Traditional power supply design focuses on maintaining a constant output voltage, regardless of load. We usually over-design the supply, to ensure a constant voltage.
But this is the opposite of what we want to drive a LED, which prefers a constant current. We go to a lot of trouble to get good constant-current drives. However, a cheap, crummy, droopy, poor load regulation power supply is inherently a somewhat constant-current drive!
For example, I recently ordered a variety of cheap wall-wart type power supplies from MPJA: http://www.mpja.com/products.asp?dept=37&main=1. The "regulated" ones they sent were pretty good, with mostly just some sloppiness in the listed OD/ID sizes of the coaxial connector. But the unregulated supplies were pretty awful, with open-circuit outputs way over the nameplate voltage, and very droopy. There was no way to predict how they would actually perform, but in generally one could expect to get vaguely the nameplate voltage out under the full nameplate load current condition.
The real joker in the deck was the DV-1000 "Universal AC/DC Adapter" China http://www.mpja.com/prodinfo.asp?number=17369+PD "MULTI OUTPUT PLUG-IN WALL SUPPLY" 17369 PD rated at 1.5, 3, 4.5, 6, 7.5, 9, 12VDC @ 1,000mA Selectable. When I tested this with a variety of loads, it was so droopy and under-powered that it just seemed universally useless. (Marlin P. Jones has impressive customer service -- they refunded my money, no arguments.) This is certainly not a "1000mA" power supply -- it would be lucky to ever put out that much current, even under short-circuit conditions. It is really only good for a few humdred milliamps, and quite droopy at that. The actual output voltage has hardly any connection to the labelling on the voltage selector switch.
But, this is just the kind of power supply we might want for driving LEDs!
In the "6V" mode, it outputs:
0mA 5.33V, 45 4.66, 169 3.65, 261 3.06, 370mA 2.41V
this looks like quite a good match to drive a 3V 350mA 1W class white LED.
In the "12V" mode it outputs:
0mA 10.06V, 83 8.54, 295 6.29, 439 5.07, 596mA 3.88V
which looks like quite a good match to drive two hi-power white 3V LEDs in series.
The world is full of old little wall-wart power supplies getting discarded and trashed. It could be hard to find just the right one you want, but if you collect a bunch you might have some luck... The key to this approach is that you want an un-regulated, under-powered unit with poor output regulation. Since you will be stressing it, it might get hot, so be careful. And if you use it long-term, not just for experimenting, make sure you mount it so that it gets good air flow for cooling, and is well separated from combustable materials, in case it does burn out/up! A fuse in the primary circuit might be a good idea.
Which leads to wondering, can we be more intentional and serious about this approach... Can we design AC line-powered LED drivers with just a simple cheap little transformer, full-wave bridge, and perhaps some filter capacitance, relying on the transformer to be under-rated and provide appropriate load current-limiting, by design?
The Wikipedia article says:
"A leakage transformer, also called a stray-field transformer, has a significantly higher leakage inductance than other transformers, sometimes increased by a magnetic bypass or shunt in its core between primary and secondary, which is sometimes adjustable with a set screw. This provides a transformer with an inherent current limitation due to the loose coupling between its primary and the secondary windings. The output and input currents are low enough to prevent thermal overload under all load conditions—even if the secondary is shorted.
Leakage transformers are used for arc welding and high voltage discharge lamps (neon lamps and cold cathode fluorescent lamps, which are series-connected up to 7.5 kV AC). It acts then both as a voltage transformer and as a magnetic ballast.
Other applications are short-circuit-proof extra-low voltage transformers for toys or doorbell installations."
Since this is not what we usually look for from a transformer, even those of us who think we know electronics may not know much about this esoteric niche. Googling "leakage transformer" is not very fruitful, and detailed specs of doorbell types are not easy to find. "Door chime transformer". They seem to be 16VAC, 10-15W output; 10V 7W, 24V 15W... Do most of them really have safety-rated, short-circuit-proof outputs? What other readily available transformers might be suitable? Any tips to mod/wind your own? Do any commercial LED bulbs use this driver approach? The circuit simplicity might tend to make it efficient, but transformer losses in this mode combined with running at line freq instead of SMTP high-freqs might make it too inefficient for most uses?
Could most dinky little line transformers be used in this mode? Would it generally be feasible to passively cool them well enough that they would not burn out? (Will they make obnoxious humming noise...)
Note: this approach is only appropriate if you want cheap-simple-sloppy good-enough drive, with LEDs that have plentiful heatsinking and you accept not having fine control over the drive current level.
But this is the opposite of what we want to drive a LED, which prefers a constant current. We go to a lot of trouble to get good constant-current drives. However, a cheap, crummy, droopy, poor load regulation power supply is inherently a somewhat constant-current drive!
For example, I recently ordered a variety of cheap wall-wart type power supplies from MPJA: http://www.mpja.com/products.asp?dept=37&main=1. The "regulated" ones they sent were pretty good, with mostly just some sloppiness in the listed OD/ID sizes of the coaxial connector. But the unregulated supplies were pretty awful, with open-circuit outputs way over the nameplate voltage, and very droopy. There was no way to predict how they would actually perform, but in generally one could expect to get vaguely the nameplate voltage out under the full nameplate load current condition.
The real joker in the deck was the DV-1000 "Universal AC/DC Adapter" China http://www.mpja.com/prodinfo.asp?number=17369+PD "MULTI OUTPUT PLUG-IN WALL SUPPLY" 17369 PD rated at 1.5, 3, 4.5, 6, 7.5, 9, 12VDC @ 1,000mA Selectable. When I tested this with a variety of loads, it was so droopy and under-powered that it just seemed universally useless. (Marlin P. Jones has impressive customer service -- they refunded my money, no arguments.) This is certainly not a "1000mA" power supply -- it would be lucky to ever put out that much current, even under short-circuit conditions. It is really only good for a few humdred milliamps, and quite droopy at that. The actual output voltage has hardly any connection to the labelling on the voltage selector switch.
But, this is just the kind of power supply we might want for driving LEDs!
In the "6V" mode, it outputs:
0mA 5.33V, 45 4.66, 169 3.65, 261 3.06, 370mA 2.41V
this looks like quite a good match to drive a 3V 350mA 1W class white LED.
In the "12V" mode it outputs:
0mA 10.06V, 83 8.54, 295 6.29, 439 5.07, 596mA 3.88V
which looks like quite a good match to drive two hi-power white 3V LEDs in series.
The world is full of old little wall-wart power supplies getting discarded and trashed. It could be hard to find just the right one you want, but if you collect a bunch you might have some luck... The key to this approach is that you want an un-regulated, under-powered unit with poor output regulation. Since you will be stressing it, it might get hot, so be careful. And if you use it long-term, not just for experimenting, make sure you mount it so that it gets good air flow for cooling, and is well separated from combustable materials, in case it does burn out/up! A fuse in the primary circuit might be a good idea.
Which leads to wondering, can we be more intentional and serious about this approach... Can we design AC line-powered LED drivers with just a simple cheap little transformer, full-wave bridge, and perhaps some filter capacitance, relying on the transformer to be under-rated and provide appropriate load current-limiting, by design?
The Wikipedia article says:
"A leakage transformer, also called a stray-field transformer, has a significantly higher leakage inductance than other transformers, sometimes increased by a magnetic bypass or shunt in its core between primary and secondary, which is sometimes adjustable with a set screw. This provides a transformer with an inherent current limitation due to the loose coupling between its primary and the secondary windings. The output and input currents are low enough to prevent thermal overload under all load conditions—even if the secondary is shorted.
Leakage transformers are used for arc welding and high voltage discharge lamps (neon lamps and cold cathode fluorescent lamps, which are series-connected up to 7.5 kV AC). It acts then both as a voltage transformer and as a magnetic ballast.
Other applications are short-circuit-proof extra-low voltage transformers for toys or doorbell installations."
Since this is not what we usually look for from a transformer, even those of us who think we know electronics may not know much about this esoteric niche. Googling "leakage transformer" is not very fruitful, and detailed specs of doorbell types are not easy to find. "Door chime transformer". They seem to be 16VAC, 10-15W output; 10V 7W, 24V 15W... Do most of them really have safety-rated, short-circuit-proof outputs? What other readily available transformers might be suitable? Any tips to mod/wind your own? Do any commercial LED bulbs use this driver approach? The circuit simplicity might tend to make it efficient, but transformer losses in this mode combined with running at line freq instead of SMTP high-freqs might make it too inefficient for most uses?
Could most dinky little line transformers be used in this mode? Would it generally be feasible to passively cool them well enough that they would not burn out? (Will they make obnoxious humming noise...)
Note: this approach is only appropriate if you want cheap-simple-sloppy good-enough drive, with LEDs that have plentiful heatsinking and you accept not having fine control over the drive current level.
