LEDs in Series with LM334 Current Source

[John Popelish]

Ok. To review.

The goal here is to use a current source. Initially the thought was to
use the LM334 as a current source to drive each string of LED's. Then the
idea became a mirrored Current source. But the LM334 current limits put
it outside the realm of driving LED's that need 20mA to reach their
brightness.

A constant current source is prefered due to the variations that can occur
within the automobile charging system, and the delicate nature of LED's

Information from CFoley1064 and Robert Monsen has been very helpful. So
now it's time to look at the preliminary design and look for issues.

Below is diagram using a LM317 voltage regulator in combination with a
resistor to ensure a fixed current source regardless of changes in
automobile voltage output.

Here are the planned devices to use:

12 Volts (as high as 14 v is possible, as low as 11)
1-EA LM317 (http://www.national.com/pf/LM/LM317.html)
1-EA 62ohm Resistor .1 watt 1% tolerance
6-EA 2N2905A PNP Transistors (Has Vceo of 60)
20-EA RL5-R3545 Super-Red LED Foward Voltage of
2.2 typical, 2.6 max (http://www.superbrightleds.com/leds/r2_specs.htm)

When wired as shown below, the goal is to achieve a current level of 20mA
across all of the LED's in a fairly even manner to achieve near equal
brightness across all of them. The theory being that changing the
resistor value can impact the amount of current flowing as it will impact
the setpoint of the voltage regulator.

Questions:
1) Are the devices listed above appropriate for the goals?
2) What is a good source for the above components. (LED's I know where to
go)
3) Any other comments or changes to the circut below recommended?

Thanks

VCC +12
|
+--------+-------+---------+--------+------+-------+
| | | | | |
| | | | | |
|< |< |< |< |< |<
+-+-------+---------+--------+------+-------+-|
| |\ |\ |\ |\ |\ |\
| | | | | | |
| | | | | | |
+-----+ V V V V V
| - \\ - \\ - \\ \\ - \\
+----------+ | | | | |
| ____ | | | | |
| | | V V V V V
+--|LM317---+ - \\ - \\ - \\ - \\ -\\
|____| | | | | | |
| .-. V V V V V
| | | - \\ - \\ - \\ - \\ -\\
| | | | | | | |
| '-' V V V V V
| | - \\ - \\ - \\ - \\ -\\
+------+ | | | | |
| | | | | |
+---------+------+---------+--------+------+-------+
|
|
===
GND -12
created by Andy´s ASCII-Circuit v1.24.140803 Beta www.tech-chat.de

I would eliminate the LM317 and add a resistor to each current source
emitter to make them track each other much better. I would set the
base voltage to one of them to force about .6 volts across its emitter
resistor.

Something like:

VCC +12
|
+--------+-------+---------+--------+------+-------+
| | | | | |
| .-. .-. .-. .-. .-.
| | | | | | | | | | |
| | | 33 | | 33 | |33 | | 33 | |33

| '-' '-' '-' '-' '-'

|---- + | | | |
/| | | | | |
| |< |< |< |< |<
+-----+---------+--------+------+-------+
| |\ |\ |\ |\ |\
| | | | | |
| | | | | |
| V V V V V
| - \\ - \\ - \\ \\ - \\
| | | | | |
| | | | | |
| V V V V V
| - \\ - \\ - \\ - \\ -\\
| | | | | |
.-. V V V V V
| | - \\ - \\ - \\ - \\ -\\
| |1k | | | | |
'-' V V V V V
| - \\ - \\ - \\ - \\ -\\
| | | | | |
| | | | | |
+--------+-------+---------+--------+------+-------+
|
|
===
GND -12

No need for as spiffy a transistor as the 2N905A. I would probably go
with 2N3905. They have a plastic case that is much cheaper and they
are common as dirt. Even Radio Shack should have them. 2N4403 is
beefier but that is not much of a factor with 20 ma of collector
current.

 

[Winfield Hill]

John Popelish wrote...

Mike H wrote

I would eliminate the LM317 and add a resistor to each current source
emitter to make them track each other much better. I would set the
base voltage to one of them to force about .6 volts across its emitter
resistor.

VCC +12 [ edited by w.h. ]
|
+-+------+-------+--------+--------+-------+-------+--- etc
| | | | | | |
| .-. .-. .-. .-. .-. .-.
| | | | | | | | | | | | |
| | | | | 33 | | 33 | |33 | | 33 | |33
| '-'100 '-' '-' '-' '-' '-'

| | | | | | |

|--- | ------+ | | | |
/| | | | | | |
| | |< |< |< |< |<
| +-----+--------+--------+-------+-------+----- etc
| | |\ |\ |\ |\ |\
| |< | | | | |
+----| | | | | |
| |\ V V V V V
| | - \\ - \\ - \\ \\ - \\
| | | | | | |
| | | | | | |
| | V V V V V
| | - \\ - \\ - \\ - \\ -\\
| 2.2k | | | | | |
.-. | V V V V V
| | | 220 - \\ - \\ - \\ - \\ -\\
| | .-. 1W | | | | |
'-' | | V V V V V
| | | - \\ - \\ - \\ - \\ -\\
| '-' | | | | |
| | | | | | |
+-+------+-------+--------+--------+-------+-------+--- etc
|
|
GND

No need for as spiffy a transistor as the 2N905A. I would probably go
with 2N3905. They have a plastic case that is much cheaper and they
are common as dirt. Even Radio Shack should have them. 2N4403 is
beefier but that is not much of a factor with 20 ma of collector
current.

Radio Shack has the 2n3904 (npn) and 2n3906 (pnp). They have the 2n4401
(npn), but not the 2n4403. They also have the TO-92 mps2222 and mps2907.

I enhanced your design just a bit with another pnp trannie, so it can
handle more LED strings if desired.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Mike H]

Mike H wrote: .... ....

I would eliminate the LM317 and add a resistor to each current source
emitter to make them track each other much better. I would set the
base voltage to one of them to force about .6 volts across its emitter
resistor.

Something like:

VCC +12
|
+--------+-------+---------+--------+------+-------+
| | | | | |
| .-. .-. .-. .-. .-.
| | | | | | | | | | |
| | | 33 | | 33 | |33 | | 33 | |33
|---- + | | | |
/| | | | | |
| |< |< |< |< |<
+-----+---------+--------+------+-------+
| |\ |\ |\ |\ |\
| | | | | |
| | | | | |
| V V V V V
| - \\ - \\ - \\ \\ - \\
| | | | | |
| | | | | |
| V V V V V
| - \\ - \\ - \\ - \\ -\\
| | | | | |
.-. V V V V V
| | - \\ - \\ - \\ - \\ -\\
| |1k | | | | |
'-' V V V V V
| - \\ - \\ - \\ - \\ -\\
| | | | | |
| | | | | |
+--------+-------+---------+--------+------+-------+
|
|
== GND -12

In the design you propose, what limits the voltage fluctuation and thus
the current change? Say if the 12v is suddenly 14v for a moment. The
point of the LM317 was it's ability to maintain a set output voltage
regardless of the input.

Thanks for the info on the devices

 

[Mike H]

....

VCC +12 [ edited by w.h. ]
|
+-+------+-------+--------+--------+-------+-------+--- etc
| | | | | | |
| .-. .-. .-. .-. .-. .-.
| | | | | | | | | | | | |
| | | | | 33 | | 33 | |33 | | 33 | |33
| '-'100 '-' '-' '-' '-' '-'

| | | | | | |

|--- | ------+ | | | |
/| | | | | | |
| | |< |< |< |< |<
| +-----+--------+--------+-------+-------+----- etc
| | |\ |\ |\ |\ |\
| |< | | | | |
+----| | | | | |
| |\ V V V V V
| | - \\ - \\ - \\ \\ - \\
| | | | | | |
| | | | | | |
| | V V V V V
| | - \\ - \\ - \\ - \\ -\\
| 2.2k | | | | | |
.-. | V V V V V
| | | 220 - \\ - \\ - \\ - \\ -\\
| | .-. 1W | | | | |
'-' | | V V V V V
| | | - \\ - \\ - \\ - \\ -\\
| '-' | | | | |
| | | | | | |
+-+------+-------+--------+--------+-------+-------+--- etc
|
|
GND

....
I enhanced your design just a bit with another pnp trannie, so it can
handle more LED strings if desired.

....

Same question for you. In the design above, what ensures that the current
across the LED's is constant and does not change when the 12v moves,
possibly as high as 14v?

 

[John Popelish]

In the design you propose, what limits the voltage fluctuation and thus
the current change? Say if the 12v is suddenly 14v for a moment. The
point of the LM317 was it's ability to maintain a set output voltage
regardless of the input.

Thanks for the info on the devices

The current through the 1k resistor has to be more than enough to
supply all the bases of all the LED current sources. As soon as the
current through the left one causes a .6 volt drop across its 33 ohm
emitter resistor, the extra transistor on the left comes on and sucks
up all the excess current from the 1k. This holds the base voltage
for all the current regulators at that which will cause a .6 volt drop
across each of their emitter resistors. 0.6/33=18.2 ma of emitter
current. The exact current through the 1k resistor is not very
important, since only a fraction of it is expected to involve the
current regulator base currents. If the regulator transistors have a
gain of 50, then 5 of them would need about 5*(0.018/50)=0.0018 amps
(1.8 milliamps). At 14 volts, there will be about
(14-1.2)/1000=0.0128amps (12.8 milliamps) passing through the 1k
resistor, so the assumption that only a fraction of its current goes
to base drive is safe.

This circuit is a little temperature sensitive, because the turn on
voltage of the left transistor goes down as temperature rises, but
this is a good idea for LEDs that also need current derating as
temperature rises.

 

[Mike H]

....

The current through the 1k resistor has to be more than enough to
supply all the bases of all the LED current sources. As soon as the
current through the left one causes a .6 volt drop across its 33 ohm
emitter resistor, the extra transistor on the left comes on and sucks
up all the excess current from the 1k. This holds the base voltage
for all the current regulators at that which will cause a .6 volt drop
across each of their emitter resistors. 0.6/33=18.2 ma of emitter
current. The exact current through the 1k resistor is not very
important, since only a fraction of it is expected to involve the
current regulator base currents. If the regulator transistors have a
gain of 50, then 5 of them would need about 5*(0.018/50)=0.0018 amps
(1.8 milliamps). At 14 volts, there will be about
(14-1.2)/1000=0.0128amps (12.8 milliamps) passing through the 1k
resistor, so the assumption that only a fraction of its current goes
to base drive is safe.

This circuit is a little temperature sensitive, because the turn on
voltage of the left transistor goes down as temperature rises, but
this is a good idea for LEDs that also need current derating as
temperature rises.

I think I understand. I'm trying to model this in Linear Technology's
free LTspice tool but their LED Model doesn't fit with the ones I'm
looking at so the values don't work out quite right.

Am I correct then that to change the current flowing across the LED's, I
need to change the 33ohm resistors?

I'm shooting for 20ma across the LED's. But I may want to shoot for lower
and then have a switch trigger an additional increase in current to give a
moment of brighter light.

 

[John Popelish]

I think I understand. I'm trying to model this in Linear Technology's
free LTspice tool but their LED Model doesn't fit with the ones I'm
looking at so the values don't work out quite right.

Just use something like 10 1N4148 diodes, instead. As long as there
is still a volt or two dropped across the regulator transistors, the
current should be pretty independent of the supply voltage.

Am I correct then that to change the current flowing across the LED's, I
need to change the 33ohm resistors?
Yes.

I'm shooting for 20ma across the LED's. But I may want to shoot for lower
and then have a switch trigger an additional increase in current to give a
moment of brighter light.

When you have a good idea what the 2 current setpoints are, I might
have some suggestions no how to do the switch.

 

[Mike H]

When you have a good idea what the 2 current setpoints are, I might
have some suggestions no how to do the switch.

Thanks. I'll do that. I'm going to head over to Radio Shack here shortly
to get some parts to start trying things out and prototype.

Mike

 

[Rich Grise]

Thanks. I'll do that. I'm going to head over to Radio Shack here shortly
to get some parts to start trying things out and prototype.

There was talk once of an LED dimming when a cap is discharged through
it. It does. But this particular one, an ordinary RS red LED, dropped
about 2V or whatever it's supposed to be, and from 20-10 mA, it stayed
just as bright to the eye as at 20 - from 10 it seemed to dim linearly,
smoothly down to imperceptibility, and the current tracked right along.

But I wouldn't depend on a current set point for bright/dim - I'd use
PWM.

Cheers!
Rich

 

[Mike H]

There was talk once of an LED dimming when a cap is discharged through
it. It does. But this particular one, an ordinary RS red LED, dropped
about 2V or whatever it's supposed to be, and from 20-10 mA, it stayed
just as bright to the eye as at 20 - from 10 it seemed to dim
linearly, smoothly down to imperceptibility, and the current tracked
right along.

But I wouldn't depend on a current set point for bright/dim - I'd use
PWM.

Not having looked into it yet my thought is that there may be some
architectural differences in LED that can impact their behavior when
driving them outside their specifications.

Somehow I feel that pulse wave modulation would be a much more complex
design to create. Not that I woundn't mind giving it a try, but I need to
walk before I run.

Thanks.