 lamps

close-up of one lamp (low exposure time)

Unfortunately, I cannot provide a light colour comparison with the former incandescent lamps, but I have taken a photograph of a rug illuminated by daylight and by the two LED lamps.

rug under daylight

rug under LED lamp light

This demonstrates the parts missing from the spectrum (mostly blue and red, I think).

My first self-built solar lamp

Preset

I was always fascinated by using the light of the Sun as a source for electrical energy. About ten years ago, after my course in electronic at The university, I bought some small photovoltaic cells to build a solar lamp. I never got around to finish it and after the recent surge in popularity of solar lamps, I thought it would not be worth the effort if you can buy them cheaply.

A few months ago, I found a big photo-voltaic (PV) panel (open circuit voltage ~ 21 V, voltage under load ~ 17 V, short circuit current ~ 200 mA) at my parents' home, intended to power the pump of a fountain for the garden pond. As I thought this was a rather wasteful use for the PV panel and the whole solar-powered pump system was never used anyway, I decided to build a solar lamp with it.

 Design

After a lot of circuit simulation, searching for suitable parts and soldering of test circuits, I came up with this design:

schematics

V1 (PV panel voltage) dependent DC simulation

 
You can download the QUCS circuit simulation file here. For the simulation to work you need to put the files in this archive in the sub-directory named ".qucs" in your user directory.

V1 is the PV panel (with Rsolar representing the inner resistor) and V2 is the battery.
You can see that the switching of the light is done with a voltage divider (R1 and R2) and two transistors (T1 and T2). This eliminates the need for a light dependent resistor (LDR). If the voltage at the base of T1 drops below 0.7 V, it blocks and the voltage at the base of T2 increases towards the battery voltage and it conducts. This switches on the LED lights.
Charging of the battery pack is current-regulated by a LM317 IC with a 10 Ohm resistor (R6) as reference. This limits the charging current to about 125 mA. The diode D1 prevents discharge of the battery over the PV panel.
Unfortunately, I did not find a way to model LEDs in QUCS other representing them by five diodes in series.

Implementation

I bought ten warm white SMD LEDs (IF = 60-80 mA, max 14 lm) and decided to use them all to have several lamps connected to a single central regulation unit. I was not sure how many of these LEDs I should connect in series. First, I bought a NiMH battery pack with 7.2 V (2000 mAh) and I found out that with this pack I could only connect two LEDs in series. This would have resulted in several lights with two LEDs in parallel with their own current regulation. So, I bought another 7.2 V battery pack to get 14.4 V. After destroying four of these nimble SMD LEDs due to my poor soldering skills, I settled on having two lights with three LEDs in series for each.

the circuit (battery pack in background)

one of the lights (three LEDs in series)

I know that my soldering skills are really poor, if non-existent. The 30 Ohm resistors in series with the LEDs are done using three 10 Ohm resistors.

Result

I placed the PV panel together with the battery pack and the circuit on the garage roof and placed one of the lights on the ground (inside a decorative candle housing connected via a 5 m cable) and the other under the garage roof.

up under the garage roof

ground light

The lights illuminate the area in front of the garage which normally is a bit dark, especially because the motion sensor responsible for switching the illumination of the house behind the garage (visible in the photo on the left) is not really working properly.
The solar lights are switched on by the circuit at early dawn when it is not completely dark.

Issues

• Leaving efficiency issues aside (having a 14.4 V battery pack with three LEDs in series with a maximum voltage drop of about 11 V wastes some electrical energy if no efficient step-down converter is used), the main problem of the circuit is that the battery pack is not sufficiently charged during the course of the day - at least with middle European sunlight conditions. The diode and LM317 take away a few volts from the PV panel voltage and so the charge voltage is high enough only with really good lighting conditions. This leaves the battery pack not fully charged after an average day.
• Another point is the current regulation for the LED lights. The simple resistor in series results in a very bright light at dusk with brightness slowly decreasing as the battery pack discharges and its voltage drops. At midnight, the brightness has dropped drastically. A real constant-current regulation has to be used.
• With my poor soldering skills I should not have bought the tiny SMD LEDs, but should have used readily available 12 V LED lights (e.g. G4 modules).

I plan to address these issues and improve the circuit soon. All in all, this project has cost a lot compared to readily available solar lamps (more than €50 without the already available PV panel). But it was worth it just to see that my design ideas can be put into practice.

Replacing 12V MR16 halogen light bulbs with LED lamps

I have replaced the three standard low voltage 20W MR16 (GU5.3) light bulbs I had in the kitchen with LED lamps to save energy. Here is the data of LED lamps:
LED number and type: 20 x 3-Chip SMD
light colour: warm white
luminous flux: 300lm
electrical power: 3.6W
beam angle: 120°
CRI: ~70

Here is one of the new lamps:

I bought the LED lamps over the Internet. At about €9 per piece they are not cheap, but with an energy saving potential of nearly 50W (~83%) for the lamp system the saved energy will make up for the price. The new lamps work without problems with the old 60W transformer (output: 12V AC) and they reach their maximum brightness instantly.
Here are some pictures:

with standard 20W halogen light bulbs

with LED lamps

under halogen light

under LED light

The photo pairs have been taken with the camera in manual mode to use the same camera settings (exposure time, aperture, ISO number, etc.).

I reckoned that the original halogen lamps had a wide beam angle, so I bought LED replacements with the widest angle I could get. As it turned out, this was too much. The LED lamps spread their light over a much larger area. This explains the darker image under LED light (bottom row). All in all though, the LED lamps seem to emit a larger amount of light because of the area they cover. A side effect of this is that the area outside the window is now better illuminated:

The light of the LED lamps is not quite equivalent to that of the incandescent halogen lamps and seems to have a greenish yellow hue. This shows that lamps with a higher CRI than 70 are desirable. But after all, the light of the new lamps is acceptable and they seem to be a good buy.