Lighting the lamps

The circuit for 3 leds in series which is on page 492 of the current edition has an error the left headlamp should be reversed.

However the method shown is very poor electronics, an led is a forward biased diode which should be supplied from a constant current source which will limit the current to the required amount in this case about 20mA for a voltage of 3v across each diode. A small increase in voltage of say 0.2 v would increase the current flow Considerably and would risk damaging the diodes. A new PP3 battery can give 9.6v I.e. 3.2v per led.

A better solution is to either supply the leds in series with a current limiting resistor using a higher voltage supply or wire the leds in parallel each one with its own resistor

The LEDs look to be correctly connected to me.

Agreed that it's not the most elegant circuit, but given that they're white LEDs with typical forward voltages of around 3.4V it'll probably work without blowing anything up, even though the V/I characteristic is exponential.

Andrew

Well to me they are all facing the same side of the battery so look to be the right way round, the rest is over my head

Looks the right way round to me too.

In other circumstances Peter makes a good point about powering LEDs. However in this application the circuit has the supreme advantage of simplicity. I think it's a reasonable solution to the problem described. Also, paralleling the LEDs with a resistor isn't advantageous in this case because voltage dropping wastes battery power.

Dave

Edited By SillyOldDuffer on 27/09/2017 21:09:06

Everything I thought I knew about LEDs is wrong. I was brought up on the constant voltage mantra, but white LEDs are hugely more forgiving. You can run some white LEDs straight off a LiPo cell at over 4.2V without trouble with no limiting resistor.

It shouldn't work, but in practice the I/V curve for some brands is much flatter than for others.

For my bench light I used an cupboard light designed to run off three 1.5V batteries with no limiting resistors, on a fully charged LiPO it is quite happy - 4.2v is less than three brand new alkaline AAs.

Neil

Apologies yes the leds are correct, my mistake I better go to spec savers.

Constant voltage or constant current?

Of course white LEDs are really blue LEDs with more class. Some of the blue light is absorbed by phosphors and re-emitted as red and green light; the combination of all three giving the impression of white light.

From a marketing point of view your product is dead in the water these days if it doesn't have at least one blue LED.

Andrew

As the graph in Neil's post shows once the junction reaches a certain point the voltage across it remains close to a constant but a slight increase in voltage produces a large increase in current. I agree that the design published will work I was simply pointing out that it isn't the best or safest way to use LEDs.

Edited By 61962 on 27/09/2017 23:11:57

How about link 61962? I've tried searching te proboards forum with no success

Depend how you define it. A LED acts as a poor constant-voltage regulator which is what I meant, which means it's normal to feed them consant current...

Here's one I made earlier

1. Leds are the right way round.

2. It will work after a fashion.

3. Why on earth don't you just wire one of these in series with the each LED and wire the whole lot in parallel or in series if you bundle the device with the power supply (battery)

**LINK**

10mA device shown (shewn) other values available.

All LED will be the same brightness (subject to binning for the pedantic) and you get to be a lot more flexible with your supply voltage to boot.

PS the lamps do look nice despite the electronics which as you say does work which is the main point.

regards Martin

Edited By Martin Kyte on 28/09/2017 09:35:36

I am amused.

This is a classic example of something that works, not because it is good design practice, but because it is pragmatic and takes advantage of the working tolerances of all the components.

There is absolutely no need to use a more complex arrangement.

In the case of white leds for a LiPO you will need to have one that are more voltage tolerant, but the 3LED+PP3 should work with any white LEDS.

.

um, I hope I said that in 1, and 2. perhaps 3. should have started, if you are bothered etc.

regards Martin

Am I missing something? According to Neil's graphs, if the battery gets down to say 8v, 2.67 volts per LED, there will be little or no forward current. If this is the case then it only works with a fresh battery. Martin's rather elegant chips would drive 2 LEDs in series even with a not new battery, but as they are surface mount they are perhaps not to everyone's taste. Is the a DIL version, perhaps having 4 drivers?

They are surface mount BUT they are also 4 mm long with the solder tags extending out from the chip and not under it which makes them easy to solder to a sliver of veroboard if you don't want to merely tack wires on. They are happy to run in parallel so 2 gets you 20mA, 3 3mA etc for the devices shown (shewn). As I mentioned you could tuck them out of the way with the battery.

I will admit to being a bit professional about this as I build a lot of high output LED drivers for the Lab using buck, boost or buck/boost switching regulators which are pulse width dimmable to maintain the wavelength profile of the emitters whilst allowing fast turn on/off at 500ish kHz. Totally over the top as far as home use is concerned. However it does make me aware of the range of simple building block LED drivers of the type I indicated.

I hope I'm not overcomplicating things and I recognise that the original cct does indeed work which was what was intended but for notionally one extra component especially insted of where some would have placed a current control resistor I think it was worth pointing out the devices so I hope that them that wants to get to use them.

regards Martin

Excellent advice, Martin

Thanks

MichaelG.

Quite a good example of 'value engineering' in action. A point often missed about 'quality' is that you have stop somewhere. If not you end up making expensive stuff the customer doesn't want or need. Don't be surprised he buys cheaper alternatives even if you think they're all crap and he's an idiot!

Just as an example, Martin's chip is indeed a jolly good thing, but reading the Application Notes reveals some issuettes that many engineers would be keen to address:

• the chip and LEDs should be protected by a rectifier diode.
• heat dissipation is a concern so it may not be a good idea just to solder wires on to the chip. Big tags on an SMD chip are often used to help cool the device by fixing it to a copper plane. In other words this baby should be on a PCB designed to keep it cool.
• Best of all is the suggestion that the chip and LED string should be Pulse Width Modulated to reduce dissipation. That's another level of complication and cost.

You have to ask if 'best practice' is necessary or even appropriate. I think Ron's article is spot on. He doesn't claim his circuit is a design template. It's minimalist and it works, job done. I'd argue a more complicated circuit would be overkill on a model locomotive. Different story entirely if the LEDs are for use in a car or an aircraft. I usually use resistor droppers on LEDs, but I'm very taken with Martin's chip - I'm a convert, and would most likely use it without any bells and whistles.

Very interesting comments in this thread. It's all good stuff that keeps the old grey cells well stirred!

Dave

Edited By SillyOldDuffer on 28/09/2017 13:53:55

Edited By SillyOldDuffer on 28/09/2017 13:54:17

That is the biggest weakness of the approach.

Use the 'flat' battery in your multimeter, you should get six months out of it