# Just checking if my thinking is right on an automotive LED lighting design

Discussion in 'Electronic Basics' started by JazzMan, Mar 13, 2005.

1. ### JazzManGuest

I want to make a set of amber DRLs for my car using LEDs in
arrays. The LED is an AND185HAP, specs are If 50 mA MAX,
iF 20mA typ @ 2.0-2.4V. Measured output on my alternator
is 14.5V typ, so I want to run four parallel strings of
six LEDs each, no resistors.

Does this sound workable?

JazzMan
--
**********************************************************
Curse those darned bulk e-mailers!
**********************************************************
"Rats and roaches live by competition under the laws of
supply and demand. It is the privilege of human beings to
live under the laws of justice and mercy." - Wendell Berry
**********************************************************

2. ### Larry BrasfieldGuest

I believe you would be disappointed with the
result and could lose (destroy) the LED's often
enough to wish you had done it differently.

The problem is that the LED current will vary
quite a bit with battery voltage, which ranges
from 11-12 Volts during engine off conditions
to 13-16 Volts when the engine is on and the
battery is being charged after a start. By the
time you keep the upper end of that range
from hurting your LED's (by using a limiting
resistor), they will be too dim near the lower
end of that range. Without the limiting resistor,
over-voltage transients (exceeding the range
mentioned) may well destroy or degrade your
LED strings.

The LED has the same exponential current
versus voltage function as other semiconductor
junctions, with a relatively small series resistance.
If you can get a plot of the E/I characteristic, you
will see the problem with voltage variation.

If efficiency is your concern, rather than simplicity,
a switching converter with controlled current
output might be attractive. If you are not so
concerned with efficiency, 1 or 2 fewer LED's
per string and limiting resistor is a good route.

3. ### JazzManGuest

Efficiency is irrelevant to me, what I'm after is maximum
brighness. The cost of the LEDs is trivial. The existing front
marker light is a dual filament bulb that uses the dim filament
for a parking light and the bright filament for intermittent use
when signaling. A friend of mine bought a module that
inverted that function and ran both filaments all the time
with lights off, cycled the bright filament when signalling,
and reverted to dim-on and bright-signal when the headlights
are on. The problem he had was that his light housing melted
because of the extra heat from running the bright filament
full time as a DRL. My thought was to replace the bulb with
as many LEDs as I can physically fit into the housing, which
looks like it will take a 4" x 1.5" circuit board. Now that
I think of it, I can probably run twenty strings of 6 LEDs
if I pack them in edge to edge like the new traffic signal
LED bulbs are built. 120 high-output LEDs should be bright
enough.

My experience with electronics beyond basic DC and some TTL
is very, very limited from a design standpoint. I'm really
good at laying out boards and fabricating complex circuits,
if I have a schematic and parts list to work from, but I'm
fairly deficient when starting from scratch.

Anyone feel like referring me to a relevant schematic?

JazzMan
--
**********************************************************
Curse those darned bulk e-mailers!
**********************************************************
"Rats and roaches live by competition under the laws of
supply and demand. It is the privilege of human beings to
live under the laws of justice and mercy." - Wendell Berry
**********************************************************

4. ### Larry BrasfieldGuest

Given your requirement, I would just cut the string
length to 4 LED's and use a 1/2W 180 to 240 Ohm
limiting resistor for each string. This will typically drive
your LED's at 24 to 33 mA depending on the resistor.

If you want better efficiency and are willing to build a
power converter, take a look at this SMPS IC to get an
idea of what would be necessary:
http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1705
The datasheet shows a few more typical hookups, at:
http://pdfserv.maxim-ic.com/en/ds/MAX629.pdf
The part has a 250 mA current strapping option, and
if you were to divide your LED collection into N strings,
where N = 0.25 / (IperLED), and use a small resistor
in each string to equalize the current sharing among the
strings, your circuit would not be much more than what
can be seen at the above links. (You choose IperLED
according to how hard you want to drive the parts.)

If you decide to go with a convertor, come back for
a few tips on layout, or a review of your planned
circuit prior to committing layout effort or buying parts.

5. ### Larry BrasfieldGuest

A closer study of the MAX629 datasheet shows that
it is not rated for the input voltage you need for an
automotive application.

If you are eager to build a power converter to run
your LED's, the MAX668 or MAX669, depicted at:
http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1901
and whose datasheet is at:
http://pdfserv.maxim-ic.com/en/ds/MAX668-MAX669.pdf
can be made to do it.

Its current limit is not well-defined in the datasheet,
so, if you are interested in this alternative, say so and
I (or somebody faster on the draw) will outline or
explain how to set the current sense resistor to get
approximately the current you want. This current
should be predictable and stable enough for your
purposes, although it is not strictly regulated.

Replace these two links with above recommendation:

6. ### PeteSGuest

Well, if you are putting these in a vehicle, be warned that you are
plugging into the power supply from hell (to quote an app note from
Linear Tech).

As a designer of equipment that ends up in vehicles, I can assure you
that the 'nominal voltage' you get is there *most* of the time, but can
be as high as 60V or so under load dump conditions (when loads get
suddenly disconnected the reverse emf on the alternator windings
generate a significant voltage), to say nothing of having large spikes
at most times anyway.

Because of that, I'd put a current limiting resistor in the sequence,
otherwise as noted by others, you will regularly (and spectacularly if
there's a really good load dump - as happens quite regularly at cold
crank, engine pre-stall, accelerator kick). Most cars will dump the A-C
if it's on for a maximum acceleration request, known to most of us as

We'd usually generate our own power from the vehicle with a nice
protected supply and use that, although you can use a string of LEDs
with current limit.

Cheers

PeteS

7. ### JazzManGuest

Ok, I bought 100 LEDs yesterday to do this project. Specs
on the LEDs are as follows:

Vf @ 20mA 2.0V, max 2.4V
output 11,150mcd

How do I get one of these chips?

JazzMan
--
**********************************************************
Curse those darned bulk e-mailers!
**********************************************************
"Rats and roaches live by competition under the laws of
supply and demand. It is the privilege of human beings to
live under the laws of justice and mercy." - Wendell Berry
**********************************************************

8. ### Robert MonsenGuest

If you want to drive them with 2.0V @ 20mA, then you can have 5 in a
string, giving 10V total. That will require 20mA for each string, which
means 400mA total current. Each string will require current limiting.
You can do that with a resistor, but as others have pointed out, spikes
will damage the LEDs over time.

Another way to do it would be to use transistors to limit the current. A
trivial 'current mirror' scheme might work here.

A TL431A, with the reference connected to the cathode, serves as a nice
2.5V voltage reference. Using that, you can build a little current
source by setting the base of an NPN transistor to 2.5V, and using a
resistor to limit current. This circuit is independent of the voltage
input, and will draw a dependable current.

By using that current reference, you can then use a set of PNP
transistors to limit current across those strings. The base of all of
these PNP transistors are tied together, and referenced using the
current source above.

12V -o---------o-------------o---------o--------------------.
| | | | |
| .-. .-. .-. .-.
| | |10 | |10 | |10 | |10
| | | | | | | | |
| '-' '-' '-' '-'
| | | | |
| | | |
| | | | |
.-. >| |< |< |<
| | 10k |----o----| .--| .. .---|
| | /| | |\ | |\ | |\
'-' | | | | | | |
| | o------)----o----)--------------' |
| | | | | |
| | |< | | |
| o----| Q1 | | |
| | |\ LEDS LEDS LEDS
| | |
| |/ |
o-------| |
TL431A | |> |
.---o | |
| | K | |
| R - .-. |
'-- ^ | | 82 |
| A | | |
| '-' |
| | |
----o---------o------o-------------------------------------------o
GND
(created by AACircuit v1.28.5 beta 02/06/05 www.tech-chat.de)

If there are spikes of 60V, the transistors and 10 ohm resistors will
absorb them. I simulated a spike of 60V, and the LEDs went from 20 to
22mA. That is well within their tolerances.

You can adjust the brightness by adjusting that 82 ohm resistor, which
sets the current limit for all of the strings. The transistors will be
slightly different, in which case the current may be a bit different in
the strings. However, the 10 ohm resistors will tend to even things out
a bit. You could even go up to 22 ohm resistors without affecting things
too badly, which would give you a larger margin for error, and also a
bit more protection against spikes.

--
Regards,
Robert Monsen

"Your Highness, I have no need of this hypothesis."
- Pierre Laplace (1749-1827), to Napoleon,
on why his works on celestial mechanics make no mention of God.