Maybe I can use a parallel design and use 4 D batteries?

That would give you a longer run time.

What is your requirement for variation in brightness? If it's not critical, you don't need to use a current source. With your existing circuit, you're still going to dissipate significant power in the current sense resistor and whatever the LEDs don't use will be dissipated in the MOSFET.

 

Thanks for pointing that out. I suspected that the built-in heat sinks would only suffice for short tests, but they may need extra dissipation for >1h operation.

What you and many others refer to as "heat sinks" are actually only spreaders.

And, your short tests better be real short.

 

Unless portability is important in your application, consider a laptop battery charger as a power source. Many of them have a 19 V output, they are good for at least 2 amps, and a used on is only a couple of dollars on ebay.

ak

 

What is your requirement for variation in brightness? If it's not critical, you don't need to use a current source. With your existing circuit, you're still going to dissipate significant power in the current sense resistor and whatever the LEDs don't use will be dissipated in the MOSFET.

It's not critical. So after all, a simple resistor would be the most energy efficient option?

Unless portability is important in your application, consider a laptop battery charger as a power source. Many of them have a 19 V output, they are good for at least 2 amps, and a used on is only a couple of dollars on ebay.

I have thought about that, and I might have a working 12V one. Maybe not lighter than 4 D batteries, and probably too bulky to fit comfortably in a portable item, but worth a try.

 

What you and many others refer to as "heat sinks" are actually only spreaders.

I had thought about anchoring them to an aluminum bar. I don't want to use glue that acts as an insulator. The notches might work with screws but they are too close to the contact pads, so PVC screws might be the only option.

 

So after all, a simple resistor would be the most energy efficient option?

That's the route I'd go. Once you determine your actual operating voltage, you can calculate a current limiting resistor value.

BTW, 9V NiMH batteries are actually going to have nominal voltages of 7.2, 8.4, or 9.6V. Nominal cell voltage for NiMH is 1.2V.

 

When mounting those "star" LEDs I use either nylon machine screws or Arctic thermal adhesive. (other plastic screws will work as well)

I have used adhesive silicon pads on occasion, but only when the current is very limited and I can insure they won't lose contact.

And don't forget the thermal compound, if you do use screws.

 

So after all, a simple resistor would be the most energy efficient option?

As above, there is no "most". All linear options are identical in power dissipation.

However, the constant-current circuit you are considering has an operational advantage. As the battery discharges and its terminal voltage decreases, the LED brightness will decrease in the 1-transistor circuit. With the constant-current circuit, the LED brightness will remain constant until there is not enough voltage headroom to run the circuit.

However, that constant-brightness performance will discharge the batteries faster.

The total energy in the batteries is what it is. The two circuits use it differently.

ak

 

All, this is a lot of incredibly useful info—thanks a lot.

 

I made this video. It shows how different LEDs react with each other when in parallel. If your diodes are in series then I'm not certain exactly what will happen. I suspect the one with the lower Vf will be brighter and hotter than the one with the higher Vf.

 

In series, all get the same current. Which is brighter would depend on the output vs current curves of each.

The ones with the highest Vf would use more power and therefore run hotter.

Bob

 

It shows how different LEDs react with each other when in parallel.

Cheap/poorly designed circuits will put LEDs in parallel without a ballast resistor; usually to save the expense of using resistors (buying and/or wiring). Under the right conditions, the LED with the lowest forward voltage will hog current and die. That can lead to a theoretical cascading failure where the new lowest forward voltage will hog current and die. Until all have burned out.

If your diodes are in series then I'm not certain exactly what will happen. I suspect the one with the lower Vf will be brighter and hotter than the one with the higher Vf.

If you operate LEDs of the same type (and bin because manufacturers have been binning by brightness since LEDs were first mass produced) at the same current, they'll have similar brightness. No LED will become hot enough to burn out if maximum continuous current isn't exceeded (and recommended heat sinking is used for high power varieties).

When you have multiple colors with different brightnesses and you want to match brightness, things get more complicated.

 

Cheap/poorly designed circuits will put LEDs in parallel without a ballast resistor; usually to save the expense of using resistors (buying and/or wiring). Under the right conditions, the LED with the lowest forward voltage will hog current and die. That can lead to a theoretical cascading failure where the new lowest forward voltage will hog current and die. Until all have burned out.

Absolutely true. Did I not state that in the video?

 

Absolutely true. Did I not state that in the video?

I don't watch many videos claiming to explain something about electronics. It saves me from wasting my time.

Some people just post long videos to monetize other peoples' time. There was a video of a baby biting his brother's finger and it netted $1M. I read a story about it...

 

In the video I connect two LEDs in parallel. One red and one green. At first, the green is illuminated. But when I put the red LED in the green goes out. In the video I explain that the red is hogging all the current. I also go on to show six (or more) blue LED's in parallel all off the same single resistor. As more and more blue LED's are added the brightness of each decreases slightly. That might have been due to my 4.5V power source. It might have been due (in part) to current sharing. Maybe one day I'll try it with a much bigger supply. Something like an SLA 12V 20Ah battery. In the video I used an old wallwart.

And no - I don't get paid for advertising. I'm not advertising on my channel. I don't think I'd make much more than a nickel.

 

Maybe one day I'll try it with a much bigger supply.

I've thought about paralleling a bunch of LEDs (50-100) and operating them at the maximum continuous current spec to see if I could create the theoretical cascading failure. So far, I've always had better things to do.

 

Cheap LEDs are not binned. Their specs vary all over the place.
I have a cheap flashlight with 24 LEDs directly in parallel and they all have the same brightness. They probably paid somebody a bowl of rice to test and group them.

Cheap dim LEDs are put into a focused package to have brighter but narrow beam.

 

What I am inferring from the conversation so far (please let me know if I'm off the mark):

1. If I put LEDs with different forward voltages values in series, the ones with lowest value will die (that's probably what happened to the red LED in my original circuit).
2. I can safely group series of LEDs with the same FV.
3. If I parallel connect series of LEDs with different individual FV, but consistent within each serial chain, either each chain should have a total voltage drop equal to the others, or I shall adjust that by using resistors of different values to balance.
4. I should always use a resistor between a LED and the power source (why? Does the resistor act as a buffer against voltage peaks?). However, since efficiency is a concern here, I would want to use a source voltage as close as possible to the voltage drop of the LEDs (and hence a small Ohm value). Is there an ideal resistor value that would waste the least power while acting effectively as a life-saver?
5. The effective VF of a LED may be different from the nominal one. I shall always measure it to be sure.

 

Each LED has a forward voltage drop, therefore the supply voltage must be higher that all of the voltage drops combined.

LEDs operate on current (not voltage), therefore there needs to be a current limiting resistor to control the current or the LED will burn out.

READ THE PDF! Know exactly what parameters must be adhered to.

Notes on LEDs (gizmology.net)

 

READ THE PDF! Know exactly what parameters must be adhered to.

Unfortunately there is no PDF for the LEDs I bought. Just a voltage range and current rating.

Notes on LEDs (gizmology.net)

This was very useful and partly answered my question about the resistor.