Hello, first time here.
I am trying to make a small wind turbine that will illuminate some LEDs when it's in motion.
The generator makes plenty of power from maybe 50 rpm, but power output rises as the wind increases, and overpowers the LEDs. The range is 0 to 8 volts or so: I have not tested the amperage, but presume it to be quite low as the generator windings are 32 ga. If I wire the LEDs to run at 8 volts, I only see them light up in hurricane winds. I really want them on as soon as possible, but protected from higher voltages as the wind speed changes.
i figured a voltage regulator would be the solution, so I added an adjustible unit and used a pot on the adjustable leg to trim the high-speed output. This circuit works great to prevent anything over 2.5 volts from reaching my LEDs, but it brought another issue...it chokes off the low-speed output from my generator.
Now the generator output is ok going INTO the regulator, but much lower coming out, even at levels lower than I have the regulator set at. I was hoping that the regulator would let voltage pass through unchanged until it got to my pre-set max and then begin shunting.

I'm hoping that someone here has the answer. I'm new to electronics and it's easy for me to overlook what may be a simple fix.
Here's how I wired the regulator: I tried different values of resistor (1Kohm and 250 ohm ) and got the same results. To the left of the image is my generator, and to the right are my LEDs.
Thanks for looking

 

Hi Wallaby, and welcome to the Forums.

I have some kind of sad news for you. The LM317 has what is known as a "dropout voltage", which is the difference from the input to the output, of 1.7v. There's just no getting around it with that regulator. So, where your LEDs used to start lighting up when your generator's output got up to 2.5v, now it has to be at least 1.7v higher, or 4.2v, before they will be the same intensity as before.

Another thing is that LEDs should be regulated by the current they get instead of the voltage across them. If you want to use a voltage regulator, that's OK - but you also need to have a resistor in series with each LED to limit the current. This will use some power, and it will also require a voltage drop of at least 0.5v in order to work properly.

But, you haven't told us about your LEDs - what are their specifications (I'm looking for the typical Vf @ rated current here, or a datasheet), and how many are you operating?

 

Yes, voltage regulators of this kind have a fairly substantial "drop-out voltage" i.e. minimum output lost from input to output. See graph on page 6 of data-sheet for LM317, even at only 20mA at 25°C this is typically 1.5V, and may be worse. http://www.national.com/ds/LM/LM117.pdf

There is no easy fix for this in my opinion. You would need some other kind of regulator with less voltage drop-out if you do not want to lose the low-speed output. Perhaps a lower drop-out current regulator would be better than a voltage regulator if all you want is to light a few LEDs.

Edit: SgtWookie has beaten my geriatric typing!

 

Alright...thanks for the help so far.
I have seen other discussions about using resistors with every LED to regulate the power that passes through them, but those have all been dealing with a constant voltage source. I fear that if I use the resistors this way, they will also serve to limit the low voltage performance I am after.

To answer your questions, I now have 6 LEDS...I could get by with 4 if that would help. (I thought LEDs used very little power). Anyway, the LEDs I have show:
Power dissipation: 60mW
forward current: 28mA max
forward voltage @ 10mA: 2.6 volts max

I'm probably trying to engineer this project backwards...I needed both red and green LEDs, and chose these because both colors had similar ratings. I then re-wound my generator to provide what they needed...then rewound again to get better low speed performance and then had to add a regulator...LOL

 

With the LM317 being a linear voltage regulator you have an inherent loss of several volts,
as much as 3V. This means that you need at least 6V minimum from your solar cells before you will turn on your LED's. Thus if your solar panel varies between 0V and 8V you don't have much head room. Ultimately you want to regulate the current to the LED(s). I am thinking of the TL431A You might want to read the data sheet.

 

I would try using a high gain transistor as a current regulator. This will require a stable reference voltage at the base using 2 diodes in series connected from the base to the output. A resistor connected fom the emitter to the output. The resistor should be of a value to have a voltage of .6 v. at the required output current. Connect a 2 K resistor from the collector to the base to supply current to the biasing diodes. I think this might work down to about 2 volts, then the biasing
string will be shunting the regulator to be the only current supply to your L.E.D.
load.

 

I wonder if a bargraph display with an increasing number of LEDs wouldn't be more fun. I haven't built such a circuit but there are many threads here on, for instance the LM3915.

 

That bargraph driver seems to have what I need...at least it is engineered to light up the LEDs over a wide range of input voltage.

Thanks for the thoughts so far. I am not familliar with transistors, so I can't really visuallize how that circuit would work, but so far I'm keeping up with everything else.

 

I see that even with the correct part, there are external resistors and caps needed to complete the job...so there is calculations and experimentation left in the mix....I guess the puzzle is kinda fun in a sick sort of way.
Ok, here comes the cascade of follow-up questions:
I have reviewed the data sheets for the suggested parts and still have questions. With the http://www.datasheetcatalog.com/datasheets_pdf/T/L/4/3/TL431A.shtml what is the dropout voltage? I don't see that anywhere. this is the value I'm trying to reduce, right?

Now about the option of using the bargraph driver http://www.national.com/mpf/LM/LM3915.html#Overview ,
would it be possible to connect 4 LEDs to pin #1 for example?, and leave the other pins unused? Or would I have to use multiple drivers to acheive this result?

 

Ok, here comes the cascade of follow-up questions:
I have reviewed the data sheets for the suggested parts and still have questions. With the http://www.datasheetcatalog.com/datasheets_pdf/T/L/4/3/TL431A.shtml what is the dropout voltage? I don't see that anywhere. this is the value I'm trying to reduce, right?

Since it is not a "voltage regulator" this part really doesn't have a "Drop Out Voltage" spec in it's datasheet.

Now about the option of using the bargraph driver http://www.national.com/mpf/LM/LM3915.html#Overview ,
would it be possible to connect 4 LEDs to pin #1 for example?, and leave the other pins unused? Or would I have to use multiple drivers to achieve this result?

The typical current per output pin on the LM3915 is 10mA, which is about 50% less than most modern run-of-the-mill LED's thus the LED's brightness will be cut from the get-go. placing two or more LED's in series on one output pin would not change the brightness, but would require 4x the voltage, which your 8V solar PV can not handle (12V). Placing 4 LED's in parallel off of the first output pin may require only 3V for the forward voltage requirements, but would only supply 2.5mA to each LED...barely visible. if that. Using multiple 3915's will just increase your total power requirements, which even at 4V(PV output) may not be enough to turn on your LED's.

NOTE Added: The output current for the 3915 is actually programmable from 1mA to 30mA, thus two resistors in series drawing 15mA each would not be bad, but the voltage drop would be near 6V, a voltage you may not get for long during the course of the day.

Talk about a quagmire ehh!

Now here's an interesting twist on the LED current regulation that you seek. It is a current regulation problem more than a voltage regulation issue that you need.

What you need is an increasing resistance with increasing voltage, one that will keep the current through the LED's within the operating specs.

I googled "voltage controlled resistance" and did get some hits. One of which was Using a FET as a Voltage Controlled Resistor. The math on this page looks like a page from the NASA math laboratory.

See: Using a FET as a Voltage Controlled Resistor I know absolutely nothing about this concept, nor if it will serve your purposes or not. Perhaps some of the members who have an understanding of this concept will offer a brief bit theory for us.

I have also seen circuits, non of which I can recall at present, that use the LED's own brightness to send feedback to the circuit that in turn adjusts the current to the LED to stay within the operating specs. I saw this on EDN's website.
A micro controller could probably do this with a digital pot and maintain a linear voltage/resistance relationship.


One question from my end: You are using a PV cell to light up LED's...what do you need the light for in the daylight?

 

Well, it's a wind turbine actualy. It started out as a whirlygig, but then I decided to have it DO something. I fashioned it into the shape of an airplane, and added a generator...now I want to add some "navigation" lights at the wingtips.

I guess I should really check to see what my generator output is. I know the voltage is enough, but never checked the amperage. I thought about using small incandesent flashlight bulbs as they seem to be more tolerant of voltage changes, but only have a life of 3 hrs!

I do have a question about the LEDs: do I need to be controlling both the voltage and amperage?

"Breathe deep, the gathering gloom...watch lights fade from every room"

 

@ Wallaby: For LEDs, the current is what needs to be controlled, and the voltage must be left to look after itself. One of the commonest mistakes is to try to run LEDs with a fixed voltage without a series resistance. Regulate the current, let the LED Vf do its own thing as long as you allow enough headroom for it.

@ Ionic:- This may not be so helpful to the OP, but I thought you might like to know this.

The "voltage controlled resistance" region of a FET output characteristic only extends to low voltages, below pinch-off. It's sometimes referred to as the "triode" region by analogy with the shape of a triode vacuum-tube characteristic. This region of the characteristic actually has low output resistance and won't simulate a constant current, except perhaps in a closed loop with a current feedback amplifier.

The FETs inherently constant-current output characteristic region is referred to as the "saturation " region - very confusingly for those of us who cut their teeth on BJTs, where saturation is synonymous with very low values of VCE. To add a bit more confusion, this is sometimes called the "pentode" region, again an old vacuum tube analogy. That does not mean that the device has miraculously grown a couple of extra electrodes, just that its output curve has flattened out at higher voltages.

http://graffiti.virgin.net/ljmayes.mal/comp/vcr.htm

 

Here is a simulation of a crude transistor current limiter. Upper curve is LED voltage, lower is LED current, both plotted against input voltage.

This is not a great circuit - the current level would vary quite a bit with temperature, and the minimum voltage drop is not that low - about 1.4V. It illustrates the idea though.

 

Alright...I guess my topic should be labeled "current regulation".

Would some gizmo like a BuckPuck be a place for me to start? I'm not sure what the drop-out voltage is for those. I see info in the data sheet that indicates an input voltage of 5v min, and 30v max, but is this device made to deal with a fluxuating input, or it has to be calibrated for a certain constant input between 5-30v?

Another consideration is the elimination of added resistance in the circuit. I didn't consider that. I see the data sheets say to keep things close as possible....all my LEDs are located 2 feet or more from the source (no way around that) but I only used 24ga magnet wire to try and keep the wiring less visible. I guess form should follow function.
Right now, I have the LEDs wired in parallel. I was thinking if one quit, the other would still work...would wiring them in series be a better choice for bringing down the power requirements?

 

Alright...I guess my topic should be labeled "current regulation".

Would some gizmo like a BuckPuck be a place for me to start? I'm not sure what the drop-out voltage is for those. I see info in the data sheet that indicates an input voltage of 5v min, and 30v max, but is this device made to deal with a fluxuating input, or it has to be calibrated for a certain constant input between 5-30v?

Another consideration is the elimination of added resistance in the circuit. I didn't consider that. I see the data sheets say to keep things close as possible....all my LEDs are located 2 feet or more from the source (no way around that) but I only used 24ga magnet wire to try and keep the wiring less visible. I guess form should follow function.
Right now, I have the LEDs wired in parallel. I was thinking if one quit, the other would still work...would wiring them in series be a better choice for bringing down the power requirements?

Regarding the power requirements... The same load means the same power requirements. P = VI thus in series you have 3 LED's dropping 3V at 20mA ((3V x 3) x .02A) = 180mW or for a parallel configuration you have 3V but 3 x 20mA (3V x (.02A x 3)) = 180mW.

The BuckPuck would work with the varying voltage input, but you need the current to drive both the BuckPuck and the LED's plus a minimum of 5V. There's not a lot of room (5V - 8V).

I think maybe a very LDO 3.3V Linear regulator with the LED's wired in parallel. This way if you have about 4V from the generator and sufficient current the LED's will be on. In series you may never see them turn on as you might require as much as 9V from a system you guessed could only produce 8V in hurricane winds.


2 IC's to consider:

LP2950CZ-3.0, T0-92
100mA, 3V, Low Power Low Dropout Voltage Regulator

LTC1515CS8-3/5
Step-Up/Step-Down Switched Capacitor DC/DC Converters with Reset

I'd probably favor the first one as it's input voltage max is not so close to your generator voltage.

 

Thank you. This unit looks like a direct replacement for my "inefficient" unit.

Now the question is if this regulates current?
I'm having trouble with deciphering all the tech info.
From what I can gather, I can choose a fixed 3v output...so I'd use a resistor at each LED to trim it to 2.5? Are there additional items I'd need external of this unit to make it work?

I'm starting to feel bad about all these stupid questions that take up your time and exhibit my ignorance. I'm trying to make this project work and learn something at the same time. I thought I was heroic by calculating and re-winding my generator...but it looks as if I have a long ways to go.

 

Thank you. This unit looks like a direct replacement for my "inefficient" unit.

Absolutely more efficient...

Now the question is if this regulates current?
I'm having trouble with deciphering all the tech info.
From what I can gather, I can choose a fixed 3v output...so I'd use a resistor at each LED to trim it to 2.5? Are there additional items I'd need external of this unit to make it work?

After you choose the fixed output of 3V you will manage the current through each LED with a series resistor. In this case where the voltage is so close to the Vf (Forward Voltage) of the LED a very small resister can be used...as small as 6.8 Ohms.

For the LP2950(T0-92 pkg.)
Something like this:

Thus, with such a low drop out voltage you might be able to see the LED's from 3.5V - 8V

I'm starting to feel bad about all these stupid questions that take up your time and exhibit my ignorance. I'm trying to make this project work and learn something at the same time. I thought I was heroic by calculating and re-winding my generator...but it looks as if I have a long ways to go.

If you think you feel stupid asking question, just think what it feel like when you make a stupid answer! Questions are born of ignorance with the desire to learn, answers on the other hand, wrong ones, are born out of arrogance and fermented in ignorance and wind up as a deception. ...and I've made my share here on AAC!



Another idea I had, to really confuse you is by using a Zener diode to regulate the voltage. This is where I make the "stupid answer", that I hope other will correct me on. A basic circuit is below.

 

Can you recommend a place to buy the LP2950CZ-3.0 ?
My local parts house carries the NTE brand, but their cross-over doesn't show a match.
NTE does show this as a low voltage regulator: http://www.nteinc.com/specs/1900to1999/pdf/nte1904.pdf

But alas, they don't stock that number in-house.

 

Can you recommend a place to buy the LP2950CZ-3.0 ?
My local parts house carries the NTE brand, but their cross-over doesn't show a match.
NTE does show this as a low voltage regulator: http://www.nteinc.com/specs/1900to1999/pdf/nte1904.pdf

But alas, they don't stock that number in-house.

Digikey- LP2950CZ-3.0-ND

.

 

Low drop-out linear regulators are great when the input voltage is close to the desired output voltage. Where you start to run into trouble is where the input voltage is significantly higher than the output, as more and more power is dissipated in the regulator itself.

For example, let's say you have 5 LEDs, each with a current of 20mA, so the regulator will need to pass 100mA to keep the LEDs happy. We'll say just for this example that the LEDs and their resistors need 3.0v across them to reach full brightness. Also, let's say that your input voltage starts at 4v, then increases to 8v.

In the beginning, you have 100mA * 3v = 300mW power total being dissipated in the LEDs, and about 100mA * (4v-3v) = 100mW power being dissipated in the regulator.

Then the input voltage goes to 8v.
You still have 100mA * 3v = 300mW power being dissipated in the LEDs, but now you have 100mA * (8v-3v) = 500mW power being dissipated in the regulator! 60% of the power is being used to heat up the regulator. This is also getting close to the maximum power dissipation allowed (625mW) for a TO92 package; if the input voltage gets much higher (9.25v) the regulator will either have to shut down or melt.

This is something that you simply have to be aware of with linear regulators; they get hot when they are required to have a large voltage across themselves while passing current.