how do you drive leds?

[valentin tihomirov]

We have to control hundreds of typical colour leds (a couple of mA) and
optocouplers (1mA). The device is not portable; however, I would like 1 -
not to waste energy on the ballast resistors and 2 - reduce amount of the
resistors. Can you point to the guidelines on the issue? In addition, the
optocouplers have open collector outputs; that is, the outputs have to be
pulled up by more power and resistors.

 

[Tim Wescott]

Hi,

What are you driving the LED's from? Computer, controller board,
microcontroller? Any way you hack it, it will take 1 digital output per
LED/OPTO; and this takes a lot of driver chips.

-- unless you're turning on LED's, Opto's or both in groups -- then you
could gang things together.

You can get optos with different outputs -- do some web searches.

 

[Luhan Monat]

We have to control hundreds of typical colour leds (a couple of mA) and
optocouplers (1mA). The device is not portable; however, I would like 1 -
not to waste energy on the ballast resistors and 2 - reduce amount of the
resistors. Can you point to the guidelines on the issue? In addition, the
optocouplers have open collector outputs; that is, the outputs have to be
pulled up by more power and resistors.

Hi,

What are you driving the LED's from? Computer, controller board,
microcontroller? Any way you hack it, it will take 1 digital output per
LED/OPTO; and this takes a lot of driver chips.

 

[James Meyer]

We have to control hundreds of typical colour leds (a couple of mA) and
optocouplers (1mA). The device is not portable; however, I would like 1 -
not to waste energy on the ballast resistors and 2 - reduce amount of the
resistors. Can you point to the guidelines on the issue? In addition, the
optocouplers have open collector outputs; that is, the outputs have to be
pulled up by more power and resistors.

Welcome to the REAL world.

Jim

 

[Anders F]

Basically the "old" way with resistors or better: Constant current drivers
are easily optimized: Lower the supply as much as you can... (To save the
energy wasted over drivers/resistors)

Alternatively you go for something switched. Typically a current is built up
in an inductor and discharged into the LEDs. The uC companies will have
app-notes on this (Microchip for sure) and you can find dedicated LED
switchers from National, Maxim, Linear, Semtec, Zetex etc. etc.

What are you driving the LED's from? Computer, controller board,
microcontroller? Any way you hack it, it will take 1 digital output per
LED/OPTO; and this takes a lot of driver chips.

What's wrong with multiplexing? For indication on closerhand this will
actually also save power: The eye is slow, so short high intensity burst
will seem bright in terms of brighness/energy...

I'd go for this approach ;-)

/A

 

[Ken Taylor]

We have to control hundreds of typical colour leds (a couple of mA) and
optocouplers (1mA). The device is not portable; however, I would like 1 -
not to waste energy on the ballast resistors and 2 - reduce amount of the
resistors. Can you point to the guidelines on the issue? In addition, the
optocouplers have open collector outputs; that is, the outputs have to be
pulled up by more power and resistors.

Use resistor packs to reduce component count. Is the power saving really
critical enough to worry about? How much do you want to save?

Ken

 

[Ol' Duffer]

We have to control hundreds of typical colour leds (a couple of mA) and
optocouplers (1mA). The device is not portable; however, I would like 1 -
not to waste energy on the ballast resistors and 2 - reduce amount of the
resistors. Can you point to the guidelines on the issue? In addition, the
optocouplers have open collector outputs; that is, the outputs have to be
pulled up by more power and resistors.

LED's and optocouplers are somewhat amenable to pulse width
modulation, assuming the switching frequency is sufficient
to prevent the junctions from overheating. Most makers
don't publish specifications for this mode of operation,
because it's easy to damage the LED's if the modulation
mechanism is not highly reliable. It only has to stick
on once at, say, 5 Volts to turn that little 2 Volt junction
into something much less useful. I'd look to incorporate
some kind of fail-soft if I were contemplating such an
arrangement.

 

[Pooh Bear]

We have to control hundreds of typical colour leds (a couple of mA) and
optocouplers (1mA). The device is not portable; however, I would like 1 -
not to waste energy on the ballast resistors and 2 - reduce amount of the
resistors. Can you point to the guidelines on the issue? In addition, the
optocouplers have open collector outputs; that is, the outputs have to be
pulled up by more power and resistors.

Keep the supply voltage low !

With a 2.5V supply, a typical led might have a Vf of 1.6V - the o/c
transistor a Vce of 0.2 - 0.3 V, so only around 0.6 - 0.7 V is wasted across
the current limit R.

The optocoupler *outputs* can be tied to whatever logic supply voltage you're
using. No need to use especially low value pull-ups unless you're switching
at very high speeds - in which case you should be using a dedicated coupler.

Do the math and see if it's worth using a simple switching regulator to
supply that 2.5V if you're main system only uses 5V.


Graham

 

[valentin tihomirov]

Use resistor packs to reduce component count.

IMO, the packs violate the main design concept - divide and conquer.

Is the power saving really critical enough to worry about?
How much do you want to save?

Personally, I do not like useless wastes. As a noob, Im just asking about
adopted practices. The power is 3.3v, diodes drop 1v..1.5v, the rest drops
at the ballast. I suppose 50..100%. In one circuit there are 32 optocoupler
diodes and source voltage is 30v; that is, 30 times more power will be burnt
than required to activate optocoupler. So I think, the total reduction can
be about 4 times.

 

[Mac]

We have to control hundreds of typical colour leds (a couple of mA) and
optocouplers (1mA). The device is not portable; however, I would like 1 -
not to waste energy on the ballast resistors and 2 - reduce amount of the
resistors. Can you point to the guidelines on the issue? In addition, the
optocouplers have open collector outputs; that is, the outputs have to be
pulled up by more power and resistors.

Usually I just put a resistor in series with the LED and connect the low
side to the output of an inverter. When the inverter output is low, the
LED is on.

Occasionally, when the difference between the forward Voltage drop on the
LED and the power supply voltage is small, I use some other approach
(e.g., a simple transistor current source).

These both waste about the same amount of power.

If you really want to save power, you could use a switch-mode current
supply to do the job. It is very hard for me to believe that this would be
worthwhile in most cases, because switch mode supplies use inductors to
store energy, and I just can't envision using an inductor for each LED.

Elsewhere in the thread, someone mentioned simple pulse-width modulation
of the LED's. I don't know whether this would actually save power but
it could make the circuitry more simple by eliminating the series
resistors.

--Mac

 

[Ken Taylor]

IMO, the packs violate the main design concept - divide and conquer.

What do you mean by that/ You didn't mention anything like that before.

Personally, I do not like useless wastes. As a noob, Im just asking about
adopted practices. The power is 3.3v, diodes drop 1v..1.5v, the rest drops
at the ballast. I suppose 50..100%. In one circuit there are 32 optocoupler
diodes and source voltage is 30v; that is, 30 times more power will be burnt
than required to activate optocoupler. So I think, the total reduction can
be about 4 times.

Maybe you should look at the design again and decide if you are approaching
it correctly. If power usage and component count are so critical you should
look from top down again.

Ken

 

[valentin tihomirov]

IMO, the packs violate the main design concept - divide and conquer.

What do you mean by that/ You didn't mention anything like that before.

Nothing special. The major design concept which reduces amount of
information enabling creation of complex systems is hierarchy. The the
partition incapsulates functionality into hierarchy modules. The purpose of
the partition, is to break down a solid block of N elements (complexity =
number of interconnections is about N^2) into loosely interconnected tight
pieces. The resistor packs will increase connectivity between modules. In
addition, my CAD does not support multple sections of a component spread
over design modules enforcing the HW hierarcy.