independent kiwi spirit of invention.
list by Ian Mander started 1 February 2008. Added to this site (Aqualab)
26 November 2008. Database released 27 May 2009.
Please note that the date mentioned below that the database code was last updated
is not the date the data itself was last updated.
Driver List Database code 20 November 2018
Footnotes 10 August 2016
All prices in US$ (except where dual prices are listed in US$ and €
for some European retailers).
All driver boards from DealExtreme and KaiDomain include shipping.
Information is unfortunately not guaranteed to be correct.
any updates, corrections, omissions, etc.
However, please don't bother sending me an email to tell me about
your company's LED products. It will be treated as spam. I really
don't like spam, and SpamCop is busy enough as it is without having
to process your email as well. Putting "Re" in the front of your spam's subject does not make it any less likely your spam will be sent to SpamCop.
Recommended drivers highlighted in green.
They have a good combination of price, features and efficiency.
Drivers no longer available (sold out
or backordered) are highlighted in grey.
Recommended drivers no longer available
are highlighted in a darker green.
Drivers listed at those resellers as "Backordered" etc
for more than a month are deemed to be discontinued (although I'm happy
to later be proven wrong).
Don't connect drivers that have capacitors across their outputs to
LEDs while the driver is powered. An explanation
(on CPF) why not.
No mains driver will be completely waterproof. Those that are water resistant mostly have an IP rating (eg, IP67).
on AMC7135 linear regulator(click to expand/contract)
The AMC7135 (datasheet)
is a linear regulator, which means it acts like a variable resistor changing
its value to try to keep the current constant. Like a resistor, any dropped
voltage is burnt off as heat. Boards include a polarity protection diode and
can easily be PWM-driven for lower modes.
Vin must be at least 0.12 V above Vf of LED to stay
in regulation, although they drop out of regulation quite gracefully, not suddenly. The
graph in the AMC7135 datasheet (Jan 2006) has the 0.1 and 1 volt vertical
lines missing. Each AMC7135 provides constant current, about 1/3 amp (actually 300-380 mA depending on particular version; I've generally assumed
330-335 mA for above listings). Boards come with one to eight AMC7135s, and
single mode up to 20 mode. Boards can be paralleled to give greater output, or connected with one multimode board controller providing the modes for several boards.
The AMC7135 is very efficient when input voltage is close to output
voltage but not particularly good when input voltage is significantly higher. Average efficiency for 3x NiMH or 1x Li-ion can be well
over 90% with an LED with the right Vf. Test
results and discussion for 3 and 4 chip boards.
Since the AMC7135 just burns off excess input volts as heat, the more volts
fed into them the hotter they'll get. One guy claimed that his got so hot they slid right off the board (ie, >183-190 °C
melting range of 60/40 solder). The AMC7135 has built-in thermal protection
(which will cause dropouts or a flickering effect if it gets too hot) but
the multi-mode control chips used on the multi-mode boards are much
less rugged. (And here.) If using with an input voltage above 4.5 V or so you can expect them to get hot!
To get multiple modes typical microcontrollers used are the Atmel ATtiny13 (or
13A or 13V) and the Microchip PIC12F629.
These both have a 5.5 V maximum, while the AMC7135 linear regulator has
a 6.0 V maximum. This means that multimode drivers will have a slighty lower
maximum voltage than single mode boards.
Tip 1: To get reliable operation at low voltages, especially with only
one AMC7135 chip being used, you may need to short out (and maybe remove)
the polarity protection diode(s)*. This is because the AMC7135 in series
with a polarity protection diode needs a minimum 2.7 V + 0.6 V (silicon
diode) = 3.3 V to stay in regulation. The Vf of LEDs
at 330-350 mA can easily be quite a bit lower than 3.3 V so will not be
running in regulation. Note that if a germanium or Schottky diode was used the drop
could be as low as 0.3 V instead of 0.6 V.
* However, I found with one multimode board this caused the board to
go unstable (don't know exactly why) but I found that inserting a small
value resistor instead of the diode was enough to get the driver stable
again. Because the drive current through that point in the circuit is
so low (6 mA for mine) there's very little voltage drop across the resistor
- much less than across the diode - so it still serves the purpose of
saving ~0.6 V.
Tip 2: If the input voltage is too high you may be able to use
another LED in series with the board to drop the voltage - it beats burning it all off as heat. (The set current
is <1 mA for single mode boards so both LEDs will get practically identical
and much discussion of use with multiple Seoul P7s and multi-mode boards.)
More than one extra LED appears to be not a good idea for use with the lower
modes of multi-mode boards since the Vf of the extra LEDs decreases
too much at the low current to protect the driver from the battery voltage.
(Many of the multi-mode boards have a capacitor on the output.) Flashing modes appear unsuited to this technique.
AMC7135-based driver options are discussed here,
or an inexpensive multimode AMC7135 driver here.
on PT4105 and alternative driver chips (PT4115, AX2002, CL6807)(click to
of this driver IC - as used in the Kennan and MR16 base drivers described above
- has been terminated. The manufacturer doesn't even have a publicly displayed
link to the datasheet any more, which is the weirdest part of it. This from
Micro Bridge (now removed from their site; try to ignore
the punctuation and spacing):
The PT4105 which the manufacture has already officially stopped producing,and
the subsequent instead item is the PT4115,AX2002 and FP6101 Also,The PT4115,AX2002
and FP6101 has superior performance over ,wider input range and more current
than the PT4105.
I look forward to the PT4115 being available in low cost LED drivers (by
its numbering the apparent successor to the PT4105), I note that it needs
an input of at least 8 V (and has under voltage lock out at 6.8 V), so isn't
nearly as well suited to low voltage torches as the PT4105 was. It will,
however, have its uses for 3x Li-ion torches and automotive purposes. The
chip has a DIM pin which gives it the ability to very easily
be dimmed. Efficiency is about 80-82% for 1 LED, up to 93% for 3 LEDs, and
apparently up to 98% for 7 LEDs. Maximum output current 1.2 A.
driver chip from AXElite looks extremely interesting. It will
accept a minimum 3.6 V input and has a maximum switched current of 2.5 A,
although it tends to overheat at more than 2 A. It includes thermal protection
over current protection, short circuit protection, and has a PWM control
circuit. Its efficiency
is good too, with an output of 2 A @ 5 V it's an impressive 91% efficient
(with 12 V input). Driving a Cree XR-E at 1 amp will give an efficiency of
about 87-88% (with 12 V input). Efficiency is not quite as good at low currents with a single LED, dropping under 80%.
AX2002 drivers can also easily be configured as a constant voltage power supply. The load is connected straight to ground and the 0.25 V reference voltage is used to control a voltage divider with a couple of moderately high value resistors to give a fixed multiple of 0.25V at VOUT.
For example, for 5 V, 5 = 20 * 0.25, so a 10 kΩ resistor is placed between ground and FB (the feedback pin), and a 190 kΩ resistor between FB and VOUT (making the total of those resistors between VOUT and ground of 200 kΩ).
When used in this way, to give stability the current through the resistors probably just needs to be comfortably greater than the feedback pin bias current of (0.1 µA typical, 0.5 µA maximum). If two exact resistor values for the voltage divider are not available it's easiest to use a single resistor for the sense resistor (between ground and FB), while the other value (between FB and VOUT) uses two resistors in series or parallel. For series, one of those two resistors will be as close as possible to the desired value, and just under it, while the other will be a much smaller resistor to tweak the total resistance up for the output voltage wanted. For parallel, the main resistor is just over the actual value wanted while the other resistor with about ten times the resistance tweaks the total resistance down. If that resistor is getting into megaohms you should probably revise your values.
Some AX2002 drivers (such as DX 3256sadly no longer an AX2002 driver) come with a 1 A Schottky diode, which will need to be changed if increasing the output current over 1 A. See the Schottky diode notes below for links.
AX2002 also has a big brother, the AX2003, which
has a maximum switched current specification of 4 amps – easily enough to
drive a Seoul P7, or a Cree MC-E with the dice in parallel. No drivers with
the AX2003 are presently known. The spec sheets of the AX chips could do
with a few more graphs showing how constant the output current is, etc.
Chinese LED driver, 1 A maximum output current, 6-35 V input, 0.1 V high side sense voltage. Claims to be able to provide up to 35 W output power. Dimmable with 0.5-2.5 V PWM signal.
So there are some nice driver chip options, but it still leaves a gap of a high efficiency,
really low voltage, low current
on Schottky diodes(click to
Schottky diodes are diodes that have a low voltage drop across them. 0.3 V is a typical figure, compared to around 0.60-0.65 V for a typical silicon diode. This makes Schottky diodes good for rectifiers and LED drivers where high efficiency is required. Drivers that use the AX2002 such as DX 3256 can easily be modified for higher output current but the Schottky diode needs to be replaced if the output current is to exceed 1 A.
Inexpensive Schottky diodes are available from these sources:
These MR16 drivers have four 1 A (SS14) Schottky diodes on them used for the rectifier plus another for the driver (links jump to driver info in table above):