Driving LEDs - high-power flash

[Leon Heller]

I'm trying to drive some high-brightness red LEDs to provide a really bright
flash. The schematic of the driver circuit is here:

http://www.leonheller.com/led.gif

The two-transistor circuit in the supply provides a constant current to the
capacitor

I'm driving it with a 38 kHz PWM signal generated with an AVR, lasting a
total of 40 ms, but can't get anything like the brightness I need, probably
because I'm getting a lot of ringing (about 330 kHz) on the trailing edge of
the pulse, when I put the scope across the LEDs.

Has anyone got any ideas?

Leon

 

[Fred Bloggs]

I'm trying to drive some high-brightness red LEDs to provide a really bright
flash. The schematic of the driver circuit is here:

http://www.leonheller.com/led.gif

The two-transistor circuit in the supply provides a constant current to the
capacitor

I'm driving it with a 38 kHz PWM signal generated with an AVR, lasting a
total of 40 ms, but can't get anything like the brightness I need, probably
because I'm getting a lot of ringing (about 330 kHz) on the trailing edge of
the pulse, when I put the scope across the LEDs.

Has anyone got any ideas?

Leon

It looks like there is a good chance you are saturating your core- so
the magnetic energy storage is being limited to far less than you
calculated.

 

[Dave Garnett]

If the eye integrates the light, then the apparent brightness will be
determined by the average current, not the peak ...

Dave

 

[Leon Heller]

It looks like there is a good chance you are saturating your core- so the
magnetic energy storage is being limited to far less than you calculated.

Thanks, it looks like you might be correct. Peak voltage across the inductor
is about 28V. It has a DC resistance of 30R, giving nearly 1 A through it.
The max. current in the spec. is 100 mA. The ringing is probably caused by
the reduction in inductance when the inductor is saturated.

I'll try using a different inductor.

Leon

 

[[email protected]]

Has anyone tried to drive LEDs using a miniature pulse-forming LC
network such as is used to supply radar pulses? It would certainly
give a better waveform output than a single solitary inductor.

Stepan

 

[Leon Heller]

Has anyone tried to drive LEDs using a miniature pulse-forming LC
network such as is used to supply radar pulses? It would certainly
give a better waveform output than a single solitary inductor.

I don't think the waveform matters all that much. The inductor is charged up
whilst the input to the driver MOSFET is high, and then discharged into the
LEDs when the input goes low and the MOSFET turns off. I changed the 1 mH
inductor to 330 uH, adjusted the pulse frequency and it worked, producing a
quite blinding flash (about 200 mA through the LEDs). This was just in time
for my client to demonstrate a working prototype to his customer today.

Leon

 

[Paul Burridge]

I don't think the waveform matters all that much. The inductor is charged up
whilst the input to the driver MOSFET is high, and then discharged into the
LEDs when the input goes low and the MOSFET turns off. I changed the 1 mH
inductor to 330 uH, adjusted the pulse frequency and it worked, producing a
quite blinding flash (about 200 mA through the LEDs). This was just in time
for my client to demonstrate a working prototype to his customer today.

What's the intended application for this, Leon?

 

[Leon Heller]

What's the intended application for this, Leon?

LED flashing beacon. There's a microcontroller in it as well, generating the
pulses, and interfacing some switches.

Leon

 

[[email protected]]

Human perception of brightness is proportional to pulse width for
pulses shorter than approx 10ms. In other words a 1ms pulse appears
much dimmer than a 10ms pulse, but a 100ms pulse does not appear
brighter than a 10ms pulse.

Here is a circuit that demonstrates the phenomenon:

http://members.shaw.ca/novotill/LedStroboscope/index.htm

White LEDs become inefficient at high currents. This is because the
fluorescent material saturates. Use non-white LEDs to avoid that.

Stepan

 

[Winfield Hill]

[email protected] wrote...

Human perception of brightness is proportional to pulse width for
pulses shorter than approx 10ms. In other words a 1ms pulse appears
much dimmer than a 10ms pulse, but a 100ms pulse does not appear
brighter than a 10ms pulse.

It should be pointed out (once again) that there's a repetition
frequency region that's especially noticeable (irritating) to
humans. We're very sensitive to the 10 to 35Hz region, because
this is faster than the slow adaptation region that's ignored
(an important biologically-adaptive fact of psychophysics, where
bright irradiance is ignored, yet a quick reflex response saves
lives), and slower than a "too fast" response that our receptors
simply can't respond to.

And I'd like to also point out this is the flicker region where
flashing lights are "in your face" and drive one nuts saying,
"Turn that off!" It's certainly NOT a matter of a nice steady
light that appears brighter because one selected a magic duty
cycle and frequency.

 

[Leon Heller]

LED flashing beacon. There's a microcontroller in it as well, generating
the pulses, and interfacing some switches.

I've just heard that the customer was very impressed with the brightness of
the beacon, the sound levels we are achieving and the power consumption,
which is gratifying. There are some problems with high-frequency noise and
current surges getting back into the supply, which we have to address (they
use the supply for signalling).

Leon

 

[R.Lewis]

Human perception of brightness is proportional to pulse width for
pulses shorter than approx 10ms. In other words a 1ms pulse appears
much dimmer than a 10ms pulse, but a 100ms pulse does not appear
brighter than a 10ms pulse.

a 100mS pulse does not appear brighter than a 10mS pulse???
complete and utter balderdash.

see huge amouts of scientific (not mumbo jumbo) work by messrs Talbot, Rey,
and Schmidt-Clausen for starters.

 

[[email protected]]

Can we at least agree that a 1 second pulse is not brighter than a
100ms pulse? If not, then where do you draw the line?
Personally I draw it at apprix 15ms but that is my own judgement.
Some literature draws it at 10ms.

s


On Sat, 25 Sep 2004 11:59:10 +0100, "R.Lewis" <h.lewis-not this

 

[[email protected]]

[email protected] wrote...
And I'd like to also point out this is the flicker region where
flashing lights are "in your face" and drive one nuts saying,
"Turn that off!" It's certainly NOT a matter of a nice steady
light that appears brighter because one selected a magic duty
cycle and frequency.

Additionally, I find that a random strobing is much more annoying and
difficult to tune out than a steady rate, (and that the 10 - 30 Hz
thing seems to hold true based upon my own experiments; Thanks for the
explanation of why that is so Win)

Based upon my own experimentation, I beleive that in order to get the
physiological effect per Joule of energy delivered to the LED, the
pulse width should be approx 15ms and the repetition rate should be
random centered around 25Hz. This is important for battery operated
circuits. The following circuit can do that(posted once before):

http://members.shaw.ca/novotill/FireLightFlicker/index.htm

Stepan

 

[[email protected]]

I've just heard that the customer was very impressed with the brightness of
the beacon, the sound levels we are achieving and the power consumption,
which is gratifying. There are some problems with high-frequency noise and
current surges getting back into the supply, which we have to address (they
use the supply for signalling).

Leon

I wonder if the FET is switching too fast and giving some HF ringing
and noise.
Did you try putting a small resistance in series with the gate lead of
the FET right at the FET?

Stepan

 

[R.Lewis]

Can we at least agree that a 1 second pulse is not brighter than a
100ms pulse? If not, then where do you draw the line?

If you bother to learn something about the subject you would not need to ask
such silly questions.

Personally I draw it at apprix 15ms but that is my own judgement.
Some literature draws it at 10ms.

Draw the line where you like - just don't believe it means anything.


I think a lot of people would like to see the reference " Some literature
draws it at 10ms." - especially Ms Goodman at the CIE whose (Div 2) report
begins ...

Division 2. Physical measurement of light and radiation

UK Representative: Miss T M Goodman. Tel: 020 8943 6863.
e-mail:[email protected]

A major activity this year was the joint Division 1/Division 2 symposium on
ìTemporal and Spatial Aspects of Light and Colour Perception and
Measurementî, held in Hungary in August. This brought together a large
number of experts in the fields of vision, measurement, signalling and
imaging, to discuss issues relating to the perception and measurement of
light sources and images whose properties vary with time or space ........

.........The symposium was followed by the annual meetings of CIE Divisions 1
and 2, commencing with a joint workshop session during which 4 topics of
common interest were reviewed: mesopic photometry; LEDs; colorimetry; and
measurement of flashing lights. ...... <snip>

 

[Ken S. Tucker]

I wonder if the FET is switching too fast and giving some HF ringing
and noise.
Did you try putting a small resistance in series with the gate lead of
the FET right at the FET?

Stepan

Build an EXIT sign from LED's, make them flash triggered
by a smoke detector system. Maybe patrons will notice and
find the exits. Recall that Bar fire in R.I (IIRC), people
just did't notice the word EXIT.
Ken

 

[Leon Heller]

[email protected] wrote in message


Build an EXIT sign from LED's, make them flash triggered
by a smoke detector system. Maybe patrons will notice and
find the exits. Recall that Bar fire in R.I (IIRC), people
just did't notice the word EXIT.

http://www.deegee.com/beacons/ipn/ipn003.htm

 

[John Woodgate]

I read in sci.electronics.design that Ken S. Tucker
oogle.com>) about 'Driving LEDs - high-power flash', on Sun, 26 Sep
2004:

Build an EXIT sign from LED's, make them flash triggered by a smoke
detector system. Maybe patrons will notice and find the exits. Recall
that Bar fire in R.I (IIRC), people
just did't notice the word EXIT.

People don't take any notice of bells and sirens, and they don't 'see'
signs. But a **voice announcement** gets immediate results and can
impart far more information.

'BING-BONG. Emergency announcement. There is an incident at the front of
the building. Please leave the building immediately in an orderly manner
by the side and rear exits.'

Repeated in other languages if necessary. Try that with a bell!

Then: 'BING-BONG. Muster at the far end of the rear car-park for roll-
call. Please assist any disabled or confused people.'

 

[Jonathan Kirwan]

On Sun, 26 Sep 2004 10:49:36 +0100, "R.Lewis" <h.lewis-not this

I think a lot of people would like to see the reference " Some literature
draws it at 10ms." - especially Ms Goodman at the CIE whose (Div 2) report
begins ...

Division 2. Physical measurement of light and radiation

UK Representative: Miss T M Goodman. Tel: 020 8943 6863.
e-mail:[email protected]

A major activity this year was the joint Division 1/Division 2 symposium on
ìTemporal and Spatial Aspects of Light and Colour Perception and
Measurementî, held in Hungary in August. This brought together a large
number of experts in the fields of vision, measurement, signalling and
imaging, to discuss issues relating to the perception and measurement of
light sources and images whose properties vary with time or space ........

........The symposium was followed by the annual meetings of CIE Divisions 1
and 2, commencing with a joint workshop session during which 4 topics of
common interest were reviewed: mesopic photometry; LEDs; colorimetry; and
measurement of flashing lights. ...... <snip>

A quick google yielded:
http://cie2.nist.gov/documents/CIE D2 quad rep 2003.pdf

Jon