High power adjustable current regulator

[yngndrw]

Hi,
I would like to build a quick and cheap adjustable current regulator for LEDs.

I have a LM317 here so it would be nice to be able to use this. The problem is that I want to be able to drive both 0.5W LEDs and 10W LEDs. (As well as anything in between.)

The 0.5W LED requires 150mA and has a voltage drop of aprox 4V.
The 10W LED requires 1.5A and has a voltage drop of aprox 8V.

Therefore I need a 0 -> 1.5A adjustable current source powered of a 12V supply to make things nice and easy.

I worked out the resistor required for both of these LEDs. The 0.5W LED requires around 8Ohms and the 10W LED requires around 0.8Ohms. The problem is that the resistor needs to be around 3Watt in order to cope with the 10W LED. (It comes out at 1.8W if I use V * I and 2.8W if I use I^2 / R.)

Obviously this would be unreasonable to expect from a variable resistor, so what can I do ?

Thanks,
-Andrew.

 

[chemelec]

Use an LM350

Hi,
I would like to build a quick and cheap adjustable current regulator for LEDs.

I have a LM317 here so it would be nice to be able to use this. The problem is that I want to be able to drive both 0.5W LEDs and 10W LEDs. (As well as anything in between.)

The 0.5W LED requires 150mA and has a voltage drop of aprox 4V.
The 10W LED requires 1.5A and has a voltage drop of aprox 8V.

Therefore I need a 0 -> 1.5A adjustable current source powered of a 12V supply to make things nice and easy.

I worked out the resistor required for both of these LEDs. The 0.5W LED requires around 8Ohms and the 10W LED requires around 0.8Ohms. The problem is that the resistor needs to be around 3Watt in order to cope with the 10W LED. (It comes out at 1.8W if I use V * I and 2.8W if I use I^2 / R.)

Obviously this would be unreasonable to expect from a variable resistor, so what can I do ?

Thanks,
-Andrew.

 

[crutschow]

It's I^2 * R giving 1.8W, the same as V*I.

Actually, you can buy low resistance rheostats that will handle that power and current but they tend to be expensive (>$20).

It would be cheaper to use fixed power resistors and switch them in and out of the circuit. Of course that only gives fixed values of current.

 

[crutschow]

Use an LM350

That won't help. It's the 1.2V reference voltage that causes the power loss in the sense resistor that's the problem.

 

[Roff]

Could you use something like this?

 

[yngndrw]

I always get the power formula mixed up when it has resistance in it, my bad.

I did want a continuously variable output so I think Roff's circuit could work well. It shows the LM350 but from the datasheet I think the LM317 would work just as well.

Shame that it needs an op-amp but oh well, would a 741 be fine ? I have no other use for the 741's that I have lying around here ..

How big would the heatsink need to be for a LM317 with in input voltage of 12V and an output voltage of 8V @ 1.5A ? Would those tiny clip on heatsinks work ?

Thanks for all of your help guys.

 

[Roff]

I always get the power formula mixed up when it has resistance in it, my bad.

I did want a continuously variable output so I think Roff's circuit could work well. It shows the LM350 but from the datasheet I think the LM317 would work just as well.

Shame that it needs an op-amp but oh well, would a 741 be fine ? I have no other use for the 741's that I have lying around here ..

How big would the heatsink need to be for a LM317 with in input voltage of 12V and an output voltage of 8V @ 1.5A ? Would those tiny clip on heatsinks work ?

Thanks for all of your help guys.

If you use an LM317, all the packages except TO-3 barely reach 1.5A. With 8V across the LED and 1.5V across the sense resistor, you will only have 2.5V across the regulator, which is below the spec. minimum of 3V. You could change the resistor to a lower value, like 0.1Ω. You would also have to change the pot range to 0-150mV.
You can't use a 741 because its inputs won't work at zero volts.
The thermal resistance of the TO-220 package is 4 deg C/watt. You have to add that to the thermal resistance of the heat sink to get the total thermal resistance. You need a maximum thermal resistance of (125-25)/6W≈17W.deg C. I would play it safe, and use (110-40)/6W, which means you would need a total thermal resistance of about 12 deg/W. Subtract 4 deg/W for the package, and you need a heat sink of 8 deg/W or less. You can go **broken link removed** to see some heat sinks. Click on the pictures to see the thermal resistance vs air flow curves.

 

[yngndrw]

From what I can see, my heatsink is about 21 deg/W so I'll have to find a better one. I guess that if I get a much better heatsink, the fact that I'd be running a LM317 to it's 1.5A limit shouldn't be a problem - It's for testing anyway.

The fact that the heatsink and op-amp needs to be bought is a shame, but on the other hand it means that I can get a more ideal heatsink and op-amp. I'll play with the resistances and put a regulated voltage reference in while I'm at it.

You can't use a 741 because its inputs won't work at zero volts.

By this do you mean that if I use a -Vs of 0V it would not work or do you mean something else ? What field in the datasheet specifies this as I cannot find anything about it.

Thank you very much for your help, I've learnt much.

 

[Roff]

In the Fairchild LM741 datasheet, see Input Voltage range in the specs section, and also see Fig. 2. Keep in mind that if the datasheet says the input range is ±12V for ±15V supplies, this really means that the inputs only work if they are 3 or more volts above the lower supply rail and below the upper supply rail..

 

[yngndrw]

Ah I see, the National datasheet which I was looking at doesn't seem to have that information. Thanks again.

 

[yngndrw]

Would the LM311 be fine ? It says in the datasheet that it can work off single ended supplies and as far as I can see it would be right. http://www.datasheetcatalog.org/datasheet2/c/0hk0a528fi1yzjwj5rxkj21i4wwy.pdf

The problem with the LM321 is that I can't get it where I wanted to source the parts (Maplin.) and it's not in my simulator.

I have put together a full circuit diagram including the new values for my application. Could you please check it for me ?

I also added a regulator for the voltage reference so that I can run it off whatever supply I have lying around. (Including 12V batteries.)

Thanks for your help.

 

[Roff]

Would the LM311 be fine ? It says in the datasheet that it can work off single ended supplies and as far as I can see it would be right. http://www.datasheetcatalog.org/datasheet2/c/0hk0a528fi1yzjwj5rxkj21i4wwy.pdf

The problem with the LM321 is that I can't get it where I wanted to source the parts (Maplin.) and it's not in my simulator.

I have put together a full circuit diagram including the new values for my application. Could you please check it for me ?

I also added a regulator for the voltage reference so that I can run it off whatever supply I have lying around. (Including 12V batteries.)

Thanks for your help.

Good job on your schematic, except LM311 is not an op amp, it's a comparator. if you try to use it, your circuit will probably oscillate. You can use LM324, which is the quad version of LM321. LM324 is very common. Maplin has it under the part number LM324N. If you use it, you need to connect all the unused noninverting inputs to ground, to disable the unused sections.

BTW, Maplin has a piss-poor search engine. It says LM324 is discontinued, but you can find it if you search for LM324N.

 

[Hero999]

Why use efficient LEDs with a power wasting linear regulator?

Go for a switching regulator, some of the more modern ICs are cheap and you can build easilly build an efficient unregulated switching supply using a 555 timer.

 

[yngndrw]

Am I right in saying that a comparator is different to an op-amp becuase the output of a comparator is either one rail or the other while the op-amp's output is analogue ?
Or by saying that a comparator is an op-amp with a very high gain ?

It would be a shame to waste three op-amps but it happens. Would it not be best to tie one input high and one input low for the unused op-amps, in order to make sure that they stay in one state ?

Maplin doesn't have the best site in the world and their range is poor. Then there's the prices .. It is however convenient which is why I use them.

Hero: The reason for this is just for testing. While I said it's for LEDs, it's actually for anything. I will also use it for laser diodes and such. I commonly find that I'd like to limit the current of things quickly and this seems the cheapest way.

Come to think of it, if my assumptions about comparators are correct - Could I not use a comparator and a MOSFET instead of the regulators so that it becomes a PWM chopper ? I'll probably keep it as it is, but was just wondering.

Cheers.

 

[Roff]

An op amp generally has internal frequency compensation to prevent oscillation when connected in a negative feedback loop. This compensation makes it work slowly as a comparator, and you generally have to add external circuitry to get the logic levels or output drive that you need.
A comparator is not compensated internally, because it is not intended to have negative feedback around it, so it generally switches faster than an op amp. Its output is also optimized to drive logic gates or some other type of switching device. Otherwise, they are similar, with high-impedance differential inputs low input bias currents, and high open loop gain.
You can use PWM for LED brightness control, and a sawtooth (or triangle) generator, a comparator, and a switching transistor (MOSFET or bipolar) is one way to do it.

 

[Hero999]

It's possible to use a comparator as an op-amp if you add a frequency compensation capacitor but it wouldn't be very good.

Use the LM358 or cheaper TS358 which comes in the same 8-pin packages as the LM311 if you don't want to waste a 14 pin DIL.
**broken link removed**

Maplin are generally pretty crapy, the only thing they're good for is convenience, there's one in my town which is a 10 miniutes drive for me.

 

[yngndrw]

Ah I see thanks for the explanation.

I might use the TS358, smaller and cheaper.

The circuit is a but counter-intuitive to me. The regulator will try and keep a 1.25V difference between the output and the adjust pin, however the regulator is being treated more like a transistor. I'm just wondering if the circuit would perform better with a transistor instead of the regulator and would also be cheaper ?

I don't use transistors much so forgive any errors. If you was to replace the regulator with a PNP and connect the base through a resistor to ground via a NPN. You then connect the NPN's base to the op-amp via another resistor. Would that perform the same job ?

I'm not going to mess with the circuit, I'll built the designed circuit as soon as I get a chance, I'm just wondering that's all.

 

[Hero999]

The only criticism I have with this design is that it requires a regulated supply.

You could easilly add a differiential amplifier so you use the LM317's referance for the current source at the cost of rour extra resistors (the other op-amp is already on the chip).

 

[yngndrw]

Can you explain how that would work. You're using the difference between the regulator's input and output as the reference, but I can't see why it would be constant.

I was just going to use a 78L05 for the reference.

 

[Hero999]

Sorry I made a mistake with the schematic and have now fixed it.

The original circuit used a potential divider taken off the power supply. This circuit uses a differential amplifier to tap into the reference voltage of the LM317.