Visibility-duty cycle relation with LEDs

[Pimpom]

I want to make a 1W red LED blink at ~1 Hz. Providing the best
possible visibility from a distance is vital and takes precedence
over consideration of power consumption. Nevertheless, I'd like
to reduce the average power as much as possible.

Is there a generally accepted optimum figure for the duty cycle
and/or pulse duration to that end? Adaptability of the human
senses being what it is, I don't want to rely solely on my own
observations with a test setup.

 

[Maynard A. Philbrook Jr.]

In article <lu8vk4$r84$1@news.albasani.net>, Pimpom@invalid.invalid
says...

I want to make a 1W red LED blink at ~1 Hz. Providing the best
possible visibility from a distance is vital and takes precedence
over consideration of power consumption. Nevertheless, I'd like
to reduce the average power as much as possible.

Is there a generally accepted optimum figure for the duty cycle
and/or pulse duration to that end? Adaptability of the human
senses being what it is, I don't want to rely solely on my own
observations with a test setup.

The only thing I can suggest is a buck switcher for a current
regulator and maybe a CMOS version of the 555 timer with a
driver tranny to switch the LED in at 1p/s///

btw,
The part number is like ICM7555, and a LMC555 I think it is.
The quiescent current is around 60uA, so that shouldn't be too
much of a drain on you.

Jamie

 

[John Larkin]

On Thu, 4 Sep 2014 11:38:01 +0530, "Pimpom" <Pimpom@invalid.invalid>
wrote:

I want to make a 1W red LED blink at ~1 Hz. Providing the best
possible visibility from a distance is vital and takes precedence
over consideration of power consumption. Nevertheless, I'd like
to reduce the average power as much as possible.

Is there a generally accepted optimum figure for the duty cycle
and/or pulse duration to that end? Adaptability of the human
senses being what it is, I don't want to rely solely on my own
observations with a test setup.

Wild guess: pulse it for 50 ms, at 1 Hz, with as much current as your
power budget allows. Check that the LED is still reasonably linear at
that current.

But your own eyeball testing is important.

Is red necessary? Other colors would be more efficient.


--

John Larkin Highland Technology, Inc

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

On Thu, 04 Sep 2014 15:32:01 -0700, John Larkin
<jlarkin@highlandtechnology.com> wrote:

On Thu, 4 Sep 2014 11:38:01 +0530, "Pimpom" <Pimpom@invalid.invalid
wrote:

I want to make a 1W red LED blink at ~1 Hz. Providing the best
possible visibility from a distance is vital and takes precedence
over consideration of power consumption. Nevertheless, I'd like
to reduce the average power as much as possible.

Is there a generally accepted optimum figure for the duty cycle
and/or pulse duration to that end? Adaptability of the human
senses being what it is, I don't want to rely solely on my own
observations with a test setup.


Wild guess: pulse it for 50 ms, at 1 Hz, with as much current as your
power budget allows. Check that the LED is still reasonably linear at
that current.

But your own eyeball testing is important.

Is red necessary? Other colors would be more efficient.

Greetings John,
When learning to program microcontrollers the first program was the
typical blinking LED. I meesed around with the timing and as I recall
50 ms on time was easy for me to see. I fiddled around with the timing
to see just how short a pulse of light I could detect reliablly and 50
ms was a lot longer than I needed. I'm sure that the pulse duration
will need to be much longer if the LED isn't very bright when viewed
from a distance, but when looking at a standard LED from no more than
3 feet your 50 ms time was more than adequate.
Eric

 

[John Larkin]

On Thu, 04 Sep 2014 17:05:28 -0700, etpm@whidbey.com wrote:

On Thu, 04 Sep 2014 15:32:01 -0700, John Larkin
jlarkin@highlandtechnology.com> wrote:

On Thu, 4 Sep 2014 11:38:01 +0530, "Pimpom" <Pimpom@invalid.invalid
wrote:

I want to make a 1W red LED blink at ~1 Hz. Providing the best
possible visibility from a distance is vital and takes precedence
over consideration of power consumption. Nevertheless, I'd like
to reduce the average power as much as possible.

Is there a generally accepted optimum figure for the duty cycle
and/or pulse duration to that end? Adaptability of the human
senses being what it is, I don't want to rely solely on my own
observations with a test setup.


Wild guess: pulse it for 50 ms, at 1 Hz, with as much current as your
power budget allows. Check that the LED is still reasonably linear at
that current.

But your own eyeball testing is important.

Is red necessary? Other colors would be more efficient.
Greetings John,
When learning to program microcontrollers the first program was the
typical blinking LED. I meesed around with the timing and as I recall
50 ms on time was easy for me to see. I fiddled around with the timing
to see just how short a pulse of light I could detect reliablly and 50
ms was a lot longer than I needed. I'm sure that the pulse duration
will need to be much longer if the LED isn't very bright when viewed
from a distance, but when looking at a standard LED from no more than
3 feet your 50 ms time was more than adequate.
Eric

A flash of some given time*current product will look about the same
below (very) roughly 50 ms. So 10 mA at 50 ms will look about like 100
mA at 5 ms. The limit (1 amp for 500 us, 10 amps for 50 us) is that
the LED efficiency will start to drop off somewhere at high currents.
The LED data sheet should have the graphs you need.




--

John Larkin Highland Technology, Inc

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Pimpom]

John Larkin wrote:

On Thu, 04 Sep 2014 17:05:28 -0700, etpm@whidbey.com wrote:

On Thu, 04 Sep 2014 15:32:01 -0700, John Larkin
jlarkin@highlandtechnology.com> wrote:

On Thu, 4 Sep 2014 11:38:01 +0530, "Pimpom"
Pimpom@invalid.invalid
wrote:

I want to make a 1W red LED blink at ~1 Hz. Providing the
best
possible visibility from a distance is vital and takes
precedence
over consideration of power consumption. Nevertheless, I'd
like
to reduce the average power as much as possible.

Is there a generally accepted optimum figure for the duty
cycle
and/or pulse duration to that end? Adaptability of the human
senses being what it is, I don't want to rely solely on my
own
observations with a test setup.


Wild guess: pulse it for 50 ms, at 1 Hz, with as much current
as
your power budget allows. Check that the LED is still
reasonably
linear at that current.

But your own eyeball testing is important.

Is red necessary? Other colors would be more efficient.
Greetings John,
When learning to program microcontrollers the first program
was the
typical blinking LED. I meesed around with the timing and as I
recall
50 ms on time was easy for me to see. I fiddled around with
the
timing to see just how short a pulse of light I could detect
reliablly and 50 ms was a lot longer than I needed. I'm sure
that
the pulse duration will need to be much longer if the LED
isn't very
bright when viewed from a distance, but when looking at a
standard
LED from no more than 3 feet your 50 ms time was more than
adequate.
Eric

A flash of some given time*current product will look about the
same
below (very) roughly 50 ms. So 10 mA at 50 ms will look about
like 100
mA at 5 ms. The limit (1 amp for 500 us, 10 amps for 50 us) is
that
the LED efficiency will start to drop off somewhere at high
currents.
The LED data sheet should have the graphs you need.

LED datasheets indicate that linear scaling of current with
reduced duty cycle is not permissible. For example, while 1W LEDs
are typically rated for a maximum continuous current of 350 mA,
the max peak current for 100usec at 1kHz is 500-700 mA. None of
the datasheets I've looked at give ratings for pulse lengths in
the tens of milliseconds.

Moreover, circumstances force me to use generic LEDs. So I want
to be conservative with respect to peak current and limit it to
400 mA or thereabouts. Given that and the repetition rate of
about 1 Hz, what I have to determine is pulse length versus
visibility over the greatest possible distance. Instinct tells me
that a high duty cycle like 0.5 will not have any significant
advantage and will just be a waste of power.

 

[whit3rd]

On Thursday, September 4, 2014 2:08:01 AM UTC-4, Pimpom wrote:

I want to make a 1W red LED blink at ~1 Hz. Providing the best

possible visibility from a distance is vital and takes precedence

Then, try to arrange a good, black, background: a hood over the lamp,
and a red filter, will help. One can also use a circular polarizer
for a window (light from outside shines in, becomes right-circular-polarized,
reflects, becomes left-circular-polarized, and can't get out).

There's tens of percents available with electronic tricks, but
there's orders of magnitude available with stray-light
management.

 

[Jasen Betts]

On 2014-09-05, Pimpom <Pimpom@invalid.invalid> wrote:

John Larkin wrote:
On Thu, 04 Sep 2014 17:05:28 -0700, etpm@whidbey.com wrote:

LED datasheets indicate that linear scaling of current with
reduced duty cycle is not permissible. For example, while 1W LEDs
are typically rated for a maximum continuous current of 350 mA,
the max peak current for 100usec at 1kHz is 500-700 mA. None of
the datasheets I've looked at give ratings for pulse lengths in
the tens of milliseconds.

Moreover, circumstances force me to use generic LEDs. So I want
to be conservative with respect to peak current and limit it to
400 mA or thereabouts. Given that and the repetition rate of
about 1 Hz, what I have to determine is pulse length versus
visibility over the greatest possible distance. Instinct tells me
that a high duty cycle like 0.5 will not have any significant
advantage and will just be a waste of power.

for that, 500ms is a good starting point, although you may have to
reduce the current a bit.

2 to 4 Hz at 50% duty cycle may be more visible,

--
umop apisdn


--- news://freenews.netfront.net/ - complaints: news@netfront.net ---

 

[John Larkin]

On Fri, 5 Sep 2014 15:36:00 +0530, "Pimpom" <Pimpom@invalid.invalid>
wrote:

John Larkin wrote:
On Thu, 04 Sep 2014 17:05:28 -0700, etpm@whidbey.com wrote:

On Thu, 04 Sep 2014 15:32:01 -0700, John Larkin
jlarkin@highlandtechnology.com> wrote:

On Thu, 4 Sep 2014 11:38:01 +0530, "Pimpom"
Pimpom@invalid.invalid
wrote:

I want to make a 1W red LED blink at ~1 Hz. Providing the
best
possible visibility from a distance is vital and takes
precedence
over consideration of power consumption. Nevertheless, I'd
like
to reduce the average power as much as possible.

Is there a generally accepted optimum figure for the duty
cycle
and/or pulse duration to that end? Adaptability of the human
senses being what it is, I don't want to rely solely on my
own
observations with a test setup.


Wild guess: pulse it for 50 ms, at 1 Hz, with as much current
as
your power budget allows. Check that the LED is still
reasonably
linear at that current.

But your own eyeball testing is important.

Is red necessary? Other colors would be more efficient.
Greetings John,
When learning to program microcontrollers the first program
was the
typical blinking LED. I meesed around with the timing and as I
recall
50 ms on time was easy for me to see. I fiddled around with
the
timing to see just how short a pulse of light I could detect
reliablly and 50 ms was a lot longer than I needed. I'm sure
that
the pulse duration will need to be much longer if the LED
isn't very
bright when viewed from a distance, but when looking at a
standard
LED from no more than 3 feet your 50 ms time was more than
adequate.
Eric

A flash of some given time*current product will look about the
same
below (very) roughly 50 ms. So 10 mA at 50 ms will look about
like 100
mA at 5 ms. The limit (1 amp for 500 us, 10 amps for 50 us) is
that
the LED efficiency will start to drop off somewhere at high
currents.
The LED data sheet should have the graphs you need.

LED datasheets indicate that linear scaling of current with
reduced duty cycle is not permissible.

It works up the the point that the current gets high enough that
efficiency starts to drop off. For basically all LEDs, you can pulse
them and increase visibility. In a recent thread on s.e.d., we
discussed pulsing white LEDs as a warning indicator, with only
microamps of current available.



For example, while 1W LEDs

are typically rated for a maximum continuous current of 350 mA,
the max peak current for 100usec at 1kHz is 500-700 mA. None of
the datasheets I've looked at give ratings for pulse lengths in
the tens of milliseconds.

Moreover, circumstances force me to use generic LEDs. So I want
to be conservative with respect to peak current and limit it to
400 mA or thereabouts. Given that and the repetition rate of
about 1 Hz, what I have to determine is pulse length versus
visibility over the greatest possible distance. Instinct tells me
that a high duty cycle like 0.5 will not have any significant
advantage and will just be a waste of power.

For a given budget of average current, low duty cycle flashing does
make an LED more visible.

Try it.


--

John Larkin Highland Technology, Inc

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

I think your bst bet is to use multiple LEDs and not push them so hard.

I do see where you were misunderstood, you want to flash at one Hertz.

If you want 50 % on and 50 % off, that does not men you can go to double the current. Your illustration proves that. What is wrong with tewo LEDs ? Much easier for the bang for the buck. Putting them in series should work just fine.

One thing though, put Zeners or something across them if you want reliablity. If one opens up, the other will still ligh but at a lower level and alert you to the fact that you need to replace one.

Of course the option of just using one means if it goes out you got nothing.. But putting them in parallel is even worse when it comes to drive.

You oculd use resistors across them. They go what like two volts ? So at four volts choose it so it is not a significant load, but is enough to light the other LED in case of failure. Just enough. Then you take your bagfull of generics and stick one in.

When you push LEDs hard they do not last forever. And I thinkk that is what you were asking. "If I only pulse this LED, how much harder can I pump it ?". Well maybe not...

I would go with the lower power thart it can handle continuous and blink it OFF. If you really want huans to see it and now it is not a natural phenomenon or whatever, blink it off three times every other second, and two or four times on the opposite seconds. Very unikely to occur in nature. Therefore unnatural enough to catch the attention.

So what is htis all about ? pople kep hitting your mailbox out front ? I got a guy fixs that. He builds to a brick mailbox type of thing. A regualr mailbox can be put on it or custom. but they would brign a bunch of constrution debris and fill the walls, ;up to where the mailbox goes. They put lights on some of them, but the drunks owuld still hit them.

But NOW with that brick and mortar at least the fucked up their cars ! mwahahaha.

Anyway, just curious as to the application here. AND, are you trying to get this thing seen during the day ? If so, well it gets pretty bright out there so you are fighting alot. In the US, here anyway, there are traffic lights that are damn hard to see, and what latitude is Ohio ?

Over in design there is a guy named Jim Thompson who ight know more about this. He designed a bunch of boards that display on those signs I guess outside stadiums and shit, and maybe when there is road work. I'm sure you've seen them if in the US, or probably a bunch of other countries.

Also, in the US, buses and trucks are now (not all) equippd with LED brae lights with "enuncitaors". Every time the driver hits the brake pedal, when the array of LEDs goed on, it purposely flickers, I think twice. that is designed to attract attention. Nothing in nature goes on and off like :

Ż|_|Ż|_|ŻŻŻŻŻŻŻŻŻŻŻŻŻŻ|_|Ż|_|ŻŻŻŻŻŻŻŻ

I think the DOT requires that on new big trucks and shit. Not sure, but if they don't, the only reason to put them in is a break from the insurance companies and when they speak, people listen.

So maybe :

Ż|_Ż|_|ŻŻŻŻ|______|ŻŻŻŻŻ|_|Ż|_Ż|_|ŻŻŻŻ|_

Something like that.

In other words, you can only get so much attention out of brighness. soetimes it is the timing.