I left the Nichia LED on the power supply overnight, at about 85 to 90 mA, and it didn’t seem to have any problem. I decided that I had to design a circuit to drive it from a low voltage, such as two AA rechargeable cells. I drew up a circuit, which had two transistors for mainly one reason: the output voltage must be more than 20 volts in order to drive this LED. Using two transistors makes the circuit oscillate without having a second feedback winding on the coil. So all that’s needed is a choke or inductor.
Well, I was thinking, why should I build a circuit when I have several of the Night Joule Thief boards from The LED Artist, Akimitsu Sadoi. All I have to do is modify one to get it to handle more voltage, current and power. I found the schematic by using Google to search for night joule thief. Its the first schematic that ‘s in the images. More info about it can be found on Instructables.
The output transistor is going to have to handle quite a bit of current, about a half amp or more. It will also have to withstand more than 20 volts peak. I thought that the BD135 would be a good choice, and should run cool enough to not need a heat sink.
1 deficiency of this circuit is that the output transistor is driven bye an NPN transistor. This means that all of the current that turns on the output transistor has to be supplied by the resistor to the collector of the driver transistor. So this resistor has to be very low and it wastes some power. If a PNP transistor was used to drive the output transistor then all of the current is supplied through the PNP transistor and the resistor from its collector to negative can be much higher and waste less power. The PNP transistor can supply much higher current than the resistor, and also turn on the output transistor much better.
I already had a Night JT PC board partly assembled so I used it. I swapped out the transistor and put leads on the BD135 to make them fit into the holes. The BD1135 leads were too thick to fit by themselves. I removed the inductor and used a 56 μH inductor used for high current. It’s surface mount with a ferrite bobbin surrounded with a square body of ferrite and has an air gap between the two. It’s very low DC resistance, only 0.35 ohm, and can handle 0.75 amp. I had to add two short leads to it for the holes in the circuit board.
I used the following values for the following parts:
- CdS photocell = omitted (for now)
- C1 = 1000 pF Ceramic Disk
- Vr1 = replaced with a 33k fixed resistor.
- R1 = 2.2k
- R2 = 220 ohms
- R3 = 10 k
- Q1 = BC327-25
- Q2 = PN2222A
- Q3 = BD135
- D2 = replaced with a 1 ohm resistor for measuring the LED current
- D1 = Nichia NS6W183 3.3W, 225 Lumen, 73 Cd, 20V, 115 mA,
- 6 chip White LED, goldmine-elec.com #G19461A
At first I used the original values for R1 and R2 but the LED current, even at 3V, was too low. So I then replaced them with the values given in the parts list above. The LED current went up to about 19 mA at 2.5V and 27 mA at 3V. At 1.5V the LED current was only about 4 mA – it was clear that this circuit has to be powered by 2.5 or more volts. The supply current at 2.5V was about 250 mA and at 3.0V it was over 400 mA. A pair of AA rechargeables is not going to last long at that current.
The LED is reasonably bright, but even 27 mA is less than 1/4 of its maximum current of 115 mA. So I have to do some more experimenting to optimize the circuit and increase the LED current. The BD135 gets warm, but not hot – it’s running well within its limits without a heat sink. The frequency is below 100 kHz, around 91 or so depending on the voltage.
Update Apr 10
I pulled out the 1000 picofarad capacitor and replaced it with a 680 pF and a variable capacitor. When I tuned the variable I got a very broad peak at 920 picofarads. So 1000 picofarads is just about right, and I put it back in.
I still wasn’t satisfied with 27 milliamps through the LED. I connected another resistor across the 220 ohm R1. The LED current jumped up to 32 milliamps. I measured the two resistors and got 110 ohms, so I removed the 2 resistors and replaced them with a 110 ohm resistor. Now I was getting 5 more milliamps, for a total of 32. Better, but still only about a third of the way. The supply current was 430 mA at 2.5V. Two AA rechargeable cells should last for about 4 hours. If it drew more current the battery lifetime will be even lower, so I think this is a good compromise.
Efficiency
The input power is 2.5V times .43Amps, or 1.075 watts. Output power is 19 V times .032 Amps, or .608 watts. The efficiency is .608 / 1.075 or 56.6 percent. Not all that great but a little bit better than the typical joule thief