Zener + transistor to regulate voltage

Josey

Flashlight Enthusiast
Joined
Jan 5, 2004
Messages
1,015
Location
NW Rainforest
I'm building a 12V multi-LED light and need to control the voltage so I don't overdrive the bulbs. Most people simply use a resistor, but that consumes energy. I'm off-grid and have only solar power, so efficiency is important, at least in the winter. Plus, my household voltage varies from 12.2V to 14.5V during the course of the day, so no resistor value will be correct for all voltage conditions.

I used to know a simple circuit that uses a zener diode and a transistor to control voltage without consuming much energy and without much reguard to small changes in input voltage. But I can't remember it and can't find my old notes.

Can anybody help? Thanks.

Josey
 

Doug Owen

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
1,992
In general, using a transistor (and related stuff) as a substitute for a resistor doesn't change the fact that "energy is being wasted". The trick is to use as much as possible to make light (maximize efficiency).

In general, you should regulate *current* (not voltage) in LEDs. It's easily possible to run strings of 3 LEDs in series (I assume you're using white?) on a single regulator. You can also use parallel strings for more light. Such regulators will compensate for changing battery voltage, giving constant light output for all within range inputs.

What is it you want to do?

Doug Owen
 

Josey

Flashlight Enthusiast
Joined
Jan 5, 2004
Messages
1,015
Location
NW Rainforest
Thanks, Doug:

I guess I could use a regulator to maintain constant voltage, it is just more expensive and consumes more energy than a zener/transitor. My understanding is that if I overshoot the rated voltage of the 3-bulb series string, I'll also increase current and shorten the life of the bulbs.

I had purchased some 36-LED lights that used only a resistor (for each 3 LEDs in series)to control voltage, but the factory had made a mistake, installing 62 ohm resistors instead of 82 ohm resistors. The result was LEDs reaching 200 degrees F.

The teacher who showed me the circuit said it was used by NASA on space flights to conserve energy.

Josey
 

HarryN

Flashlight Enthusiast
Joined
Jan 22, 2004
Messages
3,976
Location
Pleasanton (Bay Area), CA, USA
HI - I am not sure how much current you need, but I recently purchased a simple board from cpf'r georges80. He can send you one with any constant current setup you want from small (say 50 ma) up to 1 amp. His setup can take up to 35 volts input voltage.

I think he also has one model with dimming. I paid less than $ 20 for the board - around 1x 1 inch.
 

bikeNomad

Newly Enlightened
Joined
Sep 8, 2001
Messages
166
Location
Stanwood WA
[ QUOTE ]
Josey said:
I guess I could use a regulator to maintain constant voltage, it is just more expensive and consumes more energy than a zener/transitor.

[/ QUOTE ]

For the same amount of power going to the LEDs, a zener/transistor combination will require more power than a linear IC regulator. If the voltage is enough higher than the LED requirement (which it sounds like it is), then a switching regulator will be more efficient than either of those choices.

At a fixed voltage, a plain resistor actually takes less power than a zener/transistor that allows the same current to flow through the LEDs (because of the required bias current through the Zener, and the base current through the transistor if it's a BJT).

[ QUOTE ]
My understanding is that if I overshoot the rated voltage of the 3-bulb series string, I'll also increase current and shorten the life of the bulbs

[/ QUOTE ]

Actually, you have to watch the maximum current, not the voltage. You can't really regulate the voltage to the LEDs very successfully.

[ QUOTE ]
The teacher who showed me the circuit said it was used by NASA on space flights to conserve energy

[/ QUOTE ]

That's unlikely for any zener/transistor combinations I can think of, unless the input voltage is very close to what the LEDs will be running at.

Look at it this way: three typical white LEDs might have a forward voltage of (say) 10V at your desired current. Let's say that current is 20mA, for example.

Now your input voltage ranges from 12.2 to 14.5V, so a perfect linear regulator
(i.e. one which takes no power to run itself) would be wasting between (12.2-10)/12.2 and (14.5-10)/14.5 of the incoming power, or between 18% and 31%.

So the linear regulator, no matter how you make it (zener/transistor, IC, whatever) is at best 82% efficient in this application, and at worst only 69% efficient.

Typical step-down switching regulators have efficiencies between 85% and 95%, for a <font color="red">reduction of between 3% and 26% in the input power</font> (and energy) required to run these LEDs at the same brightness.

Whether it's worth it to you to spend the money on a step-down switching regulator is your choice, but you should do the math.
 

Josey

Flashlight Enthusiast
Joined
Jan 5, 2004
Messages
1,015
Location
NW Rainforest
Thanks HarryN and BikerNomad.

I tried a board like you mentioned and it failed. And I bought a commercial 12V LED light, but it was mismanufactured, and I've been waiting since the first of the year for the correct light.

I'll do some more work like you suggested.

Thanks. Josey
 

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,144
Location
New Jersey
Hi there Josey,

Using a resistor and three LEDs in series,
your average efficiency is about 75 percent.
Using a general switching regulator you'll get
about 85 percent, with a carefully designed one
you might see 95 percent.

What does this mean in terms of light performance?

Normalizing to 75 percent efficiency:

If your lights normally run for 5 hours:
Going to 85% will give you 30 minutes more run time.
Going to 95% will give you 1 hour more run time.

If your lights normally run for 10 hours:
Going to 85% will give you 1 hour more run time.
Going to 95% will give you 2 hours more run time.

If your lights normally run for 20 hours:
Going to 85% will give you 2 hour more run time.
Going to 95% will give you 4 hours more run time.

This can help you decide whether or not to go to
a switching regulator or not, and what type you'll
want.

Take care,
Al
 

Doug Owen

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
1,992
As I said before, you'd be better advised to try to regulate *current* not voltage.

And as another poster has pointed out, there's little if any advantage of resistors over a regulator in terms of efficiency.

If I understand your problem with the light you now own, it can easily be 'solved' with the addition of a series resistor to make up for the 'missing ohms'. In this case, you want 20/12 ohms or 1.7 ohms. It will need to be fairly high power, nearly a watt. Say six ten ohm, 1/4 watt resistors in parallel?

That's still a lot of current per LED (60 mA), you may want even more series resistance.

Doug Owen
 

Harrkev

Enlightened
Joined
Aug 30, 2001
Messages
443
Location
Colorado Springs, CO
Let me summarize the three main options:

1) Resistor. Wastes some energy. There is a trade-off. With a lot of LEDs in series and a large resistor, you get fairly constant brightness over a large voltage range, but the efficiency is poor. If you put as many LEDs as you can fit in series and use a small resistor, then you get better efficienty, but the brightness more with voltage.

2) Linear regulator. Think of this as a "smart resistor." It is a resistor that adjusts it resistance to maintain the proper voltage or current. All you need it one adjustable regulator and two resistors to make this work. These need at least a volt or so of overhead in order to work, unless you get the LDO (low drop-out) version, but LDOs need caps and proper board layout to keep from oscillating.

3) Switching regulator. This gives you be absolute best efficienty (if done properly), but is somewhat of a black art, and is not an ideal project for a beginner.

If you don't mind spending some money for something pre-made, check this out: http://www.techass.com/el/versalux/ulm/ulm.php
 

Josey

Flashlight Enthusiast
Joined
Jan 5, 2004
Messages
1,015
Location
NW Rainforest
Thanks MrAl, Doug and Harrkey:

I do hear you Doug, about current regulation. And Harrkey your points are great. I need at least 20 LEDs for a good reading light, so the cost at $50+/10LED cluster is steep.

I'll take all of your advice and build a couple diffent types and measure the efficieny; I'll probably end up with results like what MrAl said.

Thanks again. Josey
 

Doug Owen

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
1,992
Josey,

I'm not so sure you need 20 LEDs. I've been working on a reading/room light using 12 in 4 strings of 3 (with a 10 ohm resistor each) in parallel. After some fiddling, I ended up grinding the round end off each lens (dressing it with a file afterwards), the result is wide angle artifact free light entirely suitable for reading or other close in tasks. This is at 100 mA total (25 mA per LED). The LEDs and parts for the (three level) regulator cost about $6, more or less.

BTW, the cicuit you couldn't recall is an emitter follower. It was neither 'invented by NASA' (that was the power factor controller, for AC not DC) nor really useful here.

Anyway, send me a PM, I'll email you a couple of photos.

Doug Owen
 

Doug Owen

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
1,992
[ QUOTE ]
Josey said:
Thanks Doug:

Sounds like you created a bunch of sawed-off photon shotguns.


[/ QUOTE ]

Interesting perspective.

I'm sure after you work with this a bit you'll come to the same conclusion I did WRT beam angle. That is for reading or general lighting at close ranges we want a very wide beam angle and flat, artifact free illumination. If you look at the beam from the Attitude you can clearly see that each of the three LEDs contributes to the total. Looking at the SL 7 LED light tells us that overlapping light patterns can mask the artifacts from an individual device. This means you don't have to be too careful when you cut them back, small errors don't matter.

So, I like your analogy. In fact, I'll go you one better (at least IMO), it's a *battery* of 'sawed-off photon shotguns'. Gets back to one of my pet peeves, the way otherwise knowledgeable folks refer to a single cell as a 'battery'. For the record, one cell (like one cannon) is just that. A *battery* is one or more cells (or cannons) connected together and working as a single entity. A box of 123's from Surefire is a dozen *cells*, even in the box (since they're not connected and functioning as one), they are not a battery. One alone is *never* a battery. The Arc AAA uses an AAA cell, not 'AAA batteries'.

So that's my rant for the evening, care to share yours?

Doug Owen
 

idleprocess

Flashaholic
Joined
Feb 29, 2004
Messages
7,197
Location
decamped
A 9V battery is a proper battery. There's either a stack of raw flat cells encased in plastic, or 6 "AAAA" batteries.

Power supply design, in general, is a black art to those not in the know. There's a lot to account for - capacitance, reactance, getting the exact right oscillation, saturation points on transistors, etc - an a heck of a lot of it is more involved than Ohm's law.

The series resistors in-line with LEDs are almost universally referred to as "current-limiting" resistors, since most LEDs have a range of voltages that they can operate over. The forward voltagfe that's quoted is almost always typical. Drop below the minimum, and nothing will happen; exceed the maximum, and you'll release the magic smoke.

A current supply will generally attempt to supply a given current. V=I*R, so it will adjust the voltage (within limits) until it supplies the current it's happy with.

With your situation, you could almost use a simple voltage regulator (usually a largish 3-pin device, can't remember what the standard package is called) to regulate your fluctuating 12V to, say 9 or 6V. If you find a combination of LEDs whose voltage drop is just less then the rgeulator's output, you waste little other than the small drop across the resistor and the regulator's inefficiency (I confess that I don't know what the efficiency of a voltage regulator is).
 

Negeltu

Enlightened
Joined
Aug 28, 2003
Messages
724
Location
Oregon, USA
"A *battery* is ONE OR more cells (or cannons) connected together and working as a single entity"

One or More? Don't you mean More than One ?
 

Doug Owen

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
1,992
[ QUOTE ]
Harrkev said:
Instead of sawing off LEDs, you can just lightly sand the surface. Less work, and it still gives a very diffuse light.

[/ QUOTE ]

Yup. "sawed-off" was not my term, it actually refereed to shotguns when Josey used it.

You'll note I said 'grinding', actually I used a small power sander, as a grinder would quickly foul with debris. I dressed the surface with a file (a worn 6 inch mill ******* for those taking notes) afterwards. As you suggest, the surface need not be perfect, in fact there seems to be an advantage to it not being so. The ones I polished didn't work as well in my modest testing.

As it is, it takes only a few minutes to do the dozen needed for each array.

Doug Owen
 

Doug Owen

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
1,992
[ QUOTE ]
Negeltu said:
"A *battery* is ONE OR more cells (or cannons) connected together and working as a single entity"

One or More? Don't you mean More than One ?

[/ QUOTE ]

Exactly right, thank for catching it.

See what happens when you rant?

OTOH, it's nice to know I'm being read......

Doug Owen
 

Doug Owen

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
1,992
[ QUOTE ]
idleprocess said:
A 9V battery is a proper battery. There's either a stack of raw flat cells encased in plastic, or 6 "AAAA" batteries.



[/ QUOTE ]

Here I would have to disagree. We're starting out at a nominal 12 Volts, easily capable of running three LEDs in series, and you're suggesting regulating it down to 9 Volts, which while 'doable' will only drive two LEDs. This automatically means this system will use 50% more current (3 strings needed at 9 Volts to do what two at 12 could have done).

I suggest four strings of three at 25 mA per string for a total 100 mA from the source. You propose six strings of two, for a total current of 150 mA for the same light.

I think the solution lies in a driver that can take full advantage of the available battery. "Giving away" a third of the power just getting started seems like a bad way to go. I favor a current regulator, running straight off the battery driving an array suited to the task.

Doug Owen
 

MrAl

Flashlight Enthusiast
Joined
Sep 9, 2001
Messages
3,144
Location
New Jersey
Hi there,


I've always said that a current regulator is better
for LED's too. There is no advantage to using
a voltage regulator over a current regulator that i
know of. The LED is happy with a certain level of
current maintained regardless of it's nominal voltage,
so we regulate the current. The LED is not very happy
with a certain voltage across it maintained, as the
current will vary with ambient temperature, so we dont
want to regulate it's voltage. The LED is not that happy
with a small resistance in series with a regulated voltage
either, because again variations in LED voltage with
temperature will cause the current to vary (think of a
small resistance in series with a voltage regulator still
a voltage regulator, while a large resistance looks more
like a constant current source).

Sometimes it's nice to regulate current while not
having a voltage overhead that is much higher then
the LED nominal voltage. This means the battery can
run pretty low before the LED current starts to drop.
In this case a circuit that doesnt require any overhead is
used like this one:

http://hometown.aol.com/xaxo/page2.html

The actual overhead voltage is something like 35mv
for an LS 1 watter.

If you have some two or more volts to spare, the
usual LM317 circuit should work just fine. The
supply voltage will have to be at least 2 or so volts
above the LED voltage for this to work well however.
(Someone may wish to look up the required overhead for
the LM317).

In any case, if you're going with linear regulation,
go with current regulation over voltage regulation
for LEDs. You'll be happier knowing your LEDs are
much happier :)

Take care,
Al
 
Top