LED mark space

A little bit off metal bashing, but of relevance to our signalling system. I've got all the gremilns out (famous last words), the usual cause of failure is now flat battery (no-one seems to remember to charge them up). I've read somewhere many years ago that turning an LED on and off very rapidly reduces the power cosumption, but hardly reduces its apparent brightness. Tried google to no avail. Anyone got any info on best mark space and best frequency?

I think you might try 1/20th of a second which is a figure I remember for persistence of vision,

I always believed this but some googling suggests the effect is debatable and marginal, needing an extreme brightness for the flashes. a 5% duty cycle seems to be indicated and would still need at least half the normal power (roughly. 10x the current for 1/20 the time) to attain similar brightness, so it hardly seems worth the effort.

60Hz to avoid flicker when seen by the rods of the eye, but 20-20 should be OK.

What about a small wind/solar generator instead?

Neil

Edited By Neil Wyatt on 28/10/2015 17:52:54

I expect the information you want will be on the data sheet for the LED's your using

for example , one white LED I looked at -

Average Forward Current = 30 mA

Peak Forward Current = 100 mA （Duty＝0.1，1kHz）

the Joule thief may give you ideas -

**LINK**

**LINK**

John

This is a link to an application note that explains how better efficiency can be got from a LED.

But I think your real problem is not how to make the battery last longer, it is how to remind people that it needs recharging. How about fitting a low battery alarm ? They aren't difficult to make if you can't buy a ready made module.

Cheers,

Dave

I'm on with a low battery alarm, but I confidently expect the drivers to ignore it, after all they ignore the fact that the green LED has died without mentioning it (the green LED array seems to give up the ghost first). We're also looking into a solar panel. SillyOldDuffer's link looks just the ticket, thanks. I'll post the rsukts, but don't hold your breath, need to wait till running season is over before undertaking any major rebuild of the unit.

The other route is just to make them dimmer. The human eye follows a log law for sensitivity, so that for a given drop in current the drop in brightness will be perceived as rather smaller. Or use smaller lights to increase the 'surface brightness' and therefore the contrast with the background.

Neil

.

I didn't specify the means by which the dimming is achieved.

Neil

.

No you didn't ... but you did mention 'the other route'

MichaelG.

I make a lot of opto-genetics Led systems for the Lab and I use PWM for dimming control. I tend to go for base frequencies of around 5 to 10kHz. I generally use National Semiconductor switching regulator chips of which the LM3404 is probably the simplest and use a frequency of around 500kHz for the switching frequency of the regulator.

These little switching regulators have a good efficiency figure of 80-90%.

What is more to the point with your signals - You should also have a look at the Led emitters you are using so you get a high light ouput to current ratio and if this is a signalling system ensuring that directional emitters are employed. There is no point in genearating lots of photons most of which end up no where near the eye. Lots of emitter beam shapes to choose from and plenty of lenses too. I presume you only need the signal to be in the approach direction of the track so you only need a narrow beam angle. This makes the LEDs brighter and so you can reduce the current.

regards Martin

One thing to watch for in switched LEDs is radio frequency interference (RFI), the LED spot light I have in my workshop has a major problem which I cured by running it through an isolating transformer, RFI on the power cable, it's a mains powered unit.

Ian S C

I'm always amazed by the knowledge of people using this site, and how helpful they are. I've found an article (ME 2009) about using a lens to make the apparent diameter of the light bigger, and to make it more directional, so I'm going to experiment with this and pulsing. As the whole thing is driven by a microprocessor pulsing is software mods, so quick and easy. The built in routine for pwm runs at <1kHz so I'm not anticipating RFI. As an amusing aside I once worked on a plant which had a high power RF heating system. We were asked by 'the management' to prove that we weren't interfering with everyone's radio, so I consulted a chap who had worked for the BBC for advice. He advised me to tune a transistor radio to Radio Blackpool (the nearest station to our operating frequency) and wander round our site and the local area to see if there was a problem. No fancy tests or meters, I like solutions like that. I'll let you all know how it goes, but once again thanks to all who have replied. At least I now know why the green gives up first.

By the way Neil, I can't make the present red LEDs dimmer as they have a bult in switch mode power supply, if you reduce the input volts the current goes up (steeply) and the brightness stays the same until you reach a very low voltage, when it just gives up the ghost. I'm not planning on using these in te new manifestation!

Edited By duncan webster on 29/10/2015 12:38:56