New Constant Current LED Driver I Promised

[MrAl]

Hello there,

Here's the new constant current LDO LED driver circuit i
promised. You can adjust the output current by
changing R3. The dropout voltage is very low, always
less then 0.1 volt.

The efficiency is very high for three batteries in
series (4.5v max).

When driving from higher voltages you will have to
pay attention to power dissipation in the transistor.

When driving the 5watt LS at 700ma you should also
pay close attention to the power dissipation in the
transistor.

The turn on transient overshoot is about 500ma for 10us
max with a 6v input which should be ok for the LS.

The circuit supplies a constant current to the LS
over a range of voltages. The lowest voltage to
maintain perfect regulation is less then 0.1 volt
above the LS's normal voltage drop. This means if
your LS drops 3.2 volts at 350ma then you will still
observe full output current even at a voltage as low
as 3.3 volts.

Here's the schematic:
http://hometown.aol.com/xaxo/page2.html


Good luck with your LED circuits,
Al

 

[Doug Owen]

Al,

I don't understand the note about the switches, can you please explain it?

TIA

Doug Owen

 

[MrAl]

Hi there Doug,

Yes, you can use either switch, you dont need both.
If you have a high current switch available, then
use that and short out the low current switch.
If you have a low current switch available, use that
and short out the high current switch.

The other nice thing about this circuit is if you want
to use a very small low current switch you can do that
by shorting out the other switch and just using the
low current switch to turn on and off with.

Good luck with your LED circuits,
Al

 

[TripleDouble]

Looks great! One question though, how does the efficiency of this circuit compare w/ the ZLT? Thanks for all the good info.

 

[Doug Owen]

Thanks, Al, makes perfect sense.....now.

Somehow, my simple mind (and cheap nature) likes your LN334 circuit better.

Thanks for explaining.

Doug Owen

 

[MrAl]

Hello again,

TripleDouble:
The efficiency compares very well at starting voltages
less then about 4.5v or so. Above that, the efficiency
starts to drop off because the efficiency is lower at
higher voltages. The efficiency for set voltage levels
(like for example a car 12v battery) is going to be
very poor because the voltage never drops any lower.
The reason to use this kind of circuit is mainly for
battery operation with hand held equipment or when you
want to insure that the load always sees a set current
level. After all, this is really a precision current
reference where you can vary the level by changing R3.
Of course if you have an input source less then 3.3v
you wouldnt be able to use this circuit either; you
would have to use a switcher.

Doug:
The parts used in this new circuit are relatively cheap,
like the LM358 which is also widely available. I think
they run about 50 cents each. The reference diode might
be a little more though.

If you want to use a switcher like the ZLT with current
regulation, the newest circuit now will be the ZLT420
circuit which i hope to post soon also. This new circuit
will provide the same efficiency as the ZLT with the same
benefits but will have only one additional resistor to
sense current. The circuit will provide perfect current
regulation so once set up, it will run with any LS you
decide to hook up on the output without the need for
adjustment.
The circuit will use the new Zetex 420 chip. The data
sheet is also available on the Zetex site.


Good luck with your LED circuits,
Al

 

[Jonathan]

[ QUOTE ]
MrAl said:
Hello again,

TripleDouble:
The efficiency compares very well at starting voltages
less then about 4.5v or so. Above that, the efficiency
starts to drop off because the efficiency is lower at
higher voltages. The efficiency for set voltage levels
(like for example a car 12v battery) is going to be
very poor because the voltage never drops any lower.


[/ QUOTE ]

It looks like the LM358 could deal with supply voltages well above 12V, so this circuit could work in a car. As you note the efficiency will be poor since Vin is much greater than Vout, and lots of power is being dissipated in the pass transistor. However it would work quite well and quite efficiently for running several LEDs in series; eg. 3 white Luxeons in series from an automotive supply would be better than 70% efficient.

Al, always love the designs that you post.

-Jon

 

[MrAl]

Hello there Jon,

I hadnt thought about the series connections like that.
If someone wanted to use three of them in series that
would work very well as you said.

I also meant to post a chart of power dissipation in the
transistor as well as efficiency for various voltages from
maybe 3.3v to 14v. I guess it would be a good idea to
chart series combos as well. The transistor in the original
circuit can be upgraded to a higher power unit if needed
without any changes too.

Take care for now,
Al

 

[MrAl]

Hello again,

Here are the charts for power dissipation in the transistor
with various LS connections and constant input voltages.
This applies for any cc linear regulator as well.

V_PS is the voltage of the power supply.
p11 is the power with one 1 watt LS connected.
p15 is the power with one 5 watt LS connected.
p21 is the power with two 1 watt LS's connected in series.
p31 is the power with three 1 watt LS's connected in series.

<font class="small">Code:</font><hr /><pre>
SINGLE 1 WATT AND 5 WATT LS:
V_PS=3.3 p11=0.035
V_PS=4.3 p11=0.385
V_PS=5.3 p11=0.735
V_PS=6.3 p11=1.085
V_PS=7.3 p11=1.435 p15=0.63
V_PS=8.3 p11=1.785 p15=1.33
V_PS=9.3 p11=2.135 p15=2.03
V_PS=10.3 p11=2.485 p15=2.73
V_PS=11.3 p11=2.835 p15=3.43
V_PS=12.3 p11=3.185 p15=4.13
V_PS=13.3 p11=3.535 p15=4.83
V_PS=14.3 p11=3.885 p15=5.53



SERIES CONNECTED 1 WATT LS's:
V_PS=7.3 p21=0.315
V_PS=8.3 p21=0.665
V_PS=9.3 p21=1.015
V_PS=10.3 p21=1.365 p31=0.245
V_PS=11.3 p21=1.715 p31=0.595
V_PS=12.3 p21=2.065 p31=0.945
V_PS=13.3 p21=2.415 p31=1.295
V_PS=14.3 p21=2.765 p31=1.645
</pre><hr />

Good luck with your LED circuits,
Al

 

[papasan]

the power dissapation is rated in watts, right? so with a 1W LS and a 6V source you're only getting 50% eff?

any plans for a simple switching/digital regulator? something where the transistor is either full on or full off and sinking less heat? willie hunt's switching regulators claim 95%+ although they are voltage based not current.

by the way, i don't want to sound critical. it's good work, better than i could do for sure =). just voicing my thoughts.

 

[Jonathan]

There are a several switching designs out there. I've posted a boost converter design at http://borealis.com/~winnie/LED_BOOST/index.htm , and I am working on a simple modification that makes it into a buck converter.

However switchers are more costly, are electrically noisy, and are not more efficient when the input voltage closely matches the output voltage. Linears can also easily be built with over-rated components to tolerate very noisy supply lines. So this discussion of simple linear supplies for the LS is quite useful.

I'm planning on using a linear regulator to run 3 luxeons in series in an automotive dome light. The efficiency when the alternator is running will only be 60% or so, but just on the 12V battery the efficiency jumps to 80% because of the lower input voltage and thus lower loss in the pass transistor.

-Jon

 

[MrAl]

Hello again,

papasan:
Hello there; you are right about the poor eff of the
circuit with a 6v input supply...so are the con's of the
linear regulator, but you also have to look at the pro's:
when the circuit is used with a single LS and a battery
that starts out at 4.2v and drains down to 3.3v you get
perfectly constant current though the LS and an AVERAGE
efficiency of 85 percent! Pretty good i think. This
is what you get using a 3.6v lithium rechargable battery
and that was the target battery type that prompted the
design in the first place.

Also, with a 6v battery you get about 69 percent AVERAGE
efficiency because the battery voltage continues to drop
while the circuit is running.

Using three NiCd's in series for a nominal 3.6 volts, the
average efficiency runs about 89 percent! That's because
the voltage is very close to the nominal 3.2v of the
single white LS voltage drop. The circuit continues
to run at voltages as low as 3.3v because of the special
transistor. If you decide to replace it with a run of
the mill type it wont work that well at all. The very low
Vsat is very important here.

The above is one of the reasons i posted the power diss
for various inputs, and also so people can figure out
how much of a heatsink they need and even if a higher
powered transistor is needed. Im glad you noticed
that; i was prompted to post clearer details.

On the downside, you can easily see that running on a
constant 12v supply will be even much worse :-)

If you notice anything else please feel free to voice it too.

Jon:
That's a good point, about the voltage dropping from 14 to
12v. That makes the efficiency look even better :-)


Good luck to you guys with this circuit,
Al

 

[MrAl]

[ QUOTE ]
papasan said:

any plans for a simple switching/digital regulator? something where the transistor is either full on or full off and sinking less heat?


[/ QUOTE ]
Yes, the new Zetex 420 circuit. Should be able to post it
very soon now.

[ QUOTE ]

willie hunt's switching regulators claim 95%+ although they are voltage based not current.


[/ QUOTE ]
I tend to not believe claims of eff above 90%. There are
too many factors that eat up power and there is a lot
of misunderstanding about pulsing regulators that dont
have inductors or capacitors as energy storage elements.
If you have a url i'll take a look too.

Good luck with your LED circuits,
Al

 

[papasan]

http://www.cs.indiana.edu/~willie/lvr.html

if you look down at the feature/spec chart he actually claims 99.8% eff.

i do believe the light bulb itself is used as a power storage element, something you couldn't do with LEDs. still, a simple switching regulaor should get better eff. then a linear regulator unless the voltages are real close. and the linear reg. drops a bit of voltage, something the switching should be able to get by.

so the new zetex circuit will be a buck/boost or just a buck? i haven't had a look at the data sheet.

 

[tonyb]

Zetex 420-Boost Only!

 

[MrAl]

Hello there,

papasan:
The circuit in that url is used with a bulb, not an LED.
Bulbs act differently then LED's in pulsing circuits.

It is quite possible to see high efficiency with a circuit
regulating the bulb voltage, but only if that voltage is
always maintained at the 'design' voltage of the bulb.
In other words, dimming the bulb eats up efficiency, not
because of the circuit itself, but because the bulb eff
drops off as the voltage is reduced. The effect can be
quite dramatic: for a bulb reduced to half light output
the power consumption only drops by about 25%, meaning
the bulb itself (not the circuit) looks like it's only
66% efficient.

Zetex 420 circuit:
Yes, this circuit will be boost only, at least for now.


Good luck with your LED circuits,
Al

 

[papasan]

yes, i realize all of this...that's why i stated that it was a voltage based bulb circuit. i'll ask again, do you have any plans for a simple current based switching buck regulator? or is this just complete non-feesable and if i want a buck circuit for LEDs then the only practicle way to go would be a linear regulator?

i was under the assumption that a 420 circuit would be boost only, that is why i'm wondering were all this boost talk came from in a buck thread?

 

[MrAl]

Hello again papasan,

I dont have any plans for the buck converter just yet,
but perhaps in the future.

When you say 'buck thread' you dont mean this one do you?
I dont really consider a constant current regulator
a buck type. I consider a buck type one that uses a
switching circuit with energy storage inductor for
converting a higher voltage to a lower voltage.

Take care for now,
Al

 

[papasan]

when i think buck i think of a downward voltage regulating circuit, when i think boost i think of an upward regulating circuit. since i'm pretty much completely self-taught these definitions are probably off a bit.

i personally would classify the linear regulator that is discussed here as a buck circuit although it's non-switching with no coil its output voltage is lower then the input voltage hence my designation.

 

[NewsFlash]

MrAl,

Pardon my ignorance, (some days I do well to find the batteries at Wal-Mart!), but just what is a 1.25v reference diode? I tried the part search at Digikey's website with no results. I am new to regulation as far as Luxeon Stars go, but am convinced the benefits outweigh the drawbacks (if there are any!).

Also, which version of the LM358 did you use (and which manufacturer)? At this point I'm guessing it doesn't really matter.

I like the idea of having a higher starting voltage circuit (buck?) than a lower starting voltage circuit (boost?).

Thanks for any help you can provide. /ubbthreads/images/graemlins/help.gif