I bought a pair of LED "IS" tubes. Fixture takes just one, put it in - lights briefly then goes out. Box says I need an "Instant Start" ballast. No way am I buying that retail...
So I busted one open, looks like a full wave bridge with the 80 count LED string between + and - output, with a 22uf filter cap in parallel. Then a 0.1 uf film cap on one leg of the input, a small inductor on the other.
Looked on line and finally found a pdf where the LED current of one of these tubes was mentioned, 50 mA with 60 max. Using a 100 Ohm series resistor and a variac, I dialed up the voltage until I got 50 mA of AC current - as read by my DMM. This lit the string nicely. The voltage across the string was 93V, so 118-93/0.05 gave 500 Ohms.
Deciding I wanted to use a cap instead of a resistor (as per original), I calculated 5 uF as giving 530 Ohms at 60 Hz. So I replaced the resistor with a 5uF film cap and my current reads significantly more than 50 mA - on the DMM. Fiddling, I ended up with 3.2uF and just OK brightness for a ~50mA DMM reading.
Do I need to use the oscilloscope to look at the current waveform, to ensure I'm putting 50mA RMS through the string? I'd like as much light as I can get, without killing the reliability.
So I busted one open, looks like a full wave bridge with the 80 count LED string between + and - output, with a 22uf filter cap in parallel. Then a 0.1 uf film cap on one leg of the input, a small inductor on the other.
Looked on line and finally found a pdf where the LED current of one of these tubes was mentioned, 50 mA with 60 max. Using a 100 Ohm series resistor and a variac, I dialed up the voltage until I got 50 mA of AC current - as read by my DMM. This lit the string nicely. The voltage across the string was 93V, so 118-93/0.05 gave 500 Ohms.
Deciding I wanted to use a cap instead of a resistor (as per original), I calculated 5 uF as giving 530 Ohms at 60 Hz. So I replaced the resistor with a 5uF film cap and my current reads significantly more than 50 mA - on the DMM. Fiddling, I ended up with 3.2uF and just OK brightness for a ~50mA DMM reading.
Do I need to use the oscilloscope to look at the current waveform, to ensure I'm putting 50mA RMS through the string? I'd like as much light as I can get, without killing the reliability.
Last edited:
Design a RC network for 80V or so, and remember those devices failed in the market because high voltage drivers like that fail too frequently.
RC is the simplest and most reliable, or go for four pin Mosfet design, those Mosfets have an internal opto coupler, and set the feedback as needed.
A constant current design is more suitable for LED.
In fact, some companies have withdrawn the product here, as the replacement rate under warranty was too high for LED tube lights.
Oddly enough, bulbs and panel lamps are very reliable, and they do stay within 40V DC.
The LED tubes sold here used proper LED drivers, not the crude RC design, I am surprised those are sold in your area.
In India, the Indian Standards are now strict, and nothing in lighting can be sold without certification.
Ceiling fan regulators using film (MPP) caps to reduce the voltage are common, you could start with 1 mfd or so. Plastic case, about the size of an eraser, 400+volts. Connected in series with the motor. Regular AC motor, not DC as in inverter.
Of course, the old fans drew 50 + watts, the new inverter fans are supposed to take only 22W, but those are HV DC fans.
Maybe one of those inverter fan speed controllers can work for you.
Even a fridge contoller for small inverter may be made to work.
Torches here use a 390k in parallel with a 455J/400V or so film cap to drop the voltage, then rectifier, to charge the 4V lead acid batteries inside.
RC is the simplest and most reliable, or go for four pin Mosfet design, those Mosfets have an internal opto coupler, and set the feedback as needed.
A constant current design is more suitable for LED.
In fact, some companies have withdrawn the product here, as the replacement rate under warranty was too high for LED tube lights.
Oddly enough, bulbs and panel lamps are very reliable, and they do stay within 40V DC.
The LED tubes sold here used proper LED drivers, not the crude RC design, I am surprised those are sold in your area.
In India, the Indian Standards are now strict, and nothing in lighting can be sold without certification.
Ceiling fan regulators using film (MPP) caps to reduce the voltage are common, you could start with 1 mfd or so. Plastic case, about the size of an eraser, 400+volts. Connected in series with the motor. Regular AC motor, not DC as in inverter.
Of course, the old fans drew 50 + watts, the new inverter fans are supposed to take only 22W, but those are HV DC fans.
Maybe one of those inverter fan speed controllers can work for you.
Even a fridge contoller for small inverter may be made to work.
Torches here use a 390k in parallel with a 455J/400V or so film cap to drop the voltage, then rectifier, to charge the 4V lead acid batteries inside.
Last edited:
The LED tube light was a cottage industry here, kits being sold, maybe you can find the schematic on line, we have 220 or so mains volts, you need designs for 110V.
All the bits like LED strip, extrusions, end caps, PCB, parts for PCB...sold in bulk, the populated circuit was only 25 Rupees - 30 cents - or so.
When we made LED tubes, we got ceramic block drivers from China, AC in, DC out, and the 16W version ran well, the 18W died in three months, this was after putting larger heat sinks, which is inside a constrained space with little ventilation.
Got a mosquito killer, circuit is crude, one series cap to lamp, another to grid.
$1.25 in flea market.
Wires open, will repair, no hurry.
All the bits like LED strip, extrusions, end caps, PCB, parts for PCB...sold in bulk, the populated circuit was only 25 Rupees - 30 cents - or so.
When we made LED tubes, we got ceramic block drivers from China, AC in, DC out, and the 16W version ran well, the 18W died in three months, this was after putting larger heat sinks, which is inside a constrained space with little ventilation.
Got a mosquito killer, circuit is crude, one series cap to lamp, another to grid.
$1.25 in flea market.
Wires open, will repair, no hurry.
Oscilloscope says 84 mA (84mVAC across 1 Ohm...). Waveform looks like a very distorted sine wave; Crest factor is 2.3. But not the typical FW bridge into filter caps current waveform. The 22u is out of the circuit. I tried to get "inside" the bridge to the DC side of it, but couldnt.
I assume on the DC side of the bridge, the waveform would look like the same pulse shape, but with the negative flipped up. Still 84 mA RMS?
I assume on the DC side of the bridge, the waveform would look like the same pulse shape, but with the negative flipped up. Still 84 mA RMS?